diff --git a/analysis/current/analysis-report-hybrid.json b/analysis/current/analysis-report-hybrid.json index eab9566632bb78bd07097889e282d386e01730b2..897c9b9b60a8cc24ee44da58330c5fd67ecfc7af 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": "20260703T000025Z", - "generated_at": "2026-07-03T00:11:01Z", + "snapshot_id": "20260703T060023Z", + "generated_at": "2026-07-03T06:11:39Z", "evidence_quality": "full", "llm_enrichment": true, "meta_bugs": [ @@ -1049,35 +1049,6 @@ } ] }, - { - "cluster_id": "cluster-43451-2", - "summary": "Cluster of 2 related pull requests linked by soft_similarity.", - "status": "open", - "confidence": 0.5, - "canonical_issue_number": null, - "canonical_pr_number": 43451, - "issue_numbers": [], - "pr_numbers": [ - 43451, - 46196 - ], - "evidence_types": [ - "soft_similarity" - ], - "pr_comparisons": [ - { - "left_pr_number": 43451, - "right_pr_number": 46196, - "code_similarity": 0.167, - "size_similarity": 0.741, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.122, - "shared_filenames": [], - "shared_file_areas": [] - } - ] - }, { "cluster_id": "cluster-41211-3", "summary": "Cluster of 1 issues and 2 PRs centered on issue #41211.", @@ -1393,137 +1364,6 @@ } ] }, - { - "cluster_id": "cluster-46907-5", - "summary": "Cluster of 5 related pull requests linked by soft_similarity.", - "status": "open", - "confidence": 0.5, - "canonical_issue_number": null, - "canonical_pr_number": 46907, - "issue_numbers": [], - "pr_numbers": [ - 46907, - 46949, - 46970, - 46986, - 47006 - ], - "evidence_types": [ - "soft_similarity" - ], - "pr_comparisons": [ - { - "left_pr_number": 46907, - "right_pr_number": 46949, - "code_similarity": 0.169, - "size_similarity": 0.846, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.0, - "shared_filenames": [], - "shared_file_areas": [] - }, - { - "left_pr_number": 46907, - "right_pr_number": 46970, - "code_similarity": 0.005, - "size_similarity": 0.023, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.006, - "shared_filenames": [], - "shared_file_areas": [] - }, - { - "left_pr_number": 46907, - "right_pr_number": 46986, - "code_similarity": 0.165, - "size_similarity": 0.788, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.048, - "shared_filenames": [], - "shared_file_areas": [] - }, - { - "left_pr_number": 46907, - "right_pr_number": 47006, - "code_similarity": 0.183, - "size_similarity": 0.885, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.039, - "shared_filenames": [], - "shared_file_areas": [] - }, - { - "left_pr_number": 46949, - "right_pr_number": 46970, - "code_similarity": 0.004, - "size_similarity": 0.019, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.001, - "shared_filenames": [], - "shared_file_areas": [] - }, - { - "left_pr_number": 46949, - "right_pr_number": 46986, - "code_similarity": 0.133, - "size_similarity": 0.667, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.0, - "shared_filenames": [], - "shared_file_areas": [] - }, - { - "left_pr_number": 46949, - "right_pr_number": 47006, - "code_similarity": 0.191, - "size_similarity": 0.957, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.0, - "shared_filenames": [], - "shared_file_areas": [] - }, - { - "left_pr_number": 46970, - "right_pr_number": 46986, - "code_similarity": 0.01, - "size_similarity": 0.029, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.026, - "shared_filenames": [], - "shared_file_areas": [] - }, - { - "left_pr_number": 46970, - "right_pr_number": 47006, - "code_similarity": 0.008, - "size_similarity": 0.02, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.026, - "shared_filenames": [], - "shared_file_areas": [] - }, - { - "left_pr_number": 46986, - "right_pr_number": 47006, - "code_similarity": 0.144, - "size_similarity": 0.697, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.028, - "shared_filenames": [], - "shared_file_areas": [] - } - ] - }, { "cluster_id": "cluster-43240-3", "summary": "Cluster of 1 issues and 2 PRs centered on issue #43240.", @@ -2275,15 +2115,15 @@ ] }, { - "cluster_id": "cluster-46658-2", + "cluster_id": "cluster-46836-2", "summary": "Cluster of 2 related pull requests linked by soft_similarity.", "status": "open", "confidence": 0.5, "canonical_issue_number": null, - "canonical_pr_number": 46658, + "canonical_pr_number": 46836, "issue_numbers": [], "pr_numbers": [ - 46658, + 46836, 47003 ], "evidence_types": [ @@ -2291,62 +2131,86 @@ ], "pr_comparisons": [ { - "left_pr_number": 46658, + "left_pr_number": 46836, "right_pr_number": 47003, - "code_similarity": 0.039, - "size_similarity": 0.044, + "code_similarity": 0.034, + "size_similarity": 0.034, "file_overlap": 0.0, "area_overlap": 0.0, - "patch_similarity": 0.202, + "patch_similarity": 0.182, "shared_filenames": [], "shared_file_areas": [] } ] }, { - "cluster_id": "cluster-46554-2", + "cluster_id": "cluster-43656-2", "summary": "Cluster of 2 related pull requests linked by soft_similarity.", "status": "open", "confidence": 0.5, "canonical_issue_number": null, - "canonical_pr_number": 46558, + "canonical_pr_number": 43656, "issue_numbers": [], "pr_numbers": [ - 46554, - 46558 + 43656, + 43842 ], "evidence_types": [ "soft_similarity" ], "pr_comparisons": [ { - "left_pr_number": 46554, - "right_pr_number": 46558, - "code_similarity": 0.433, - "size_similarity": 0.429, + "left_pr_number": 43656, + "right_pr_number": 43842, + "code_similarity": 0.405, + "size_similarity": 0.5, "file_overlap": 1.0, - "area_overlap": 0.1, - "patch_similarity": 0.081, + "area_overlap": 0.0, + "patch_similarity": 0.036, "shared_filenames": [ - ".github/workflows/pr-ci-security-gate.yml" + "src/transformers/cli/serve.py" ], "shared_file_areas": [ { - "filename": ".github/workflows/pr-ci-security-gate.yml", + "filename": "src/transformers/cli/serve.py", "left_ranges": [ [ - 16, - 45 + 11, + 18 + ], + [ + 30, + 36 + ], + [ + 315, + 323 + ], + [ + 665, + 671 + ], + [ + 931, + 937 + ], + [ + 1843, + 1849 + ], + [ + 1868, + 1874 ] ], "right_ranges": [ [ - 13, - 19 + 54, + 61 ], [ - 57, - 63 + 587, + 593 ] ] } @@ -2661,13 +2525,13 @@ "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-43451-2", - "canonical_pr_number": 43451, + "cluster_id": "cluster-43656-2", + "canonical_pr_number": 43656, "duplicate_pr_numbers": [ - 46196 + 43842 ], "target_issue_number": null, - "reason": "PRs in cluster-43451-2 are treated as duplicates because they share soft_similarity evidence." + "reason": "PRs in cluster-43656-2 are treated as duplicates because they share soft_similarity evidence." }, { "cluster_id": "cluster-43698-3", @@ -2741,22 +2605,13 @@ "reason": "PRs in cluster-46159-3 are treated as duplicates because they converge on issue #46159 with closing_reference evidence." }, { - "cluster_id": "cluster-46554-2", - "canonical_pr_number": 46558, - "duplicate_pr_numbers": [ - 46554 - ], - "target_issue_number": null, - "reason": "PRs in cluster-46554-2 are treated as duplicates because they share soft_similarity evidence." - }, - { - "cluster_id": "cluster-46658-2", - "canonical_pr_number": 46658, + "cluster_id": "cluster-46836-2", + "canonical_pr_number": 46836, "duplicate_pr_numbers": [ 47003 ], "target_issue_number": null, - "reason": "PRs in cluster-46658-2 are treated as duplicates because they share soft_similarity evidence." + "reason": "PRs in cluster-46836-2 are treated as duplicates because they share soft_similarity evidence." }, { "cluster_id": "cluster-46846-3", @@ -2785,18 +2640,6 @@ ], "target_issue_number": 46894, "reason": "PRs in cluster-46894-6 are treated as duplicates because they converge on issue #46894 with closing_reference, shared_issue_target, soft_similarity evidence." - }, - { - "cluster_id": "cluster-46907-5", - "canonical_pr_number": 46907, - "duplicate_pr_numbers": [ - 46949, - 46970, - 46986, - 47006 - ], - "target_issue_number": null, - "reason": "PRs in cluster-46907-5 are treated as duplicates because they share soft_similarity evidence." } ], "best_issue": { diff --git a/analysis/current/analysis-report-hybrid.llm-reviews.json b/analysis/current/analysis-report-hybrid.llm-reviews.json index 0c64824fcb0f34f523c27338d870ad5b2d0e7323..f73e9329e6981145c3632796df01e5b4f38c53a3 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": "20260703T000025Z", - "generated_at": "2026-07-03T00:11:01Z", + "snapshot_id": "20260703T060023Z", + "generated_at": "2026-07-03T06:11:39Z", "model": "gpt-5.4-mini?service_tier=flex", "reviews": [ { @@ -102,8 +102,8 @@ "issue:34928|issue:44805", "issue:45685|issue:46869", "issue:45859|issue:45902", - "issue:34928|issue:36331", "issue:43638|issue:45161", + "issue:34928|issue:36331", "issue:43596|issue:45161", "issue:43716|issue:45507", "issue:45874|issue:46294", @@ -132,14 +132,14 @@ "issue:45237|issue:46015", "issue:43039|issue:44805" ], - "prepared_review_unit_hash": "83942f1bcc61375b546d18aebc87245146c3c142f3ae0e219f7349186e4016cd", + "prepared_review_unit_hash": "b55cc8dde3b86133a4f9417694cb8ff71a12da67f9d06e947c7d74d33a0eff18", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, @@ -154,396 +154,324 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly heterogeneous. The only clear issue-duplicate subclusters are the DeepSpeed ZeRO-3 BERT IndexError reports, the Qwen3.5 MoE rotary-embedding device/materialization reports, and a torch 2.4.1 compatibility pair. Most other soft edges are cross-model or cross-subsystem false positives.", - "confidence": 0.74, - "canonical_issue_reason": "issue:43596 is the clearest and broadest representative of the strongest duplicate pair: the ZeRO-3/BERT index-out-of-bounds failure.", + "summary": "Mostly a mixed cluster of unrelated runtime, model-specific, and docs issues. Only a couple of true duplicates stand out: the DeepSpeed ZeRO-3/BERT index-0 failure pair and the torch 2.4.1 compatibility wording pair.", + "confidence": 0.73, + "canonical_issue_reason": "issue:43596 is the clearest duplicate-worthy bug in the set: it matches issue:43638 on the same DeepSpeed ZeRO-3 / BERT \"index 0 is out of bounds\" failure path and is phrased more generally.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43596 best fits the cluster as the most concrete, generalizable issue in the only very strong duplicate subcluster.", + "best_issue_reason": "issue:43596 is the best issue anchor because it represents one of the few exact duplicate pairs here and is more canonical than the typo-ridden follow-up report.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44322", "right": "issue:45464", "accept": false, - "reason": "Different model families and failures: Qwen3.5 config attribute error vs streaming chat/completions failure." + "reason": "Different Qwen3.5 bugs: missing config field vs streaming chat/completions failure." }, { "left": "issue:45237", "right": "issue:45685", "accept": false, - "reason": "Unrelated hardware/backend bugs: AMD GPU failure vs MPS histogram on Int." + "reason": "Different models and failure modes: AMD GPU support vs MPS histogram op missing." }, { "left": "issue:43716", "right": "issue:45237", "accept": false, - "reason": "Different models and symptoms: Mistral-3 dtype mismatch vs GPT-OSS on AMD GPUs." + "reason": "Different model/runtime problems: image preprocessor dtype mismatch vs GPT-OSS on AMD GPUs." }, { "left": "issue:43828", "right": "issue:45237", "accept": false, - "reason": "Autocast dtype mismatch in Phi-tiny-MoE is not the same bug as GPT-OSS AMD GPU failure." + "reason": "Both are hardware/runtime issues, but the concrete bug and code path differ." }, { "left": "issue:45237", "right": "issue:46468", "accept": false, - "reason": "Different models and code paths: GPT-OSS AMD GPU issue vs Nemotron example breakage." + "reason": "Different models/examples and no shared concrete failure path." }, { "left": "issue:45237", "right": "issue:45507", "accept": false, - "reason": "No shared concrete bug; one is GPU backend failure, the other is an invalid-method model-call issue." + "reason": "Different MoE/device issues; not the same underlying bug." }, { "left": "issue:30277", "right": "issue:35545", "accept": false, - "reason": "Different problems: ZeRO-3 collective mismatch vs ONNX export error." + "reason": "DeepSpeed collective mismatch vs ONNX export error are unrelated." }, { "left": "issue:35545", "right": "issue:42915", "accept": false, - "reason": "ModernBERT ONNX export and Qwen3Moe FP8 failure are unrelated model-specific bugs." + "reason": "Different failures: ONNX export vs FineGrainedFP8Config runtime issue." }, { "left": "issue:35545", "right": "issue:41720", "accept": false, - "reason": "ONNX export error vs Qwen3 auto device-map cuda assert are different failures." + "reason": "Different code paths: export error vs auto device-mapping CUDA assert." }, { "left": "issue:45901", "right": "issue:46015", "accept": false, - "reason": "Table QA crash and outdated RAG examples are unrelated issues." + "reason": "Task crash vs outdated RAG docs; not the same issue." }, { "left": "issue:35545", "right": "issue:36010", "accept": false, - "reason": "ONNX export failure is unrelated to the GenerationMixin import error." + "reason": "Export failure and GenerationMixin import failure are unrelated." }, { "left": "issue:44509", "right": "issue:44512", "accept": false, - "reason": "Both are docs regressions, but they concern different removed items and different doc cleanup scopes." + "reason": "Both are v5 docs cleanup items, but they target different removed docs content." }, { "left": "issue:44918", "right": "issue:46165", "accept": false, - "reason": "TRL SFT embedding unpacking and resume-training with trust_remote_code are different problems." + "reason": "Different training issues: TRL SFT embedding unpacking vs resume/trust-remote-code behavior." }, { "left": "issue:46165", "right": "issue:46869", "accept": false, - "reason": "Resume-training/trust_remote_code and YOLO multi-GPU device-map failures are unrelated." + "reason": "Different scenarios and failures: resume-training remote code vs YOLO multi-GPU device map." }, { "left": "issue:44805", "right": "issue:45823", "accept": false, - "reason": "Mask shape mismatch is not the same bug as Gemma4 PLE device mismatch." + "reason": "Mask-shape indexing error vs device mismatch in Gemma4 PLE." }, { "left": "issue:43638", "right": "issue:44805", "accept": false, - "reason": "Both involve index errors, but one is ZeRO-3 BERT loading and the other is a mask/tensor shape mismatch." + "reason": "Both involve shape/indexing errors, but the specific bug and stack path differ." }, { "left": "issue:43638", "right": "issue:46165", "accept": false, - "reason": "ZeRO-3 BERT IndexError and resume-training/trust_remote_code are unrelated." + "reason": "Completely different training issues." }, { "left": "issue:43596", "right": "issue:44805", "accept": false, - "reason": "Different failure modes: ZeRO-3 empty-shard index error vs mask/indexed-tensor shape mismatch." + "reason": "Different failures: ZeRO-3/BERT init index error vs generic mask/tensor mismatch." }, { "left": "issue:44805", "right": "issue:46294", "accept": false, - "reason": "Mask/tensor shape mismatch is unrelated to Qwen3.5 video chat template token mismatch." + "reason": "Different multimodal and tensor-shape bugs." }, { "left": "issue:44805", "right": "issue:45161", "accept": false, - "reason": "Shape mismatch in indexing is unrelated to GPT-OSS MoE tensor-parallel issues." + "reason": "Different failures: mask indexing vs GPT-OSS TP/MoE issue." }, { "left": "issue:46165", "right": "issue:46468", "accept": false, - "reason": "Resume-training issue and Nemotron example breakage do not share a concrete code path." + "reason": "Different problems and code paths." }, { "left": "issue:44560", "right": "issue:46165", "accept": false, - "reason": "Video StopIteration in chat template processing is unrelated to training resume failures." + "reason": "Video chat-template failure vs training resume issue." }, { "left": "issue:34634", "right": "issue:35545", "accept": false, - "reason": "BarkProcessor voice_preset and ModernBERT ONNX export are different subsystems." + "reason": "Different subsystems: BarkProcessor voice preset vs ONNX export." }, { "left": "issue:30277", "right": "issue:43638", "accept": false, - "reason": "Both mention distributed/deepspeed, but the collective mismatch and BERT zero3 index error are not the same bug." + "reason": "DeepSpeed collective mismatch and BERT zero3 index error are not the same bug." }, { "left": "issue:43596", "right": "issue:43638", "accept": true, - "reason": "Same ZeRO-3 BERT loading bug: both report index 0 out of bounds with BERT under DeepSpeed ZeRO-3." + "reason": "Same DeepSpeed ZeRO-3/BERT index-0 out-of-bounds failure; these look like duplicate reports." }, { "left": "issue:43039", "right": "issue:45161", "accept": false, - "reason": "Liger cross_entropy being called and GPT-OSS tensor-parallel failure are unrelated." + "reason": "Different bugs: Liger Kernel cross-entropy being called vs GPT-OSS TP failure." }, { "left": "issue:43975", "right": "issue:44457", "accept": false, - "reason": "Incorrect detokenization and LoRA merge/save mismatch are separate issues." + "reason": "Different issues: detokenization regression vs LoRA merge/save mismatch." }, { "left": "issue:44292", "right": "issue:44457", "accept": false, - "reason": "Qwen-3 NVFP4 runtime error and LoRA merge/save output mismatch are unrelated." + "reason": "Different model/runtime bugs." }, { "left": "issue:44841", "right": "issue:46661", "accept": false, - "reason": "Voxtral processor failure and remote-code config_class AttributeError are different bugs." + "reason": "Different processor failure vs remote-code model_class/config_class error." }, { "left": "issue:43716", "right": "issue:46869", "accept": false, - "reason": "Mistral-3 dtype mismatch and YOLO multi-GPU device-map failure are unrelated." + "reason": "Different model and failure mode; only broad device-map similarity." }, { "left": "issue:24643", "right": "issue:30277", "accept": false, - "reason": "DeepSpeed weight-shape error and collective mismatch are different distributed-training failures." + "reason": "Both mention DeepSpeed, but the concrete runtime errors are different." }, { "left": "issue:34928", "right": "issue:44805", "accept": false, - "reason": "FSDP activation-checkpointing tensor-size mismatch is not the same as mask indexing shape mismatch." + "reason": "Both are tensor-size/shape issues, but not the same underlying bug." }, { "left": "issue:45685", "right": "issue:46869", "accept": false, - "reason": "MPS histogram kernel issue and YOLO device-map auto failure are unrelated." + "reason": "Different hardware/runtime errors with no shared code-path bug." }, { "left": "issue:45859", "right": "issue:45902", - "accept": true, - "reason": "Same Qwen3.5 MoE rotary-embedding device-transfer bug around `inv_freq` and offload/materialization." - }, - { - "left": "issue:34928", - "right": "issue:36331", "accept": false, - "reason": "Activation-checkpointing/FSDP size mismatch and CustomTrainer compute_loss signature mismatch are unrelated." + "reason": "Same class name, but one is CPU-offload compatibility and the other is uninitialized-memory materialization." }, { "left": "issue:43638", "right": "issue:45161", "accept": false, - "reason": "ZeRO-3 BERT index error and GPT-OSS tensor-parallel issue do not share a concrete failure path." + "reason": "Different bugs: ZeRO-3/BERT index error vs GPT-OSS TP issue." + }, + { + "left": "issue:34928", + "right": "issue:36331", + "accept": false, + "reason": "Activation-checkpointing/FSDP shape mismatch vs Trainer API signature mismatch." }, { "left": "issue:43596", "right": "issue:45161", "accept": false, - "reason": "Different distributed-training bugs: BERT ZeRO-3 index error vs GPT-OSS TP failure." + "reason": "Different bugs and models." }, { "left": "issue:43716", "right": "issue:45507", "accept": false, - "reason": "Mistral-3 image-preprocessor dtype mismatch and GraniteMoEHybrid invalid-method call are unrelated." + "reason": "Different model-specific runtime failures." }, { "left": "issue:45874", "right": "issue:46294", "accept": false, - "reason": "Gemma4 reverse-embedding lookup and Qwen3.5 video token mismatch are different multimodal bugs." + "reason": "Different Gemma3/Gemma4 feature bugs; not the same failure." }, { "left": "issue:44292", "right": "issue:45237", "accept": false, - "reason": "Qwen-3 NVFP4 runtime error and GPT-OSS AMD GPU failure are different model/runtime issues." + "reason": "Different model/runtime errors." }, { "left": "issue:39290", "right": "issue:45081", "accept": false, - "reason": "Gemma3 sliding-window-pattern vLLM error and Mistral regex patch crash are unrelated." + "reason": "Gemma3 vLLM config issue vs Mistral tokenizer regex patch crash." }, { "left": "issue:45823", "right": "issue:45874", "accept": false, - "reason": "Gemma4 device mismatch and Gemma4 reverse-embedding slowdown are different problems." + "reason": "Different Gemma4 issues: device mismatch vs expensive reverse embedding lookup." }, { "left": "issue:45800", "right": "issue:46515", "accept": true, - "reason": "Same torch 2.4.1 compatibility problem; the second title is just a shorter restatement." + "reason": "Same torch 2.4.1 compatibility regression; effectively duplicate wording." }, { "left": "issue:34928", "right": "issue:41720", "accept": false, - "reason": "FSDP checkpointing mismatch and Qwen3 auto device-map cuda assert are unrelated." + "reason": "Different bugs: FSDP checkpointing mismatch vs Qwen3 auto device-map CUDA assert." }, { "left": "issue:43425", "right": "issue:44292", "accept": false, - "reason": "Torch version incompatibility and Qwen-3 NVFP4 runtime error are different issues." - }, - { - "left": "issue:43828", - "right": "issue:45161", - "accept": false, - "reason": "Autocast dtype mismatch in Phi-tiny-MoE is not the same as GPT-OSS TP failure." - }, - { - "left": "issue:43541", - "right": "issue:43716", - "accept": false, - "reason": "Mixtral torch-dynamo grouped_mm error and Mistral-3 dtype mismatch are unrelated." + "reason": "Generic torch compatibility vs Qwen3 NVFP4 runtime error are not the same issue." }, { "left": "issue:43828", - "right": "issue:44292", - "accept": false, - "reason": "Different model-specific numeric/runtime failures with no shared code path." - }, - { - "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": "Qwen-3 NVFP4 runtime error and Qwen3.5 embedding unpacking failure are different bugs." - }, - { - "left": "issue:45507", - "right": "issue:45685", - "accept": false, - "reason": "Invalid-method model call and MPS histogram kernel issue are unrelated." - }, - { - "left": "issue:43716", - "right": "issue:44292", - "accept": false, - "reason": "Different model/runtime failures; no evidence of the same underlying bug." - }, - { - "left": "issue:43716", "right": "issue:45685", "accept": false, - "reason": "Mistral-3 dtype mismatch and MPS Int histogram limitation are unrelated." - }, - { - "left": "issue:44918", - "right": "issue:45161", - "accept": false, - "reason": "TRL SFT embedding unpacking and GPT-OSS tensor-parallel failure are unrelated." + "reason": "Different hardware-specific dtype vs MPS op-availability issues." }, { "left": "issue:43039", "right": "issue:44918", "accept": false, - "reason": "Liger kernel cross_entropy invocation and Qwen3.5 SFT embedding unpacking are unrelated." + "reason": "Liger Kernel cross-entropy call vs TRL embedding unpacking are unrelated." }, { "left": "issue:41762", "right": "issue:43638", "accept": false, - "reason": "Both are ZeRO-3 index errors, but the model families and likely code paths differ; not enough to treat as the same bug." + "reason": "Both involve ZeRO-3, but one is Gemma3 loading and the other is BERT init; different bugs." }, { "left": "issue:43827", "right": "issue:44512", "accept": false, - "reason": "Both are docs regressions, but they concern different removed features and should not be merged as the same issue." + "reason": "Both are docs regressions from v5 removals, but they cover different removed content." }, { "left": "issue:43828", "right": "issue:45685", "accept": false, - "reason": "Phi autocast dtype mismatch and MPS histogram kernel issue are unrelated." - }, - { - "left": "issue:43039", - "right": "issue:43828", - "accept": false, - "reason": "Liger cross_entropy handling and Phi autocast dtype mismatch are different problems." - }, - { - "left": "issue:30277", - "right": "issue:36331", - "accept": false, - "reason": "DeepSpeed collective mismatch and compute_loss signature change are unrelated." - }, - { - "left": "issue:34928", - "right": "issue:36010", - "accept": false, - "reason": "FSDP checkpointing mismatch and GenerationMixin import failure are unrelated." - }, - { - "left": "issue:45237", - "right": "issue:46015", - "accept": false, - "reason": "GPT-OSS AMD GPU failure and outdated RAG examples are unrelated." + "reason": "Different runtime problems; same broad area but not the same bug." }, { "left": "issue:43039", "right": "issue:44805", "accept": false, - "reason": "Liger kernel cross_entropy call and mask/tensor shape mismatch are unrelated." + "reason": "Different failure modes and code paths." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative on the soft edges. The three accepted duplicate pairs are plausible from the titles, and the remaining rejections avoid overstating weak cross-model similarities." + "feedback": "The summary is grounded in the packet and stays conservative. The two accepted soft-edge pairs are well supported by the issue titles and are plausible duplicates, and the rejected pairs are phrased cautiously without overstating evidence. No mergeability concerns apply since these are issue pairs, not PR pairs." }, "evaluator_used": true, "retried": false, @@ -657,20 +585,20 @@ "issue:43572|issue:45070", "issue:45800|issue:45823" ], - "prepared_review_unit_hash": "904421a97ac9623a2af06c38ac3f91d781dd688631a5d15f97922fea70a8c67e", + "prepared_review_unit_hash": "e6090c211c2a079bc0810899150ad46a26ebcbeea8e335c1d0adf051387c4562", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 42, - "serialized_chars": 40043, + "serialized_chars": 40042, "estimated_input_tokens": 10011, "estimated_eval_tokens": 20278 }, @@ -679,273 +607,273 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a heterogeneous set of unrelated issue reports. The tokenizer regression pair (45812/45920) is the only clearly supported duplicate subgroup. The ZeRO-3 index-error pair (41762/43596) looks related at a high level, but the evidence here is not strong enough to call it a true duplicate; treat it as a possible same-bug candidate only.", - "confidence": 0.86, - "canonical_issue_reason": "Issue 45920 is the best canonical issue because it captures the broader v5 AutoTokenizer token-ID regression across multiple model families and clearly subsumes the narrower Granite-only report 45812.", + "summary": "This cluster is heterogeneous rather than one true duplicate set. The only strong duplicate-like subclusters are the DeepSpeed ZeRO-3 loading IndexErrors and the AutoTokenizer wrong-token-ID regressions; most other pairs are just same-subsystem or same-error-shape but different bugs.", + "confidence": 0.73, + "canonical_issue_reason": "Issue 45920 is the best canonical anchor for the tokenizer regression subcluster: it states the broadest, most user-facing form of the bug and covers multiple affected model families, making it a better umbrella than the narrower Granite-only report.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45920 is the strongest representative issue in this set: it has the broadest impact and is the only pair here with clear duplicate-level support.", + "best_issue_reason": "Issue 45920 is the strongest overall issue candidate because it describes a broad v5 regression with clear symptom wording and multi-model impact, which makes it the most representative item among the actual duplicate-like reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45800", "right": "issue:46304", "accept": false, - "reason": "Both are transformers compatibility complaints, but they describe different breakages and different triggers (torch version vs USE_DISTRIBUTED=0)." + "reason": "Both are version-compatibility complaints, but they concern different incompatibilities and different failure modes; no shared concrete code-path bug." }, { "left": "issue:43866", "right": "issue:46402", "accept": false, - "reason": "Different models and failures: corrupted checkpoint vs missing text-encoder weights." + "reason": "Checkpoint corruption vs missing text-encoder weights are unrelated model-loading problems." }, { "left": "issue:43296", "right": "issue:46304", "accept": false, - "reason": "vLLM/PaddleOCR-VL load failure is unrelated to the distributed-build regression." + "reason": "Different stacks and symptoms: vLLM/PaddleOCR-VL load failure versus PyTorch distributed support regression." }, { "left": "issue:41762", "right": "issue:43596", - "accept": false, - "reason": "Both are ZeRO-3 init/loading index errors, but the packet only supports title-level similarity and a shared generic error family. That is not enough to call them a definite duplicate; at most they are a possible same-bug candidate." + "accept": true, + "reason": "Both report the same ZeRO-3 loading failure pattern: index 0 out of bounds during model init/load, plausibly the same underlying bug in the shared load path." }, { "left": "issue:39290", "right": "issue:41720", "accept": false, - "reason": "Different models and different failure modes: Gemma3 config attribute error vs Qwen3 auto device-map CUDA assert." + "reason": "Missing config attribute in Gemma3/vLLM is unrelated to Qwen3 auto-device-map cuda assert." }, { "left": "issue:43901", "right": "issue:44512", "accept": false, - "reason": "Both are docs issues, but they concern different outdated references and different APIs." + "reason": "Both touch docs, but they are different documentation issues about different APIs and different removed behaviors." }, { "left": "issue:44589", "right": "issue:46304", "accept": false, - "reason": "Float8 storage lookup failure is unrelated to the torch/distributed compatibility issue." + "reason": "Float8 storage lookup failure is unrelated to the PyTorch USE_DISTRIBUTED=0 regression." }, { "left": "issue:45237", "right": "issue:45901", "accept": false, - "reason": "Different subsystems: AMD GPU model runtime vs table-question-answering crash." + "reason": "Different features and different failures; no shared code-path bug." }, { "left": "issue:43975", "right": "issue:46015", "accept": false, - "reason": "Tokenizer detokenization bug is unrelated to outdated RAG examples documentation." + "reason": "Model detokenization regression and outdated RAG examples are unrelated." }, { "left": "issue:43039", "right": "issue:44560", "accept": false, - "reason": "Liger Kernel cross-entropy call and Qwen3-vl video StopIteration are unrelated issues." + "reason": "Liger-kernel cross-entropy dispatch and Qwen3 VL video StopIteration are unrelated." }, { "left": "issue:45823", "right": "issue:46294", "accept": false, - "reason": "Different model families and different symptoms: device mismatch vs video-token/template mismatch." + "reason": "Both involve device/input handling, but one is a device mismatch in forward and the other is a video feature/token mismatch during chat templating." }, { "left": "issue:44841", "right": "issue:45823", "accept": false, - "reason": "Processor failure for Voxtral is unrelated to Gemma4 device-map mismatch." + "reason": "Processor failure for Voxtral and Gemma4 device mismatch are different model-path bugs." }, { "left": "issue:43866", "right": "issue:46772", "accept": false, - "reason": "Corrupted Ovis2 checkpoint and DiffusionGemma NVFP4 compatibility are different problems." + "reason": "Checkpoint corruption is not the same as NVFP4/modelopt compatibility." }, { "left": "issue:44863", "right": "issue:46015", "accept": false, - "reason": "NemotronH checkpoint-loading bug is unrelated to RAG docs." + "reason": "NemotronH checkpoint loading and RAG docs are unrelated." }, { "left": "issue:45020", "right": "issue:46304", "accept": false, - "reason": "Remote-code model breakage is broader but not the same as the USE_DISTRIBUTED=0 torch support regression." + "reason": "Remote-code breakage and distributed-build breakage are different regressions." }, { "left": "issue:45812", "right": "issue:45920", "accept": true, - "reason": "Same underlying tokenizer regression: wrong token IDs from AutoTokenizer across v5, with 45920 being the broader report." + "reason": "Both describe the same AutoTokenizer v5 regression producing wrong token IDs, just reported against different model families; they plausibly share one fix." }, { "left": "issue:44292", "right": "issue:46015", "accept": false, - "reason": "Qwen-3 NVFP4 runtime error is unrelated to documentation examples." + "reason": "NVFP4 model execution failure and RAG documentation are unrelated." }, { "left": "issue:43866", "right": "issue:44863", "accept": false, - "reason": "Different model-loading failures for different models; no shared concrete bug." + "reason": "Corrupted checkpoint and checkpoint loading implementation issue are not the same bug." }, { "left": "issue:43872", "right": "issue:46304", "accept": false, - "reason": "bitsandbytes constructor incompatibility is unrelated to the distributed-build regression." + "reason": "bitsandbytes constructor incompatibility is unrelated to the USE_DISTRIBUTED regression." }, { "left": "issue:39290", "right": "issue:43531", "accept": false, - "reason": "Both mention sliding-window behavior, but they involve different models and different failure contexts." + "reason": "Both mention sliding_window, but one is a missing config attribute in vLLM and the other is a model-specific sliding_window issue." }, { "left": "issue:41093", "right": "issue:43638", "accept": false, - "reason": "Both are ZeRO-3-related index errors, but the specific failures differ enough that they do not look like the same bug." + "reason": "Both are IndexErrors, but the shapes and triggering paths differ; this is too generic to be a duplicate." }, { "left": "issue:43792", "right": "issue:45901", "accept": false, - "reason": "Whisper loading failure and table-question-answering crash are unrelated." + "reason": "Whisper load/runtime failure and table-question-answering crash are unrelated." }, { "left": "issue:43854", "right": "issue:43866", "accept": false, - "reason": "Different model reports with different failure modes; not the same bug." + "reason": "Model loading test failure and checkpoint corruption are different problems." }, { "left": "issue:43425", "right": "issue:45070", "accept": false, - "reason": "Torch-version incompatibility and PretrainedConfig pydantic regression are different breakages." + "reason": "Torch version incompatibility and pydantic/config regression are unrelated." }, { "left": "issue:43975", "right": "issue:45901", "accept": false, - "reason": "Tokenizer detokenization bug is unrelated to table QA crash." + "reason": "Detokenization regression and table QA crash do not share the same bug." }, { "left": "issue:45362", "right": "issue:45464", "accept": false, - "reason": "Same model family, but different code paths and symptoms; not enough to treat as one bug." + "reason": "Different Qwen3.5 issues with different symptoms and code paths." }, { "left": "issue:43792", "right": "issue:45070", "accept": false, - "reason": "Whisper loading failure is unrelated to the config-field regression." + "reason": "Whisper failure and pydantic field breakage are unrelated." }, { "left": "issue:43596", "right": "issue:45800", "accept": false, - "reason": "ZeRO-3 index error is unrelated to the torch/transformers version incompatibility report." + "reason": "Both mention loading/runtime problems, but one is ZeRO-3 index error and the other is a torch/transformers compatibility complaint." }, { "left": "issue:43541", "right": "issue:46869", "accept": false, - "reason": "Different model families and failures: grouped_mm tracing error vs multi-GPU device-map issue." + "reason": "Different model families and different errors; no shared fix is plausible." }, { "left": "issue:44292", "right": "issue:45901", "accept": false, - "reason": "Qwen NVFP4 runtime error is unrelated to table QA crashes." + "reason": "Different task and failure mode." }, { "left": "issue:39290", "right": "issue:45823", "accept": false, - "reason": "Different model-specific runtime failures; no shared underlying bug." + "reason": "Gemma3 sliding-window config error is unrelated to Gemma4 device-mapping mismatch." }, { "left": "issue:45464", "right": "issue:45685", "accept": false, - "reason": "Chat streaming failure and MPS histogram Int support error are unrelated." + "reason": "Chat/completions streaming failure and MPS histogram error are unrelated." }, { "left": "issue:36010", "right": "issue:39290", "accept": false, - "reason": "ImportError for GenerationMixin and Gemma3 config attribute error are unrelated." + "reason": "Import path breakage and Gemma3/vLLM config mismatch are unrelated." }, { "left": "issue:43541", "right": "issue:45685", "accept": false, - "reason": "Different MoE/runtime failures with no common concrete code-path bug." + "reason": "Different MoE/runtime problems with no shared concrete code path." }, { "left": "issue:43541", "right": "issue:43828", "accept": false, - "reason": "Both involve model execution, but grouped_mm tracing and autocast dtype mismatch are distinct issues." + "reason": "Grouped_mm tracing error and autocast dtype mismatch are different problems despite both involving MoE models." }, { "left": "issue:44841", "right": "issue:45084", "accept": false, - "reason": "Processor failure for Voxtral is unrelated to compiler/template-node errors." + "reason": "Processor failure and template-node compile error are unrelated." }, { "left": "issue:43404", "right": "issue:43866", "accept": false, - "reason": "Weight-tying bug in Mistral3 is unrelated to corrupted Ovis2 checkpoint loading." + "reason": "Weight-tying bug in Mistral3 is unrelated to a corrupted checkpoint report." }, { "left": "issue:45507", "right": "issue:46869", "accept": false, - "reason": "Invalid method in GraniteMoEHybrid and YOLO multi-GPU device-map failure are different bugs." + "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 ZeRO-3 index-related, but the concrete symptoms differ and do not clearly point to the same bug." + "reason": "Both are IndexErrors, but the failure shapes and model-loading contexts differ enough that this is not the same bug." }, { "left": "issue:45399", "right": "issue:46661", "accept": false, - "reason": "Flash-attn fallback logic and remote-code config_class attribute error are unrelated." + "reason": "Flash-attn fallback gating and remote-code AutoModel config_class errors are different bugs." }, { "left": "issue:43572", "right": "issue:45070", "accept": false, - "reason": "Missing pad_token_idx in StableLmConfig and pydantic PretrainedConfig regression are different config issues." + "reason": "Missing pad_token_idx and pydantic field breakage are unrelated." }, { "left": "issue:45800", "right": "issue:45823", "accept": false, - "reason": "Torch/transformers compatibility issue and Gemma4 device-mismatch issue are unrelated." + "reason": "Torch/transformers incompatibility and Gemma4 device mismatch are different regression classes." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The summary is mostly grounded, but one soft-edge verdict is too strong for the evidence. The 41762/43596 ZeRO-3 pair is supported only by title-level similarity and a generic shared error family; the packet does not give enough evidence to call it a true duplicate with confidence. That should be phrased more conservatively as a possible duplicate or same-bug candidate. The tokenizer pair (45812/45920) is well-supported." + "accept": true, + "feedback": "The summary is grounded in the packet: it uses only the issue titles to identify two plausible duplicate-like subclusters (DeepSpeed ZeRO-3 index errors and AutoTokenizer v5 wrong-token regressions) and keeps the rest conservative. The accepted soft edges are supported by strong title-level similarity, and the rejected pairs do not overclaim a shared bug based on the available evidence." }, "evaluator_used": true, - "retried": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -1044,220 +972,220 @@ "issue:45183|issue:46463", "issue:44918|issue:45507" ], - "prepared_review_unit_hash": "7ef0b5f8f711ed8abdd6b4ca8673a4c11e824cd127e30166042778000f38524b", + "prepared_review_unit_hash": "da132711c161a507e1b51a4df9b205a91effcad86edc673a0b928ba8b7f91a6f", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 31, - "serialized_chars": 36237, - "estimated_input_tokens": 9060, - "estimated_eval_tokens": 18376 + "serialized_chars": 36241, + "estimated_input_tokens": 9061, + "estimated_eval_tokens": 18378 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a loose collection of unrelated Transformers/vLLM bug reports. A few pairs share broad themes like v5 regressions, custom-code loading, or dtype mismatches, but none look like the same underlying bug/change.", - "confidence": 0.88, - "canonical_issue_reason": "issue:45005 is the broadest, most representative v5 regression report in the set and the closest thing to an umbrella issue, though the cluster is not truly duplicate-heavy.", + "summary": "This cluster is mostly heterogeneous model/regression reports with a loose overlap around Transformers v5 loading/custom-code behavior. None of the soft pairs look like true duplicates; they share themes but not the same concrete bug.", + "confidence": 0.23, + "canonical_issue_reason": "Issue 45020 is the broadest representative of the cluster\u2019s strongest recurring theme: recent Transformers versions breaking remote-code/custom-model loading.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45005 is the strongest standalone representative because it clearly describes a concrete regression class with cross-model impact and is easier to use as a cluster anchor than the narrower, family-specific reports.", + "best_issue_reason": "45020 is the best single issue to anchor the cluster because it is broad, recent, and central to the loading-regression pattern seen in several nearby reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43296", "right": "issue:43366", "accept": false, - "reason": "Different failures: PaddleOCR-VL load breakage vs GGUF architecture support. Same general area, not the same bug." + "reason": "Different failures: PaddleOCR-VL load regression vs GGUF gpt-oss support request." }, { "left": "issue:44805", "right": "issue:44918", "accept": false, - "reason": "Both mention shape-related errors, but one is mask/tensor indexing and the other is Qwen3.5 input-embedding unpacking. Not the same code path." + "reason": "Both are shape/embedding related, but the concrete bugs are different." }, { "left": "issue:43541", "right": "issue:45161", "accept": false, - "reason": "Both involve Mixtral/MoE-style models, but one is torch dynamo grouped_mm tracing and the other is GPT-OSS tensor parallel support. Distinct issues." + "reason": "Both involve MoE/parallelism, but one is grouped_mm tracing and the other is TP support." }, { "left": "issue:43828", "right": "issue:45464", "accept": false, - "reason": "Autocast dtype mismatch on Phi-tiny-MoE is not the same as Qwen3.5 streaming chat/completions failure." + "reason": "Different code paths: autocast dtype mismatch vs streaming chat/completions failure." }, { "left": "issue:44387", "right": "issue:45005", "accept": false, - "reason": "One is int4 quantization memory growth/OOM, the other is tied-weights behavior in translation models. Related release area, different bug." + "reason": "OOM under int4 quantization is unrelated to tied-weights regressions." }, { "left": "issue:43366", "right": "issue:45464", "accept": false, - "reason": "GGUF architecture support and Qwen3.5 streaming inference are unrelated concrete problems." + "reason": "gpt-oss GGUF support and Qwen3.5 streaming failures are unrelated." }, { "left": "issue:44291", "right": "issue:45070", "accept": false, - "reason": "Different breakages: init_empty_weights argument handling vs pydantic PretrainedConfig field changes." + "reason": "init_empty_weights argument handling is distinct from the pydantic PretrainedConfig regression." }, { "left": "issue:43531", "right": "issue:45070", "accept": false, - "reason": "Sliding_window MoE behavior and pydantic config regression are separate bugs." + "reason": "Qwen3-MoE sliding_window bug is unrelated to config-field breakage." }, { "left": "issue:41093", "right": "issue:43596", "accept": false, - "reason": "Both are IndexErrors, but one is a mask-length mismatch and the other is a zero-sized tensor access during ZeRO-3 init. Not duplicates." + "reason": "Both are IndexErrors, but the tensor/mask mismatch and zero3 init failure are different bugs." }, { "left": "issue:43716", "right": "issue:46294", "accept": false, - "reason": "Image-preprocessor dtype mismatch for Mistral-3 is unrelated to Qwen3.5 video token/template mismatch." + "reason": "Image-preprocessor dtype mismatch is not the same as video token/feature mismatch." }, { "left": "issue:45020", "right": "issue:45800", "accept": false, - "reason": "Remote-code model breakage in recent transformers is not the same as a torch 2.4.1 / transformers 5.8.0 compatibility report." + "reason": "Both are version regressions, but one is remote_code breakage and the other is torch 2.4.1 incompatibility." }, { "left": "issue:43054", "right": "issue:43994", "accept": false, - "reason": "Different symptom and scope: SigLIP2 text embedding quality regression vs loading/nonsensical results for a specific SigLIP2 base model." + "reason": "Both mention SigLIP2, but one is embedding quality and the other is bad AutoModel/pipeline outputs." }, { "left": "issue:44479", "right": "issue:46294", "accept": false, - "reason": "Both are Qwen video-related, but one is a v5.3.0 regression across multiple models while the other is a specific apply_chat_template token/feature count mismatch." + "reason": "General v5.3.0 video regression is different from a specific Qwen3.5-9B template mismatch." }, { "left": "issue:46468", "right": "issue:46869", "accept": false, - "reason": "Nemotron example breakage and YOLO multi-GPU device-map failure are unrelated model families and code paths." + "reason": "Broken example docs and multi-GPU device-map failure are not the same bug." }, { "left": "issue:45759", "right": "issue:45902", "accept": false, - "reason": "One is from_config not unwrapping text_config; the other is uninitialized rotary embedding memory after to_empty. Different bugs." + "reason": "One is text_config unwrapping, the other is uninitialized rotary embedding memory." }, { "left": "issue:43596", "right": "issue:46165", "accept": false, - "reason": "ZeRO-3 init index error and resume-training with trust_remote_code are distinct failure modes." + "reason": "Zero3/BertModel indexing and remote-code resume issues do not share the same root cause." }, { "left": "issue:46640", "right": "issue:46661", "accept": false, - "reason": "Flash Attention 2 issue and remote-code model_class/config_class AttributeError are not the same underlying bug." + "reason": "Flash Attention fallback and remote-code model_class/config_class errors are unrelated." }, { "left": "issue:43716", "right": "issue:44918", "accept": false, - "reason": "Both touch multimodal-style inputs, but one is dtype mismatch in preprocessing and the other is Qwen3.5 embedding unpacking with TRL SFT. Not duplicates." + "reason": "Different multimodal/training failures: image dtype mismatch vs Qwen3.5 embedding unpacking." }, { "left": "issue:39186", "right": "issue:44918", "accept": false, - "reason": "FSDP weight-dimension error and Qwen3.5 embedding unpacking are unrelated." + "reason": "FSDP weight dimensionality and Qwen3.5 embedding unpacking are different problems." }, { "left": "issue:40990", "right": "issue:44960", "accept": false, - "reason": "Perplexity regression on gpt-oss-20b and GLM5 are different model-specific problems." + "reason": "Perplexity regression on gpt-oss and GLM5 model support are unrelated." }, { "left": "issue:43645", "right": "issue:45399", "accept": false, - "reason": "Jupyter custom-model initialization and flash-attn2 fallback gating are separate issues." + "reason": "Jupyter custom-model initialization and flash-attn fallback blocking are different issues." }, { "left": "issue:45632", "right": "issue:45698", "accept": false, - "reason": "Both involve remote code/custom modules, but one is cache-path collision and the other is wrong module loading after save_pretrained. Different root causes." + "reason": "Both involve remote-code loading, but cache-path collision and wrong module loading after save are not the same bug." }, { "left": "issue:43404", "right": "issue:45005", "accept": false, - "reason": "lm_head tying failure in Mistral3 multimodal generation is not the same as translation-model tied-weight regressions in v5." + "reason": "Tied lm_head in Mistral3 and translation-model tied weights regressions are separate concrete cases." }, { "left": "issue:42915", "right": "issue:45399", "accept": false, - "reason": "Qwen3MoE FineGrainedFP8Config failure and flash-attn2 fallback blockage are unrelated." + "reason": "Qwen3-MoE FP8 failure is unrelated to flash-attn fallback logic." }, { "left": "issue:43873", "right": "issue:45005", "accept": false, - "reason": "Quantization offloading behavior and tied-weight translation regressions do not look like the same bug." + "reason": "Quantization offloading behavior and tied-weights regression are different code paths." }, { "left": "issue:44360", "right": "issue:44512", "accept": false, - "reason": "A discussion about the DSA indexer/ReLU is unrelated to stale docs mentioning the removed transformers run command." + "reason": "DSA indexer activation bug and docs wording change are unrelated." }, { "left": "issue:44530", "right": "issue:45005", "accept": false, - "reason": "PagedAttentionCache linear_attention crash in Qwen3.5 models is a different concrete failure than tied weights on translation models." + "reason": "PagedAttentionCache group-type crash is unrelated to tied-weights issues." }, { "left": "issue:44492", "right": "issue:44512", "accept": false, - "reason": "Typo in cache strategies and stale run-command docs are distinct documentation issues." + "reason": "A cache-strategy typo and a removed-doc-command note are not duplicates." }, { "left": "issue:44863", "right": "issue:45005", "accept": false, - "reason": "NemotronH checkpoint loading failure is not the same as tied weights on translation models." + "reason": "NemotronH checkpoint loading and translation-model tied weights are unrelated." }, { "left": "issue:45183", "right": "issue:46463", "accept": false, - "reason": "Generic XOR input validation bug and T5 decoder-input validation messaging are related only at a high level; not the same concrete fix." + "reason": "Both are validation-message bugs, but one is XOR logic across models and the other is T5 decoder-input validation." }, { "left": "issue:44918", "right": "issue:45507", "accept": false, - "reason": "Qwen3.5 embedding unpacking and GraniteMoEHybrid invalid method calls are different model-specific problems." + "reason": "Qwen3.5 embedding unpacking and GraniteMoEHybrid invalid method are different failures." } ] }, @@ -1344,10 +1272,10 @@ "issue:45987|issue:46858", "issue:38175|issue:43054", "issue:43950|issue:45902", - "issue:41553|issue:43883", "issue:44910|issue:45859", "issue:43054|issue:43493", "issue:42947|issue:44387", + "issue:41553|issue:43883", "issue:45663|issue:45902", "issue:44403|issue:44451", "issue:43975|issue:45520", @@ -1361,220 +1289,220 @@ "issue:43824|issue:43872", "issue:43782|issue:44960" ], - "prepared_review_unit_hash": "4b011528037295966e8f431e50d03690f7aef494c1c30f724f423a0dd614d8e5", + "prepared_review_unit_hash": "cf0634f2e240521ccb9c9033aea1b9081b0b8e2f85d35144ac5e1ce08ba85312", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 36285, - "estimated_input_tokens": 9072, - "estimated_eval_tokens": 18400 + "serialized_chars": 36283, + "estimated_input_tokens": 9071, + "estimated_eval_tokens": 18398 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous, but there are a few clear duplicate subgroups: non-persistent buffer corruption in v5, SigLIP2 implementation mismatch/quality regression, and static-cache seq_length return-type issues. Most other pairs are related only at a broad subsystem level and should stay separate.", - "confidence": 0.72, - "canonical_issue_reason": "issue:43950 is the strongest canonical anchor: it describes the broad v5 regression around non-persistent buffers being corrupted, and it has the clearest nearby duplicates (e.g. 44534, likely 45902).", + "summary": "This cluster is mostly a loose set of unrelated Transformers bug reports that only share broad themes like loading, tokenization, cache handling, and quantization. A few pairs are true duplicates, especially the non-persistent-buffer regression, the StaticCache get_seq_length contract breakage, and the Gemma4 quantized embedder dtype-cast bug.", + "confidence": 0.82, + "canonical_issue_reason": "43950 is the strongest canonical anchor: it describes the broadest and clearest regression class (non-persistent buffers being corrupted/junk-filled in Transformers 5.x), and it best captures the shared underlying bug seen in the closest duplicate pair.", "canonical_pr_reason": null, - "best_issue_reason": "43950 is the best overall issue to keep as the representative bug report because it states the underlying regression generically rather than a single model-specific manifestation.", + "best_issue_reason": "43950 is the best representative issue because it is broadly worded, regression-focused, and central to one of the clearest duplicate groups in the set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43975", "right": "issue:44292", "accept": false, - "reason": "Different models and failures: DeepSeek tokenizer detokenization bug vs Qwen-3-8B-NVFP4 load/run error." + "reason": "Different failures on different models: DeepSeek-Coder detokenization vs Qwen NVFP4 load/runtime error. Same broad area, not the same bug." }, { "left": "issue:46489", "right": "issue:46491", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different models and describe different concrete bugs (DeepSeek-Coder output vs CodeLlama leading-space round-trip)." + "reason": "Both are tokenizer regressions, but they describe different concrete model-specific symptoms and no shared code path is evident from the titles alone." }, { "left": "issue:39692", "right": "issue:43054", "accept": false, - "reason": "Same SigLIP2 area, but one is a docs/example mismatch + quantization issue and the other is a text-embedding quality regression." + "reason": "Both mention SigLIP2, but one is a docs/example mismatch and quantization failure, while the other is a model quality/embedding discrepancy." }, { "left": "issue:43950", "right": "issue:44534", "accept": true, - "reason": "Same underlying v5 bug: non-persistent buffers are being populated/corrupted with junk during loading/materialization." + "reason": "Same underlying non-persistent-buffer regression: both describe v5 filling/corrupting register_buffer(persistent=False) state with junk." }, { "left": "issue:36331", "right": "issue:41762", "accept": false, - "reason": "Unrelated trainer API mismatch vs Gemma3 DeepSpeed ZeRO-3 model loading failure." + "reason": "Unrelated errors: custom trainer compute_loss kwarg mismatch vs Gemma3 loading failure under DeepSpeed ZeRO-3." }, { "left": "issue:43643", "right": "issue:43950", "accept": false, - "reason": "Different regressions: missing fields from trust_remote_code AutoConfig vs non-persistent buffer corruption." + "reason": "Remote-code config field loss and non-persistent-buffer corruption are different bugs affecting different parts of the loading stack." }, { "left": "issue:43872", "right": "issue:44589", "accept": false, - "reason": "Both are quantization/storage-related, but they are distinct incompatibility errors with different failing code paths." + "reason": "Different quantization/storage failures: bitsandbytes Int8Params constructor incompatibility vs missing Float8 storage object." }, { "left": "issue:46899", "right": "issue:46932", "accept": true, - "reason": "Same Gemma4 quantized vision/audio embedder dtype-casting bug; 46932 is explicitly a fix for the same issue." + "reason": "These are the same Gemma4 quantized vision/audio embedder dtype-casting bug; the second explicitly calls it the same issue." }, { "left": "issue:43828", "right": "issue:44928", "accept": false, - "reason": "Different Qwen/Phi training failures: dtype mismatch under autocast vs 3D position_ids causing flash-attn/SDPA issues." + "reason": "Different model and failure mode: autocast dtype mismatch vs RLHF NaN/SDPA fallback from 3D position_ids." }, { "left": "issue:45920", "right": "issue:46489", "accept": false, - "reason": "Both are tokenizer regressions in v5, but the affected models and symptoms are different enough that they do not look like one concrete bug." + "reason": "Both are tokenizer regressions, but the first is a broad AutoTokenizer token-id issue across many models while the second is a specific DeepSeek-Coder output bug tied to a prior fix gap." }, { "left": "issue:43054", "right": "issue:45701", "accept": false, - "reason": "SigLIP2 embedding-quality regression vs a broad tokenization-version-change report; not the same underlying bug." + "reason": "SigLIP2 embedding quality complaint vs general tokenizer-change report; no concrete same bug is shown." }, { "left": "issue:45987", "right": "issue:46858", "accept": true, - "reason": "Same static-cache seq_length contract bug: it returns a tensor instead of an int, which breaks downstream arithmetic." + "reason": "Same StaticCache.get_seq_length contract breakage: one report says it returns a shape-(1,) tensor, the other says it returns a device tensor and breaks slice arithmetic." }, { "left": "issue:38175", "right": "issue:43054", "accept": false, - "reason": "Both mention SigLIP2, but one is zero-probabilities behavior and the other is embedding quality; not the same bug." + "reason": "Both involve SigLIP2 behavior, but one reports zero probabilities and the other lower text-embedding quality; too little evidence of the same bug." }, { "left": "issue:43950", "right": "issue:45902", - "accept": true, - "reason": "Same non-persistent-buffer corruption family; 45902 is a model-specific manifestation after meta -> to_empty(cuda) materialization." - }, - { - "left": "issue:41553", - "right": "issue:43883", "accept": false, - "reason": "Bad AutoTokenizer error message for Voxtral vs Molmo attribute error during loading; unrelated failures." + "reason": "Related around buffer initialization, but one is a broad from_pretrained non-persistent-buffer corruption regression and the other is a specific model/materialization uninitialized-memory read." }, { "left": "issue:44910", "right": "issue:45859", "accept": false, - "reason": "Both involve Qwen3.5, but one is flash-attn illegal memory access from 3D position_ids and the other is CPU-offload incompatibility in rotary embedding." + "reason": "Different problems: Flash Attention crash from packed-sequence misinterpretation vs CPU-offload incompatibility in a Qwen3_5MoE rotary embedding." }, { "left": "issue:43054", "right": "issue:43493", - "accept": true, - "reason": "Same SigLIP2 underlying problem: HF implementation diverges from the original JAX behavior, which plausibly explains the embedding-quality regression." + "accept": false, + "reason": "Both discuss SigLIP2, but one is about degraded embeddings and the other is a discrepancy versus JAX implementation; not clearly the same concrete bug." }, { "left": "issue:42947", "right": "issue:44387", "accept": false, - "reason": "Different memory problems: LoRA gradient checkpointing inefficiency vs int4 quantization reserved-memory OOM." + "reason": "Different training pathologies: LoRA gradient checkpointing ineffective vs int4 quantization causing higher CUDA reserved memory/OOM." + }, + { + "left": "issue:41553", + "right": "issue:43883", + "accept": false, + "reason": "Unrelated model-loading issues: bad AutoTokenizer error message for Voxtral vs missing all_tied_weights_keys on MolmoForCausalLM." }, { "left": "issue:45663", "right": "issue:45902", "accept": false, - "reason": "Different models and code paths: Gemma4 FSDP2 shared_kv_states bug vs Qwen3_5Moe non-persistent buffer initialization." + "reason": "Different code paths and symptoms: Gemma4 FSDP2 shared_kv_states KeyError vs non-persistent buffer materialization reading junk." }, { "left": "issue:44403", "right": "issue:44451", "accept": false, - "reason": "Generic loading noise vs a specific model failing to load; not the same underlying bug." + "reason": "One is about noisy loading logs, the other is a model-loading failure for ScandiBERT." }, { "left": "issue:43975", "right": "issue:45520", "accept": false, - "reason": "DeepSeek tokenizer detokenization bug vs Python 3.13 flash_attn import error; unrelated." + "reason": "Different issues: tokenizer detokenization bug vs Python 3.13 import_utils flash_attn KeyError." }, { "left": "issue:24643", "right": "issue:28530", "accept": false, - "reason": "Training DeepSpeed weight-dimension error vs early-stopping metric lookup issue; unrelated." + "reason": "Training with DeepSpeed weight shape error vs early-stopping metric configuration; unrelated failures." }, { "left": "issue:42947", "right": "issue:43856", "accept": false, - "reason": "Both are training/memory-related, but one is LoRA checkpointing inefficiency and the other is Qwen3 MoE memory usage; not the same bug." + "reason": "Both concern memory/training efficiency, but they are different model/workflow problems and not the same concrete bug." }, { "left": "issue:44403", "right": "issue:44488", "accept": false, - "reason": "Loading noise vs an actual model loading failure for a different checkpoint; too broad to merge." + "reason": "Different symptom severity and scope: noisy transformer loading vs actual loading failure for a specific model." }, { "left": "issue:43643", "right": "issue:45020", "accept": false, - "reason": "Both mention remote_code, but one is missing fields in AutoConfig and the other is a broader version breakage report; not the same concrete bug." + "reason": "Remote-code handling is the common theme, but one is missing fields in AutoConfig output while the other is a broad report of model breakage; not the same concrete defect." }, { "left": "issue:41762", "right": "issue:42915", "accept": false, - "reason": "Different model/runtime failures: Gemma3 DeepSpeed ZeRO-3 loading vs Qwen3Moe FineGrainedFP8Config failure." + "reason": "Different load-time failures on different models and configurations: Gemma3 ZeRO-3 index error vs Qwen3Moe FineGrainedFP8Config failure." }, { "left": "issue:41553", "right": "issue:43881", "accept": false, - "reason": "Bad Voxtral tokenizer error message vs glm-4v-9b loading failure; unrelated." + "reason": "Unrelated model-loading complaints: Voxtral tokenizer error message vs glm-4v-9b load failure." }, { "left": "issue:44991", "right": "issue:45399", "accept": false, - "reason": "Tokenizer-loading regression for a specific model vs flash-attn fallback blocking logic; unrelated." + "reason": "Tokenizer loading regression vs flash-attn fallback gating; different subsystems and failure modes." }, { "left": "issue:43824", "right": "issue:43872", "accept": false, - "reason": "Import error for a missing Qwen2_5_VL class vs bitsandbytes Int8Params constructor incompatibility; unrelated." + "reason": "Import error for a missing model class is unrelated to the bitsandbytes Int8Params constructor incompatibility." }, { "left": "issue:43782", "right": "issue:44960", "accept": false, - "reason": "Qwen3VL weight_only loading error vs a separate GLM5 issue; no concrete duplicate signal." + "reason": "Different models and failure modes: weight_only=True loading error vs a generic GLM5 issue." } ] }, "evaluator_result": { "accept": true, - "feedback": "The analyst output is mostly grounded in the packet and the pairwise verdicts are generally conservative. The main duplicate clusters identified by the titles\u2014non-persistent buffer corruption, SigLIP2 implementation mismatch/quality regression, and the StaticCache seq_length contract bug\u2014are plausible, and the false pairs are mostly well-justified. I did not see a clear overstatement strong enough to reject." + "feedback": "The summary stays grounded in the packet and is appropriately conservative about the cluster being mostly loose/unrelated with only a few likely duplicate pairs. The accepted soft-edge pairs are supported by closely matching titles, and the rejected pairs avoid overclaiming beyond the evidence available in the packet." }, "evaluator_used": true, "retried": false, @@ -1666,28 +1594,28 @@ "issue:43257|issue:45305", "issue:42947|issue:45923", "issue:36331|issue:41093", - "issue:41553|issue:44991", "issue:43940|issue:45313", + "issue:41553|issue:44991", "issue:43854|issue:44863", "issue:44928|issue:45588", "issue:41093|issue:44805", "issue:44479|issue:44560", "issue:24643|issue:30064" ], - "prepared_review_unit_hash": "0b46edec6952398ec2219805efe0ac812570c084117db4ee0d99aedb218ac9e5", + "prepared_review_unit_hash": "cc1d6fbcfefd55e3ff8420db373cd92525024dd532cda1e0057a7c0c7f76b7e8", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36755, + "serialized_chars": 36753, "estimated_input_tokens": 9189, "estimated_eval_tokens": 18634 }, @@ -1696,186 +1624,186 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly unrelated issues, with a few clear duplicate pairs: v5 docs cleanup for removed pipelines, identical mask-shape IndexErrors, the `_is_hf_initialized`/`init_empty_weights` compatibility break, and the training-loss double-shift bug. The remaining pairs are too model- or code-path-specific to merge safely.", - "confidence": 0.87, - "canonical_issue_reason": "issue:46894 is the clearest representative of the strongest duplicate family in this set: a concrete training-loss double-shift bug with the broader, more self-contained title and higher reference activity.", + "summary": "Mostly a heterogeneous cluster of unrelated bug reports, with a few true duplicate pockets: docs about removed pipelines, init_empty_weights/_is_hf_initialized loading failures, identical mask-shape IndexErrors, and the Florence/Cohere double-shift pattern. The rest are similarity-only false positives across different models or subsystems.", + "confidence": 0.82, + "canonical_issue_reason": "No single cluster-wide canonical issue exists because the items span many unrelated bugs; if one issue must anchor the only clearly duplicated concrete bug here, 46894 is the best choice as the earlier, more explicit report paired with 46897.", "canonical_pr_reason": null, - "best_issue_reason": "issue:46894 is the best overall issue representative here because it describes the bug plainly and maps cleanly onto the accepted duplicate pair.", + "best_issue_reason": "46894 is the strongest issue to keep as the representative anchor: it is the earlier member of the clearest true duplicate pair and states the underlying bug cleanly.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45685", "right": "issue:45823", "accept": false, - "reason": "Different failures and code paths: MPS int histogram support vs device-map mismatch in Gemma4." + "reason": "Different symptoms and code paths: MPS histogram int error vs Gemma4 device-mapping mismatch." }, { "left": "issue:45081", "right": "issue:45399", "accept": false, - "reason": "Tokenizer regex crash and flash-attn fallback gating are unrelated bugs." + "reason": "Both touch attention/tokenizer areas, but one is a Mistral regex/tokenizer crash and the other is a flash-attn fallback blocker." }, { "left": "issue:45020", "right": "issue:45823", "accept": false, - "reason": "Remote-code regression vs Gemma4 device mismatch; different root causes." + "reason": "Broad \"recent versions break remote_code\" report is not the same bug as a Gemma4 forward device mismatch." }, { "left": "issue:44661", "right": "issue:45020", "accept": false, - "reason": "Tokenizer mapping edge case vs remote-code loading break; not the same bug." + "reason": "Tokenizer mapping edge case vs remote-code model loading regressions; too different to treat as one bug." }, { "left": "issue:43881", "right": "issue:43883", "accept": false, - "reason": "Different models and different load-time failures; not a shared code-path bug." + "reason": "Both are model-load failures, but for different models and different missing attributes." }, { "left": "issue:43828", "right": "issue:45823", "accept": false, - "reason": "Autocast dtype mismatch and device-map mismatch are different issues." + "reason": "Dtype mismatch under autocast is a different failure mode from a device placement bug." }, { "left": "issue:42491", "right": "issue:44368", "accept": false, - "reason": "Different LoRA/tie-embeddings problems; same area, but not the same bug." + "reason": "Both mention Qwen/LoRA, but one is a cross-version compatibility failure and the other is just a warning about tie_word_embeddings." }, { "left": "issue:45020", "right": "issue:45356", "accept": false, - "reason": "Remote-code loading regression vs tokenizer codec handling regression." + "reason": "General remote_code regressions and a Kimi tokenizer codec regression are not the same underlying bug." }, { "left": "issue:45715", "right": "issue:46315", "accept": false, - "reason": "Both are tokenizer performance regressions, but they hit different methods and failure modes." + "reason": "Same tokenizer performance theme, but different functions and likely different hot paths." }, { "left": "issue:43872", "right": "issue:46869", "accept": false, - "reason": "Bitsandbytes constructor incompatibility vs YOLO multi-GPU device-map failure; unrelated." + "reason": "Bitsandbytes init-empty-weights incompatibility is unrelated to YOLO multi-GPU device-map failure." }, { "left": "issue:45759", "right": "issue:46661", "accept": false, - "reason": "Composite config unwrapping vs remote-code config_class lookup; different auto-model paths." + "reason": "Both involve model construction, but one is config unwrapping for composite multimodal configs and the other is remote-code config_class resolution." }, { "left": "issue:45488", "right": "issue:46489", "accept": false, - "reason": "Both involve DeepSeek tokenizers, but the concrete tokenizer regression differs." + "reason": "Both are tokenizer regressions around DeepSeek-family models, but they describe different failures and different root causes." }, { "left": "issue:43827", "right": "issue:44509", "accept": true, - "reason": "Same docs cleanup for v5 pipeline removals; both report obsolete summarization/translation pipeline references." + "reason": "Same docs cleanup: both say removed pipeline tasks are still referenced after v5 removals." }, { "left": "issue:43404", "right": "issue:43950", "accept": false, - "reason": "lm_head tying bug and non-persistent buffer corruption are unrelated regressions." + "reason": "Tied lm_head weights vs non-persistent buffer corruption are distinct model-state bugs." }, { "left": "issue:46894", "right": "issue:46897", "accept": true, - "reason": "Same training-loss double-shift bug pattern; both are model-specific reports of the same underlying loss-shift defect." + "reason": "Near-identical duplicate reports of the same training-loss double-shift bug." }, { "left": "issue:43872", "right": "issue:44291", "accept": true, - "reason": "Same `_is_hf_initialized` compatibility break when loading with `init_empty_weights` in transformers 5.x." + "reason": "Same unexpected _is_hf_initialized / init_empty_weights loading failure." }, { "left": "issue:45020", "right": "issue:46661", "accept": false, - "reason": "Remote-code regression and `config_class` AttributeError are different failure modes." + "reason": "Generic remote-code breakage and an AutoModel config_class AttributeError are related only at a high level." }, { "left": "issue:43792", "right": "issue:44220", "accept": false, - "reason": "Whisper runtime failure and `_torch_extract_fbank_features()` issue are not clearly the same code-path bug." + "reason": "Whisper not running and a specific fbank extraction issue are plausibly related, but not clearly the same concrete bug." }, { "left": "issue:43257", "right": "issue:45305", "accept": false, - "reason": "Both are DeepSpeed-related, but one is weight conversion and the other is gradient averaging; different bugs." + "reason": "Both involve DeepSpeed/Qwen training, but one is weight conversion at load time and the other is gradient averaging during training." }, { "left": "issue:42947", "right": "issue:45923", "accept": false, - "reason": "Gradient checkpointing ineffective vs missing support flag on remote-code variant; not the same issue." + "reason": "Gradient checkpointing problems, but one is PEFT LoRA ineffectiveness and the other is a missing support flag on a remote-code variant." }, { "left": "issue:36331", "right": "issue:41093", "accept": false, - "reason": "Unexpected kwarg in `compute_loss` and mask/tensor shape mismatch are unrelated." + "reason": "Unexpected compute_loss kwarg and mask-shape IndexError are unrelated training failures." }, { - "left": "issue:41553", - "right": "issue:44991", + "left": "issue:43940", + "right": "issue:45313", "accept": false, - "reason": "Bad tokenizer error message vs tokenizer loading failure for a different model; different bugs." + "reason": "Both are ZeRO-3 weight-loading failures, but on different Qwen variants with different reported failure shapes; not enough to merge." }, { - "left": "issue:43940", - "right": "issue:45313", + "left": "issue:41553", + "right": "issue:44991", "accept": false, - "reason": "Both are ZeRO-3 weight-loading problems, but the model targets and symptoms differ too much to merge as one bug." + "reason": "Bad tokenizer error message for Voxtral vs tokenizer loading failure for EMBEDDIA/est-roberta are different issues." }, { "left": "issue:43854", "right": "issue:44863", "accept": false, - "reason": "Different models and different checkpoint-loading failures." + "reason": "Different model families and different load/test failures." }, { "left": "issue:44928", "right": "issue:45588", "accept": false, - "reason": "Gradient explosion in RLHF and flash-attn `s_aux=None` crash are unrelated." + "reason": "RLHF gradient explosion from 3D position_ids is unrelated to a flash_attention s_aux=None crash." }, { "left": "issue:41093", "right": "issue:44805", "accept": true, - "reason": "Same mask/tensor shape mismatch IndexError; the failure text and bug pattern match closely." + "reason": "Same generic mask-shape mismatch IndexError; likely the same label/mask alignment bug reported twice." }, { "left": "issue:44479", "right": "issue:44560", "accept": false, - "reason": "Different video-input regressions affecting different model families and code paths." + "reason": "Both are Qwen3 VL/video regressions, but they describe different broken paths and symptoms." }, { "left": "issue:24643", "right": "issue:30064", "accept": false, - "reason": "DeepSpeed training weight-shape error and void segmentation-map processing are unrelated." + "reason": "DeepSpeed training weight shape error and void segmentation map processing 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 issue titles: the v5 docs cleanup, the mask-shape IndexErrors, the `_is_hf_initialized` / `init_empty_weights` compatibility break, and the training-loss double-shift reports all look like the same underlying bug family. The rejected pairs are generally framed cautiously enough." + "feedback": "The cluster read is grounded in the packet and stays mostly conservative. The three accepted soft pairs are well supported by highly similar titles and error patterns, and the representative choice of 46894 over 46897 is reasonable as the earlier duplicate. The summary appropriately characterizes the rest as heterogeneous without over-claiming a cluster-wide canonical issue." }, "evaluator_used": true, "retried": false, @@ -1947,8 +1875,8 @@ ], "soft_pairs": [ "issue:44530|issue:44863", - "issue:41553|issue:44488", "issue:44863|issue:46402", + "issue:41553|issue:44488", "issue:43366|issue:43828", "issue:45325|issue:45902", "issue:43366|issue:43531", @@ -1981,20 +1909,20 @@ "issue:43596|issue:44661", "issue:42491|issue:43720" ], - "prepared_review_unit_hash": "d12caa3ba706ed674ede8f8d4a7227e600a48733dc2824329edc482a518bff73", + "prepared_review_unit_hash": "dc15327b9947e058094318053db3510dd3026b90e571361f5004d185320969d8", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 38058, + "serialized_chars": 38059, "estimated_input_tokens": 9515, "estimated_eval_tokens": 19286 }, @@ -2003,222 +1931,222 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is mostly heterogeneous. I only see a few likely duplicate issue pairs: the two custom-model init regression reports, the two Siglip2 wrong-output reports, and the two Gemma4 4-bit multimodal dtype-cast reports. Everything else looks like distinct bugs or regressions in different models/code paths.", - "confidence": 0.84, - "canonical_issue_reason": "As a local anchor, issue 43994 is the clearest and most general duplicate-report candidate: it states the Siglip2 regression across AutoModel and pipeline use, which subsumes the more symptom-specific 38175 report.", + "summary": "This set is mostly heterogeneous, with only two clear duplicate subclusters: Transformers v5 custom model initialization, and SigLIP2 bad-output reports. Most other pairs share a subsystem or model family but not the same concrete bug.", + "confidence": 0.63, + "canonical_issue_reason": "issue:43646 is the broadest clean duplicate target for the true custom-model-initialization regression; issue:43645 is just a notebook-specific repro of the same problem.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43994 is the strongest representative issue in this set because it is concise, reproducible at a high level, and clearly framed around the same broken Siglip2 behavior as its duplicate counterpart.", + "best_issue_reason": "issue:43646 is the best representative of the accepted duplicate set because it names the underlying regression cleanly and broadly.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44530", "right": "issue:44863", "accept": false, - "reason": "Different models and failures: Qwen3.5 PagedAttentionCache/linear_attention vs NemotronH checkpoint loading." + "reason": "Both are model-specific loading failures, but they concern different architectures and different failure modes." }, { - "left": "issue:41553", - "right": "issue:44488", + "left": "issue:44863", + "right": "issue:46402", "accept": false, - "reason": "Both are model-loading complaints, but they target different models and likely different root causes." + "reason": "Both mention weights not loading, but they are different models and likely different code paths." }, { - "left": "issue:44863", - "right": "issue:46402", + "left": "issue:41553", + "right": "issue:44488", "accept": false, - "reason": "Different architectures and symptoms: NemotronH checkpoint loading vs Sam3Model text-encoder weights not loading." + "reason": "Both are loading-related, but one is an AutoTokenizer error message and the other is a separate model-loading regression." }, { "left": "issue:43366", "right": "issue:43828", "accept": false, - "reason": "Unrelated bug classes: GGUF architecture support vs autocast dtype mismatch in Phi-tiny-MoE." + "reason": "Different models and different runtime failures; only a loose MoE-family similarity." }, { "left": "issue:45325", "right": "issue:45902", "accept": false, - "reason": "Both involve Qwen rotary/position logic, but they are different models and different failure modes." + "reason": "Both involve positional/rotary internals, but they are on different models with different bugs." }, { "left": "issue:43366", "right": "issue:43531", "accept": false, - "reason": "No shared underlying bug; one is GGUF support, the other is a sliding_window issue in Qwen3-MoE." + "reason": "One is GGUF support for gpt-oss; the other is a Qwen3-MoE sliding_window bug." }, { "left": "issue:43596", "right": "issue:46515", "accept": false, - "reason": "Different problem domains: DeepSpeed ZeRO3/Bert indexing vs a torch version incompatibility report." + "reason": "An index error during Deepspeed/Bert init is not the same as a broad torch-version incompatibility report." }, { "left": "issue:44155", "right": "issue:45584", "accept": false, - "reason": "Both are audio-related, but the model, workflow, and failure mechanism are different." + "reason": "Both are audio-related, but one is cross-track embedding leakage and the other is empty-transcription generation failure." }, { "left": "issue:45663", "right": "issue:45823", "accept": false, - "reason": "Same family (Gemma4) but different concrete bugs: FSDP2 shared_kv_states rebuild vs device_map auto device mismatch." + "reason": "Same model family, but FSDP2 shared_kv_state rebuilding and device_map auto PLE mismatch are different bugs." }, { "left": "issue:43366", "right": "issue:44589", "accept": false, - "reason": "No meaningful overlap beyond general backend/model-support context." + "reason": "These are unrelated failures on different models; one is a support request, the other is a storage-type error." }, { "left": "issue:43638", "right": "issue:46515", "accept": false, - "reason": "Non-pretrained Bert/ZeRO3 index error is unrelated to a torch 2.4.1 compatibility complaint." + "reason": "Both mention training/runtime issues, but the specific Deepspeed/Bert index error is unrelated to torch 2.4.1 incompatibility." }, { "left": "issue:43950", "right": "issue:45834", "accept": false, - "reason": "Different bugs: non-persistent buffer corruption during from_pretrained vs long-ocr index error in Kosmos2.5." + "reason": "Different regressions: non-persistent buffer corruption versus a Kosmos2.5 OCR length/index problem." }, { "left": "issue:43824", "right": "issue:43957", "accept": false, - "reason": "ImportError for a missing class and meta-device loading breakage are different code paths and symptoms." + "reason": "ImportError for a missing class is not the same bug as meta-device loading breakage." }, { "left": "issue:42915", "right": "issue:43957", "accept": false, - "reason": "Qwen3Moe FP8Config failure is not the same underlying issue as generic meta-device loading breakage." + "reason": "Qwen3-MoE FP8 failure and meta-device loading regression are distinct code paths." }, { "left": "issue:43716", "right": "issue:44560", "accept": false, - "reason": "Different multimodal models and different errors; no evidence of a shared root cause." + "reason": "Different multimodal bugs: image-preprocessor dtype mismatch versus Qwen3-VL video StopIteration." }, { "left": "issue:44910", "right": "issue:46858", "accept": false, - "reason": "Both involve cache/attention internals, but the specific bug is different: Qwen3.5 packed position_ids vs static-cache seq_length arithmetic." + "reason": "Both touch caching/position handling, but one is a Flash Attention packed-sequence crash and the other is a device-tensor seq-length arithmetic bug." }, { "left": "issue:43645", "right": "issue:43646", "accept": true, - "reason": "These are the same custom-model initialization regression, with 43645 just adding the Jupyter-notebook context to the broader 43646 report." + "reason": "Same custom-model-initialization regression in Transformers 5.0.0; 43645 is just a Jupyter-notebook repro." }, { "left": "issue:42617", "right": "issue:43366", "accept": false, - "reason": "A script/runtime failure is unrelated to GGUF architecture support." + "reason": "A script execution failure and GGUF support request are not the same underlying issue." }, { "left": "issue:44560", "right": "issue:45685", "accept": false, - "reason": "Different backend/model issues: Qwen3-vl video StopIteration vs an MPS histogram kernel type error." + "reason": "Both are model/runtime failures, but one is Qwen3-VL video handling and the other is an MPS histogram error in MoE code." }, { "left": "issue:43550", "right": "issue:44910", "accept": false, - "reason": "Both touch attention/compile paths, but they are different models and distinct failures." + "reason": "Torch.compile+SDPA on Bamba-9B-v2 is a different failure from Qwen3.5 flash-attention position-id handling." }, { "left": "issue:36010", "right": "issue:42915", "accept": false, - "reason": "Importing GenerationMixin and Qwen3Moe FP8 failures are unrelated." + "reason": "Importing GenerationMixin and Qwen3-MoE FP8 are unrelated regressions." }, { "left": "issue:44683", "right": "issue:46515", "accept": false, - "reason": "Both mention torch versions, but one is flex_attention on torch>=2.9 and the other is a torch 2.4.1 incompatibility." + "reason": "Compiled flex_attention on torch>=2.9 is a different compatibility break than the torch 2.4.1 report." }, { "left": "issue:44560", "right": "issue:44918", "accept": false, - "reason": "Same general model family, but one is video processing StopIteration and the other is embedding unpacking with TRL SFT trainer." + "reason": "Both mention Qwen3.5 VL, but one is a video StopIteration and the other is SFT embedding unpacking." }, { "left": "issue:45812", "right": "issue:46489", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different model families and appear to be separate bugs." + "reason": "Both are tokenizer regressions, but they affect different model families and likely different mappings." }, { "left": "issue:45823", "right": "issue:46515", "accept": false, - "reason": "Gemma4 device placement mismatch is not the same as a generic torch 2.4.1 incompatibility." + "reason": "Gemma4 device-mapping and a generic torch incompatibility are not the same bug." }, { "left": "issue:44661", "right": "issue:45823", "accept": false, - "reason": "add-new-model-like/tokenizer mapping logic is unrelated to Gemma4 forward device mismatch." + "reason": "add-new-model-like mapping logic is unrelated to Gemma4 PLE device placement." }, { "left": "issue:38175", "right": "issue:43994", "accept": true, - "reason": "Same Siglip2 model regression: both report incorrect/nonsensical outputs from the same underlying model behavior, with 43994 being the clearer general report." + "reason": "Both reports are about the same SigLIP2 base model producing obviously wrong outputs; the later issue is the broader duplicate target." }, { "left": "issue:44560", "right": "issue:44805", "accept": false, - "reason": "Different Qwen3-vl video/input issue vs a generic mask-shape IndexError." + "reason": "Different symptoms and code paths: video StopIteration versus mask shape mismatch." }, { "left": "issue:45672", "right": "issue:46899", - "accept": true, - "reason": "Both describe the same Gemma4 multimodal 4-bit dtype-cast problem; 46899 is the broader formulation and 45672 is a specific audio-module symptom." + "accept": false, + "reason": "Both are Gemma4 quantization issues, but one is an audio-module uint8 torch.finfo error and the other is unsafe dtype casting in multimodal embedders." }, { "left": "issue:44877", "right": "issue:45030", "accept": false, - "reason": "Both involve stricter config validation, but the affected models and failures are different." + "reason": "Config validation blocking granite_speech and tiny-random glm4v loading are separate config regressions." }, { "left": "issue:43632", "right": "issue:45834", "accept": false, - "reason": "_is_hf_initialized regression is unrelated to Kosmos2.5 long-input indexing." + "reason": "Both are runtime errors, but one is the `_is_hf_initialized` flag regression and the other is a Kosmos2.5 OCR index error." }, { "left": "issue:43638", "right": "issue:44661", "accept": false, - "reason": "DeepSpeed ZeRO3/Bert indexing and tokenizer mapping validation are different bugs." + "reason": "Deepspeed zero3/Bert initialization and add-new-model-like mapping validation are not the same failure." }, { "left": "issue:43596", "right": "issue:44661", "accept": false, - "reason": "Zero3/Bert index error is unrelated to add-new-model-like failing in TOKENIZER_MAPPING_NAMES." + "reason": "A zero-sized parameter index error is unrelated to tokenizer-mapping validation." }, { "left": "issue:42491", "right": "issue:43720", "accept": false, - "reason": "Different areas and symptoms: Qwen3_moe LoRA compatibility vs BitNet packed-weight loading." + "reason": "Qwen3-MoE LoRA compatibility and BitNet AutoBitLinear packed-weight loading are unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet and stays conservative overall. The three accepted duplicate pairs are plausible from the titles/body scope: 43645/43646, 38175/43994, and 45672/46899. The other soft-edge rejections are appropriately cautious and do not overstate shared root causes." + "feedback": "The summary is grounded in the packet, and the soft-edge decisions stay conservative overall. The two accepted duplicate pairs are well supported by the issue titles, while the rest are reasonably rejected as similar subsystems/model families but different concrete bugs." }, "evaluator_used": true, "retried": false, @@ -2308,8 +2236,8 @@ "issue:44488|issue:44991", "issue:43856|issue:43873", "issue:42491|issue:45292", - "issue:43856|issue:44387", "issue:43854|issue:43931", + "issue:43856|issue:44387", "issue:43708|issue:44075", "issue:42491|issue:45335", "issue:44960|issue:46050", @@ -2320,220 +2248,220 @@ "issue:45362|issue:45406", "issue:44910|issue:45588" ], - "prepared_review_unit_hash": "d28fe64f2e01379b4a223ed3e55e72376291dd060d2da2b3dd85c9a9c6bc6697", + "prepared_review_unit_hash": "4eb2cb3666c77204b56937f1e960a0136953eefec77515171e3570d6455129ff", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36655, - "estimated_input_tokens": 9164, - "estimated_eval_tokens": 18584 + "serialized_chars": 36657, + "estimated_input_tokens": 9165, + "estimated_eval_tokens": 18586 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is mostly heterogeneous. Only two soft edges look like true duplicates: the Transformers 5 non-persistent-buffer regression and the resize_token_embeddings/output_embeddings regression. The rest appear to be separate model-, tokenizer-, loading-, or training-specific bugs.", - "confidence": 0.91, - "canonical_issue_reason": "issue:43644 is the strongest canonical anchor here: it is the earlier, more descriptive report for the same Transformers 5 non-persistent-buffer corruption bug, and issue:44534 is essentially a reworded duplicate.", + "summary": "This set is mostly heterogeneous and not one clean duplicate cluster. I only see a few genuine duplicate pairs: the Transformers v5 non-persistent-buffer corruption reports, the `resize_token_embeddings` output-embedding mismatch reports, and the training-loss double-shift reports.", + "confidence": 0.71, + "canonical_issue_reason": "Issue 43644 is the best canonical representative of the clearest duplicate subcluster: 44534 is essentially a reworded report of the same Transformers v5 non-persistent-buffer junk regression.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43644 is the best issue representative in this set because it has the clearest root-cause framing and the closest near-verbatim duplicate (issue:44534).", + "best_issue_reason": "43644 is the strongest overall issue to anchor because it is the most explicit, broad, and clearly duplicated report among the true duplicate subgroups.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43317", "right": "issue:44756", "accept": false, - "reason": "Different problems: dequantized model offload/loading vs mmap-related OOM avoidance." + "reason": "Different bugs: one is quantized/dequantized loading with device_map/offload, the other is Strix Halo mmap-related OOM mitigation." }, { "left": "issue:44661", "right": "issue:44805", "accept": false, - "reason": "Unrelated bugs: tokenizer mapping model-registration failure vs mask/tensor shape mismatch." + "reason": "Unrelated failures: tokenizer mapping/name handling vs an attention mask shape mismatch." }, { "left": "issue:38175", "right": "issue:39692", "accept": false, - "reason": "Both involve SigLIP2, but one is zero probabilities and the other is docs/model-processor/quantization issues; not the same bug." + "reason": "Same model family, but different concrete issues: unexpected zero probabilities vs docs example model/processor mismatch and quantization errors." }, { "left": "issue:43644", "right": "issue:44534", "accept": true, - "reason": "Same underlying Transformers v5 regression: non-persistent buffers are being filled with junk." + "reason": "Same regression, just reworded: both report Transformers v5 filling non-persistent buffers with junk." }, { "left": "issue:41720", "right": "issue:42915", "accept": false, - "reason": "Different Qwen3 failures with different triggers and code paths." + "reason": "Both involve Qwen3, but one is auto device mapping on A800 and the other is FineGrainedFP8Config; different code paths and symptoms." }, { "left": "issue:42915", "right": "issue:44756", "accept": false, - "reason": "Qwen3MoE FP8/config failure is unrelated to mmap/OOM offload behavior." + "reason": "Completely different areas: FP8/Qwen3-MoE failure versus Strix Halo mmap/OOM behavior." }, { "left": "issue:42491", "right": "issue:45276", "accept": false, - "reason": "Different bugs: LoRA checkpoint compatibility on hf5.x vs resize_token_embeddings not updating embeddings." + "reason": "Different bugs and models: LoRA compatibility across hf4.x/hf5.x for Qwen3 MoE versus Gemma4 resize-token-embedding behavior." }, { "left": "issue:44960", "right": "issue:44977", "accept": false, - "reason": "Different models and different failure modes; no shared concrete code-path bug." + "reason": "Different model-specific generation bugs; no evidence they share the same concrete code path." }, { "left": "issue:44451", "right": "issue:44488", "accept": false, - "reason": "Both are load failures, but the model names and reports are distinct; not enough evidence of the same underlying bug." + "reason": "Both are model-loading failures, but for different checkpoints with no evidence of the same root cause." }, { "left": "issue:44863", "right": "issue:46846", "accept": false, - "reason": "NemotronH checkpoint loading and Qwen3.5 TP-plan issues are unrelated." + "reason": "Different problems: NemotronH checkpoint loading versus Qwen3.5 tensor-parallel planning/OOM at TP>1." }, { "left": "issue:46459", "right": "issue:46512", "accept": false, - "reason": "Different dtype-related bugs: one is loading dtype being ignored, the other is config.dtype not being updated." + "reason": "Related to dtype handling, but one is composite checkpoint loading and the other is from_config dtype/config staleness; not the same bug." }, { "left": "issue:45313", "right": "issue:45357", "accept": false, - "reason": "Different subsystems: ZeRO-3 language_model loading vs save_pretrained visual-encoder key regression." + "reason": "Different components and code paths: ZeRO-3 weight loading for language_model versus save_pretrained key serialization for the visual encoder." }, { "left": "issue:45507", "right": "issue:46468", "accept": false, - "reason": "Invalid method call in GraniteMoEHybrid is unrelated to the Nemotron example breakage." + "reason": "Unrelated issues in different model/example flows." }, { "left": "issue:43644", "right": "issue:43645", "accept": false, - "reason": "Different regressions: buffer corruption vs Jupyter custom-model initialization breakage." + "reason": "Both are Transformers v5 regressions, but one is buffer initialization and the other is Jupyter custom-model initialization; different concrete bugs." }, { "left": "issue:39692", "right": "issue:43994", "accept": false, - "reason": "Related model family, but one is a docs/example mismatch and quantization failure while the other is bad outputs through AutoModel/pipeline." + "reason": "Same model family, but one is a docs example/mismatch report and the other is nonsensical inference outputs; not the same concrete failure." }, { "left": "issue:43257", "right": "issue:43854", "accept": false, - "reason": "Different model-loading issues with different frameworks and symptoms." + "reason": "Different models and symptoms: Qwen3 MoE conversion under accelerate+deepspeed versus GLM-4.7-Flash unit-test loading." }, { "left": "issue:44488", "right": "issue:44991", "accept": false, - "reason": "Both are loading/tokenizer failures, but the affected models and concrete failure reports differ enough to keep separate." + "reason": "Both are tokenizer-load regressions, but on different models with no direct evidence they stem from the same root cause." }, { "left": "issue:43856", "right": "issue:43873", "accept": false, - "reason": "MoE training memory usage and quantization offload are separate issues." + "reason": "Different resource-management problems: Qwen3 MoE training memory inefficiency versus quantization offload behavior." }, { "left": "issue:42491", "right": "issue:45292", "accept": false, - "reason": "Different problems: hf4.x/hf5.x LoRA compatibility vs resize_token_embeddings not updating output embeddings." + "reason": "Different architectures and failure modes; only the shared phrase `resize_token_embeddings` overlaps." }, { - "left": "issue:43856", - "right": "issue:44387", + "left": "issue:43854", + "right": "issue:43931", "accept": false, - "reason": "Different memory regressions with different quantization/training contexts." + "reason": "Different loading failures for different models: GLM-4.7-Flash test loading versus Qwen3-VL weight-shape mismatch." }, { - "left": "issue:43854", - "right": "issue:43931", + "left": "issue:43856", + "right": "issue:44387", "accept": false, - "reason": "Different model-specific loading failures for different checkpoints." + "reason": "Both are memory-related, but one is training-memory inefficiency and the other is int4 quantization reserved-memory/OOM regression." }, { "left": "issue:43708", "right": "issue:44075", "accept": false, - "reason": "Trainer resume/max_steps bug is unrelated to SGD optimizer arguments not being used." + "reason": "Trainer checkpoint-resume math bug is unrelated to optimizer argument handling." }, { "left": "issue:42491", "right": "issue:45335", "accept": false, - "reason": "Different issues: LoRA checkpoint compatibility vs token embedding resize propagation." + "reason": "Different model families and different affected embeddings; same API name alone is not enough." }, { "left": "issue:44960", "right": "issue:46050", "accept": false, - "reason": "Different model families and different save/load quantization failures." + "reason": "Different bug classes: GLM5 issue versus quantized-model save/load regression." }, { "left": "issue:44164", "right": "issue:44291", "accept": false, - "reason": "Related save/load themes, but one is extra_state handling and the other is init_empty_weights argument compatibility." + "reason": "Different serialization/loading failures: extra_state handling versus unexpected `_is_hf_initialized` argument." }, { "left": "issue:46897", "right": "issue:46901", - "accept": false, - "reason": "Both are training-loss shift bugs, but the model families and exact implementations differ; not clearly the same issue." + "accept": true, + "reason": "Same concrete bug pattern: both report the training-loss double-shift where labels are trained against a shifted slice instead of the original labels." }, { "left": "issue:45488", "right": "issue:45812", "accept": false, - "reason": "Tokenizer-pretokenizer regression and Granite AutoTokenizer ID regression are different concrete bugs." + "reason": "Both are tokenizer regressions, but they affect different model families and different tokenizer behaviors." }, { "left": "issue:45276", "right": "issue:45335", "accept": true, - "reason": "Same underlying resize_token_embeddings bug: output embeddings are not updated correctly; 45276 is a Gemma4-specific manifestation of the same code-path problem." + "reason": "Same concrete `resize_token_embeddings` failure class: the resized embedding changes are not propagated to the model's output/decoder embeddings." }, { "left": "issue:45362", "right": "issue:45406", "accept": false, - "reason": "Different crashes in different models/components; no shared underlying bug." + "reason": "Different failures: Qwen3.5 chat crashing versus Gemma4Processor missing `_tokenizer` in transformers serve." }, { "left": "issue:44910", "right": "issue:45588", "accept": false, - "reason": "Flash-attention misinterpreting 3D position_ids is unrelated to the s_aux=None AttributeError." + "reason": "Different flash-attention bugs: 3D `position_ids` misinterpreted as packed sequence versus `s_aux=None` attribute error." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative overall. The two accepted duplicate edges are well supported by the titles and strong lexical overlap, and the remaining pairwise rejections avoid overclaiming beyond the evidence. The canonical issue choice is also reasonable, with issue:43644 clearly the cleaner anchor over issue:44534." + "feedback": "The summary is grounded in the packet and stays appropriately conservative. The accepted duplicate pairs are all supported by the issue titles and the analyst avoids overclaiming on the many near-matches that lack direct evidence of the same root cause. The main cluster characterization as mostly heterogeneous is reasonable." }, "evaluator_used": true, "retried": false, @@ -2617,8 +2545,8 @@ "issue:45663|issue:45859", "issue:44292|issue:44912", "issue:36010|issue:39404", - "issue:44387|issue:44756", "issue:42947|issue:43381", + "issue:44387|issue:44756", "issue:44155|issue:45823", "issue:43957|issue:44756", "issue:43335|issue:45663", @@ -2636,14 +2564,14 @@ "issue:43828|issue:43957", "issue:43646|issue:43708" ], - "prepared_review_unit_hash": "f27eda6f9bb89312316e7b34e79bb27a7b0f2e39eaf92077dc690f2972436e55", + "prepared_review_unit_hash": "95cd7d0a21672584bab9b5a5579b38c95d76a18d9305062505c896958e90bf4f", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, @@ -2658,207 +2586,207 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Most pairs are clearly distinct bug reports. The tightest related cluster is the resize_token_embeddings/output_embeddings family: 45276 and 45292 look like the same core bug, while 45335 is very closely related but not shown in the soft-edge list. I\u2019d keep the cluster boundary conservative rather than treating 45292 as definitively subsuming all Gemma4/Gemma/T5Gemma variants from the titles alone.", - "confidence": 0.71, - "canonical_issue_reason": "issue:45292 is the cleanest representative of the resize_token_embeddings/output_embeddings bug family because its title is the most generic and least model-specific, but it should be treated as a representative of a closely related cluster rather than proven umbrella canonical for all variants.", + "summary": "Most items are unrelated bug reports. The only clear duplicate cluster is around `resize_token_embeddings` not updating model-specific embeddings/output heads, with `issue:45292` as the broadest report. `issue:45276` and `issue:45335` look like model-specific variants of that same bug.", + "confidence": 0.79, + "canonical_issue_reason": "`issue:45292` is the cleanest generic statement of the `resize_token_embeddings` bug and best represents the duplicate cluster; the other matched reports are narrower model-specific variants.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45292 is the best single representative issue for the resize_token_embeddings/output_embeddings cluster: concise title, generic symptom wording, and strong overlap with the Gemma4-specific follow-up.", + "best_issue_reason": "`issue:45292` is the best global issue candidate because it is concise, model-agnostic, and captures the shared underlying failure more generally than the other resize-token reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44402", "right": "issue:44488", "accept": false, - "reason": "Different symptoms and likely different model/loading failures; only broad tokenizer/model-loading overlap." + "reason": "Different failure modes: one is tokenizer vocab/indexing, the other is a specific model load regression." }, { "left": "issue:39692", "right": "issue:44960", "accept": false, - "reason": "Unrelated topics: SigLIP2 doc example errors vs a GLM5 issue." + "reason": "Unrelated topics: docs/example errors versus a GLM5 model issue." }, { "left": "issue:41720", "right": "issue:44155", "accept": false, - "reason": "Different code paths and failures: Qwen3 auto device mapping CUDA assert vs AudioFlamingo3 batched inference token/embedding leak." + "reason": "Different code paths and symptoms: device-map CUDA assert versus batched inference token leakage." }, { "left": "issue:43425", "right": "issue:43957", "accept": false, - "reason": "Both involve compatibility/loading problems, but the concrete bugs are different (torch version incompatibility vs meta-device loading breakage)." + "reason": "Both involve compatibility/loading, but they are distinct bugs with different triggers and failure points." }, { "left": "issue:45663", "right": "issue:45759", "accept": false, - "reason": "Distinct model families and failures: Gemma-4 FSDP2 shared_kv_states bug vs Qwen multimodal config unwrapping." + "reason": "Different concrete problems: FSDP/shared-kv state bug versus config unwrapping for multimodal Qwen." }, { "left": "issue:42947", "right": "issue:45704", "accept": false, - "reason": "Gradient checkpointing/LoRA issue is unrelated to the Apex RMSNorm memory leak in T5." + "reason": "Training checkpointing/LoRA behavior is unrelated to the T5 RMSNorm memory leak." }, { "left": "issue:43404", "right": "issue:43454", "accept": false, - "reason": "Both involve lm_head tying, but they are different model implementations and do not look like the same concrete bug." + "reason": "Same general theme of untied lm_head weights, but different model implementations and likely different fixes." }, { "left": "issue:44291", "right": "issue:44756", "accept": false, - "reason": "Different problems: init_empty_weights argument handling vs mmap/OOM behavior on Strix Halo." + "reason": "Different problems: init_empty_weights argument handling versus Strix Halo mmap/OOM behavior." }, { "left": "issue:43873", "right": "issue:44387", "accept": false, - "reason": "Related to quantization, but one is offloading behavior and the other is reserved-memory/OOM regression; not the same bug." + "reason": "Both mention quantization/memory, but one is offloading behavior and the other is int4 reserved-memory growth." }, { "left": "issue:44079", "right": "issue:46829", "accept": false, - "reason": "Generic ModelOutput key assignment bug vs a specific model output field mix-up." + "reason": "Generic ModelOutput key assignment bug versus a specific model output field mismatch." }, { "left": "issue:45663", "right": "issue:45859", "accept": false, - "reason": "Both involve Gemma/Qwen model internals, but the failures are unrelated (FSDP2 KeyError vs CPU-offload incompatibility)." + "reason": "Different model-specific offload issues with different failing components." }, { "left": "issue:44292", "right": "issue:44912", "accept": false, - "reason": "Both are quantization-loading issues, but with different formats/models and different failure modes." + "reason": "Different quantization formats and failure modes; same broad area only." }, { "left": "issue:36010", "right": "issue:39404", "accept": false, - "reason": "Import path breakage for GenerationMixin is unrelated to Whisper pipeline return_language behavior." + "reason": "Import path regression and Whisper pipeline behavior are unrelated." }, { - "left": "issue:44387", - "right": "issue:44756", + "left": "issue:42947", + "right": "issue:43381", "accept": false, - "reason": "OOM symptoms differ: int4 quantization reserved memory regression vs mmap on Strix Halo." + "reason": "Both concern gradient checkpointing, but one is ineffective LoRA training and the other is eval-mode incompatibility." }, { - "left": "issue:42947", - "right": "issue:43381", + "left": "issue:44387", + "right": "issue:44756", "accept": false, - "reason": "Same broad training area, but one is ineffective gradient checkpointing with LoRA and the other is an eval-mode restriction." + "reason": "Both end in OOM, but the causes are unrelated." }, { "left": "issue:44155", "right": "issue:45823", "accept": false, - "reason": "Different models and failure mechanisms: embedding/token leak vs device mismatch in Gemma4 PLE." + "reason": "A batched inference leak is not the same bug as a device-mapping mismatch." }, { "left": "issue:43957", "right": "issue:44756", "accept": false, - "reason": "Meta-device model loading breakage is not the same as the Strix Halo mmap OOM issue." + "reason": "Different loading/memory issues with different root causes." }, { "left": "issue:43335", "right": "issue:45663", "accept": false, - "reason": "Completely different model components: SwitchTransformers sparse-layer config bug vs Gemma-4 attention/FSDP issue." + "reason": "Sparse-layer config bug is unrelated to Gemma4 FSDP attention state handling." }, { "left": "issue:44863", "right": "issue:46468", "accept": false, - "reason": "Both mention Nemotron, but one is checkpoint loading and the other is an example breakage; not enough to be the same bug." + "reason": "Checkpoint-loading failure and broken example code are related only by model family, not by the same bug." }, { "left": "issue:45276", "right": "issue:45292", "accept": true, - "reason": "Same underlying resize_token_embeddings/output_embeddings bug; 45276 reads as a Gemma4-specific manifestation of the same failure." + "reason": "Same underlying bug: `resize_token_embeddings` does not propagate correctly to the model's output embeddings; `45276` is just a model-specific report." }, { "left": "issue:44479", "right": "issue:44877", "accept": false, - "reason": "Different regression surfaces: video-input behavior change vs strict config loading for granite_speech." + "reason": "Video-input regression and strict config loading are different issues." }, { "left": "issue:43381", "right": "issue:44928", "accept": false, - "reason": "Both touch training stability, but one is about checkpointing in eval mode and the other is RLHF NaN explosion from 3D position_ids/SDPA fallback." + "reason": "Gradient checkpointing eval-mode behavior is unrelated to the SDPA/BF16 instability." }, { "left": "issue:44164", "right": "issue:44756", "accept": false, - "reason": "Save/from_pretrained extra_state handling is unrelated to mmap/OOM on Strix Halo." + "reason": "Serialization `extra_state` handling is unrelated to mmap/OOM on Strix Halo." }, { "left": "issue:42796", "right": "issue:46326", "accept": false, - "reason": "Tokenizer BPE marker handling and continue_final_message are different tokenizer bugs." + "reason": "Both tokenizer-related, but one is BPE marker handling and the other is `continue_final_message` behavior." }, { "left": "issue:45588", "right": "issue:45663", "accept": false, - "reason": "Flash attention sinkless-model crash is unrelated to Gemma-4 FSDP2 shared_kv_states rebuilding." + "reason": "Flash-attention sink handling is a different code path from Gemma4 FSDP attention state rebuilding." }, { "left": "issue:45684", "right": "issue:45698", "accept": false, - "reason": "Both involve save/pretrained for custom models, but one is file permissions and the other is wrong module resolution after saving." + "reason": "Both involve custom-model save/load, but one is permissions propagation and the other is wrong module resolution." }, { "left": "issue:43854", "right": "issue:45216", "accept": false, - "reason": "Different Qwen/GLM loading regressions with different causes and scopes." + "reason": "Different models and different regressions: unit-test loading versus save_pretrained checkpoint correctness." }, { "left": "issue:45292", "right": "issue:45335", - "accept": false, - "reason": "Very closely related resize_token_embeddings reports, but they mention different affected submodules and different model families; not clearly the same concrete bug." + "accept": true, + "reason": "Same underlying `resize_token_embeddings` bug; `45335` is a model-specific manifestation on `decoder.embed_tokens`." }, { "left": "issue:43645", "right": "issue:44164", "accept": false, - "reason": "Jupyter/custom model initialization breakage is distinct from extra_state save/load handling." + "reason": "Custom-model notebook initialization and `extra_state` save/load are separate failures." }, { "left": "issue:43828", "right": "issue:43957", "accept": false, - "reason": "Autocast dtype mismatch in Phi-tiny-MoE is unrelated to meta-device loading regressions." + "reason": "Autocast dtype mismatch and meta-device loading are different runtime paths." }, { "left": "issue:43646", "right": "issue:43708", "accept": false, - "reason": "Both concern trainer/model initialization changes, but one is custom model init breakage and the other is checkpoint-resume max_steps miscalculation." + "reason": "Custom model initialization regression and checkpoint-resume step counting are unrelated." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The main duplicate call is plausible, but the writeup is a bit too strong for the packet. In particular, the claim that issue:45292 is the broad, Gemma4-subsuming canonical issue is not directly supported by the titles alone, and the summary\u2019s \u2018only clear duplicate cluster\u2019 wording underplays the also-high-similarity 45292\u219445335 pair. I\u2019d keep the verdicts conservative and say these issues are closely related rather than definitively the same cluster." + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The resize_token_embeddings cluster is well-supported by the titles: issue:45292 is the most generic report, and issue:45276 / issue:45335 are clearly model-specific variants of the same failure mode. The other soft-edge rejections are phrased cautiously enough based on the available title-level evidence." }, "evaluator_used": true, - "retried": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -2960,14 +2888,14 @@ "issue:44164|issue:46821", "issue:45479|issue:46897" ], - "prepared_review_unit_hash": "2fd666727160819109d5df5a0345bbe8fea5297f12ba217f9114e86664f0a483", + "prepared_review_unit_hash": "de115e4047f76b5e011da26ff98aadaaad256a450ba4ea9549822766bfa74a66", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, @@ -2982,220 +2910,217 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: tokenizer regressions, model-loading breakages, multimodal dtype issues, docs problems, and unrelated runtime errors. I found one plausible duplicate pair around the training-loss double-shift bug; the rest are not the same underlying change/bug.", - "confidence": 0.9, - "canonical_issue_reason": "No single cluster-wide canonical issue: the items span many distinct regressions. The closest true duplicate theme is the training-loss double-shift reports.", + "summary": "Loose similarity cluster of mostly distinct Transformers v5-era regression reports spanning tokenization, model loading, multimodal dtype/device issues, and a few performance bugs. None of the soft pairs look like the same concrete bug/change path, so I reject all soft edges.", + "confidence": 0.83, + "canonical_issue_reason": "Issue 43425 is the broadest, most central compatibility report in the set and is the best single anchor for this loose cluster, even though it is not a true duplicate of the others.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 46901 is the clearest standalone report of the shared double-shift training-loss bug, but it only represents that narrow subtheme, not the whole cluster.", + "best_issue_reason": "Issue 43425 is the most general and triage-friendly issue here: concise, high-level, and closer to an umbrella regression than the more model-specific reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44402", "right": "issue:44451", "accept": false, - "reason": "Both are tokenizer-loading complaints, but one is a vocab-size/token-ID mismatch and the other is a model-specific load failure; too little evidence of the same root cause." + "reason": "Both are tokenizer-loading complaints, but they point to different failure modes and different models; not enough evidence of one underlying bug." }, { "left": "issue:46894", "right": "issue:46901", - "accept": true, - "reason": "Same underlying training-loss double-shift bug: both describe labels being shifted incorrectly before loss computation, and one PR/bugfix could plausibly cover both model families." + "accept": false, + "reason": "Same broad \"double-shift\" wording, but different model implementations; likely separate bugs rather than one shared code-path fix." }, { "left": "issue:44589", "right": "issue:44756", "accept": false, - "reason": "Different problems: missing Float8 storage vs mmap/OOM behavior on Strix Halo." + "reason": "A missing Float8 storage error and an mmap/OOM issue are unrelated failure modes." }, { "left": "issue:43425", "right": "issue:43828", "accept": false, - "reason": "Generic Torch version incompatibility vs an autocast dtype mismatch in a specific model." + "reason": "Torch version incompatibility is too broad to equate with a specific autocast dtype mismatch." }, { "left": "issue:46315", "right": "issue:46491", "accept": false, - "reason": "Performance regression in convert_tokens_to_ids vs a CodeLlama encode/decode spacing regression." + "reason": "One is a performance regression in convert_tokens_to_ids; the other is an encode/decode whitespace regression." }, { "left": "issue:45812", "right": "issue:46315", "accept": false, - "reason": "Silent Granite token ID regression vs tokenizer slowdown; different symptoms and code paths." + "reason": "Both involve tokenizers, but one is wrong token IDs and the other is conversion slowness; different bugs." }, { "left": "issue:39404", "right": "issue:45701", "accept": false, - "reason": "Whisper pipeline return_language regression vs broad tokenization changes; not the same bug." + "reason": "Whisper pipeline return_language and a general tokenization regression are not the same issue." }, { "left": "issue:44530", "right": "issue:45084", "accept": false, - "reason": "PagedAttentionCache linear_attention crash vs a template-node compilation error; unrelated failures." + "reason": "Different subsystems and different exceptions; no common concrete code path is evident." }, { "left": "issue:43611", "right": "issue:43645", "accept": false, - "reason": "Base_model_prefix loading failure vs custom-model initialization in notebooks; both are v5 breakages but not the same concrete bug." + "reason": "Both are v5 model-loading regressions, but one is about base_model_prefix handling and the other about notebook-defined custom models." }, { "left": "issue:39404", "right": "issue:46665", "accept": false, - "reason": "Pipeline language output regression vs from_pretrained behavior inside classes; different code paths." + "reason": "One is a Whisper pipeline regression; the other is from_pretrained behavior inside Python classes." }, { "left": "issue:43425", "right": "issue:46869", "accept": false, - "reason": "Torch compatibility issue vs YOLO multi-GPU device-map failure; no shared root cause." + "reason": "Torch compatibility and multi-GPU device-map behavior are distinct problems." }, { "left": "issue:44683", "right": "issue:45823", "accept": false, - "reason": "Compiled flex_attention torch-version failure vs Gemma4 device mismatch during forward." + "reason": "Compiled flex_attention on newer torch and Gemma4 device mismatch are unrelated." }, { "left": "issue:43425", "right": "issue:43716", "accept": false, - "reason": "Torch version incompatibility vs image-preprocessor/model dtype mismatch." + "reason": "A generic torch incompatibility report does not match a specific image-preprocessor/model dtype mismatch." }, { "left": "issue:43425", "right": "issue:45823", "accept": false, - "reason": "Broad Torch incompatibility vs a Gemma4 device-map bug; not the same bug." + "reason": "These are different classes of failures: version incompatibility vs device placement mismatch." }, { "left": "issue:44479", "right": "issue:44805", "accept": false, - "reason": "Qwen video input regression vs a mask-shape IndexError; symptoms and likely fixes differ." + "reason": "A video-input regression and a mask-shape IndexError are too generic to treat as one bug." }, { "left": "issue:44589", "right": "issue:45084", "accept": false, - "reason": "Storage-object lookup failure vs tokenizer template compilation error; unrelated." + "reason": "Storage lookup failure and template compilation failure are clearly different." }, { "left": "issue:43901", "right": "issue:44509", "accept": false, - "reason": "Docs about return_all_scores vs docs for removed pipeline tasks; both docs issues but not duplicates." + "reason": "Both are docs-related, but they cover different removed/changed pipeline APIs." }, { "left": "issue:43906", "right": "issue:44530", "accept": false, - "reason": "Isolated reproduction of one issue vs Qwen3.5 paged-attention crash; no concrete overlap." + "reason": "An isolated reproduction of another issue and a Qwen3.5 cache crash do not share a concrete code path." }, { "left": "issue:43994", "right": "issue:44079", "accept": false, - "reason": "Wrong SigLIP2 outputs vs ModelOutput key assignment bug; distinct behaviors." + "reason": "A wrong-output multimodal model issue and a ModelOutput key-assignment bug are unrelated." }, { "left": "issue:44265", "right": "issue:44898", "accept": false, - "reason": "torch.export failure with torch_compilable_check vs Perceiver interpolation failure; different model/export paths." + "reason": "torch.export failure due to torch_compilable_check is not the same as Perceiver interpolation failure." }, { "left": "issue:46050", "right": "issue:46326", "accept": false, - "reason": "Quantized save/load breakage vs continue_final_message behavior in mistral_common tokenizers." + "reason": "Quantized save/load breakage and mistral_common continue_final_message behavior are different features." }, { "left": "issue:39404", "right": "issue:44912", "accept": false, - "reason": "Whisper pipeline regression vs MXFP4 quantization fallback; unrelated." + "reason": "Whisper pipeline behavior and MXFP4 quantization fallback are unrelated." }, { "left": "issue:36032", "right": "issue:43452", "accept": false, - "reason": "Tokenizer method-name conflict on load vs gguf_file handling in AutoTokenizer/AutoModel loading." + "reason": "Both are loading errors, but one is a T5 tokenizer method conflict and the other is gguf_file handling." }, { "left": "issue:44164", "right": "issue:45759", "accept": false, - "reason": "extra_state save/load bug vs composite-config unwrapping bug; different layers of the stack." + "reason": "Extra state persistence and composite multimodal config unwrapping are separate save/load issues." }, { "left": "issue:43906", "right": "issue:46326", "accept": false, - "reason": "Reproduction of one issue vs mistral_common continue_final_message bug; not the same code path." + "reason": "These point to different tokenizer-related regressions with no shared concrete failure." }, { "left": "issue:44671", "right": "issue:44991", "accept": false, - "reason": "CamemBERT MLM regression vs tokenizer loading failure for est-roberta; unrelated." + "reason": "Different model families and different symptoms; not the same bug." }, { "left": "issue:45823", "right": "issue:46899", "accept": false, - "reason": "Device mismatch in forward vs unsafe dtype cast in multimodal embedders; both are multimodal but not the same bug." + "reason": "Both are Gemma4 multimodal issues, but one is device placement and the other is a 4-bit dtype-cast problem." }, { "left": "issue:44991", "right": "issue:45081", "accept": false, - "reason": "Two tokenizer loading failures, but one is an est-roberta load regression and the other is a mistral regex patch crash; not the same underlying issue." + "reason": "Two tokenizer-loading regressions with distinct causes and model-specific patches." }, { "left": "issue:45479", "right": "issue:46894", "accept": false, - "reason": "Zero-loss classification bug vs a double-shift ASR training-loss bug; different loss-path issues." + "reason": "A generic classification-loss issue and an ASR double-shift issue are not the same underlying bug." }, { "left": "issue:43299", "right": "issue:43931", "accept": false, - "reason": "Related Qwen3-VL family loading problems, but the symptoms differ enough that they do not clearly share one concrete fix." + "reason": "Both mention Qwen3-VL loading, but one is a dev0 breakage and the other is a weight-shape mismatch; too different to merge." }, { "left": "issue:43856", "right": "issue:44756", "accept": false, - "reason": "MoE training memory usage vs mmap/OOM behavior on Strix Halo; different problems." + "reason": "Memory-usage inefficiency during training and an mmap OOM workaround are unrelated." }, { "left": "issue:44164", "right": "issue:46821", "accept": false, - "reason": "extra_state save/from_pretrained handling vs a TimesFM window_size AttributeError; unrelated." + "reason": "save/from_pretrained extra_state handling and TimesFM window_size AttributeError are different bugs." }, { "left": "issue:45479", "right": "issue:46897", "accept": false, - "reason": "Degenerate classification loss vs the same double-shift training-loss pattern; close in theme, but not the same bug." + "reason": "Different model families and different loss bugs; no shared code path is evident." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The summary is well grounded in the packet and stays conservative. It correctly treats the cluster as mostly heterogeneous and only accepts the one clearly supported duplicate theme around the training-loss double-shift reports (46894/46901). The other pairwise rejections are cautious and consistent with the issue titles and descriptions." - }, - "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 }, @@ -3294,52 +3219,52 @@ "issue:43883|issue:44488", "issue:43746|issue:46326" ], - "prepared_review_unit_hash": "aeed7451be657528ca2d61496e42ccea68b06dced66c328f87d069eb119463ad", + "prepared_review_unit_hash": "a0d57b71484228aa1a87f8ce6edce04d85703d6920ff7a2e403276394af5c2ba", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37012, - "estimated_input_tokens": 9253, - "estimated_eval_tokens": 18762 + "serialized_chars": 37019, + "estimated_input_tokens": 9255, + "estimated_eval_tokens": 18766 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of related-but-distinct tokenizer/model-loading regression reports, many centered on Transformers v5 and GGUF/custom-tokenizer paths. I do not see enough evidence for confirmed duplicates here; the safest read is same-area complaints, not merged bugs.", - "confidence": 0.56, - "canonical_issue_reason": "Issue 44462 is the clearest representative of the tokenizer-loading regression slice: it is concrete, user-facing, and broad enough to summarize the nearby loading complaints without claiming it is their root cause.", + "summary": "This cluster is mostly a loose set of Transformers v5 regression reports rather than one true duplicate family. Only two soft pairs look like strong duplicates: GGUF loading failures and the `_torch_extract_fbank_features()` audio regression.", + "confidence": 0.61, + "canonical_issue_reason": "Issue 43452 is the best canonical anchor: it describes the broadest concrete loading regression in the set, spanning both AutoTokenizer and AutoModelForCausalLM via `gguf_file`, and it naturally covers the more specific Qwen2.5-GGUF failure.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44462 is the best single issue to anchor the cluster because it describes a specific loading failure with broad impact, while staying grounded in an observable code path.", + "best_issue_reason": "Issue 43452 is the strongest global representative because it has the clearest, broad user-facing code-path break and the most general title among the candidates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43596", "right": "issue:44683", "accept": false, - "reason": "Different failures: DeepSpeed/Zero3 BertModel indexing vs torch>=2.9 flex_attention compilation." + "reason": "Different failures: DeepSpeed/zero3 BertModel indexing vs torch>=2.9 flex_attention compilation." }, { "left": "issue:44488", "right": "issue:44987", "accept": false, - "reason": "Both are loading regressions, but they involve different models and no shared concrete bug is shown." + "reason": "Both are tokenizer load regressions, but on different models and likely different root causes." }, { "left": "issue:44488", "right": "issue:45081", "accept": false, - "reason": "Specific model load failure vs Mistral regex/tokenizer backend crash; different code paths." + "reason": "Different tokenizer bugs: repo loading failure vs Mistral regex/backend tokenizer crash." }, { "left": "issue:43611", @@ -3351,173 +3276,191 @@ "left": "issue:42796", "right": "issue:43404", "accept": false, - "reason": "Tokenizer BPE-marker handling and lm_head weight tying are separate model-specific defects." + "reason": "BPE-marker tokenizer bug vs tied lm_head weight bug." }, { "left": "issue:44462", "right": "issue:46489", "accept": false, - "reason": "Related tokenizer-loading area, but 46489 is a specific DeepSeek-Coder regression, not clearly the same bug." + "reason": "Different tokenizer-loading regressions: ignoring tokenizer.json vs a DeepSeek-Coder v5 output regression." }, { "left": "issue:36032", "right": "issue:42915", "accept": false, - "reason": "T5 tokenizer method conflict and Qwen3Moe FP8 failure are unrelated." + "reason": "Unrelated: T5 tokenizer method conflict vs Qwen3Moe FineGrainedFP8Config failure." }, { "left": "issue:46285", "right": "issue:46402", "accept": false, - "reason": "CLIP structure inconsistency and Sam3 text-encoder weight loading are different issues." + "reason": "Different model families and symptoms: CLIP structure mismatch vs Sam3 text-encoder weights not loading." }, { "left": "issue:43716", "right": "issue:45084", "accept": false, - "reason": "Image preprocessor dtype mismatch vs template compilation error are unrelated." + "reason": "Image dtype mismatch vs template compilation TypeError; different code paths." }, { "left": "issue:39404", "right": "issue:44671", "accept": false, - "reason": "Whisper pipeline return_language regression and CamemBERT masked-LM regression are different bugs." + "reason": "Whisper return_language pipeline bug vs CamemBERT masked-LM prediction regression." }, { "left": "issue:44737", "right": "issue:45859", "accept": false, - "reason": "Both touch device/offload behavior, but the concrete failures and code paths differ." + "reason": "Both involve device placement, but the concrete bugs are different: XLNet CPU arange vs Qwen3.5 CPU-offload incompatibility." }, { "left": "issue:43643", "right": "issue:44530", "accept": false, - "reason": "AutoConfig trust_remote_code field loss is unrelated to the PagedAttentionCache group-type crash." + "reason": "trust_remote_code config-field loss vs Qwen3.5 PagedAttentionCache linear_attention crash." }, { "left": "issue:44661", "right": "issue:44683", "accept": false, - "reason": "Model-mapping registration failure and flex_attention compilation failure are unrelated." + "reason": "TOKENIZER_MAPPING_NAMES registration bug vs flex_attention compilation regression." }, { "left": "issue:42915", "right": "issue:44530", "accept": false, - "reason": "Qwen3Moe FP8 config and Qwen3.5 cache/group-type crashes are separate defects." + "reason": "Different Qwen3-family failures and different subsystems (FP8 config vs paged-attention cache)." }, { "left": "issue:40990", "right": "issue:43421", "accept": false, - "reason": "Perplexity regression on a model benchmark is unrelated to runtime special-token post-processor updates." + "reason": "Perplexity regression on a model vs tokenizer post-processor update behavior." }, { "left": "issue:44451", "right": "issue:44987", "accept": false, - "reason": "Different model-specific loading failures with no shared concrete failure mode." + "reason": "Both are model-loading failures, but on different repositories and with different failure modes." }, { "left": "issue:43541", "right": "issue:45084", "accept": false, - "reason": "Grouped_mm dynamo tracing and non-template-node compilation are distinct compiler/runtime issues." + "reason": "Different compile/tracing problems: grouped_mm dtype error vs non-template-node compilation failure." }, { "left": "issue:34567", "right": "issue:40444", "accept": false, - "reason": "Trainer token accounting and multimodal iterable dataset loading are unrelated." + "reason": "Trainer token accounting bug vs multimodal finetuning failure with IterableDataset." }, { "left": "issue:43644", "right": "issue:43646", "accept": false, - "reason": "Both are v5 initialization regressions, but they describe different symptoms and likely different fixes." + "reason": "Related v5 initialization regressions, but one is a specific buffer-state issue and the other is a broad custom-model init break." }, { "left": "issue:43582", "right": "issue:43632", "accept": false, - "reason": "Apple Silicon warmup TypeError and _is_hf_initialized breakage are unrelated." + "reason": "Apple Silicon allocator warmup TypeError vs `_is_hf_initialized` flag regression." }, { "left": "issue:40444", "right": "issue:46326", "accept": false, - "reason": "Multimodal iterable dataset failure and mistral_common continue_final_message behavior are different bugs." + "reason": "Different domains: multimodal iterable finetuning vs Whisper/mistral_common tokenizer handling." }, { "left": "issue:42175", "right": "issue:46178", "accept": false, - "reason": "Packaging/backend install issue vs missing BACKENDS_MAPPING entry are not the same defect." + "reason": "Backend-install/mapping problems, but one is missing TF backend in extras and the other is a bad BACKENDS_MAPPING entry." }, { "left": "issue:44206", "right": "issue:44220", + "accept": true, + "reason": "Same audio feature-extraction regression: both point to `_torch_extract_fbank_features()` / the same fbank pipeline bug." + }, + { + "left": "issue:44462", + "right": "issue:44987", "accept": false, - "reason": "Both are audio-feature related, but the concrete failures differ and no shared root cause is evident." + "reason": "Different tokenizer-loading regressions for different models; no shared concrete bug is shown." }, { - "left": "issue:36032", - "right": "issue:43645", + "left": "issue:43541", + "right": "issue:45084", "accept": false, - "reason": "T5 tokenizer method conflict and custom-model initialization in notebooks are unrelated." + "reason": "Different compile/tracing problems with unrelated underlying causes." + }, + { + "left": "issue:34567", + "right": "issue:40444", + "accept": false, + "reason": "Duplicate of another unrelated pair; these are different bugs." + }, + { + "left": "issue:43644", + "right": "issue:43646", + "accept": false, + "reason": "Duplicate of another unrelated pair; the failure modes differ." }, { "left": "issue:44811", "right": "issue:45584", "accept": false, - "reason": "Whisper batch_decode skip_special_tokens handling and empty-transcription generation are distinct issues." + "reason": "Both are Whisper special-token issues, but one is decode-time skip_special_tokens handling and the other is empty-transcription generation after align_special_tokens." }, { "left": "issue:43452", "right": "issue:43482", - "accept": false, - "reason": "They are both GGUF/loading regressions, but the evidence is too thin to call them the same bug." + "accept": true, + "reason": "Same GGUF load path; 43482 looks like the model-specific manifestation of the broader `gguf_file`/from_pretrained break in 43452." }, { "left": "issue:44987", "right": "issue:44991", "accept": false, - "reason": "Different model-loading regressions on different models; same area, not the same concrete failure." + "reason": "Both are loading failures after v5, but they affect different models and likely different tokenizer setup bugs." }, { "left": "issue:45084", "right": "issue:45507", "accept": false, - "reason": "Template compilation error and GraniteMoEHybrid invalid-method call are unrelated." + "reason": "Template compilation failure vs invalid method call on GraniteMoEHybrid." }, { "left": "issue:43317", "right": "issue:43482", "accept": false, - "reason": "CPU/GPU offload dequantization and Qwen2.5-GGUF loading failures follow different paths." + "reason": "Different bugs: device_map/offload dequantization vs GGUF loading." }, { "left": "issue:43883", "right": "issue:44488", "accept": false, - "reason": "Missing all_tied_weights_keys and failure to load a specific BERT model are unrelated." + "reason": "Missing `all_tied_weights_keys` vs tokenizer loading failure for a different model." }, { "left": "issue:43746", "right": "issue:46326", "accept": false, - "reason": "PEFT/local checkpoint loading and mistral_common continue_final_message behavior are unrelated." + "reason": "PEFT/local-checkpoint loading issue vs Whisper tokenizer `continue_final_message` bug." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The cluster summary is mostly cautious, but one soft-edge verdict is too strong for the packet: treating issue:43482 as a concrete manifestation of issue:43452 overstates the evidence. From the titles alone they are both gguf/loading regressions, but there is not enough grounded evidence here to call them the same bug or a mergeable single PR. The canonical issue wording also leans speculative by suggesting 44462 is plausibly the root cause of nearby complaints. A more conservative summary would keep these as same-area, not confirmed duplicates." + "accept": true, + "feedback": "The summary is grounded in the packet and stays mostly conservative. The two accepted soft pairs are plausibly supported by the issue titles, especially 44206/44220 and the GGUF loading pair 43452/43482. The canonical anchor choice for 43452 is also reasonable from the title alone. " }, "evaluator_used": true, - "retried": true, - "accepted_nontrivial_soft_edge": false, + "retried": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -3599,10 +3542,10 @@ "issue:43994|issue:44610", "issue:44987|issue:45081", "issue:42915|issue:44291", - "issue:43883|issue:44991", "issue:44008|issue:46899", "issue:36032|issue:43742", "issue:44451|issue:45081", + "issue:43883|issue:44991", "issue:44265|issue:44610", "issue:44062|issue:44987", "issue:43632|issue:46821", @@ -3614,20 +3557,20 @@ "issue:43531|issue:44877", "issue:45216|issue:45357" ], - "prepared_review_unit_hash": "734bc8016800e14c41452c403fc2e914d0f7d7631912351b86be7a26c9a4b40e", + "prepared_review_unit_hash": "5f4a048d24c49d52c62753ba0fc1a36ddd424d1ff5b007ef1421c063b5488cb1", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36634, + "serialized_chars": 36635, "estimated_input_tokens": 9159, "estimated_eval_tokens": 18574 }, @@ -3636,192 +3579,192 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is almost entirely heterogeneous issue reports. The only plausible duplicate pair is the Qwen3.5 save_pretrained checkpoint serialization bug (45216/45357); the rest are different models, subsystems, or failure modes.", - "confidence": 0.89, - "canonical_issue_reason": "issue:45216 is the better canonical issue: it is the broader, earlier report about Qwen3.5 save_pretrained producing an incorrect checkpoint, and issue:45357 looks like a narrower follow-up about the same serialization path.", + "summary": "Most candidate pairs are related only by broad subsystem or model family, not the same underlying bug. I would accept only the two Qwen3.5 save_pretrained issues as duplicates; the rest are distinct enough to keep separate.", + "confidence": 0.88, + "canonical_issue_reason": "issue:45216 is the broader, earlier regression report about Qwen3.5 save_pretrained producing an incorrect checkpoint; issue:45357 reads like a narrower follow-up symptom focused on the visual encoder keys.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45216 is the strongest representative because it captures the underlying Qwen3.5 save_pretrained regression at a higher level than the more specific visual-encoder-key follow-up.", + "best_issue_reason": "issue:45216 is the best representative issue because it captures the general Qwen3.5 save_pretrained regression without narrowing to one specific malformed key set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43262", "right": "issue:43377", "accept": false, - "reason": "Both are audio-related, but one is a chat-template sampling-rate default and the other is batched-vs-single encoder padding-mask behavior; different bugs." + "reason": "Both are audio-related, but one is a processor chat-template sampling-rate default and the other is MIMI batching/padding-mask behavior; different bug and code path." }, { "left": "issue:43208", "right": "issue:47013", "accept": false, - "reason": "Both are xLSTM training bugs, but the concrete failures differ: multiple small-model blockers vs a return_last_states unpacking crash." + "reason": "Shared xLSTM label only; one is a bundle of training blockers for small models, the other is a specific train-mode unpacking crash when return_last_states=False." }, { "left": "issue:43646", "right": "issue:44534", "accept": false, - "reason": "Both involve Transformers v5 regressions, but one is custom model init in Jupyter and the other is non-persistent buffer serialization; not the same bug." + "reason": "One is custom model initialization breaking in v5, the other is non-persistent buffers being filled with junk; different failure modes." }, { "left": "issue:43645", "right": "issue:46661", "accept": false, - "reason": "One is notebook custom-model initialization; the other is remote-code AutoModel loading missing config_class. Related area, different code paths." + "reason": "Both involve model initialization/loading, but one is Jupyter/custom model init and the other is remote-code AutoModel config_class handling." }, { "left": "issue:44403", "right": "issue:45081", "accept": false, - "reason": "Both concern model loading, but one is noisy loading behavior and the other is a Mistral regex patch crash; different failures." + "reason": "Unnecessary loading noise and a Mistral regex patch crash are not the same bug." }, { "left": "issue:43792", "right": "issue:46869", "accept": false, - "reason": "Different models and symptoms: Whisper runtime failure vs YOLO multi-GPU auto device-map failure." + "reason": "Different models and failures: Whisper runtime failure vs YOLO multi-GPU device-map issue." }, { "left": "issue:42915", "right": "issue:45084", "accept": false, - "reason": "FineGrainedFP8Config failure in Qwen3MoE is unrelated to the template-node compilation TypeError." + "reason": "Qwen3-MoE auto device mapping CUDA assert is unrelated to a templating compilation TypeError." }, { "left": "issue:44849", "right": "issue:45325", "accept": false, - "reason": "Different Qwen variants and different bugs: hidden-states output vs RoPE temporal \u0564\u056b\u0580/position scaling." + "reason": "Both are Qwen-family regressions, but one is output_hidden_states behavior and the other is still-image rope/position-id scaling; not the same bug." }, { "left": "issue:43377", "right": "issue:43819", "accept": false, - "reason": "Both are audio encoder correctness issues, but MIMI padding-mask mismatch and DAC missing-STE mismatch are separate code paths." + "reason": "Both are audio model issues, but one is missing padding-mask support in MIMI and the other is DAC.from_latents vs forward mismatch without STE." }, { "left": "issue:43404", "right": "issue:44466", "accept": false, - "reason": "Both mention lm_head/tied weights, but one is a model-specific untied-weight bug and the other is device-dependent save_pretrained serialization." + "reason": "Both mention lm_head/tied weights, but one is a missing tie in a specific model class and the other is save_pretrained serialization varying by device." }, { "left": "issue:28530", "right": "issue:34567", "accept": false, - "reason": "Unrelated trainer issues: early stopping metric lookup vs num_input_tokens_seen not updating." + "reason": "Training early-stopping metric selection and TrainerState token counter updating are unrelated." }, { "left": "issue:44464", "right": "issue:45987", "accept": false, - "reason": "Different generation/cache bugs: compiled-forward chunked generation inconsistency vs StaticCache seq-length return type." + "reason": "Both touch generation/cache behavior, but one is chunked-generation inconsistency under compiled forward and the other is a return type contract issue for StaticCache.get_seq_length()." }, { "left": "issue:43994", "right": "issue:44610", "accept": false, - "reason": "Different models and failures: SigLIP2 inference corruption vs OmDet-Turbo processor image-size mismatch." + "reason": "Different subsystems: SigLIP2 inference/pipeline correctness vs OmDet-Turbo processor image size mismatch." }, { "left": "issue:44987", "right": "issue:45081", "accept": false, - "reason": "Different tokenizer/load regressions affecting different models and different internal failures." + "reason": "Tokenizer loading for a specific model and a Mistral tokenizer regex crash are separate issues." }, { "left": "issue:42915", "right": "issue:44291", "accept": false, - "reason": "Both are loading/config issues, but one is Qwen3MoE FP8 configuration and the other is unexpected _is_hf_initialized argument handling." - }, - { - "left": "issue:43883", - "right": "issue:44991", - "accept": false, - "reason": "Molmo tied-weights attribute error and EMBEDDIA tokenizer loading failure are unrelated." + "reason": "Both involve model loading/initialization under v5, but the concrete bugs differ: CUDA assert in Qwen3-MoE vs unexpected _is_hf_initialized argument with init_empty_weights." }, { "left": "issue:44008", "right": "issue:46899", "accept": false, - "reason": "Same family name, but different versions/models and different bugs: variable-name collision vs unsafe dtype cast in multimodal embedders." + "reason": "Both are Gemma multimodal issues, but one is a variable-name collision in forward and the other is an unsafe dtype cast under 4-bit quantization." }, { "left": "issue:36032", "right": "issue:43742", "accept": false, - "reason": "Both are tokenizer/model load errors, but the concrete cause and affected model differ." + "reason": "Different tokenizer/model load failures with distinct root causes." }, { "left": "issue:44451", "right": "issue:45081", "accept": false, - "reason": "Different model-loading issues; ScandiBERT load failure is not the same as the Mistral regex backend_tokenizer crash." + "reason": "ScandiBERT loading failure and Mistral regex patch crash are unrelated." + }, + { + "left": "issue:43883", + "right": "issue:44991", + "accept": false, + "reason": "An attribute error on tied-weights metadata and a tokenizer load failure for est-roberta are not the same bug." }, { "left": "issue:44265", "right": "issue:44610", "accept": false, - "reason": "torch.export.export with torch_compilable_check is unrelated to OmDet-Turbo input-size preprocessing." + "reason": "torch.export/torch_compilable_check failure is unrelated to OmDet-Turbo processor sizing." }, { "left": "issue:44062", "right": "issue:44987", "accept": false, - "reason": "Different tokenizer regressions: AddedToken special-arg duplication vs a model load failure in transformers>=5.1.0." + "reason": "Both are tokenizer-related, but one is AddedToken argument duplication and the other is a specific model load failure." }, { "left": "issue:43632", "right": "issue:46821", "accept": false, - "reason": "Transformers v5 _is_hf_initialized breakage and TimesFM window_size AttributeError are separate regressions." + "reason": "_is_hf_initialized flag breakage and TimesFM window_size AttributeError are different regressions." }, { "left": "issue:43322", "right": "issue:43540", "accept": false, - "reason": "Different multimodal model failures: Llava Next segfault vs Qwen3OmniMoe video ValueError." + "reason": "Both are multimodal loading/processing failures, but Llava Next segfault and Qwen3OmniMoe video-processing ValueError are not the same code path." }, { "left": "issue:42915", "right": "issue:44589", "accept": false, - "reason": "Both mention FP8, but one is a model/config regression and the other is a missing Float8 storage registration error." + "reason": "Qwen3-MoE CUDA assert and Float8 storage lookup failure are unrelated." + }, + { + "left": "issue:43089", + "right": "issue:45987", + "accept": false, + "reason": "Generation performance overhead is not the same as StaticCache returning the wrong type." }, { "left": "issue:44479", "right": "issue:45325", "accept": false, - "reason": "Both are Qwen VL-related temporal/video bugs, but one is a broad video-input regression and the other is a still-image position-id scaling issue." + "reason": "These are both Qwen vision-language regressions, but one is a broad video-input regression and the other is still-image temporal position-id scaling; too different to merge confidently." }, { "left": "issue:36032", "right": "issue:41720", "accept": false, - "reason": "Unrelated failures: T5 tokenizer special-token conflict vs Qwen3 auto device-map cudaErrorAssert." + "reason": "Tokenizer construction error and Qwen3 device-map CUDA assert are unrelated." }, { "left": "issue:43531", "right": "issue:44877", "accept": false, - "reason": "Different subsystems: Qwen3-MoE sliding_window behavior vs strict config loading for granite_speech." + "reason": "Sliding-window behavior in Qwen3-MoE and strict config loading for granite_speech are different issues." }, { "left": "issue:45216", "right": "issue:45357", "accept": true, - "reason": "Same underlying Qwen3.5 save_pretrained serialization bug; 45357 is a narrower description of the incorrect checkpoint/output keys reported in 45216." - }, - { - "left": "issue:45987", - "right": "issue:47013", - "accept": false, - "reason": "Different runtime bugs: cache length return type vs xLSTM train-mode unpacking when return_last_states=False." + "reason": "Same underlying Qwen3.5 save_pretrained regression; 45357 is a narrower follow-up describing incorrect visual encoder keys, while 45216 reports the broader broken checkpoint output." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet: it treats the cluster as mostly heterogeneous and only accepts the clearly related Qwen3.5 save_pretrained pair (45216/45357). The soft-edge rejections are conservative and the reasons align with the issue titles, with no obvious overstatement of evidence." + "feedback": "The analysis is grounded in the packet\u2019s titles/body summaries, and the soft-edge calls stay conservative overall. The only accepted duplicate pair, issue:45216 and issue:45357, looks mergeable as the same Qwen3.5 save_pretrained regression. The remaining pairs are plausibly distinct bugs rather than overconfident duplicates." }, "evaluator_used": true, "retried": false, @@ -3915,8 +3858,8 @@ "issue:43632|issue:44534", "issue:43531|issue:44291", "issue:43421|issue:44568", - "issue:43482|issue:45538", "issue:42175|issue:45542", + "issue:43482|issue:45538", "issue:41720|issue:45081", "issue:44079|issue:44534", "issue:43317|issue:43856", @@ -3925,226 +3868,220 @@ "issue:43611|issue:43646", "issue:43377|issue:45491" ], - "prepared_review_unit_hash": "de7314bc4b7b4afc3de0b092796f2bf892232aa907dec5088f7cb2b73736b260", + "prepared_review_unit_hash": "733f342dc9769c5b77eb1c270f5fd5945eef771fffb742cb8f1308d9cbb3a451", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 32, - "serialized_chars": 37298, - "estimated_input_tokens": 9325, - "estimated_eval_tokens": 18906 + "serialized_chars": 37301, + "estimated_input_tokens": 9326, + "estimated_eval_tokens": 18908 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab-bag of unrelated model- and tokenizer-specific regressions. The only weak common thread is Transformers v5 loading/tokenizer breakage, but the soft pairs are generally different concrete bugs, so I would not merge them as duplicates.", - "confidence": 0.83, - "canonical_issue_reason": "issue:43611 is the broadest representative of the cluster\u2019s main theme: a Transformers 5.0 loading regression affecting model instantiation across multiple models.", + "summary": "The set is a collection of largely unrelated issues; the soft pairs are mostly superficial matches by subsystem or version. No PRs are present. If a single issue must be treated as the best representative, the broad Granite tokenizer regression is the strongest candidate.", + "confidence": 0.82, + "canonical_issue_reason": "Issue 45812 is the broadest, most centrally relevant tokenizer regression in the set (all Granite models, silent v4\u2192v5 behavior change) and has the strongest reference signal.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43611 is the best single representative because it describes a generic v5 loading failure rather than a narrowly model-specific edge case.", + "best_issue_reason": "Issue 45812 is the best overall representative because it affects an entire model family and describes a concrete, user-visible regression with cross-cutting impact.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44991", "right": "issue:46315", "accept": false, - "reason": "One is tokenizer loading failure for a specific repo; the other is a performance regression in convert_tokens_to_ids for slow tokenizers. Different bugs." + "reason": "Both involve tokenizer regressions, but one is model-specific loading and the other is a slow conversion path; different bugs and code paths." }, { "left": "issue:43425", "right": "issue:44403", "accept": false, - "reason": "Packaging/backend incompatibility vs. loading noise/logging. Same ecosystem, not the same defect." + "reason": "Torch compatibility and loading noise are unrelated problems." }, { "left": "issue:40444", "right": "issue:45325", "accept": false, - "reason": "Both are Qwen2.5-VL-related, but one is multi-image finetuning failure and the other is still-image rope/position_id scaling. Different code paths." + "reason": "Both are Qwen2.5-VL-related, but one is a multi-image iterable dataset failure and the other is still-image temporal position-id scaling; not the same issue." }, { "left": "issue:43611", "right": "issue:44534", "accept": false, - "reason": "Both are Transformers v5 regressions, but one is model loading with base_model_prefix and the other is non-persistent buffer initialization. Not the same bug." + "reason": "One is a loading failure around base_model_prefix, the other is non-persistent buffers being filled with junk; distinct regressions." }, { "left": "issue:43716", "right": "issue:46899", "accept": false, - "reason": "Both mention multimodal dtype issues, but they affect different models and different failure modes (preprocessor/model mismatch vs unsafe cast in 4-bit embedders)." + "reason": "Both mention multimodal dtype handling, but they affect different models and different components (image preprocessor vs audio/vision embedders)." }, { "left": "issue:43296", "right": "issue:43531", "accept": false, - "reason": "PaddleOCR-VL load failure and Qwen3-MoE sliding_window failure are unrelated model-specific issues." + "reason": "PaddleOCR-VL load failure and Qwen3-MoE sliding_window behavior are unrelated." }, { "left": "issue:44220", "right": "issue:44683", "accept": false, - "reason": "Audio feature extraction failure vs compiled flex_attention failure on newer torch. Different subsystems." + "reason": "Fbank feature extraction and compiled flex_attention on torch>=2.9 are unrelated issues." }, { "left": "issue:45812", "right": "issue:46491", "accept": false, - "reason": "Both are tokenizer regressions, but one corrupts Granite token IDs and the other strips a leading space in CodeLlama round-trips. Different underlying bugs." + "reason": "Both are tokenizer regressions, but one is Granite token IDs and the other is CodeLlama leading-space round-trip behavior; different models and failure modes." }, { "left": "issue:42796", "right": "issue:43746", "accept": false, - "reason": "Tokenizer BPE marker handling vs PEFT adapter checkpoint loading. No overlap in failure mechanism." + "reason": "Tokenizer BPE-marker handling and PEFT checkpoint loading are unrelated." }, { "left": "issue:44403", "right": "issue:44991", "accept": false, - "reason": "Unnecessary logging when loading a transformer is not the same as a tokenizer load regression for est-roberta." + "reason": "Generic loading noise is not the same as AutoTokenizer ignoring tokenizer.json." }, { "left": "issue:43381", "right": "issue:45923", "accept": false, - "reason": "One is eval-mode gradient checkpointing behavior; the other is a missing supports_gradient_checkpointing flag on a trust_remote_code model." + "reason": "Gradient checkpointing in eval mode and a missing supports_gradient_checkpointing flag are related only by theme, not the same bug." }, { "left": "issue:43632", "right": "issue:44756", "accept": false, - "reason": "Transformers v5 initialization flag issue vs Strix Halo mmap OOM. Different symptoms and code paths." + "reason": "_is_hf_initialized regression and Strix Halo mmap OOM are unrelated." }, { "left": "issue:33666", "right": "issue:40444", "accept": false, - "reason": "Multi-GPU training for Qwen2-VL and multi-image finetuning for Qwen2.5-VL are different problems." + "reason": "Both are Qwen-VL training issues, but they concern different setups and failures." }, { "left": "issue:40990", "right": "issue:44811", "accept": false, - "reason": "Perplexity regression on GPT-OSS vs Whisper batch_decode skip_special_tokens handling. Unrelated." + "reason": "Perplexity regression and Whisper batch_decode skip_special_tokens behavior are unrelated." }, { "left": "issue:43792", "right": "issue:44589", "accept": false, - "reason": "Whisper model runtime failure and missing Float8 storage object are different backend/runtime issues." + "reason": "Whisper runtime failure and missing Float8 storage support are unrelated." }, { "left": "issue:44291", "right": "issue:44589", "accept": false, - "reason": "init_empty_weights argument regression vs missing Float8 storage support. Both involve v5-era breakage, but not the same bug." + "reason": "Both are transformers v5 regressions, but they affect different initialization/storage code paths." }, { "left": "issue:42796", "right": "issue:46710", "accept": false, - "reason": "Tokenizer BPE-marker bug and DeepSeek accuracy regression are not the same underlying change." + "reason": "Tokenizer BPE-marker handling and an accuracy regression after upgrading Transformers are not the same bug." }, { "left": "issue:35707", "right": "issue:40444", "accept": false, - "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to multi-image finetuning failure." + "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen2.5-VL multi-image dataset loading." }, { "left": "issue:44079", "right": "issue:45663", "accept": false, - "reason": "ModelOutput key assignment bug vs Gemma4 FSDP2 shared_kv_states KeyError. Different abstractions." + "reason": "ModelOutput key assignment and Gemma4 FSDP2 shared_kv_states rebuilding are unrelated." }, { "left": "issue:43482", "right": "issue:43525", "accept": false, - "reason": "GGUF loading failure for Qwen2.5 vs missing pad_token_id on Llama4Config. Different model/config issues." + "reason": "Qwen2.5-GGUF loading and missing pad_token_id on Llama4Config are different model-specific failures." }, { "left": "issue:43632", "right": "issue:44534", "accept": false, - "reason": "These are both Transformers v5 regressions, but one is _is_hf_initialized and the other is junk-filled non-persistent buffers. Not the same defect." + "reason": "Different v5 regressions: one is initialization flag handling, the other is buffer state corruption." }, { "left": "issue:43531", "right": "issue:44291", "accept": false, - "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to init_empty_weights/_is_hf_initialized argument handling." + "reason": "Sliding-window behavior in Qwen3-MoE and init_empty_weights argument handling are unrelated." }, { "left": "issue:43421", "right": "issue:44568", "accept": false, - "reason": "Runtime special-token post-processor updates and add_special_tokens behavior for mdeberta-v3-base are related only loosely; not the same concrete bug." + "reason": "Runtime post-processor updates for special tokens and add_special_tokens BOS/EOS behavior are different tokenizer paths." }, { - "left": "issue:43482", - "right": "issue:45538", + "left": "issue:42175", + "right": "issue:45542", "accept": false, - "reason": "Qwen2.5-GGUF loading failure and CLIPTokenizer max length behavior are different tokenizer/model issues." + "reason": "Both involve TensorFlow packaging/backend, but one is missing TF in the torch extra while the other is an undefined backend error from tensorboard-only installs." }, { - "left": "issue:42175", - "right": "issue:45542", + "left": "issue:43482", + "right": "issue:45538", "accept": false, - "reason": "Packaging TensorFlow with pip extras vs undefined tf backend when only tensorboard is installed. Similar area, different failure." + "reason": "Qwen2.5-GGUF loading and CLIPTokenizer model_max_length are unrelated." }, { "left": "issue:41720", "right": "issue:45081", "accept": false, - "reason": "Qwen3 auto device_map CUDA assert and Mistral tokenizer regex patch crash are unrelated." + "reason": "Qwen3 device-map cudaErrorAssert and Mistral regex patch crashes are unrelated." }, { "left": "issue:44079", "right": "issue:44534", "accept": false, - "reason": "ModelOutput key bookkeeping and non-persistent buffer initialization are distinct Transformers v5 regressions." + "reason": "ModelOutput key assignment and non-persistent buffer initialization are unrelated." }, { "left": "issue:43317", "right": "issue:43856", "accept": false, - "reason": "GPU+CPU offload loading failure for a dequantized model vs Qwen3 MoE training memory inefficiency. Different problems." + "reason": "Dequantized model offload loading and Qwen3 MoE training memory usage are different memory-related issues with different causes." }, { "left": "issue:44462", "right": "issue:45812", "accept": false, - "reason": "Ignoring tokenizer.json in AutoTokenizer and wrong Granite token IDs are both tokenizer issues, but not the same underlying bug." + "reason": "AutoTokenizer ignoring tokenizer.json could be a contributor, but the Granite token-ID regression is a distinct, broader failure and not proven to be the same bug." }, { "left": "issue:43296", "right": "issue:46821", "accept": false, - "reason": "PaddleOCR-VL model loading and TimesFM window_size attribute handling are unrelated." + "reason": "PaddleOCR-VL loading and TimesFM window_size AttributeError are unrelated." }, { "left": "issue:43611", "right": "issue:43646", "accept": false, - "reason": "Both are broad v5 regressions, but one is base_model_prefix loading and the other is custom model initialization. Too different to merge." - }, - { - "left": "issue:43377", - "right": "issue:45491", - "accept": false, - "reason": "MIMI padding-mask batching bug and Gemma3 sliding-window NaN issue are different attention/input-processing failures." + "reason": "Both are transformers v5 regressions, but base_model_prefix loading and custom model initialization are different code-path failures." } ] }, @@ -4218,10 +4155,10 @@ ], "soft_pairs": [ "issue:44509|issue:46909", - "issue:41553|issue:43582", "issue:42175|issue:43296", "issue:43644|issue:44079", "issue:44829|issue:45479", + "issue:41553|issue:43582", "issue:40444|issue:45381", "issue:44265|issue:45072", "issue:43818|issue:46402", @@ -4248,20 +4185,20 @@ "issue:41720|issue:45823", "issue:43606|issue:43819" ], - "prepared_review_unit_hash": "306bb49a76566de1b0798642bd0e169f091d7eb1682a85cfff4b67179e09eb24", + "prepared_review_unit_hash": "70a54d523e66d4f95aabe96a9a5d5fa51be95dc4eb8315428d9c5501345aaad4", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 36269, + "serialized_chars": 36271, "estimated_input_tokens": 9068, "estimated_eval_tokens": 18392 }, @@ -4270,201 +4207,201 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly heterogeneous. The only clear duplicate-like pair is the SigLIP2 implementation/output regression; the tied `lm_head.weight` reports are related but not clearly the same bug, so that edge should be rejected.", - "confidence": 0.63, - "canonical_issue_reason": "No single canonical issue cleanly covers the whole cluster because most items are unrelated. The strongest true duplicate pair is the SigLIP2 regression, with issue:43493 as the better anchor.", + "summary": "This is a heterogeneous similarity cluster with two real duplicate-like subclusters: SigLIP2 implementation/output mismatch, and a Qwen2.5-VL rope-index/video-position regression. The rest are different bugs, docs problems, or unrelated model/runtime failures.", + "confidence": 0.69, + "canonical_issue_reason": "issue:45325 is the clearest, most actionable root-cause report in the cluster: it names the specific get_rope_index / still-image temporal position_ids bug, and issue:45381 looks like the same regression from a different reproduction.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43493 is the best issue anchor because it is the most root-cause-oriented report in the only clearly duplicate pair (HF vs JAX SigLIP2 discrepancy vs nonsensical outputs).", + "best_issue_reason": "Best representative of the cluster\u2019s strongest duplicate pair: it is specific, code-path focused, and clearly maps to the same regression as 45381.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44509", "right": "issue:46909", "accept": false, - "reason": "Both are documentation problems, but they concern different docs regressions and different examples." - }, - { - "left": "issue:41553", - "right": "issue:43582", - "accept": false, - "reason": "AutoTokenizer/Voxtral loading and Apple Silicon allocator warmup are unrelated failures." + "reason": "Both are documentation complaints, but they refer to different docs problems and different examples; not the same underlying bug." }, { "left": "issue:42175", "right": "issue:43296", "accept": false, - "reason": "Package backend dependencies and a vLLM model load failure are different bugs." + "reason": "One is missing TensorFlow in the torch extra backend; the other is a vLLM/PaddleOCR-VL load failure. Different issues." }, { "left": "issue:43644", "right": "issue:44079", "accept": false, - "reason": "One is about non-persistent buffer junk values; the other is about ModelOutput key assignment." + "reason": "Non-persistent buffer junk and ModelOutput key assignment are unrelated code paths." }, { "left": "issue:44829", "right": "issue:45479", "accept": false, - "reason": "Both describe bad training behavior, but the models and failure mechanisms differ." + "reason": "Both affect sequence classification, but one is a flash-attention-3 training regression and the other is a zero-loss classification-label bug." + }, + { + "left": "issue:41553", + "right": "issue:43582", + "accept": false, + "reason": "Tokenizer loading error message vs Apple Silicon allocator TypeError are unrelated." }, { "left": "issue:40444", "right": "issue:45381", "accept": false, - "reason": "Multiple-images finetuning failure and video position-id regression are different multimodal code paths." + "reason": "Both involve multimodal/video models, but these are different models and different failures." }, { "left": "issue:44265", "right": "issue:45072", "accept": false, - "reason": "torch.export failure from torch_compilable_check is unrelated to bfloat16 dtype mismatches." + "reason": "torch.export failures from torch_compilable_check are not the same as bfloat16 dtype mismatch bugs." }, { "left": "issue:43818", "right": "issue:46402", "accept": false, - "reason": "Video-LLaVA architectural issues and Sam3 text encoder loading are unrelated." + "reason": "Video-LLaVA architecture/weights issue vs Sam3Model text encoder loading issue; different bugs." }, { "left": "issue:43493", "right": "issue:43994", "accept": true, - "reason": "Both point to the same SigLIP2 implementation being wrong: one compares HF vs JAX, the other reports nonsensical outputs from the same model." + "reason": "Both describe the same SigLIP2 HF-vs-reference implementation mismatch leading to incorrect/nonsensical outputs." }, { "left": "issue:43818", "right": "issue:45357", "accept": false, - "reason": "Video-LLaVA architectural issues and Qwen3.5 save_pretrained key issues are unrelated." + "reason": "Video-LLaVA temporal-attention/weight-sharing bug is unrelated to Qwen3.5 save_pretrained key serialization." }, { "left": "issue:45325", "right": "issue:45834", "accept": false, - "reason": "Incorrect Qwen2.5-VL rope/position logic and Kosmos2.5 OCR index errors are different bugs." + "reason": "Qwen2.5-VL rope-index position bug and Kosmos2.5 OCR index error are different failures." }, { "left": "issue:45325", "right": "issue:45381", - "accept": false, - "reason": "They both touch Qwen2.5-VL positions, but one is still-image temporal scaling and the other is video vision_position_ids; too different to merge confidently." + "accept": true, + "reason": "Same Qwen2.5-VL video/still-image temporal-position_ids regression, described with the same underlying code path." }, { "left": "issue:43454", "right": "issue:44466", "accept": false, - "reason": "Both mention tied `lm_head.weight`, but the titles describe different failure modes: missing weight tying vs inconsistent serialization/device-dependent behavior." + "reason": "One is lm_head not being tied for generation correctness; the other is device-dependent serialization of tied weights. Related theme, different bug." }, { "left": "issue:42915", "right": "issue:43482", "accept": false, - "reason": "Qwen3Moe FP8 config failure and Qwen2.5-GGUF v5 loading failure are separate model/load-path regressions." + "reason": "Qwen3Moe FP8 config failure and Qwen2.5-GGUF loading failure are separate model/config compatibility issues." }, { "left": "issue:44291", "right": "issue:44877", "accept": false, - "reason": "Unexpected init_empty_weights argument and strict granite_speech config loading are different compatibility problems." + "reason": "init_empty_weights argument handling and strict granite_speech config loading are different loading bugs." }, { "left": "issue:43873", "right": "issue:45701", "accept": false, - "reason": "Quantized offloading misbehavior and tokenization changes across versions are unrelated." + "reason": "Quantized offloading behavior and tokenization changes across versions are unrelated." }, { "left": "issue:43701", "right": "issue:45325", "accept": false, - "reason": "Checkpoint resume key mismatch is not the same bug as Qwen2.5-VL rope-index scaling." + "reason": "Checkpoint resume key mismatch is not the same as the Qwen2.5-VL rope-index regression." }, { "left": "issue:42796", "right": "issue:45381", "accept": false, - "reason": "Tokenizer BPE-marker handling and Qwen2.5-VL video position ids are separate issues." + "reason": "Tokenizer BPE-marker handling and Qwen2.5-VL video position_ids are unrelated." }, { "left": "issue:40990", "right": "issue:45584", "accept": false, - "reason": "High perplexity on gpt-oss and Whisper empty-transcription failure do not share a code path." + "reason": "High perplexity on gpt-oss and Whisper empty-transcription failure are unrelated model behaviors." }, { "left": "issue:43606", "right": "issue:44265", "accept": false, - "reason": "CPU offload device mismatch and torch.export torch_compilable_check failure are different runtime paths." + "reason": "CPU offload device mismatch and torch.export / torch_compilable_check failure are different execution paths." }, { "left": "issue:43262", "right": "issue:43746", "accept": false, - "reason": "Audio chat-template sample rate defaulting and PEFT checkpoint loading are unrelated." + "reason": "Audio chat-template sample-rate defaulting and GraniteSpeech PEFT checkpoint loading are unrelated." }, { "left": "issue:44479", "right": "issue:45823", "accept": false, - "reason": "Video-input regression and Gemma4 PLE device mismatch are different model-specific bugs." + "reason": "Both are device/vision related, but they are different model families and different regressions." }, { "left": "issue:35707", "right": "issue:41720", "accept": false, - "reason": "inputs_embeds/past_key_values progressive generation and Qwen3 device-mapping CUDA asserts are unrelated." + "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen3 auto device-mapping CUDA asserts." }, { "left": "issue:43828", "right": "issue:46821", "accept": false, - "reason": "Autocast dtype mismatch and TimesFM window_size AttributeError are different failures." + "reason": "Autocast dtype mismatch in Phi-tiny-MoE is unrelated to a TimesFM window_size AttributeError." }, { "left": "issue:34634", "right": "issue:35707", "accept": false, - "reason": "Bark voice_preset and progressive generation issues are unrelated." + "reason": "Bark voice_preset and progressive-generation issues are different bugs." }, { "left": "issue:39692", "right": "issue:40444", "accept": false, - "reason": "SigLIP2 docs-example errors and Qwen2.5-VL multi-image finetuning failures are different problems." + "reason": "SigLIP2 docs example errors and Qwen2.5-VL iterable dataset failures are different issues." }, { "left": "issue:44186", "right": "issue:44265", "accept": false, - "reason": "Tokenizer crash and torch.export failure are not the same bug." + "reason": "Tokenizer batching crashes and torch.export failures are unrelated." }, { "left": "issue:43454", "right": "issue:43550", "accept": false, - "reason": "Tied-weight serialization/garbage text generation and Bamba torch.compile SDPA failure are unrelated." + "reason": "lm_head weight tying and Bamba torch.compile SDPA failure are unrelated." }, { "left": "issue:41720", "right": "issue:45823", "accept": false, - "reason": "Qwen3 auto device mapping CUDA asserts and Gemma4 PLE device mismatch are different models and code paths." + "reason": "Both mention device mapping, but the underlying models and failure modes differ." }, { "left": "issue:43606", "right": "issue:43819", "accept": false, - "reason": "CPU offload device mismatch and DAC.from_latents missing-STE mismatch are unrelated." + "reason": "CPU offload device mismatch and DAC STE mismatch are unrelated code-path problems." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The overall clustering summary is reasonably grounded, but one soft-edge accept is too strong for the evidence in the packet. `issue:43454` and `issue:44466` both involve tied `lm_head.weight`, yet the titles point to different failure modes (missing weight tying vs inconsistent serialization/device-dependent behavior). From the packet alone, that is only a plausible relation, not clearly the same bug or a single mergeable fix. The other rejects are suitably conservative." + "accept": true, + "feedback": "The analysis is grounded in the packet and stays mostly conservative. The two accepted pairs are plausibly the same regressions from the issue titles/body context, while the rejected pairs are generally justified by clearly different models, code paths, or failure modes. The summary does not overclaim beyond the evidence provided." }, "evaluator_used": true, - "retried": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -4535,12 +4472,12 @@ "issue:43257|issue:45923", "issue:42796|issue:46209", "issue:44568|issue:45920", - "issue:46305|issue:46315", "issue:43454|issue:45072", + "issue:46305|issue:46315", "issue:45823|issue:46821", - "issue:43881|issue:44488", "issue:40990|issue:42947", "issue:42617|issue:42915", + "issue:43881|issue:44488", "issue:43377|issue:44898", "issue:43454|issue:43720", "issue:43749|issue:44534", @@ -4553,8 +4490,8 @@ "issue:41720|issue:44977", "issue:36331|issue:42915", "issue:40990|issue:45701", - "issue:44466|issue:45357", "issue:44877|issue:44991", + "issue:44466|issue:45357", "issue:43450|issue:45701", "issue:43065|issue:46402", "issue:41720|issue:42617", @@ -4562,214 +4499,214 @@ "issue:44186|issue:45072", "issue:43873|issue:44756" ], - "prepared_review_unit_hash": "be7fbfc739d5c426b24d7887c457695c44f1e21f65b706829ba9290913f84b98", + "prepared_review_unit_hash": "fa9d62d95a5e0aed5280c59ec69609b616de56fffca83db025ee83f612240dc2", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 36059, - "estimated_input_tokens": 9015, - "estimated_eval_tokens": 18286 + "serialized_chars": 36065, + "estimated_input_tokens": 9017, + "estimated_eval_tokens": 18290 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "No duplicate pairs look safe to merge. The set spans unrelated bugs across checkpointing, tokenization, quantization/offloading, vision models, and trainer APIs, so all soft edges are false positives.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "This cluster is heterogeneous: most items are distinct tokenizer, loading, quantization, or model-specific bugs. I rejected all soft edges as non-duplicates and used the broad offloading/quantization regression issue as the best representative.", + "confidence": 0.71, + "canonical_issue_reason": "issue:43873 is the broadest representative of the repeated loading/offloading/quantization regression theme in this set, even though the cluster is not a true duplicate group.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "issue:43873 best matches the cluster\u2019s widest recurring theme (offloading/quantization behavior) and is the least specific representative among the issues.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43377", "right": "issue:44610", "accept": false, - "reason": "Different failures: MIMI batched-vs-single padding-mask behavior vs OmDet processor/model input-size mismatch." + "reason": "Different models and failure modes: padding-mask batched-output bug vs image input shape mismatch." }, { "left": "issue:43257", "right": "issue:45923", "accept": false, - "reason": "Unrelated problems: Qwen3 MOE weight conversion/loading vs missing gradient-checkpointing flag on a trust_remote_code model." + "reason": "Different subsystems: MoE weight conversion/loading vs missing gradient-checkpointing flag." }, { "left": "issue:42796", "right": "issue:46209", "accept": false, - "reason": "Tokenizer BPE-marker handling vs FP8 inference failure on B200; different code paths and symptoms." + "reason": "Tokenizer BPE-marker handling is unrelated to FP8 inference failure on B200." }, { "left": "issue:44568", "right": "issue:45920", "accept": false, - "reason": "Specific BOS/EOS insertion regression for mdeberta-v3-base vs a broad wrong-token-IDs regression across many models." - }, - { - "left": "issue:46305", - "right": "issue:46315", - "accept": false, - "reason": "Streaming mask off-by-one bug vs convert_tokens_to_ids performance regression." + "reason": "Both are tokenizer regressions, but they affect different code paths and model families." }, { "left": "issue:43454", "right": "issue:45072", "accept": false, - "reason": "lm_head weight tying/serialization issue vs bfloat16 dtype mismatch in inference." + "reason": "lm_head tying/generation corruption is unrelated to bf16 dtype mismatches." }, { - "left": "issue:45823", - "right": "issue:46821", + "left": "issue:46305", + "right": "issue:46315", "accept": false, - "reason": "Gemma4 device-mapping forward mismatch vs TimesFM window_size AttributeError." + "reason": "Different areas: sliding-window masking off-by-one vs slow token-id conversion." }, { - "left": "issue:43881", - "right": "issue:44488", + "left": "issue:45823", + "right": "issue:46821", "accept": false, - "reason": "Two separate model-loading failures for different checkpoints; no shared underlying bug is evident." + "reason": "Device-map forward mismatch is unrelated to TimesFM window_size AttributeError." }, { "left": "issue:40990", "right": "issue:42947", "accept": false, - "reason": "Benchmark perplexity issue on gpt-oss-20b vs LoRA gradient-checkpointing effectiveness; unrelated bugs." + "reason": "High perplexity on a model is not the same bug as LoRA gradient-checkpointing behavior." }, { "left": "issue:42617", "right": "issue:42915", "accept": false, - "reason": "3d_parallel.py execution failure vs Qwen3Moe loading with FineGrainedFP8Config." + "reason": "No shared concrete code path: 3D parallel script failure vs FineGrainedFP8Config failure." + }, + { + "left": "issue:43881", + "right": "issue:44488", + "accept": false, + "reason": "Both are load failures, but for different models and likely different root causes." }, { "left": "issue:43377", "right": "issue:44898", "accept": false, - "reason": "Both are shape-related vision issues, but the concrete failures differ: MIMI batching/padding vs Perceiver non-default resolution handling." + "reason": "Both involve input/shape behavior, but one is batched-vs-single encoder output and the other is image resolution handling." }, { "left": "issue:43454", "right": "issue:43720", "accept": false, - "reason": "Tied-weight serialization bug vs BitNet packed-weight unpacking during accelerate loading." + "reason": "Weight tying bug is unrelated to packed-weight unpacking during accelerate loading." }, { "left": "issue:43749", "right": "issue:44534", "accept": false, - "reason": "FSDP CPU RAM-efficient loading vs junk-filled non-persistent buffers; different mechanisms and fixes." + "reason": "Different bugs: FSDP CPU-efficient loading vs non-persistent buffer serialization junk." }, { "left": "issue:43317", "right": "issue:43873", "accept": false, - "reason": "Specific dequantized-model gpu/cpu offload load failure vs a broad quantization offloading issue; not the same concrete bug." + "reason": "Related broad area, but one is dequantized gpu+cpu offload loading and the other is general quantized offloading behavior." }, { "left": "issue:43572", "right": "issue:45759", "accept": false, - "reason": "StableLmConfig pad_token_idx regression vs composite Qwen text_config unwrapping." + "reason": "Config missing pad_token_idx is not the same as failing to unwrap text_config for composite models." }, { "left": "issue:43322", "right": "issue:46050", "accept": false, - "reason": "Llava Next loading segfault vs sinq save/load breakage after quantization." + "reason": "Segfault loading Llava Next is unrelated to sinq save/load corruption." }, { "left": "issue:41720", "right": "issue:43322", "accept": false, - "reason": "Qwen3 auto device mapping cudaErrorAssert vs Llava Next model-loading segmentation fault." + "reason": "Both are loading failures, but the model families and concrete failure modes differ." }, { "left": "issue:36331", "right": "issue:42617", "accept": false, - "reason": "CustomTrainer compute_loss signature mismatch vs 3d_parallel.py runtime failure." + "reason": "Trainer signature change is unrelated to a parallel script runtime failure." }, { "left": "issue:44387", "right": "issue:45704", "accept": false, - "reason": "int4 quantization memory regression vs T5/apex RMSNorm memory leak." + "reason": "OOM from int4 quantization is not the same as the T5 apex RMSNorm memory leak." }, { "left": "issue:41720", "right": "issue:44977", "accept": false, - "reason": "Same model family, but different concrete bugs: device-mapping assert vs flash-attention generation corruption." + "reason": "Same model family only; device-map CUDA assert and flash-attention generation bug are different issues." }, { "left": "issue:36331", "right": "issue:42915", "accept": false, - "reason": "Trainer API change bug vs Qwen3Moe fine-grained FP8 loading issue." + "reason": "Unexpected trainer keyword argument is unrelated to Qwen3MoE FP8 loading." }, { "left": "issue:40990", "right": "issue:45701", "accept": false, - "reason": "Perplexity/evaluation regression vs version-dependent tokenization changes; not the same bug." + "reason": "Perplexity regression and tokenization-version change are not the same concrete bug." }, { - "left": "issue:44466", - "right": "issue:45357", + "left": "issue:44877", + "right": "issue:44991", "accept": false, - "reason": "lm_head serialization inconsistency vs incorrect visual encoder keys saved in Qwen3.5." + "reason": "Strict config loading and tokenizer loading failures are separate problems." }, { - "left": "issue:44877", - "right": "issue:44991", + "left": "issue:44466", + "right": "issue:45357", "accept": false, - "reason": "Strict config rejection for granite_speech vs tokenizer loading failure for est-roberta." + "reason": "Both are save_pretrained serialization regressions, but they affect different tensors/keys and model paths." }, { "left": "issue:43450", "right": "issue:45701", "accept": false, - "reason": "Video processor batched-shape bug vs tokenization behavior changes; unrelated." + "reason": "Video processor batching shape bug is unrelated to tokenization behavior changes." }, { "left": "issue:43065", "right": "issue:46402", "accept": false, - "reason": "Dummy Conv2d in Sam3PixelDecoder vs Sam3Model text-encoder weights not loading." + "reason": "Same model family, but pixel-decoder dummy conv and text-encoder weight loading are different code paths." }, { "left": "issue:41720", "right": "issue:42617", "accept": false, - "reason": "Qwen3 device-mapping assert vs 3d_parallel.py failure; no shared code-path bug." + "reason": "Different models and failures: device-map CUDA assert vs a separate script execution issue." }, { "left": "issue:26413", "right": "issue:28530", "accept": false, - "reason": "resume_from_checkpoint epoch-iterator failure vs early-stopping metric lookup/disabling." + "reason": "Checkpoint resume iterator exhaustion and early-stopping metric lookup are unrelated training issues." }, { "left": "issue:44186", "right": "issue:45072", "accept": false, - "reason": "LayoutLMv2 tokenizer crash vs bfloat16 dtype mismatch in model inference." + "reason": "Tokenizer crash on NER/padding is unrelated to bf16 dtype mismatches." }, { "left": "issue:43873", "right": "issue:44756", "accept": false, - "reason": "Quantization/offloading bug vs mmap-related OOM on Strix Halo; different fixes." + "reason": "Both mention memory/offloading, but mmap-on-Strix-Halo OOM is not the same as quantized offloading behavior." } ] }, @@ -4843,23 +4780,23 @@ "issue:47020" ], "soft_pairs": [ - "issue:43526|issue:43883", - "issue:43881|issue:44991", "issue:46612|issue:46858", "issue:43577|issue:43708", + "issue:43526|issue:43883", "issue:43550|issue:43720", + "issue:43881|issue:44991", "issue:43295|issue:45701", "issue:45685|issue:46821", - "issue:44186|issue:44610", "issue:43646|issue:46665", + "issue:44186|issue:44610", "issue:44451|issue:47020", "issue:43550|issue:43606", "issue:36010|issue:42617", "issue:43404|issue:45357", "issue:44610|issue:44898", "issue:42915|issue:43317", - "issue:44734|issue:46858", "issue:44423|issue:45406", + "issue:44734|issue:46858", "issue:46821|issue:46869", "issue:44008|issue:45823", "issue:41720|issue:45161", @@ -4872,202 +4809,202 @@ "issue:36331|issue:41720", "issue:44186|issue:44361" ], - "prepared_review_unit_hash": "ea9f9f5ff44abd84928e5c9827eefe8810c5554b8ab6ac374796841d8026d69c", + "prepared_review_unit_hash": "e7e3b11a7b38b8b39b7c31479de1003b3cc8c3e6ca071a59db5076b3de8b798e", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36265, - "estimated_input_tokens": 9067, - "estimated_eval_tokens": 18390 + "serialized_chars": 36263, + "estimated_input_tokens": 9066, + "estimated_eval_tokens": 18388 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a heterogeneous bucket of unrelated issues across training, tokenization, model loading, device mapping, cache/generation, and multimodal serving. None of the soft pairs look like true duplicates of the same underlying bug or change.", + "summary": "The cluster is a loose collection of unrelated Transformers issues. None of the soft pairs look like the same underlying bug or change, so all soft edges are rejected.", "confidence": 0.96, - "canonical_issue_reason": "No single canonical duplicate target stands out for the whole cluster; if one must be chosen, issue 46858 is the most self-contained and concrete bug report (static cache + torch.compile on non-CUDA backends).", + "canonical_issue_reason": "Issue 46858 is the most self-contained and technically precise report, but there is no true duplicate cluster here; the issues span unrelated subsystems and models.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 46858 is the strongest anchor candidate because it has a precise failure mode, a narrowly defined code path, and a clear repro surface, even though the surrounding items are not duplicates of it.", + "best_issue_reason": "Issue 46858 has the clearest root-cause framing and reproduction details, making it the best single representative of this mixed set, though not a duplicate hub.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43526", - "right": "issue:43883", - "accept": false, - "reason": "Different bugs: BEiT Fast label reduction vs Molmo missing tied-weights attribute." - }, - { - "left": "issue:43881", - "right": "issue:44991", - "accept": false, - "reason": "Both are load failures, but they involve different models and different root causes." - }, { "left": "issue:46612", "right": "issue:46858", "accept": false, - "reason": "Both involve cache/generation, but one is missing beam-search cache reorder and the other is a static-cache arithmetic bug on non-CUDA backends." + "reason": "Both involve generation/cache behavior, but one is beam-search reorder skipping and the other is a static-cache seq-length arithmetic bug; different code paths." }, { "left": "issue:43577", "right": "issue:43708", "accept": false, - "reason": "Unrelated: dtype propagation during load vs trainer resume/max_steps calculation." + "reason": "One is a dtype-loading bug for Blip2, the other is Trainer checkpoint/resume max_steps logic; unrelated problems." + }, + { + "left": "issue:43526", + "right": "issue:43883", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and Molmo tied-weights metadata are separate components with different failures." }, { "left": "issue:43550", "right": "issue:43720", "accept": false, - "reason": "Different concrete failures: SDPA/torch.compile for Bamba vs packed-weight unpacking during accelerate loading for BitNet." + "reason": "Torch.compile + SDPA in Bamba and BitNet packed-weight loading with accelerate are distinct bugs." + }, + { + "left": "issue:43881", + "right": "issue:44991", + "accept": false, + "reason": "Model loading failures, but for different causes: glm-4v-9b model init versus tokenizer loading for est-roberta." }, { "left": "issue:43295", "right": "issue:45701", "accept": false, - "reason": "Both are version regressions around tokenization, but they are different APIs and symptoms, not the same bug." + "reason": "Both mention tokenization regressions, but one is a processor.tokenizer API break and the other is a broad version-related tokenization change." }, { "left": "issue:45685", "right": "issue:46821", "accept": false, - "reason": "MPS histogram-on-int failure is unrelated to TimesFM window_size attribute handling." + "reason": "An MPS histogram kernel limitation is unrelated to a TimesFM window_size attribute error." }, { - "left": "issue:44186", - "right": "issue:44610", + "left": "issue:43646", + "right": "issue:46665", "accept": false, - "reason": "Both touch preprocessing, but NER padding/truncation in LayoutLMv2Tokenizer is unrelated to OmDet-Turbo image size mismatch." + "reason": "CPU offload/device-map loading failure is unrelated to from_pretrained behavior inside Python classes." }, { - "left": "issue:43646", - "right": "issue:46665", + "left": "issue:44186", + "right": "issue:44610", "accept": false, - "reason": "Custom model initialization breakage is not the same as inconsistent from_pretrained behavior inside Python classes." + "reason": "LayoutLMv2 NER/padding crashes and OmDet-Turbo input-size mismatch are different tokenizer/processor bugs." }, { "left": "issue:44451", "right": "issue:47020", "accept": false, - "reason": "Tokenizer loading failures, but different tokenizers and different failure mechanisms." + "reason": "Both are tokenizer load issues, but the affected models and failure modes differ; not enough evidence of the same underlying defect." }, { "left": "issue:43550", "right": "issue:43606", "accept": false, - "reason": "Both are device/compile related, but SDPA compilation and CPU offload mismatch are distinct problems." + "reason": "Both are runtime failures, but one is SDPA/compile and the other is CPU offload device mismatch; unrelated." }, { "left": "issue:36010", "right": "issue:42617", "accept": false, - "reason": "Completely unrelated: GenerationMixin import error vs inability to run 3d_parallel.py." + "reason": "GenerationMixin import failure and 3d_parallel.py runtime failure are unrelated." }, { "left": "issue:43404", "right": "issue:45357", "accept": false, - "reason": "Different model-saving bugs: untied lm_head weights vs incorrect visual encoder keys." + "reason": "Both touch multimodal model internals, but one is weight tying in Mistral3 and the other is incorrect visual encoder save keys in Qwen3.5." }, { "left": "issue:44610", "right": "issue:44898", "accept": false, - "reason": "Both concern resolution handling, but they affect different models and different code paths." + "reason": "Both are vision/preprocessing issues, but one is a processor output size bug and the other is a model positional-encoding/interpolation bug." }, { "left": "issue:42915", "right": "issue:43317", "accept": false, - "reason": "Different quantization/device-map issues with different failure modes." + "reason": "Different loading paths: FineGrainedFP8Config failure versus dequantized model loading with device_map auto/offload." }, { - "left": "issue:44734", - "right": "issue:46858", + "left": "issue:44423", + "right": "issue:45406", "accept": false, - "reason": "Both mention cache, but serve KV continuation indexing is not the same bug as static cache seq-length arithmetic." + "reason": "Both affect serve with multimodal models, but the errors point to different root causes: string .to() misuse versus missing _tokenizer on Gemma4Processor." }, { - "left": "issue:44423", - "right": "issue:45406", + "left": "issue:44734", + "right": "issue:46858", "accept": false, - "reason": "Both are serve/multimodal crashes, but one is a string-to-device error and the other is a missing processor attribute." + "reason": "Both involve cache/generation internals, but one is KV-cache continuation indexing in serve and the other is static-cache seq-length arithmetic on non-CUDA backends." }, { "left": "issue:46821", "right": "issue:46869", "accept": false, - "reason": "Unrelated models and failures: TimesFM window_size vs YOLO multi-GPU device-map loading." + "reason": "TimesFM window_size attribute and YOLO multi-GPU device-map failure are unrelated." }, { "left": "issue:44008", "right": "issue:45823", "accept": false, - "reason": "Same family name, but different models and different bugs: variable collision vs device mismatch." + "reason": "Different Gemma variants and different bugs: variable name collision in forward versus device mismatch under auto device mapping." }, { "left": "issue:41720", "right": "issue:45161", "accept": false, - "reason": "Both relate to distributed/device setup, but Qwen3 auto device mapping and GPT-OSS tensor parallelism are distinct issues." + "reason": "Both involve large-model parallelism, but cudaErrorAssert on A800 and GPT-OSS MoE tensor-parallel failure are distinct issues." }, { "left": "issue:28530", "right": "issue:33666", "accept": false, - "reason": "Training early-stopping config error is unrelated to Qwen2-VL multi-GPU training." + "reason": "Early-stopping metric handling and Qwen2-VL multi-GPU training are unrelated." }, { "left": "issue:44464", "right": "issue:46858", "accept": false, - "reason": "Both involve compiled generation, but inconsistent chunked generation outputs are not the same as the static-cache length bug." + "reason": "Both mention torch.compile/generation, but one is inconsistent chunked generation output and the other is a concrete static-cache seq-length tensor bug." }, { "left": "issue:43531", "right": "issue:43572", "accept": false, - "reason": "Different config bugs in different models: sliding_window handling vs missing pad_token_idx." + "reason": "Qwen3-MoE sliding_window behavior and StableLmConfig missing pad_token_idx are separate config/model issues." }, { "left": "issue:45632", "right": "issue:45684", "accept": false, - "reason": "Both are custom-model persistence issues, but cache-path collision and read-only permission propagation are unrelated." + "reason": "Both concern save/cache behavior for custom models, but one is cache-path collision and the other is read-only permission propagation." }, { "left": "issue:44671", "right": "issue:44743", "accept": false, - "reason": "Different model-specific regressions: CamemBERT masked-LM predictions vs Qwen3.5 recurrent-state reset." + "reason": "CamemBERT masked-LM prediction regression and Qwen3.5 recurrent-state reset under cache are unrelated." }, { "left": "issue:42915", "right": "issue:46291", "accept": false, - "reason": "Quantization/runtime failure is unrelated to kernels package version requirements on pipeline import." + "reason": "FineGrainedFP8Config model failure and pipelines import dependency/version error are unrelated." }, { "left": "issue:36331", "right": "issue:41720", "accept": false, - "reason": "Trainer API incompatibility is unrelated to Qwen3 device-mapping CUDA asserts." + "reason": "Trainer compute_loss signature breakage is unrelated to Qwen3 auto device-mapping cuda assert." }, { "left": "issue:44186", "right": "issue:44361", "accept": false, - "reason": "Both are tokenizer crashes, but they involve different tokenizers and different task paths." + "reason": "Both are tokenizer crashes, but they affect different tokenizers and different failure sites; not the same bug." } ] }, @@ -5152,13 +5089,13 @@ "issue:44387|issue:45701", "issue:43550|issue:43653", "issue:44534|issue:46665", - "issue:43526|issue:43881", "issue:43531|issue:45759", "issue:43452|issue:47020", "issue:41720|issue:43746", "issue:43606|issue:43720", "issue:44492|issue:44509", "issue:42490|issue:43329", + "issue:43526|issue:43881", "issue:45406|issue:45464", "issue:43388|issue:45059", "issue:34567|issue:35707", @@ -5170,208 +5107,208 @@ "issue:43531|issue:44589", "issue:44077|issue:44625" ], - "prepared_review_unit_hash": "e22b00fd376ad9bdd4cee94f8cce516bedfdbc699d0bca70b4df0a0ab5c8d986", + "prepared_review_unit_hash": "36140365305c3c6060abf0069b801283b28396e4d7b074ab8315e40579fba7a2", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 29, - "serialized_chars": 35779, - "estimated_input_tokens": 8945, - "estimated_eval_tokens": 18146 + "serialized_chars": 35788, + "estimated_input_tokens": 8947, + "estimated_eval_tokens": 18150 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "No duplicate cluster emerges here: the items span many unrelated bugs, and every soft pair looks like a separate issue rather than the same underlying defect.", - "confidence": 0.95, - "canonical_issue_reason": "No single canonical issue is justified; the issues are heterogeneous across tokenizers, configs, multimodal processors, loading/offload, and training/runtime behavior.", + "summary": "All soft pairs look like false positives: they mention similar components (tokenizers, configs, loading, offload, Qwen, etc.) but describe different concrete failures, models, or code paths. No duplicate cluster is supported here.", + "confidence": 0.97, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "No best issue for deduplication purposes\u2014none of the items is representative of a coherent duplicate set.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43425", "right": "issue:46821", "accept": false, - "reason": "One is a Torch version compatibility problem; the other is a TimesFM `window_size` attribute error." + "reason": "Torch version compatibility and a TimesFM window_size AttributeError are unrelated bugs." }, { "left": "issue:43316", "right": "issue:44913", "accept": false, - "reason": "Different config bugs: Gemma3TextConfig API mismatch vs GPTNeoX `rotary_pct` not persisting." + "reason": "Both are config-related, but they concern different models and different fields with different failure modes." }, { "left": "issue:28530", "right": "issue:30064", "accept": false, - "reason": "Early-stopping metric lookup and image processor void segmentation maps are unrelated." + "reason": "Early stopping metric lookup and void segmentation map processing are unrelated subsystems." }, { "left": "issue:42915", "right": "issue:43322", "accept": false, - "reason": "Different failure modes and code paths: Qwen3Moe FP8 config vs Llava Next loading segfault." + "reason": "Qwen3Moe FP8 config failure and Llava Next segfault are different model-loading problems." }, { "left": "issue:33357", "right": "issue:35707", "accept": false, - "reason": "CLIP MacOS bus error is unrelated to progressive generation with `inputs_embeds` and `past_key_values`." + "reason": "MacOS CLIP bus error and progressive generation with inputs_embeds/past_key_values are distinct issues." }, { "left": "issue:43540", "right": "issue:43782", "accept": false, - "reason": "Video-input validation in Qwen3OmniMoe is distinct from Qwen3VL `weight_only=True` loading errors." + "reason": "Video processing in Qwen3OmniMoe and Qwen3VL weight_only loading error are different code paths." }, { "left": "issue:43873", "right": "issue:45704", "accept": false, - "reason": "Quantization/offload memory issues are not the same as T5's apex RMSNorm memory leak." + "reason": "Quantization/offloading memory behavior is unrelated to T5's apex RMSNorm leak." }, { "left": "issue:29159", "right": "issue:36032", "accept": false, - "reason": "Slow/fast tokenizer inconsistency is different from the T5 `add_special_tokens` name conflict." + "reason": "Slow/fast tokenizer inconsistency and T5 add_special_tokens load-time conflict are different tokenizer bugs." }, { "left": "issue:30064", "right": "issue:34634", "accept": false, - "reason": "Void segmentation map processing and Bark `voice_preset` handling are unrelated." + "reason": "Segmentation-map processing and BarkProcessor voice_preset handling do not describe the same defect." }, { "left": "issue:44387", "right": "issue:45701", "accept": false, - "reason": "Int4 quantization OOM is not the same as tokenization behavior changing across versions." + "reason": "Int4 quantization OOM and tokenization changing across versions are unrelated." }, { "left": "issue:43550", "right": "issue:43653", "accept": false, - "reason": "Torch.compile/SDPA runtime failure and BigBird tokenizer special-token registration are different bugs." + "reason": "torch.compile/SDPA failure in Bamba and BigBirdTokenizer special-token registration are unrelated." }, { "left": "issue:44534", "right": "issue:46665", "accept": false, - "reason": "Non-persistent buffer initialization junk and `from_pretrained` behavior inside Python classes do not share the same code path." - }, - { - "left": "issue:43526", - "right": "issue:43881", - "accept": false, - "reason": "BeitImageProcessorFast label reduction and glm-4v-9b loading failure are unrelated." + "reason": "Non-persistent buffer initialization in v5 is not the same as from_pretrained behavior inside Python classes." }, { "left": "issue:43531", "right": "issue:45759", "accept": false, - "reason": "Qwen3-MoE sliding-window behavior differs from unwrapping `text_config` in composite Qwen configs." + "reason": "Qwen3-MoE sliding_window and composite config unwrapping are different Qwen config issues." }, { "left": "issue:43452", "right": "issue:47020", "accept": false, - "reason": "GGUF loading issues are unrelated to XLMRobertaTokenizer passing a dict to Unigram." + "reason": "gguf_file loading support and XLMRobertaTokenizer/Unigram type mismatch are different loading failures." }, { "left": "issue:41720", "right": "issue:43746", "accept": false, - "reason": "Auto device-mapping CUDA assert and GraniteSpeech PEFT checkpoint loading are different problems." + "reason": "Qwen3 auto device mapping cudaErrorAssert and GraniteSpeech PEFT checkpoint loading are unrelated." }, { "left": "issue:43606", "right": "issue:43720", "accept": false, - "reason": "CPU-offload device mismatch in Bark is not the same as BitNet packed-weight unpacking during accelerate loading." + "reason": "CPU offload device mismatch and BitNet packed-weight unpacking are different accelerate-loading bugs." }, { "left": "issue:44492", "right": "issue:44509", "accept": false, - "reason": "A cache-strategy typo is not the same as outdated docs listing removed pipeline tasks." + "reason": "Cache strategy typo and removed pipeline-task docs are both documentation-like but not the same issue." }, { "left": "issue:42490", "right": "issue:43329", "accept": false, - "reason": "Local-path import warnings and undefined video-branch helpers in multimodal token counting are unrelated." + "reason": "Trailing-slash local path warning and missing video-branch variables in multimodal token counting are unrelated." + }, + { + "left": "issue:43526", + "right": "issue:43881", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and glm-4v-9b loading failure are different problems." }, { "left": "issue:45406", "right": "issue:45464", "accept": false, - "reason": "Gemma4Processor missing `_tokenizer` and Qwen3.5 streaming inference failures are different code paths." + "reason": "Gemma4Processor missing _tokenizer and Qwen3.5 streaming chat/completions failure are separate serving bugs." }, { "left": "issue:43388", "right": "issue:45059", "accept": false, - "reason": "Metric gathering label truncation and SAM3 text/bbox behavior are unrelated." + "reason": "Last-batch label gathering and SAM3 text/box behavior are different evaluation/model issues." }, { "left": "issue:34567", "right": "issue:35707", "accept": false, - "reason": "Trainer token-count tracking is unrelated to progressive generation with cached past key values." + "reason": "TrainerState token counter not updating is unrelated to progressive generation cache handling." }, { "left": "issue:43606", "right": "issue:43653", "accept": false, - "reason": "Device-mismatch offload bug and tokenizer special-token bug are entirely separate." + "reason": "CPU offload device mismatch and BigBirdTokenizer mask-token registration are unrelated." }, { "left": "issue:34634", "right": "issue:41720", "accept": false, - "reason": "Bark voice preset handling and Qwen3 auto device mapping failures are unrelated." + "reason": "BarkProcessor voice_preset and Qwen3 auto device mapping failures do not share a code path." }, { "left": "issue:43531", "right": "issue:46821", "accept": false, - "reason": "Qwen3-MoE sliding-window support is unrelated to TimesFM `window_size` attribute handling." + "reason": "Qwen3-MoE sliding_window and TimesFM window_size are superficially similar but different APIs and bugs." }, { "left": "issue:43540", "right": "issue:46858", "accept": false, - "reason": "Qwen3OmniMoe video-input validation is unrelated to static-cache sequence-length tensor arithmetic." + "reason": "Video-input handling in Qwen3OmniMoe and static-cache slice arithmetic on non-CUDA backends are unrelated." }, { "left": "issue:45759", "right": "issue:45859", "accept": false, - "reason": "Composite Qwen config unwrapping and CPU-offload rotary embedding compatibility are different bugs." + "reason": "Composite config unwrapping and CPU-offload rotary embedding compatibility are different Qwen issues." }, { "left": "issue:43531", "right": "issue:44589", "accept": false, - "reason": "Sliding-window behavior and missing Float8 storage handling are unrelated." + "reason": "Sliding-window behavior and missing Float8 storage support are unrelated." }, { "left": "issue:44077", "right": "issue:44625", "accept": false, - "reason": "`patchtsmixer` post-init policy and Qwen3.5 `num_labels` propagation are distinct configuration issues." + "reason": "patchtsmixer post_init policy and Qwen3.5 num_labels propagation are different model-config issues." } ] }, @@ -5454,8 +5391,8 @@ "issue:44062|issue:44451", "issue:43381|issue:43844", "issue:39692|issue:44610", - "issue:44368|issue:45357", "issue:41720|issue:45704", + "issue:44368|issue:45357", "issue:43746|issue:46050", "issue:46655|issue:46693", "issue:43381|issue:45704", @@ -5475,214 +5412,214 @@ "issue:45081|issue:45356", "issue:43421|issue:44186" ], - "prepared_review_unit_hash": "9ebdbdaf567810941167259f4509d978e8f77d30404023f54ef8f825cd3a8071", + "prepared_review_unit_hash": "75dbd990140b8466a8fe64927bc52e10281e51a8a9ed0f807616c1020bd61c41", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36801, - "estimated_input_tokens": 9201, - "estimated_eval_tokens": 18658 + "serialized_chars": 36806, + "estimated_input_tokens": 9202, + "estimated_eval_tokens": 18660 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous, with two clear duplicate subgroups: Qwen2_5_VLProcessor batched-input crash on padding=False, and HF_HUB_OFFLINE tokenizer-cache failure caused by the Mistral regex patch. The rest look like distinct bugs despite some superficial title similarity.", - "confidence": 0.8, - "canonical_issue_reason": "No single issue cleanly represents the whole heterogeneous cluster; if an anchor is needed, issue 44514 is the best canonical issue because it is the older, more detailed report in an exact-duplicate pair with 44545.", + "summary": "This cluster is mostly a heterogeneous mix of unrelated Transformers bug reports across tokenizers, multimodal processors, model loading, and training/runtime behavior. The only clear duplicate pair is the Qwen2_5_VLProcessor batched apply_chat_template crash with padding=False.", + "confidence": 0.92, + "canonical_issue_reason": "issue:44545 is the best canonical issue: it is an exact duplicate of 44514, has slightly more discussion/inbound signal, and cleanly names the concrete batched-input padding=False crash.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44514 is the strongest overall issue representative because it is part of an exact duplicate pair and has the clearest reproduction/description among the near-identical reports.", + "best_issue_reason": "issue:44545 is the strongest representative of the only true duplicate pair in this set; it captures the specific code-path failure most clearly.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45357", "right": "issue:46402", "accept": false, - "reason": "Different models and different failure modes: visual-encoder key saving vs missing text encoder weights." + "reason": "Both are multimodal save/load failures, but one is about incorrect visual encoder keys and the other about missing text encoder weights; different components and failure modes." }, { "left": "issue:44451", "right": "issue:44991", "accept": false, - "reason": "Both are loading problems, but for different repositories and likely different tokenizer/model issues." + "reason": "Both are tokenizer-loading regressions, but they affect different model repos and appear to stem from different issues." }, { "left": "issue:43644", "right": "issue:46665", "accept": false, - "reason": "Unrelated bugs: non-persistent buffer initialization vs from_pretrained behavior inside Python classes." + "reason": "Non-persistent buffer initialization junk is unrelated to from_pretrained behavior inside Python classes." }, { "left": "issue:44898", "right": "issue:45072", "accept": false, - "reason": "Different subsystems and symptoms: image-size interpolation failure vs dtype mismatch in bfloat16 inference." + "reason": "One is an image-resolution/interpolation bug; the other is a bfloat16 dtype mismatch in inference." }, { "left": "issue:44987", "right": "issue:45042", "accept": false, - "reason": "Not the same bug; one is a model-loading failure, the other is an image-processor dependency regression." + "reason": "A model-loading failure for a specific checkpoint is not the same as an image processor incorrectly requiring torchvision." }, { "left": "issue:44625", "right": "issue:45479", "accept": false, - "reason": "Both involve classification config, but one is config propagation and the other is a zero-loss behavior bug across models." + "reason": "num_labels propagation and single-label classification loss degeneracy are different bugs." }, { "left": "issue:44062", "right": "issue:44451", "accept": false, - "reason": "Tokenizer construction error vs model loading failure; no shared underlying bug." + "reason": "AddedToken keyword conflicts are unrelated to the ScandiBERT loading failure." }, { "left": "issue:43381", "right": "issue:43844", "accept": false, - "reason": "Different training/runtime issues: eval-mode checkpointing restriction vs gradient explosion with ZeRO-3." + "reason": "Eval-mode gradient checkpointing restriction is unrelated to the ZeRO-3 gradient explosion regression." }, { "left": "issue:39692", "right": "issue:44610", "accept": false, - "reason": "Different model families and different failure points; documentation example errors are not the same as processor size mismatch." + "reason": "A documentation/example mismatch is not the same as an image processor producing the wrong resolution." }, { - "left": "issue:44368", - "right": "issue:45357", + "left": "issue:41720", + "right": "issue:45704", "accept": false, - "reason": "A warning about tied embeddings is unrelated to incorrect visual-encoder key saving." + "reason": "CUDA assert under auto device mapping is unrelated to the T5 apex RMSNorm memory leak." }, { - "left": "issue:41720", - "right": "issue:45704", + "left": "issue:44368", + "right": "issue:45357", "accept": false, - "reason": "Completely different bug classes: CUDA assert in Qwen3 device mapping vs T5 apex RMSNorm memory leak." + "reason": "A warning about tie_word_embeddings does not match the Qwen3.5 save_pretrained visual-key bug." }, { "left": "issue:43746", "right": "issue:46050", "accept": false, - "reason": "Different loading paths and root causes: PEFT local checkpoint loading vs quantized save/load breakage." + "reason": "Local PEFT checkpoint loading and sinq quantized save/load breakage are different code paths." }, { "left": "issue:46655", "right": "issue:46693", "accept": false, - "reason": "Video padding calculation bug is unrelated to FA2 long-context performance regression." + "reason": "Video processor padding calculation is unrelated to the FA2 long-context performance regression." }, { "left": "issue:43381", "right": "issue:45704", "accept": false, - "reason": "No shared underlying bug; checkpointing/eval-mode behavior differs from T5 memory-leak behavior." + "reason": "Eval-mode checkpointing behavior and a memory leak in T5 normalization are not the same bug." }, { "left": "issue:43421", "right": "issue:43927", "accept": false, - "reason": "Both involve token/config state, but one is post-processor updates and the other is config save/load losing token IDs." + "reason": "Runtime post-processor updates for special tokens and DiaConfig custom-token-ID loss are distinct tokenizer/config issues." }, { "left": "issue:44918", "right": "issue:44936", "accept": false, - "reason": "Trainer unpacking failure and train-then-evaluate failure are different problems." + "reason": "TRL SFT embedding unpacking and trainer train/evaluate lifecycle failure are different trainer problems." }, { "left": "issue:45704", "right": "issue:45923", "accept": false, - "reason": "Memory leak in T5 RMSNorm is unrelated to a missing supports_gradient_checkpointing flag." + "reason": "The T5 memory leak and the missing gradient_checkpointing flag on a trust_remote_code model are unrelated." }, { "left": "issue:42886", "right": "issue:44843", - "accept": true, - "reason": "Same underlying offline-mode tokenizer failure: an unconditional Hub metadata lookup in the Mistral regex patch breaks cache-only loading." + "accept": false, + "reason": "Both involve offline mode, but one is cache loading in general while the other is an unconditional model_info() call in _patch_mistral_regex." }, { "left": "issue:45759", "right": "issue:45923", "accept": false, - "reason": "Different composite-config behavior vs a model capability flag missing on trust_remote_code variant." + "reason": "Composite Qwen config unwrapping is unrelated to a missing gradient_checkpointing capability flag." }, { "left": "issue:43746", "right": "issue:43873", "accept": false, - "reason": "PEFT local-checkpoint loading and quantization offloading behavior are not the same concrete bug." + "reason": "PEFT checkpoint loading and quantization offloading are different save/load and inference behaviors." }, { "left": "issue:44514", "right": "issue:44545", "accept": true, - "reason": "Near-verbatim duplicate reports of the same Qwen2_5_VLProcessor batched-input crash when padding=False." + "reason": "These are the same crash report: Qwen2_5_VLProcessor.apply_chat_template fails on batched input when padding=False." }, { "left": "issue:43994", "right": "issue:44671", "accept": false, - "reason": "Different model families and symptoms; both are output-quality complaints but not the same bug." + "reason": "Different models and different failure modes: nonsensical SigLIP2 predictions vs CamemBERT masked LM errors." }, { "left": "issue:39692", "right": "issue:43493", "accept": false, - "reason": "Same broad model family, but documentation/example mistakes are distinct from HF-vs-JAX implementation discrepancy." + "reason": "SigLIP2 docs/example errors are not the same as a model-implementation discrepancy versus JAX." }, { "left": "issue:46032", "right": "issue:46463", "accept": false, - "reason": "Different code paths entirely: Mamba2 use_cache correctness vs T5 decoder-input validation messaging." + "reason": "A correctness bug in use_cache for seq_len>1 is unrelated to an error-message issue in T5 decoder input validation." }, { "left": "issue:43421", "right": "issue:43653", "accept": false, - "reason": "Both concern special tokens, but one is runtime post-processor updates and the other is BigBirdTokenizer registration/decode behavior." + "reason": "Special-token post-processor refresh and BigBird mask-token registration are different tokenizer bugs." }, { "left": "issue:30064", "right": "issue:36010", "accept": false, - "reason": "Unrelated issues: image segmentation map processing vs a GenerationMixin import error." + "reason": "Segmentation-map processing and a GenerationMixin import error are unrelated." }, { "left": "issue:43782", "right": "issue:45381", "accept": false, - "reason": "Different Qwen multimodal bugs affecting different components and failure modes." + "reason": "Weight-only load failure for Qwen3VL is unrelated to the qwen2.5-vl vision_position_ids bug." }, { "left": "issue:45081", "right": "issue:45356", "accept": false, - "reason": "Both touch Mistral/Kimi tokenizer handling, but one is a crash and the other is a codec/warning regression; not clearly the same bug." + "reason": "The Mistral regex AttributeError and the Kimi-K2.5 codec/regression warning are related areas but not the same concrete bug." }, { "left": "issue:43421", "right": "issue:44186", "accept": false, - "reason": "Tokenizer special-token updates are unrelated to LayoutLMv2 NER padding/truncation crashes." + "reason": "Runtime special-token update logic is unrelated to LayoutLMv2 NER padding/truncation crashes." } ] }, "evaluator_result": { "accept": true, - "feedback": "Grounded and conservative overall. The two accepted soft pairs are well supported by near-identical titles, and the rejected pairs mostly stay cautious about superficial similarity without overclaiming shared root cause. The canonical-anchor choice for 44514 is also reasonable for this mostly heterogeneous cluster." + "feedback": "The summary is grounded in the packet and stays conservative on the soft-edge calls. The only accepted duplicate pair, issue:44514/issue:44545, is clearly supported by the identical titles and matching crash description, and the rest of the rejections do not overstate similarity." }, "evaluator_used": true, "retried": false, @@ -5735,11 +5672,11 @@ "issue:44877", "issue:44898", "issue:44991", - "issue:44995", "issue:45081", "issue:45084", "issue:45356", "issue:45381", + "issue:45507", "issue:45834", "issue:45907", "issue:45987", @@ -5753,13 +5690,13 @@ "issue:47020" ], "soft_pairs": [ - "issue:44995|issue:46858", + "issue:45507|issue:46858", "issue:45356|issue:45907", "issue:39692|issue:43873", "issue:43720|issue:43819", "issue:43653|issue:43720", - "issue:46829|issue:46832", "issue:43295|issue:45834", + "issue:46829|issue:46832", "issue:45987|issue:46829", "issue:43550|issue:43927", "issue:43550|issue:44361", @@ -5785,220 +5722,220 @@ "issue:39692|issue:45834", "issue:44821|issue:45084" ], - "prepared_review_unit_hash": "00472b22b36cede568104e6f1b5b1ed5e2e29d3a15da456af5e4dad2d3bed660", + "prepared_review_unit_hash": "5d739cea5809e06dd1a38ab8eb9430d8e354ff2b6558464b73ac464b000c1e59", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37303, - "estimated_input_tokens": 9326, - "estimated_eval_tokens": 18908 + "serialized_chars": 37261, + "estimated_input_tokens": 9316, + "estimated_eval_tokens": 18888 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Heterogeneous cluster of issue reports across tokenizers, vision/video, cache, quantization, and training/runtime regressions. I found no true duplicate pair among the soft edges, and there are no PRs in the cluster.", - "confidence": 0.96, - "canonical_issue_reason": "No single canonical issue: the items span unrelated bugs and regressions across many subsystems, so picking one as the cluster representative would be misleading.", + "summary": "No duplicate cluster here: the items are mostly unrelated bug reports spanning tokenizers, vision/image processing, cache semantics, quantization/offloading, and model-specific regressions. All soft edges look like superficial topic overlap rather than the same underlying fix. No PR candidates were present.", + "confidence": 0.92, + "canonical_issue_reason": "No clear canonical duplicate hub exists; the issues are too heterogeneous, and the apparent overlaps are only broad subsystem similarity.", "canonical_pr_reason": null, - "best_issue_reason": "No issue cleanly represents the whole cluster; the closest matches are still distinct bugs, so the safest triage is to avoid choosing a global best issue.", + "best_issue_reason": "If one issue must anchor the set, 44843 is the most generally actionable library regression report, but it still does not represent the rest of the cluster well.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44995", + "left": "issue:45507", "right": "issue:46858", "accept": false, - "reason": "Both involve cache/indexing language, but they hit different models and different code paths." + "reason": "Different bugs: GraniteMoEHybrid invalid-method call vs static-cache seq-length tensor arithmetic on non-CUDA backends." }, { "left": "issue:45356", "right": "issue:45907", "accept": false, - "reason": "Tokenizer regression vs hub API forwarding bug; no shared underlying defect." + "reason": "Unrelated regressions: Kimi tokenizer codec/fix_mistral_regex behavior vs Hub list_repo_files revision forwarding." }, { "left": "issue:39692", "right": "issue:43873", "accept": false, - "reason": "Doc/example errors and quantization offloading behavior are unrelated bugs." + "reason": "Docs/model mismatch and quantization failure are not the same bug as offloading not working with quantization." }, { "left": "issue:43720", "right": "issue:43819", "accept": false, - "reason": "Different models and failures: packed-weight loading vs DAC latent reconstruction." + "reason": "Different model/code paths: BitNet packed-weight loading vs DAC.from_latents missing STE parity." }, { "left": "issue:43653", "right": "issue:43720", "accept": false, - "reason": "Tokenizer special-token decode bug is unrelated to BitNet packed-weight loading." + "reason": "BigBird special-token registration bug is unrelated to BitNet accelerate loading/unpacking." }, { - "left": "issue:46829", - "right": "issue:46832", + "left": "issue:43295", + "right": "issue:45834", "accept": false, - "reason": "Both are vision models, but one is output packaging and the other is a shape/level mismatch in forward." + "reason": "Processor/tokenizer regression is unrelated to Kosmos2.5 long-OCR index errors." }, { - "left": "issue:43295", - "right": "issue:45834", + "left": "issue:46829", + "right": "issue:46832", "accept": false, - "reason": "Processor/tokenizer regression and long OCR input indexing are different problems." + "reason": "Both are vision-model failures, but one is an output-field mixup and the other is an RT-DETR forward TypeError with different triggers." }, { "left": "issue:45987", "right": "issue:46829", "accept": false, - "reason": "Static cache sequence length vs classifier hidden_states are unrelated." + "reason": "StaticCache seq-length contract issue is unrelated to VideoPrism output mapping." }, { "left": "issue:43550", "right": "issue:43927", "accept": false, - "reason": "torch.compile+SDPA issue is unrelated to DiaConfig save/load token ID loss." + "reason": "Different failures: torch.compile/SDPA in Bamba vs DiaConfig losing custom token IDs on save/load." }, { "left": "issue:43550", "right": "issue:44361", "accept": false, - "reason": "Different subsystems and failure modes: compile/attention vs tokenizer attribute error." + "reason": "Bamba torch.compile/SDPA regression is unrelated to MLukeTokenizer task AttributeError." }, { "left": "issue:43540", "right": "issue:46294", "accept": false, - "reason": "Both are Qwen video-input bugs, but they fail in different model families and different validation paths." + "reason": "Both involve video inputs, but the failures are different layers: Qwen3OmniMoe video processing vs chat-template token/feature mismatch." }, { "left": "issue:45381", "right": "issue:46693", "accept": false, - "reason": "Video position-id correctness and FA2 performance regression are not the same bug." + "reason": "Qwen2.5-VL video position IDs bug is unrelated to FA2 long-context performance regression." }, { "left": "issue:43531", "right": "issue:43643", "accept": false, - "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to AutoConfig missing fields." + "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to AutoConfig trust_remote_code missing fields." }, { "left": "issue:44062", "right": "issue:44488", "accept": false, - "reason": "AddedToken construction error and model load failure are distinct tokenizer issues." + "reason": "AddedToken special-arg duplication is a tokenizer construction bug, not the same as loading cjvt/sleng-bert." }, { "left": "issue:43596", "right": "issue:44220", "accept": false, - "reason": "DeepSpeed/BertModel init and audio fbank extraction are unrelated failures." + "reason": "DeepSpeed ZeRO-3/BertModel index error is unrelated to _torch_extract_fbank_features()." }, { "left": "issue:43611", "right": "issue:46665", "accept": false, - "reason": "Both concern loading, but one is base_model_prefix compatibility and the other is class-context from_pretrained behavior." + "reason": "Both concern loading behavior, but base_model_prefix breakage and from_pretrained-in-classes inconsistency are distinct code paths." }, { "left": "issue:44186", "right": "issue:44898", "accept": false, - "reason": "NER/padding crashes in LayoutLMv2Tokenizer are unrelated to Perceiver resolution handling." + "reason": "LayoutLMv2 tokenizer NER/padding issues are unrelated to Perceiver image classification interpolation failures." }, { "left": "issue:43653", "right": "issue:43819", "accept": false, - "reason": "Tokenizer special-token registration and DAC latent mismatch have no shared code path." + "reason": "Tokenizer special-token registration and DAC latent reconstruction mismatch are different bugs." }, { "left": "issue:41720", "right": "issue:44991", "accept": false, - "reason": "Qwen device-map cuda assert and a tokenizer load regression are unrelated." + "reason": "Qwen3 auto-device-map cudaErrorAssert is unrelated to loading EMBEDDIA/est-roberta tokenizers." }, { "left": "issue:30064", "right": "issue:36331", "accept": false, - "reason": "Image processor segmentation-map handling is unrelated to CustomTrainer loss signature changes." + "reason": "Image processor segmentation-map failure is unrelated to CustomTrainer.compute_loss signature changes." }, { "left": "issue:44008", "right": "issue:46858", "accept": false, - "reason": "Gemma3n variable collision and static-cache sequence length are different bugs." + "reason": "Gemma3n variable-collision forward bug is unrelated to static-cache seq-length tensor behavior." }, { "left": "issue:44220", "right": "issue:44661", "accept": false, - "reason": "Audio feature extraction and TOKENIZER_MAPPING_NAMES model registration are not the same defect." + "reason": "Audio fbank extraction and add-new-model-like tokenizer mapping are unrelated subsystems." }, { "left": "issue:45987", "right": "issue:46932", "accept": false, - "reason": "Static cache return type and Gemma4 quantized embedders are unrelated." + "reason": "StaticCache seq-length tensor return is unrelated to Gemma4 quantized vision/audio embedders dtype casting." }, { "left": "issue:43618", "right": "issue:43697", "accept": false, - "reason": "Output field assignment regression and RT-DETR differing outputs are different problems." + "reason": "Missing CLIPOutput attentions and RTDetrV2 output divergence are distinct model-output regressions." }, { "left": "issue:43335", "right": "issue:43756", "accept": false, - "reason": "Sparse-layer config bug and RoPE-layer dropping are unrelated model configuration issues." + "reason": "SwitchTransformers sparse-layer config bug and Smollm3 RoPE-layer mismatch are different model-configuration problems." }, { "left": "issue:44488", "right": "issue:47020", "accept": false, - "reason": "Model load failure and XLMRoberta Unigram constructor type error are separate tokenizer bugs." + "reason": "Model-loading failure for cjvt/sleng-bert is not the same as XLMRobertaTokenizer Unigram dict-vs-sequence initialization." }, { "left": "issue:43643", "right": "issue:44877", "accept": false, - "reason": "Both are config/loading regressions, but they affect different strictness checks and different objects." + "reason": "AutoConfig trust_remote_code missing-fields regression is unrelated to strict config blocking granite_speech loading." }, { "left": "issue:43550", "right": "issue:43819", "accept": false, - "reason": "Compile/SDPA failure and DAC from_latents mismatch do not share a concrete code path." + "reason": "Bamba torch.compile/SDPA failure is unrelated to DAC.from_latents forward mismatch." }, { "left": "issue:44843", "right": "issue:45081", "accept": false, - "reason": "Both touch _patch_mistral_regex, but one is an offline network call and the other is a missing backend_tokenizer assumption." + "reason": "Both touch _patch_mistral_regex, but one is an offline model_info call bug and the other is a backend_tokenizer attribute assumption." }, { "left": "issue:39692", "right": "issue:45834", "accept": false, - "reason": "SigLIP2 doc-example errors are unrelated to Kosmos2.5 long-input indexing." + "reason": "SigLIP2 docs/quantization errors are unrelated to Kosmos2.5 OCR length/index issues." }, { "left": "issue:44821", "right": "issue:45084", "accept": false, - "reason": "AutoImageProcessor URL loading and a template-node compile error are unrelated." + "reason": "AutoImageProcessor URL loading and template-node compilation errors are unrelated." } ] }, @@ -6074,8 +6011,8 @@ "soft_pairs": [ "issue:34634|issue:36010", "issue:44448|issue:44671", - "issue:44186|issue:44442", "issue:43540|issue:44008", + "issue:44186|issue:44442", "issue:45584|issue:45834", "issue:45183|issue:45479", "issue:43322|issue:45685", @@ -6085,221 +6022,218 @@ "issue:43782|issue:44977", "issue:43421|issue:44811", "issue:44479|issue:45834", + "issue:43582|issue:46512", "issue:43550|issue:45072", "issue:43825|issue:43994", - "issue:43582|issue:46512", "issue:43577|issue:43818", - "issue:43388|issue:46032", "issue:44561|issue:45020", + "issue:43388|issue:46032", "issue:44991|issue:46291", "issue:46294|issue:46463", "issue:43716|issue:44008", - "issue:43931|issue:46846", "issue:44805|issue:44936", + "issue:43931|issue:46846", "issue:43540|issue:45381", "issue:44610|issue:45072", "issue:28218|issue:29159" ], - "prepared_review_unit_hash": "b2a9357677ec69b5a6d4f280cb5400f4bb95cb5e0343132ce9be47f56eba818c", + "prepared_review_unit_hash": "e8db899e029a78e4534f6b211d1325eaecad8f1cbb9c2621922839e8a0bca6bf", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 27, - "serialized_chars": 36050, - "estimated_input_tokens": 9013, - "estimated_eval_tokens": 18282 + "serialized_chars": 36042, + "estimated_input_tokens": 9011, + "estimated_eval_tokens": 18278 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a loose topical mix of unrelated regression reports across tokenizers, loading, vision/video, and training. The only plausible duplicate-looking edge is the tokenizer spacing/inconsistency pair; everything else looks like distinct bugs or changes.", - "confidence": 0.77, - "canonical_issue_reason": "issue:29159 is the broadest tokenizer-related report here and best matches the specific added-token spacing symptom in issue:28218.", + "summary": "No soft pair looks like the same concrete bug/change. The set is a loose semantic cluster of tokenizer, loading, dtype, and video-input regressions that mostly share only broad subsystem vocabulary.", + "confidence": 0.91, + "canonical_issue_reason": "Issue 45020 is the broadest, most representative regression report in the set: it describes recent-version breakage for remote_code models, which best matches the cluster\u2019s loose theme without pretending the other reports are duplicates.", "canonical_pr_reason": null, - "best_issue_reason": "issue:29159 is the best representative issue for the one genuine duplicate-like theme in the cluster: tokenizer behavior inconsistency.", + "best_issue_reason": "Issue 45020 is the most general and centrally phrased report among the items, so it is the best umbrella representative of the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:34634", "right": "issue:36010", "accept": false, - "reason": "Different subsystems and failures: BarkProcessor voice_preset vs a GenerationMixin import error." + "reason": "BarkProcessor voice_preset is a model-specific processor bug; GenerationMixin import failure is a separate API/module-path issue." }, { "left": "issue:44448", "right": "issue:44671", "accept": false, - "reason": "Both are v5 regressions, but they affect different models and different behaviors; not the same concrete bug." + "reason": "Different models and symptoms: Pegasus output drift vs CamemBERT masked-LM prediction errors." }, { - "left": "issue:44186", - "right": "issue:44442", + "left": "issue:43540", + "right": "issue:44008", "accept": false, - "reason": "Both mention tokenizers, but one is a LayoutLMv2Tokenizer crash and the other is an AutoTokenizer loading failure for a different model class." + "reason": "Qwen3OmniMoe video-input processing error and Gemma3n audio_mel_mask variable collision are unrelated code paths." }, { - "left": "issue:43540", - "right": "issue:44008", + "left": "issue:44186", + "right": "issue:44442", "accept": false, - "reason": "Different model families and distinct bugs: Qwen3OmniMoe video input handling vs Gemma 3n variable shadowing." + "reason": "LayoutLMv2Tokenizer NER/padding crash is distinct from AutoTokenizer failing to load FastSpeech2ConformerTokenizer." }, { "left": "issue:45584", "right": "issue:45834", "accept": false, - "reason": "Whisper empty-transcription generation vs Kosmos2.5 long-OCR index error are unrelated failure modes." + "reason": "Whisper empty-transcription decoding after align_special_tokens is unrelated to Kosmos2.5 long-OCR index errors." }, { "left": "issue:45183", "right": "issue:45479", "accept": false, - "reason": "One is input validation for input_ids/inputs_embeds; the other is a classification loss/labeling bug." + "reason": "Validation of input_ids/inputs_embeds is a different bug from degenerate zero loss in sequence classification." }, { "left": "issue:43322", "right": "issue:45685", "accept": false, - "reason": "A segfault while loading Llava Next is unrelated to an MPS histogram error in moe code." + "reason": "Llava Next loading segfault and MPS histogram Int error are different failures in different subsystems." }, { "left": "issue:43644", "right": "issue:43749", "accept": false, - "reason": "Both are loading/runtime regressions, but one is non-persistent buffer initialization and the other is FSDP CPU RAM efficient loading." + "reason": "Non-persistent buffer corruption and FSDP CPU RAM efficient loading are both loading-related, but not the same concrete bug." }, { "left": "issue:43606", "right": "issue:45072", "accept": false, - "reason": "Different underlying issues: CPU offload device mismatch in Bark vs dtype mismatch in bfloat16 inference." + "reason": "Bark CPU-offload device mismatch and SwitchTransformers/TimmWrapperModel dtype mismatch are separate mismatch bugs." }, { "left": "issue:44991", "right": "issue:47020", "accept": false, - "reason": "Both are tokenizer loading problems, but they involve different tokenizer classes and different root causes." + "reason": "Tokenizer loading failure for EMBEDDIA/est-roberta is not the same as XLMRobertaTokenizer passing a dict to Unigram." }, { "left": "issue:43782", "right": "issue:44977", "accept": false, - "reason": "Qwen3VL weight_only loading and Qwen3.5 flash-attention generation are separate bugs." + "reason": "Qwen3VL weight-only loading error and Qwen3.5 flash-attention generation failure are different stages and code paths." }, { "left": "issue:43421", "right": "issue:44811", "accept": false, - "reason": "Runtime special-token post-processing updates vs Whisper batch_decode ignoring skip_special_tokens are different code paths." + "reason": "Runtime post-processor updates for special tokens and Whisper batch_decode skip_special_tokens handling are distinct tokenization behaviors." }, { "left": "issue:44479", "right": "issue:45834", "accept": false, - "reason": "Different multimodal models and different video/OCR failure modes." + "reason": "Broad video-input regression across Qwen variants does not match a Kosmos2.5 long-OCR indexing bug." + }, + { + "left": "issue:43582", + "right": "issue:46512", + "accept": false, + "reason": "Apple Silicon warmup TypeError is unrelated to from_config(dtype=...) leaving config.dtype stale." }, { "left": "issue:43550", "right": "issue:45072", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba is unrelated to bfloat16 dtype mismatch issues." + "reason": "Bamba torch.compile SDPA failure and bfloat16 dtype mismatches in other models are not the same bug." }, { "left": "issue:43825", "right": "issue:43994", "accept": false, - "reason": "A bad pipeline error message is not the same as SigLIP2 producing wrong outputs." - }, - { - "left": "issue:43582", - "right": "issue:46512", - "accept": false, - "reason": "Apple Silicon TypeError in warmup and stale config.dtype after from_config are different bugs." + "reason": "pipeline translation error messaging and SigLIP2 nonsensical outputs are different problems." }, { "left": "issue:43577", "right": "issue:43818", "accept": false, - "reason": "BLIP2 dtype retention and Video-LLaVA temporal-attention/weight-sharing issues are unrelated." + "reason": "BLIP2 dtype reporting mismatch and Video-LLaVA missing temporal attention/shared weights are unrelated model bugs." }, { - "left": "issue:43388", - "right": "issue:46032", + "left": "issue:44561", + "right": "issue:45020", "accept": false, - "reason": "Metrics gathering drops labels; Mamba2Mixer use_cache returns incorrect results. Different components and symptoms." + "reason": "Both concern recent-version breakage around remote_code, but one is a specific API removal impact and the other is a broader compatibility report; not the same concrete bug." }, { - "left": "issue:44561", - "right": "issue:45020", + "left": "issue:43388", + "right": "issue:46032", "accept": false, - "reason": "Both concern remote_code regressions, but one is specifically removal of is_torch_fx_available and the other is a broader umbrella report." + "reason": "gather_for_metrics label truncation and Mamba2Mixer use_cache correctness are separate training/runtime issues." }, { "left": "issue:44991", "right": "issue:46291", "accept": false, - "reason": "Both are tokenizer-related, but one is a specific model tokenizer load regression and the other is a kernels package import constraint." + "reason": "Tokenizer loading for a specific model and a kernels dependency/version requirement during pipelines import are different failures." }, { "left": "issue:46294", "right": "issue:46463", "accept": false, - "reason": "Qwen3.5 video/chat-template mismatch and T5 decoder input validation are unrelated." + "reason": "Qwen3.5 video/chat-template mismatch and T5 decoder-input validation errors are unrelated model-family bugs." }, { "left": "issue:43716", "right": "issue:44008", "accept": false, - "reason": "Different model families and different causes: image preprocessor/model dtype mismatch vs variable-name collision." + "reason": "Mistral-3 preprocessor/model dtype mismatch is not the same as Gemma3n forward-pass variable collision." }, { - "left": "issue:43931", - "right": "issue:46846", + "left": "issue:44805", + "right": "issue:44936", "accept": false, - "reason": "Both are Qwen-related load/shape problems, but one is a specific model weight mismatch and the other is a TP-plan omission causing OOM/shape errors." + "reason": "A mask-vs-tensor shape IndexError and trainer.evaluate() failing after trainer.train() are too generic and likely separate causes." }, { - "left": "issue:44805", - "right": "issue:44936", + "left": "issue:43931", + "right": "issue:46846", "accept": false, - "reason": "A tensor-mask shape IndexError is not the same as trainer.evaluate failing after train()." + "reason": "Model weight-shape mismatch on load and missing linear_attn entries in TP plan causing OOM/shape error are different loading/parallelism bugs." }, { "left": "issue:43540", "right": "issue:45381", "accept": false, - "reason": "Video-input processing for Qwen3OmniMoe is unrelated to the generic input_ids/inputs_embeds validation regression." + "reason": "Qwen3OmniMoe video-input processing and Qwen2.5-VL/Qwen3 video_position_ids issues are related only by the broad video domain, not the same bug." }, { "left": "issue:44610", "right": "issue:45072", "accept": false, - "reason": "OmDet-Turbo processor size mismatch and dtype mismatch in inference are different bugs." + "reason": "Processor output size mismatch for OmDet-Turbo and dtype mismatches in unrelated models are not duplicates." }, { "left": "issue:28218", "right": "issue:29159", - "accept": true, - "reason": "Both describe the same tokenizer inconsistency around added-token handling; 29159 is the broader report and 28218 is a specific symptom of it." + "accept": false, + "reason": "Added-token extra-space behavior may be related to tokenizer parity, but 29159 is only a broad slow-vs-fast inconsistency report and not clearly the same concrete bug." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The analysis is grounded in the packet and stays appropriately conservative. The only accepted soft edge (28218 \u2194 29159) is well supported by the titles and the tokenizer-added-token symptom, while the rejected pairs are generally justified as distinct bugs across different models/subsystems. No obvious overstatement or mergeability issue is present." - }, - "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 }, @@ -6377,110 +6311,110 @@ "issue:46697|issue:46728", "issue:45003|issue:46285", "issue:43782|issue:45325", - "issue:43653|issue:44912", "issue:43262|issue:43819", + "issue:43653|issue:44912", "issue:43582|issue:44857", "issue:45372|issue:46326", "issue:44589|issue:45042", "issue:43742|issue:45020", "issue:43550|issue:44912", - "issue:43257|issue:43329", "issue:36010|issue:43645", - "issue:43526|issue:43697", + "issue:43257|issue:43329", "issue:36010|issue:36331", + "issue:43526|issue:43697", "issue:43906|issue:46209", "issue:45081|issue:45406", "issue:43452|issue:44991", "issue:43329|issue:45600" ], - "prepared_review_unit_hash": "862d37f0139edb8ef255d58f8b282c58059c213a1463f27dc20cb669d2826a13", + "prepared_review_unit_hash": "05bdffb7ca6b07d377cd82d6f5010a4d51de4197b7ee36b6d9c3592195ad5c70", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 27, - "serialized_chars": 35650, - "estimated_input_tokens": 8913, - "estimated_eval_tokens": 18082 + "serialized_chars": 35660, + "estimated_input_tokens": 8915, + "estimated_eval_tokens": 18086 }, "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 many unrelated regressions across loading, tokenization, vision/audio processors, quantization, and trainer APIs. None of the soft-edge pairs looks like the same concrete bug/change, so I would keep them separate. If one issue must anchor the set, #45020 is the broadest umbrella regression report.", - "confidence": 0.82, - "canonical_issue_reason": "#45020 is the broadest, most umbrella-style regression report in an otherwise heterogeneous set, so it is the best canonical anchor if one must be chosen.", + "summary": "This cluster is highly heterogeneous: the soft pairs are mostly unrelated model-, tokenizer-, loading-, or processor-specific regressions, not true duplicates. I would not merge any of the candidate edges.", + "confidence": 0.91, + "canonical_issue_reason": "Issue 45020 is the best representative umbrella item because it describes the broadest recurring theme in the set: recent Transformers versions breaking model loading/behavior. The other issues are much narrower and model-specific.", "canonical_pr_reason": null, - "best_issue_reason": "#45020 has the widest scope and is the most likely to represent a central regression theme rather than a narrow edge case.", + "best_issue_reason": "Issue 45020 is the strongest overall representative for the cluster because it captures a broad, user-facing v5 regression pattern and is more general than the other reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43749", "right": "issue:44291", "accept": false, - "reason": "Both concern model loading, but one is FSDP CPU RAM efficient loading and the other is an init_empty_weights/_is_hf_initialized signature error; different failure points." + "reason": "Both are loading-related, but one is FSDP CPU RAM-efficient loading and the other is an init_empty_weights/_is_hf_initialized TypeError. Different code paths and root causes." }, { "left": "issue:43720", "right": "issue:45072", "accept": false, - "reason": "Accelerate loading of packed BitNet weights is unrelated to a bfloat16 inference dtype mismatch in SwitchTransformers/TimmWrapperModel." + "reason": "Different models and failures: BitNet packed-weight loading vs dtype mismatch in SwitchTransformers/TimmWrapperModel inference. Not the same bug." }, { "left": "issue:43645", "right": "issue:46665", "accept": false, - "reason": "Both mention custom model instantiation, but one is a Transformers 5.0 Jupyter notebook breakage and the other is from_pretrained behavior inside Python classes; not the same bug." + "reason": "Both mention custom usage patterns, but one is about defining/initializing custom models in Jupyter notebooks and the other is from_pretrained inside Python classes. Too different to be the same issue." }, { "left": "issue:43825", "right": "issue:43827", "accept": false, - "reason": "One is a runtime error message about translation tasks, the other is stale docs referencing pipeline(); same topic, different change." + "reason": "One is a runtime error message about translation tasks in pipeline(), the other is docs still referencing pipeline() after removals. Related area, but not the same underlying bug/change." }, { "left": "issue:44671", "right": "issue:45701", "accept": false, - "reason": "CamemBERT masked-LM prediction regression and general tokenization changes are not the same code-path issue." + "reason": "CamemBERT masked-LM prediction regressions and general tokenization changes are different symptoms with different likely causes." }, { "left": "issue:43577", "right": "issue:43582", "accept": false, - "reason": "BLIP2 dtype propagation and an Apple Silicon caching_allocator_warmup TypeError are unrelated." + "reason": "Blip2 dtype propagation and Apple Silicon caching_allocator_warmup TypeError are unrelated." }, { "left": "issue:34634", "right": "issue:36331", "accept": false, - "reason": "BarkProcessor voice_preset and CustomTrainer.compute_loss() signature handling are completely different subsystems." + "reason": "BarkProcessor voice_preset and CustomTrainer.compute_loss keyword handling are completely separate bugs." }, { "left": "issue:42175", "right": "issue:45081", "accept": false, - "reason": "Package extra/backend composition is unrelated to a Mistral tokenizer regex patch crash." + "reason": "Package backend dependency selection and a Mistral tokenizer regex patch crash are unrelated." }, { "left": "issue:46175", "right": "issue:46932", "accept": false, - "reason": "Both are Gemma4-related, but one is a huge vision patch-embedder memory issue and the other is quantized vision/audio dtype casting." + "reason": "Gemma4VisionPatchEmbedder one-hot materialization and Gemma4 quantized embedder dtype casting are different issues in different parts of the model stack." }, { "left": "issue:46697", "right": "issue:46728", "accept": false, - "reason": "Swin2SR over-padding and Aria patch-count logic are different image-processor bugs." + "reason": "Both are image-processing bugs, but they affect different models and different computations: Swin2SR padding vs Aria patch counting." }, { "left": "issue:45003", @@ -6492,97 +6426,97 @@ "left": "issue:43782", "right": "issue:45325", "accept": false, - "reason": "Qwen3VL weight_only loading failure and Qwen2.5-VL rope-index scaling are distinct VL model bugs." + "reason": "Qwen3VL weight_only loading failure and Qwen2.5-VL rope index scaling are different models and different code paths." }, { - "left": "issue:43653", - "right": "issue:44912", + "left": "issue:43262", + "right": "issue:43819", "accept": false, - "reason": "BigBird tokenizer special-token registration and BitNet packed-weight loading are not the same problem." + "reason": "Audio chat-template sampling-rate defaults and DAC.from_latents STE mismatch are separate audio-processing bugs." }, { - "left": "issue:43262", - "right": "issue:43819", + "left": "issue:43653", + "right": "issue:44912", "accept": false, - "reason": "Audio chat-template sampling-rate defaults and DAC latent reconstruction/STE mismatch are unrelated." + "reason": "BigBirdTokenizer special-token registration and MXFP4 quantization fallback in git-oss-20b are unrelated." }, { "left": "issue:43582", "right": "issue:44857", "accept": false, - "reason": "Apple Silicon allocator warmup and CUDA AMP loss crashes are different platform-specific failures." + "reason": "Apple Silicon allocator warmup TypeError and LwDetrImageLoss AMP/CUDA crash are different runtime failures." }, { "left": "issue:45372", "right": "issue:46326", "accept": false, - "reason": "A mistral_common import regression in Gemma4 processor loading is not the same as continue_final_message behavior." + "reason": "Gemma4 processor import failure from mistral_common and continue_final_message behavior for mistral_common tokenizers are not the same bug." }, { "left": "issue:44589", "right": "issue:45042", "accept": false, - "reason": "Float8 storage lookup failures and torchvision dependency requirements are unrelated." + "reason": "Float8 storage lookup failure and PIL image processors requiring torchvision are unrelated." }, { "left": "issue:43742", "right": "issue:45020", "accept": false, - "reason": "A MobileLLM key error and a broad remote_code regression are different issues." + "reason": "KeyError loading a specific model and broad remote_code breakage are different failures." }, { "left": "issue:43550", "right": "issue:44912", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba is unrelated to MXFP4 quantization fallback in git-oss-20b." + "reason": "torch.compile/SDPA failure in Bamba-9B-v2 is unrelated to MXFP4 quantization fallback in git-oss-20b." + }, + { + "left": "issue:36010", + "right": "issue:43645", + "accept": false, + "reason": "GenerationMixin import failure and custom-model Jupyter notebook initialization are unrelated." }, { "left": "issue:43257", "right": "issue:43329", "accept": false, - "reason": "MOE weight conversion during accelerate+deepspeed loading has nothing to do with undefined video token helpers." + "reason": "Qwen3 MoE weight conversion during accelerate/deepspeed loading and a multimodal token counting video-branch NameError are different bugs." }, { "left": "issue:36010", - "right": "issue:43645", + "right": "issue:36331", "accept": false, - "reason": "GenerationMixin import failure and notebook custom-model initialization are different loading/import bugs." + "reason": "ImportError for GenerationMixin and CustomTrainer.compute_loss keyword mismatch are separate API regressions." }, { "left": "issue:43526", "right": "issue:43697", "accept": false, - "reason": "BEiT label reduction and RTDetrV2 output differences are separate model/processor regressions." - }, - { - "left": "issue:36010", - "right": "issue:36331", - "accept": false, - "reason": "An import error from transformers.generation is unrelated to a Trainer compute_loss keyword mismatch." + "reason": "BeitImageProcessorFast reduce_labels behavior and RTDetrV2 output differences are unrelated." }, { "left": "issue:43906", "right": "issue:46209", "accept": false, - "reason": "An isolated reproduction of an older issue and an FP8 inference failure on B200 are not the same underlying bug." + "reason": "A reproduction of an older issue and FP8 inference failure on B200 are not the same bug." }, { "left": "issue:45081", "right": "issue:45406", "accept": false, - "reason": "Mistral regex patching and Gemma4Processor missing _tokenizer are different tokenizer/processor failures." + "reason": "Mistral tokenizer regex patch crash and Gemma4Processor missing _tokenizer in transformers serve are different processor/tokenizer issues." }, { "left": "issue:43452", "right": "issue:44991", "accept": false, - "reason": "Generic gguf_file loading breakage and a specific tokenizer load failure for one model are not the same issue." + "reason": "gguf_file loading regressions and EMBEDDIA/est-roberta tokenizer loading failures are different tokenizer/model-load problems." }, { "left": "issue:43329", "right": "issue:45600", "accept": false, - "reason": "Undefined multimodal token helpers and stale auto_mappings references are separate maintenance bugs." + "reason": "Undefined multimodal token helper names and removed Sam3LiteText auto-mapping references are unrelated CI/runtime issues." } ] }, @@ -6658,15 +6592,15 @@ "soft_pairs": [ "issue:44060|issue:44368", "issue:43526|issue:44554", - "issue:45084|issue:45538", "issue:43819|issue:43927", - "issue:43638|issue:44936", "issue:43531|issue:45042", + "issue:43638|issue:44936", + "issue:45084|issue:45538", "issue:42175|issue:46291", - "issue:42490|issue:43257", "issue:43746|issue:44811", - "issue:43643|issue:45759", + "issue:42490|issue:43257", "issue:43854|issue:44977", + "issue:43643|issue:45759", "issue:39692|issue:44912", "issue:38175|issue:44610", "issue:43720|issue:44610", @@ -6685,202 +6619,202 @@ "issue:42915|issue:44821", "issue:29159|issue:34567" ], - "prepared_review_unit_hash": "c0dd3fb47ac9d0114abef03b19ccf2373ce88a1616d5df35eea7b627a36ea171", + "prepared_review_unit_hash": "c82d672d74af73fcf5673cd0978c94dbbdf0ce3caa59e06dc3d1848b2ce27d5c", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36285, - "estimated_input_tokens": 9072, - "estimated_eval_tokens": 18400 + "serialized_chars": 36284, + "estimated_input_tokens": 9071, + "estimated_eval_tokens": 18398 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The items are broadly heterogeneous; none of the soft pairs look like true duplicates or mergeable PR-equivalents. Most are same-subsystem false positives (tokenizers, cache, Qwen, processors, training) but with different concrete bugs.", - "confidence": 0.71, - "canonical_issue_reason": "Issue 46421 is the best cluster anchor: it is open, well-scoped, and concrete, with a closely adjacent cache-related report (46424), even though the pair is not a duplicate.", + "summary": "These items are a heterogeneous set of unrelated issues; none of the proposed soft pairs look like the same underlying bug or change, so I rejected all duplicate links.", + "confidence": 0.92, + "canonical_issue_reason": "No single canonical issue stands out because the set spans unrelated bugs across tokenizers, training, multimodal processing, quantization, and model loading.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 46421 is the strongest standalone issue here: it is current, actionable, and describes a specific regression path rather than a broad symptom report.", + "best_issue_reason": "Issue 45042 is the clearest standalone report to anchor on if one must be chosen: it is specific, reproducible, and has external references, but it is not a duplicate hub for the rest of the set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44060", "right": "issue:44368", "accept": false, - "reason": "Both mention tied-weights warnings in Qwen models, but one is a mistaken tie across `dt_bias` and the other is a config warning during LoRA fine-tuning; different underlying bugs." + "reason": "Both mention tied-weights warnings, but they concern different Qwen variants and different causes; not the same fix." }, { "left": "issue:43526", "right": "issue:44554", "accept": false, - "reason": "BeitImageProcessorFast label reduction is unrelated to the MPS attention correctness bug; no shared code path." + "reason": "One is a BeitImageProcessorFast label-reduction bug; the other is an MPS attention correctness issue." }, { - "left": "issue:45084", - "right": "issue:45538", + "left": "issue:43819", + "right": "issue:43927", "accept": false, - "reason": "`torch.compile` template-node failure and CLIPTokenizer max_length behavior are unrelated tokenizer/compiler issues." + "reason": "Different subsystems and failures: DAC latent reconstruction vs DiaConfig save/load token ID corruption." }, { - "left": "issue:43819", - "right": "issue:43927", + "left": "issue:43531", + "right": "issue:45042", "accept": false, - "reason": "DAC latent/STE mismatch and DiaConfig token-ID persistence are different model/config bugs." + "reason": "Qwen3-MoE sliding-window handling and PIL-backend torchvision dependency are unrelated bugs." }, { "left": "issue:43638", "right": "issue:44936", "accept": false, - "reason": "A DeepSpeed ZeRO-3 empty-index error during training is not the same as evaluate-after-train state corruption." + "reason": "Different training paths: a Zero3/non-pretrained Bert IndexError versus evaluate() failing after train()." }, { - "left": "issue:43531", - "right": "issue:45042", + "left": "issue:45084", + "right": "issue:45538", "accept": false, - "reason": "Qwen3-MoE sliding-window logic and PIL backend torchvision dependency are unrelated." + "reason": "Compilation/template-node error and CLIPTokenizer max_length bug are distinct code paths." }, { "left": "issue:42175", "right": "issue:46291", "accept": false, - "reason": "Backend extras missing TensorFlow and pipelines import failing on kernels version/revision requirements are separate packaging/import issues." - }, - { - "left": "issue:42490", - "right": "issue:43257", - "accept": false, - "reason": "Trailing-slash local model path warning and Qwen3 MOE weight conversion with accelerate/deepspeed are unrelated loading problems." + "reason": "Packaging/backend install issue is unrelated to kernels import-time dependency validation." }, { "left": "issue:43746", "right": "issue:44811", "accept": false, - "reason": "PEFT adapter checkpoint loading and Whisper batch_decode skip_special_tokens handling do not share the same code path." + "reason": "Local checkpoint loading for PEFT adapters is unrelated to Whisper batch_decode skip_special_tokens handling." }, { - "left": "issue:43643", - "right": "issue:45759", + "left": "issue:42490", + "right": "issue:43257", "accept": false, - "reason": "`trust_remote_code` losing config fields and `from_config` not unwrapping `text_config` are different config-construction bugs." + "reason": "A trailing-slash path warning is not the same bug as Qwen3 MOE weight conversion with accelerate+deepspeed." }, { "left": "issue:43854", "right": "issue:44977", "accept": false, - "reason": "GLM-4.7-Flash load/test failure and Qwen3.5 flash-attention generation failure are separate model-specific regressions." + "reason": "GLM-4.7-Flash test loading failure and Qwen3.5 flash-attention generation regression are different model-specific issues." + }, + { + "left": "issue:43643", + "right": "issue:45759", + "accept": false, + "reason": "AutoConfig trust_remote_code field loss and AutoModelForCausalLM text_config unwrapping are separate config-loading bugs." }, { "left": "issue:39692", "right": "issue:44912", "accept": false, - "reason": "SigLIP2 docs example errors and gpt-oss MXFP4 fallback are unrelated." + "reason": "SigLIP2 doc-example mistakes are not the same as GPT-oss MXFP4 quantization falling back to bf16." }, { "left": "issue:38175", "right": "issue:44610", "accept": false, - "reason": "Zero probabilities in SigLIP2 and OmDet-Turbo 640px-vs-224px input sizing are different symptoms and code paths." + "reason": "Unexpected zero probabilities in SigLIP2 is unrelated to OmDet-Turbo's input-size mismatch." }, { "left": "issue:43720", "right": "issue:44610", "accept": false, - "reason": "BitNet packed-weight loading and OmDet-Turbo processor shape mismatch are unrelated bugs." + "reason": "BitNet packed-weight loading and OmDet-Turbo processor sizing are different failure modes." }, { "left": "issue:43262", "right": "issue:44811", "accept": false, - "reason": "Audio `apply_chat_template` sample-rate defaulting and Whisper `batch_decode` special-token handling are different processor behaviors." + "reason": "Audio chat-template sampling-rate default and Whisper batch_decode token skipping are different processor behaviors." }, { "left": "issue:43257", "right": "issue:43696", "accept": false, - "reason": "Qwen3 MOE weight conversion under accelerate/deepspeed and GPT-oss OOM are not the same failure mode." + "reason": "MOE weight conversion under distributed loading is unrelated to a GPT-oss CUDA OOM report." }, { "left": "issue:44387", "right": "issue:44617", "accept": false, - "reason": "Both are OOM-related, but one is a Transformers 5.x int4 reserved-memory regression and the other is a Sam3Video model memory demand issue." + "reason": "Quantization-related CUDA reserved-memory regression is not the same as Sam3Video's model-specific OOM." }, { "left": "issue:43329", "right": "issue:43450", "accept": false, - "reason": "Undefined video-token helpers in multimodal token counting and batched video processor shape errors are distinct video-processing bugs." + "reason": "Undefined video-token-count helpers and batched video output shape are separate multimodal processor bugs." }, { "left": "issue:43257", "right": "issue:43931", "accept": false, - "reason": "Accelerate/deepspeed Qwen3 MOE conversion and Qwen3-VL weight-shape mismatch are different loading failures." + "reason": "Qwen3 MOE conversion/loading and Qwen3-VL weight-shape mismatch are different model-loading problems." }, { "left": "issue:43329", "right": "issue:44479", "accept": false, - "reason": "The undefined-helper bug in multimodal token counting is not the same as the v5.3.0 video input regression across Qwen VL variants." + "reason": "The undefined video-branch helper bug is not the same as the v5.3.0 video regression across multiple model families." }, { "left": "issue:42915", "right": "issue:43742", "accept": false, - "reason": "FineGrainedFP8Config failure in Qwen3 MOE and MobileLLM key error are unrelated model-loading bugs." + "reason": "FineGrainedFP8Config failure for Qwen3Moe and a MobileLLM key error are unrelated." }, { "left": "issue:34567", "right": "issue:40990", "accept": false, - "reason": "TrainerState token counter not updating and GPT-oss perplexity being too high are unrelated training vs evaluation symptoms." + "reason": "TrainerState token counter update and GPT-oss perplexity are not the same bug." }, { "left": "issue:45459", "right": "issue:45803", "accept": false, - "reason": "Both involve exception handling hiding errors, but they are in different components and different failure scenarios; not the same bug." + "reason": "Both concern exception handling, but they affect different functions and different error paths, so they are not duplicates." }, { "left": "issue:44291", "right": "issue:45042", "accept": false, - "reason": "`init_empty_weights`/_is_hf_initialized argument handling and PIL backend torchvision dependency are unrelated." + "reason": "init_empty_weights argument incompatibility is unrelated to the PIL-backend torchvision dependency regression." }, { "left": "issue:46421", "right": "issue:46424", "accept": false, - "reason": "Both are cache-related, but one is lazy-init causing spurious recompilation and the other is `max_cache_length` only applying to Quantized Cache; separate fixes." + "reason": "Chunked-prefill recompilation and max_cache_length only applying to quantized cache are distinct cache issues." }, { "left": "issue:43650", "right": "issue:43906", "accept": false, - "reason": "A placeholder issue titled 'ADD THE DATA' is not a duplicate of an isolated reproduction report for another issue." + "reason": "The placeholder 'ADD THE DATA' issue is not the same as an isolated reproduction of issue 38071." }, { "left": "issue:42915", "right": "issue:44821", "accept": false, - "reason": "Qwen3 MOE FP8 loading and AutoImageProcessor-from-URL loading are different loader paths." + "reason": "Qwen3Moe FP8 loading and AutoImageProcessor URL loading are unrelated." }, { "left": "issue:29159", "right": "issue:34567", "accept": false, - "reason": "Tokenizer slow/fast inconsistency and TrainerState token-count tracking are unrelated subsystems." + "reason": "Tokenizer slow/fast inconsistency and TrainerState token counting are different subsystems and bugs." } ] }, @@ -6904,8 +6838,10 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-506", "nodes": [ + "issue:24643", "issue:28282", "issue:29127", + "issue:29159", "issue:29870", "issue:33357", "issue:34567", @@ -6926,7 +6862,6 @@ "issue:43749", "issue:43824", "issue:43927", - "issue:44062", "issue:44291", "issue:44479", "issue:44568", @@ -6935,7 +6870,6 @@ "issue:44912", "issue:44936", "issue:44987", - "issue:44991", "issue:45081", "issue:45127", "issue:45137", @@ -6954,8 +6888,8 @@ "issue:47020" ], "soft_pairs": [ - "issue:43749|issue:44756", "issue:43606|issue:45859", + "issue:43749|issue:44756", "issue:43540|issue:46701", "issue:28282|issue:29870", "issue:43454|issue:43577", @@ -6973,212 +6907,212 @@ "issue:42175|issue:43824", "issue:28282|issue:29127", "issue:45538|issue:46869", - "issue:42915|issue:45538", "issue:45081|issue:47020", + "issue:42915|issue:45538", "issue:44479|issue:46693", "issue:43742|issue:44291", "issue:44936|issue:46821", "issue:43452|issue:43582", - "issue:42915|issue:43749", - "issue:44062|issue:44991", - "issue:43577|issue:46512" + "issue:43577|issue:46512", + "issue:24643|issue:29159", + "issue:42915|issue:43749" ], - "prepared_review_unit_hash": "e672abb40d4e33a69f5be37362a0fd75148b7b5c4c9d7ecd6ffbb0e4ab320702", + "prepared_review_unit_hash": "03e40d38477441ae49ac8409571db61d3fb68b70b08f46b8693640e9ff2b7a8a", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36331, - "estimated_input_tokens": 9083, - "estimated_eval_tokens": 18422 + "serialized_chars": 36318, + "estimated_input_tokens": 9080, + "estimated_eval_tokens": 18416 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These issues are loosely related by model/loading regressions and tokenizer/config edge cases, but they describe different models, failure modes, and code paths. None of the soft pairs look like true duplicates or the same fix target.", - "confidence": 0.28, - "canonical_issue_reason": "Issue 43749 has the strongest cross-cutting signal in this set (high discussion and inbound references) and is a concrete runtime-loading failure, but the cluster is too heterogeneous for a true single canonical duplicate.", + "summary": "The set is heterogeneous: most items are distinct model-loading, tokenizer, dtype, or training bugs rather than duplicates. I rejected all soft edges because they do not look like the same underlying bug/change.", + "confidence": 0.92, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue 43749 is the best representative among these issues because it is a clearly described, high-impact loading bug with substantial discussion and references.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43749", - "right": "issue:44756", + "left": "issue:43606", + "right": "issue:45859", "accept": false, - "reason": "Different bugs: FSDP CPU RAM-efficient loading vs Strix Halo mmap OOM. Related to loading, but not the same underlying defect or fix." + "reason": "Both mention CPU offload, but they affect different models/code paths and different failures; not the same bug." }, { - "left": "issue:43606", - "right": "issue:45859", + "left": "issue:43749", + "right": "issue:44756", "accept": false, - "reason": "Device mismatch in CPU offload for a specific model vs Qwen3.5Moe rotary embedding incompatibility with CPU offload. Same broad area, different code paths." + "reason": "One is FSDP CPU RAM efficient loading, the other is mmap/OOM on Strix Halo; separate runtime issues." }, { "left": "issue:43540", "right": "issue:46701", "accept": false, - "reason": "Video input processing error in Qwen3OmniMoe vs a pipeline/output integration issue for DiffusionGemmaGenerationOutput. Not the same bug." + "reason": "Both are multimodal/pipeline failures, but on different models and distinct error conditions." }, { "left": "issue:28282", "right": "issue:29870", "accept": false, - "reason": "PyTorch import error vs BART generate() decoder_attention_mask unbound-local error. Completely different failure modes." + "reason": "ImportError about missing PyTorch is unrelated to Bart generate() unbound local error." }, { "left": "issue:43454", "right": "issue:43577", "accept": false, - "reason": "Weight-tying/generation bug in AyaVision vs BLIP2 dtype propagation issue. Both involve model loading, but the defects are unrelated." + "reason": "Both involve model loading, but one is untied lm_head weights and the other is dtype propagation in BLIP2; different bugs." }, { "left": "issue:43525", "right": "issue:46291", "accept": false, - "reason": "Llama4Config pad_token_id attribute error vs transformers.pipelines import failure due to kernels version/revision requirements. Different subsystems and causes." + "reason": "Missing pad_token_id in a config is unrelated to pipelines import failing on kernels version/revision checks." }, { "left": "issue:43927", "right": "issue:44912", "accept": false, - "reason": "Custom token IDs lost on save/load vs MXFP4 quantization fallback to bf16. Both are model config-related, but not the same bug." + "reason": "DiaConfig save/load token ID loss is a different problem from MXFP4 quantization fallback behavior." }, { "left": "issue:28282", "right": "issue:34689", "accept": false, - "reason": "Missing PyTorch dependency import error vs Llama 3.2 Vision model loading breakage in 4.46.2. Distinct import/loading issues." + "reason": "Both are loading-related, but one is missing PyTorch and the other is a specific regression loading Llama 3.2 vision." }, { "left": "issue:44568", "right": "issue:45356", "accept": false, - "reason": "Tokenizer special-token behavior regression vs Kimi-K2.5 codec/mistral-regex regression. Similar area, different concrete defect." + "reason": "Both concern tokenizers, but the affected models and the underlying regression are different." }, { "left": "issue:44987", "right": "issue:45538", "accept": false, - "reason": "Model loading regression for a specific model vs CLIPTokenizer model_max_length bug. Different tokenizer/model behavior." + "reason": "Model loading failure for one repo is not the same as CLIPTokenizer's model_max_length bug." }, { "left": "issue:33357", "right": "issue:34567", "accept": false, - "reason": "MacOS bus error with community CLIP model vs TrainerState token-count tracking not updating. No shared root cause." + "reason": "A MacOS CLIP bus error and TrainerState token counter not updating are unrelated." }, { "left": "issue:43425", "right": "issue:46291", "accept": false, - "reason": "Torch version incompatibility vs kernels import/version requirement failure. Both dependency-related, but not the same issue." + "reason": "Torch compatibility and pipeline import dependency resolution are separate issues." }, { "left": "issue:44857", "right": "issue:46832", "accept": false, - "reason": "Float16 AMP CUDA crash in LwDetrImageLoss vs RT-DETR forward TypeError for feature-level mismatch. Different model components and triggers." + "reason": "Different DETR-family crashes with different triggers and code paths." }, { "left": "issue:43697", "right": "issue:45357", "accept": false, - "reason": "RTDetrV2 output mismatch in v5 vs Qwen3.5 save_pretrained wrong visual encoder keys. Both regressions, but unrelated code paths." + "reason": "RT-DETR output mismatch and Qwen3.5 save_pretrained key mapping are unrelated." }, { "left": "issue:43404", "right": "issue:45127", "accept": false, - "reason": "Mistral3 lm_head tying bug vs LoRA merge with extended vocabulary collapse. Both concern embeddings/weights, but not the same bug." + "reason": "Both touch tied embeddings/lm_head, but one is a model init bug and the other is LoRA merge with extended vocab; not the same concrete bug." }, { "left": "issue:45137", "right": "issue:45405", "accept": false, - "reason": "DeepSpeed ZeRO3 deque error vs PEFT minimum-version bump. No underlying bug overlap." + "reason": "DeepSpeed ZeRO3 deque error and an unreleased PEFT version bump are unrelated." }, { "left": "issue:42175", "right": "issue:43824", "accept": false, - "reason": "Tensorflow missing from torch install extra vs import failure for Qwen2.5 VL class. Different packaging/import problems." + "reason": "Backend packaging issue and missing Qwen2.5-VL import are different failure modes." }, { "left": "issue:28282", "right": "issue:29127", "accept": false, - "reason": "PyTorch missing import error vs LayoutLMv3 insufficient-box-info error message. Both are user-facing errors, but not the same defect." + "reason": "Missing PyTorch import error is unrelated to LayoutLMv3 box-validation messaging." }, { "left": "issue:45538", "right": "issue:46869", "accept": false, - "reason": "CLIPTokenizer max length bug vs YOLO multi-GPU device-map failure. Different model families and failure modes." + "reason": "CLIP tokenizer max_length bug and YOLO multi-GPU device-map failure are different problems." }, { - "left": "issue:42915", - "right": "issue:45538", + "left": "issue:45081", + "right": "issue:47020", "accept": false, - "reason": "FineGrainedFP8Config failure in Qwen3Moe vs CLIPTokenizer max length issue. Unrelated features." + "reason": "These are both tokenizer bugs, but with different tokenizer classes and distinct failure mechanisms." }, { - "left": "issue:45081", - "right": "issue:47020", + "left": "issue:42915", + "right": "issue:45538", "accept": false, - "reason": "Mistral regex patch crash on tokenizer object vs XLMRobertaTokenizer passing a dict to Unigram. Both tokenizer-loading bugs, but distinct root causes." + "reason": "FineGrainedFP8 quantization failure is unrelated to CLIPTokenizer's max_length sentinel issue." }, { "left": "issue:44479", "right": "issue:46693", "accept": false, - "reason": "Video-input regression for Qwen VL variants vs performance regression with FA2 and long-context generation. Same broad release area, different symptoms and fixes." + "reason": "A video-input regression and a long-context FA2 performance regression are different bugs." }, { "left": "issue:43742", "right": "issue:44291", "accept": false, - "reason": "Key error loading MobileLLM-125M vs init_empty_weights TypeError from unexpected _is_hf_initialized arg. Both are load-time errors, but not the same issue." + "reason": "KeyError loading one model and init_empty_weights argument mismatch in another are distinct loading paths." }, { "left": "issue:44936", "right": "issue:46821", "accept": false, - "reason": "trainer.evaluate() after train() failure vs TimesFM window_size AttributeError. Different training/evaluation paths." + "reason": "Trainer evaluate/train state bug and TimesFM window_size AttributeError are unrelated." }, { "left": "issue:43452", "right": "issue:43582", "accept": false, - "reason": "gguf_file breaks tokenizer/model loading vs Apple Silicon caching_allocator_warmup TypeError. Not the same underlying bug." + "reason": "gguf_file loading issues and Apple Silicon allocator warmup TypeError are separate bugs." }, { - "left": "issue:42915", - "right": "issue:43749", + "left": "issue:43577", + "right": "issue:46512", "accept": false, - "reason": "Qwen3Moe FP8 config failure vs FSDP CPU RAM-efficient loading broken. Both touch advanced loading, but different mechanisms and model paths." + "reason": "Both are dtype-related, but one is incorrect loaded tensor dtypes and the other is stale config.dtype after from_config." }, { - "left": "issue:44062", - "right": "issue:44991", + "left": "issue:24643", + "right": "issue:29159", "accept": false, - "reason": "AddedToken special-argument TypeError vs tokenizer loading failure for EMBEDDIA/est-roberta. Different tokenizer construction bugs." + "reason": "DeepSpeed training weight-shape error and slow/fast tokenizer inconsistency are unrelated." }, { - "left": "issue:43577", - "right": "issue:46512", + "left": "issue:42915", + "right": "issue:43749", "accept": false, - "reason": "BLIP2 dtype mismatch on load vs from_config dtype leaving config stale. Related to dtype handling, but different observable bug and fix scope." + "reason": "FP8 config failure and FSDP CPU RAM efficient loading are different features and failure modes." } ] }, @@ -7202,10 +7136,8 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-506", "nodes": [ - "issue:24643", "issue:27957", "issue:28282", - "issue:29159", "issue:34567", "issue:36331", "issue:38175", @@ -7219,6 +7151,7 @@ "issue:43716", "issue:43742", "issue:43756", + "issue:43906", "issue:44062", "issue:44291", "issue:44479", @@ -7233,6 +7166,7 @@ "issue:44898", "issue:44964", "issue:44987", + "issue:44991", "issue:45042", "issue:45072", "issue:45081", @@ -7247,20 +7181,21 @@ "issue:46399", "issue:46463", "issue:46491", + "issue:46693", "issue:46701", "issue:46821" ], "soft_pairs": [ - "issue:24643|issue:29159", + "issue:44062|issue:44991", "issue:44849|issue:45701", "issue:27957|issue:28282", "issue:41720|issue:43540", "issue:43540|issue:44560", "issue:44987|issue:45907", - "issue:44479|issue:45381", "issue:42175|issue:44488", - "issue:34567|issue:36331", + "issue:44479|issue:45381", "issue:43540|issue:44479", + "issue:34567|issue:36331", "issue:38175|issue:44898", "issue:44792|issue:44964", "issue:44568|issue:45488", @@ -7278,210 +7213,217 @@ "issue:44737|issue:46463", "issue:43577|issue:45072", "issue:44568|issue:46491", - "issue:45823|issue:46463" + "issue:45823|issue:46463", + "issue:43906|issue:46693" ], - "prepared_review_unit_hash": "3ec37c98f2918cf4940b7d3b68e594925baf4f4bc7a6fb2bc90ce18c0459d2d3", + "prepared_review_unit_hash": "fb861cc3c8bbf6cb2a6a5e5a723da098c061a23e634d3769783d1888c8932d94", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 28, - "serialized_chars": 35622, - "estimated_input_tokens": 8906, - "estimated_eval_tokens": 18068 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36512, + "estimated_input_tokens": 9128, + "estimated_eval_tokens": 18512 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Most pairs are unrelated Transformer bugs affecting different models or code paths. Only the Qwen video-position pair looks like the same underlying regression; the rest should stay separate.", - "confidence": 0.79, - "canonical_issue_reason": "issue:44479 is the best umbrella report for the shared Qwen video-input regression, and issue:45381 reads like a narrower manifestation of the same bug.", + "summary": "This set is mostly heterogeneous; there is no strong cluster-wide duplicate theme. The only likely true duplicate pair is the Qwen2.5-VL video-position-id regression, where one report is the broader umbrella and the other is a focused reproduction.", + "confidence": 0.61, + "canonical_issue_reason": "issue:44479 is the best available anchor because it is the broader, more general report for the one clear duplicate subcluster (Qwen video-input regression); the rest of the items are too diverse to form a single coherent canonical issue.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44479 is the clearest representative because it describes the broader failing code path across Qwen video models, while 45381 is model-specific.", + "best_issue_reason": "issue:44479 is the strongest issue to keep as the representative report for the small Qwen VL video regression cluster, but it is not a good canonical for the full set because most other items are unrelated bugs.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:24643", - "right": "issue:29159", + "left": "issue:44062", + "right": "issue:44991", "accept": false, - "reason": "Different failures: DeepSpeed training weight-shape error vs tokenizer inconsistency. No shared bug path." + "reason": "Both are tokenizer-related, but one is an AddedToken keyword collision and the other is a tokenizer-loading failure for a specific model; different bugs." }, { "left": "issue:44849", "right": "issue:45701", "accept": false, - "reason": "One is a Qwen3.5 hidden-states bug; the other is a broad tokenization-version report. Too different in scope and symptom." + "reason": "Different symptoms and code paths: Qwen3.5 hidden-state output bug vs a broad tokenization regression report." }, { "left": "issue:27957", "right": "issue:28282", "accept": false, - "reason": "XLMRoberta flash-attention issue and missing PyTorch dependency are unrelated." + "reason": "Unrelated failures: FlashAttention2 support for XLMRoberta vs PyTorch import detection in AutoModel." }, { "left": "issue:41720", "right": "issue:43540", "accept": false, - "reason": "Different Qwen models and different failure modes: device-mapping assert vs video-input value error." + "reason": "Different model families and failures: cudaErrorAssert during auto device mapping vs a video-input ValueError in Qwen3OmniMoe." }, { "left": "issue:43540", "right": "issue:44560", "accept": false, - "reason": "Both mention video inputs, but they affect different Qwen model families and appear to hit different code paths." + "reason": "Both mention video inputs, but the affected models and error paths differ; no evidence of the same underlying bug." }, { "left": "issue:44987", "right": "issue:45907", "accept": false, - "reason": "Unrelated: model loading regression vs repo file listing not forwarding revision." + "reason": "One is a model loading regression, the other is a hub API forwarding bug; unrelated." + }, + { + "left": "issue:42175", + "right": "issue:44488", + "accept": false, + "reason": "Packaging/dependency issue vs a model-loading failure for cjvt/sleng-bert; not the same problem." }, { "left": "issue:44479", "right": "issue:45381", "accept": true, - "reason": "Both point to the same Qwen2.5-VL/Qwen video-position regression in transformers 5.3.0; 45381 looks like a specific instance of the broader bug in 44479." + "reason": "Same Qwen2.5-VL video-position-id regression family in v5.3.0; one is the broader regression report and the other is a focused description of the same underlying bug." }, { - "left": "issue:42175", - "right": "issue:44488", + "left": "issue:43540", + "right": "issue:44479", "accept": false, - "reason": "Package backend dependency problem vs a specific model-loading failure; no common underlying bug." + "reason": "Both are video-related Qwen issues, but they target different models and different failures; not mergeable as one bug." }, { "left": "issue:34567", "right": "issue:36331", "accept": false, - "reason": "Trainer state accounting bug and custom compute_loss signature breakage are distinct trainer issues." - }, - { - "left": "issue:43540", - "right": "issue:44479", - "accept": false, - "reason": "Different model families and symptoms; not enough evidence they share the same concrete bug." + "reason": "TrainerState token-count tracking vs a CustomTrainer compute_loss signature error are separate training bugs." }, { "left": "issue:38175", "right": "issue:44898", "accept": false, - "reason": "Different vision-model failures with no clear shared code path." + "reason": "Both are image-classification-related, but one is zero probabilities on SigLIP2 and the other is a Perceiver non-default-resolution failure; distinct root causes." }, { "left": "issue:44792", "right": "issue:44964", "accept": false, - "reason": "Janus test failure and Phi-4 multimodal model load failure are unrelated." + "reason": "Janus image-generation test failure vs Phi-4 multimodal loading failure; same broad area, different issues." }, { "left": "issue:44568", "right": "issue:45488", "accept": false, - "reason": "Both are tokenizer regressions, but they involve different tokenizer implementations and different concrete breakages." + "reason": "Tokenizer regressions, but one concerns add_special_tokens behavior and the other overrides the pre-tokenizer; not the same bug." }, { "left": "issue:44291", "right": "issue:44821", "accept": false, - "reason": "Init-empty-weights argument handling and AutoImageProcessor URL loading are unrelated." + "reason": "Unexpected _is_hf_initialized argument during init_empty_weights vs loading AutoImageProcessor from a URL; unrelated." }, { "left": "issue:43257", "right": "issue:43756", "accept": false, - "reason": "MoE weight conversion under accelerate/deepspeed and SmolLM3 RoPE-layer truncation are unrelated." + "reason": "Qwen3 MOE weight conversion under accelerate/deepspeed vs Smollm3 RoPE-layer count mismatch; different models and fixes." }, { "left": "issue:44062", "right": "issue:45081", "accept": false, - "reason": "Different tokenizer crashes with different root causes; not the same bug." + "reason": "Both involve tokenizer internals, but one is an AddedToken duplicate-kwarg error and the other is a Mistral regex-patching attribute error." }, { "left": "issue:43582", "right": "issue:46821", "accept": false, - "reason": "Apple Silicon allocator TypeError and TimesFM window_size AttributeError are unrelated." + "reason": "Apple Silicon warmup TypeError vs TimesFM window_size AttributeError; unrelated backend/model bugs." }, { "left": "issue:45042", "right": "issue:45081", "accept": false, - "reason": "Image-processor torchvision dependency regression vs Mistral tokenizer regex crash." + "reason": "Image-processor dependency requirements vs Mistral tokenizer regex patching; completely different code paths." }, { "left": "issue:43742", "right": "issue:44589", "accept": false, - "reason": "MobileLLM key error and missing Float8 storage type are different loading failures." + "reason": "A key-error while loading MobileLLM vs a Float8 storage lookup TypeError; different failures." }, { "left": "issue:43756", "right": "issue:46399", "accept": false, - "reason": "SmolLM3 RoPE-layer issue and Qwen3.5 kernel registration error are unrelated." + "reason": "Both mention rotary/rope terms, but Smollm3 RoPE-layer dropping and Qwen3.5 kernel registration are unrelated bugs." }, { "left": "issue:43756", "right": "issue:45325", "accept": false, - "reason": "Different models and different rotary-position bugs." + "reason": "Both are RoPE-related, but they affect different model families and different logic paths; not the same underlying issue." }, { "left": "issue:43716", "right": "issue:46701", "accept": false, - "reason": "Mistral-3 dtype mismatch and DiffusionGemma pipeline failure are unrelated." + "reason": "Image-preprocessor dtype mismatch vs a DiffusionGemma pipeline failure; distinct problems." }, { "left": "issue:43531", "right": "issue:44821", "accept": false, - "reason": "Qwen3-MoE sliding-window bug vs AutoImageProcessor URL loading; no shared underlying change." + "reason": "Sliding-window bug in Qwen3-MoE vs AutoImageProcessor URL loading failure; unrelated." }, { "left": "issue:45405", "right": "issue:45600", "accept": false, - "reason": "Version pin problem and stale auto-mapping reference are separate CI issues." + "reason": "Dependency-version bump not yet on PyPI vs stale auto_mappings references; one is packaging, the other is CI cleanup." }, { "left": "issue:44737", "right": "issue:46463", "accept": false, - "reason": "CPU device placement bug in XLNet vs T5 validation error message; unrelated." + "reason": "XLNet CPU placement bug in relative position encoding vs an incorrect T5 error message; different areas." }, { "left": "issue:43577", "right": "issue:45072", "accept": false, - "reason": "Both are dtype-related, but they affect different model stacks and different forward paths." + "reason": "Both are dtype-related, but they affect different models and different mismatches; not one concrete bug." }, { "left": "issue:44568", "right": "issue:46491", "accept": false, - "reason": "Separate tokenizer regressions for different models; not the same concrete bug." + "reason": "Tokenizer behavior regressions, but one is special-token insertion and the other is encode/decode leading-space round-trip; separate issues." }, { "left": "issue:45823", "right": "issue:46463", "accept": false, - "reason": "Gemma4 device-mapping mismatch and T5 decoder-input validation message are unrelated." + "reason": "Gemma4 device-map mismatch during forward vs a T5 decoder-input validation message bug; unrelated." + }, + { + "left": "issue:43906", + "right": "issue:46693", + "accept": false, + "reason": "An isolated reproduction of one prior issue vs an FA2 long-context performance regression; no shared bug." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary stays grounded in the packet and the soft-edge calls are conservative. The one accepted pair, issue:44479 and issue:45381, is plausibly the same Qwen video-position regression in v5.3.0; the rest are reasonably kept separate." + "feedback": "The summary is grounded in the packet and stays conservative. The heterogeneous-cluster conclusion is supported by the issue titles, and the one accepted pair (44479/45381) is a reasonable duplicate/near-duplicate within the Qwen2.5-VL video regression subcluster. No soft-edge verdict appears to overstate evidence." }, "evaluator_used": true, "retried": false, @@ -7552,7 +7494,6 @@ "issue:46899" ], "soft_pairs": [ - "issue:43906|issue:46693", "issue:45672|issue:45674", "issue:24643|issue:29870", "issue:43906|issue:46468", @@ -7572,204 +7513,198 @@ "issue:44466|issue:45005", "issue:45200|issue:46463", "issue:43643|issue:45375", - "issue:41762|issue:44936", "issue:44493|issue:46050", + "issue:41762|issue:44936", "issue:43688|issue:43749", "issue:44361|issue:44442", "issue:38175|issue:43295", "issue:43653|issue:44361", "issue:43452|issue:44843" ], - "prepared_review_unit_hash": "b1340e65a086def0fb614a73d5c2f527beaea725096904d91aa7936bfbe0d120", + "prepared_review_unit_hash": "0dc7f04f8bb14aa216de80a131d6a95bfc82a3c1c91b821af2e3fa851eb1f4b3", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 27, - "serialized_chars": 36429, - "estimated_input_tokens": 9108, - "estimated_eval_tokens": 18472 + "soft_pair_count": 26, + "serialized_chars": 36123, + "estimated_input_tokens": 9031, + "estimated_eval_tokens": 18318 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a heterogeneous set of distinct Transformers bug reports across tokenizers, configs, quantization, and model-specific inference paths. Only the Qwen3.5 save_pretrained regression pair looks like a real duplicate; the rest are too different to collapse safely.", - "confidence": 0.74, - "canonical_issue_reason": "Issue 45216 is the best anchor: it reports the core Qwen3.5 save_pretrained checkpoint-serialization regression, and 45357 reads like a later confirmation that the same visual-encoder key bug still exists in 5.5.3.", + "summary": "Most pairs are clearly different bug reports across unrelated models or code paths. The only plausible duplicate pair is the Qwen3.5 `save_pretrained` regression reports (45216/45357); everything else should stay separate.", + "confidence": 0.62, + "canonical_issue_reason": "Issue 45216 is the earliest, broadest report of the Qwen3.5 `save_pretrained` regression; 45357 reads like a later, narrower follow-up on the same serialization path.", "canonical_pr_reason": null, - "best_issue_reason": "45216 is the clearest root report for the Qwen3.5 save_pretrained corruption, with 45357 as a follow-up manifestation of the same problem.", + "best_issue_reason": "Issue 45216 is the best representative issue here because it captures the core checkpoint corruption problem with the broadest framing and strongest standalone signal.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43906", - "right": "issue:46693", - "accept": false, - "reason": "One is an isolated reproduction of #38071; the other is a FA2 long-context performance regression report. Same broad area, but not the same concrete bug." - }, { "left": "issue:45672", "right": "issue:45674", "accept": false, - "reason": "Both are Gemma4 4-bit inference bugs, but one is a uint8/torch.finfo dtype issue and the other is llm_int8_skip_modules clearing the lm_head exclusion. Different code paths." + "reason": "Both involve 4-bit inference, but the root causes differ: unsafe dtype casting in Gemma4 audio modules vs. BitsAndBytes skip-module handling." }, { "left": "issue:24643", "right": "issue:29870", "accept": false, - "reason": "DeepSpeed training weight-shape error vs Bart generate decoder_attention_mask UnboundLocalError; unrelated failures." + "reason": "DeepSpeed weight-shape error during training is unrelated to Bart generate() using `decoder_attention_mask` before assignment." }, { "left": "issue:43906", "right": "issue:46468", "accept": false, - "reason": "A reproduction of another issue does not match the official Nemotron example breakage; no concrete shared bug is visible." + "reason": "An isolated reproduction of an old issue is not the same as a broken Nemotron example; no shared concrete bug is evident." }, { "left": "issue:44625", "right": "issue:44877", "accept": false, - "reason": "Qwen3.5 num_labels propagation vs granite_speech strict config loading; similar config theme, but different defects." + "reason": "Different config-loading failures on different models: `num_labels` propagation in Qwen3.5 vs strict config blocking `granite_speech`." }, { "left": "issue:43540", "right": "issue:46693", "accept": false, - "reason": "Qwen3OmniMoe video-processing ValueError vs FA2 long-context generation slowdown/regression; different problems." + "reason": "Different multimodal/video-generation failures and different likely code paths; only broad feature similarity." }, { "left": "issue:43452", "right": "issue:44488", "accept": false, - "reason": "Both concern loading, but gguf_file breakage and cjvt/sleng-bert load failure appear to be different loader issues." + "reason": "Both are load failures, but one is about `gguf_file` handling for tokenizer/model loading and the other is a specific model that currently won't load." }, { "left": "issue:45663", "right": "issue:46846", "accept": false, - "reason": "Gemma4 FSDP2 shared_kv_states KeyError and Qwen3.5 TP-plan missing linear_attn entries are unrelated model-specific bugs." + "reason": "Both involve large-model parallelism, but one is an FSDP2 `shared_kv_states` bug and the other is missing `linear_attn` entries in tensor-parallel planning." }, { "left": "issue:43697", "right": "issue:46829", "accept": false, - "reason": "RTDetrV2 output drift and VideoPrism output field mismatch are different model-output bugs." + "reason": "Different models and symptoms: RTDetrV2 output drift in v5 vs. VideoPrism returning the wrong tensor in `hidden_states`." }, { "left": "issue:28282", "right": "issue:32090", "accept": false, - "reason": "Missing PyTorch import error vs Trainer _gpu_broadcast_one NoneType error; no shared root cause." + "reason": "One is a missing PyTorch import environment error; the other is a Trainer broadcast failure on `NoneType`." }, { "left": "issue:44737", "right": "issue:45588", "accept": false, - "reason": "XLNet CPU arange performance bug and flash_attention s_aux=None crash are unrelated." + "reason": "XLNet CPU-device placement bug and flash-attention `s_aux=None` crash are unrelated code paths." }, { "left": "issue:46463", "right": "issue:46899", "accept": false, - "reason": "T5 decoder-input validation messaging issue vs Gemma4 4-bit dtype-cast failure; different code paths." + "reason": "T5 decoder-input validation messaging is unrelated to Gemma4 4-bit dtype-cast failures." }, { "left": "issue:44913", "right": "issue:46121", "accept": false, - "reason": "GPTNeoX rotary_pct reload regression and rope validation TypeError are both config-related but not the same bug." + "reason": "GPTNeoX config reload behavior is unrelated to rope-validation JSON list type handling." }, { "left": "issue:45357", "right": "issue:45920", "accept": false, - "reason": "Qwen3.5 save_pretrained key corruption and tokenizer token-ID regressions are separate failures." + "reason": "Different regressions in different APIs: Qwen3.5 `save_pretrained` vs. broad AutoTokenizer token-ID errors across many models." }, { "left": "issue:43526", "right": "issue:45479", "accept": false, - "reason": "BeitImageProcessorFast reduce_labels and sequence-classification zero-loss with num_labels=1 are unrelated." + "reason": "`reduce_labels` truncation in BeitImageProcessorFast is not the same as zero-loss behavior for `num_labels=1` classification." }, { "left": "issue:45200", "right": "issue:46463", "accept": false, - "reason": "Gemma4 mm_token_type_ids defaulting and T5 decoder-input validation are unrelated." + "reason": "Gemma 4 missing `mm_token_type_ids` defaults is unrelated to T5 decoder-input validation messaging." }, { "left": "issue:43643", "right": "issue:45375", "accept": false, - "reason": "Generic AutoConfig trust_remote_code field loss and Qwen3_5MoeVisionConfig strict-dropping a specific field are not the same defect." + "reason": "Both mention missing fields, but they affect different configs/models and different strictness behaviors." }, { - "left": "issue:41762", - "right": "issue:44936", + "left": "issue:44493", + "right": "issue:46050", "accept": false, - "reason": "ZeRO-3 loading IndexError and trainer.evaluate() after train failure are different runtime issues." + "reason": "Unexpected position-id keys and sinq quantization save/load breakage are different issues." }, { - "left": "issue:44493", - "right": "issue:46050", + "left": "issue:41762", + "right": "issue:44936", "accept": false, - "reason": "Unexpected position-id key warnings vs sinq quantized save/load breakage are different bugs." + "reason": "Gemma3 ZeRO-3 loading failure is not the same as `trainer.evaluate()` breaking after `trainer.train()`." }, { "left": "issue:43688", "right": "issue:43749", "accept": false, - "reason": "Auxiliary-loss normalization and FSDP CPU-RAM-efficient-loading regressions are unrelated." + "reason": "Aux-loss normalization in MoE models and `FSDP_CPU_RAM_EFFICIENT_LOADING` are unrelated." }, { "left": "issue:44361", "right": "issue:44442", "accept": false, - "reason": "MLukeTokenizer task AttributeError and FastSpeech2ConformerTokenizer loading failure are different tokenizers and different failure modes." + "reason": "Different tokenizer-loading failures for different models; same broad symptom, not the same bug." }, { "left": "issue:38175", "right": "issue:43295", "accept": false, - "reason": "siglip2 zero-probability outputs and processor.tokenizer/image-passing regression are separate issues." + "reason": "Siglip2 zero probabilities and the v4.57.5 processor/tokenizer regression are distinct multimodal issues." }, { "left": "issue:43653", "right": "issue:44361", "accept": false, - "reason": "BigBird special-token registration bug and MLukeTokenizer AttributeError are both tokenizer-related but not the same underlying bug." + "reason": "BigBird special-token registration/decode bug is unrelated to MLukeTokenizer task failures." }, { - "left": "issue:43720", - "right": "issue:44898", + "left": "issue:43452", + "right": "issue:44843", "accept": false, - "reason": "BitNet accelerate-loading packed-weight bug and Perceiver interpolate_pos_encoding failure are unrelated." + "reason": "Both touch tokenizer loading, but one is a `gguf_file` loader break and the other is an offline-mode regression from an unconditional hub query." }, { "left": "issue:44466", "right": "issue:45005", "accept": false, - "reason": "Both mention tied weights in v5, but one is lm_head serialization varying by device and the other is a broader translation-model report; not enough evidence of one concrete bug." + "reason": "Both mention tied weights, but one is device-dependent `lm_head.weight` serialization while the other is a broader translation-model tied-weights issue." + }, + { + "left": "issue:45672", + "right": "issue:46899", + "accept": false, + "reason": "Both concern Gemma4 and quantization, but one is audio/vision embedders failing on `torch.finfo(uint8)` and the other is a different unsafe cast path." }, { "left": "issue:45216", "right": "issue:45357", "accept": true, - "reason": "Both report the same Qwen3.5 save_pretrained regression with incorrect checkpoint keys; 45357 is a later version confirming the same breakage." - }, - { - "left": "issue:43452", - "right": "issue:44843", - "accept": false, - "reason": "gguf_file loading breakage and unconditional model_info() offline-mode breakage are different AutoTokenizer issues." + "reason": "Same Qwen3.5 `save_pretrained` regression: the second report is a later, narrower manifestation of the same incorrect checkpoint serialization path." } ] }, @@ -7853,8 +7788,8 @@ "issue:43295|issue:45356", "issue:45542|issue:46178", "issue:29159|issue:33357", - "issue:43653|issue:44568", "issue:43688|issue:45704", + "issue:43653|issue:44568", "issue:43688|issue:45701", "issue:43646|issue:43749", "issue:43452|issue:43577", @@ -7862,26 +7797,26 @@ "issue:44464|issue:44743", "issue:42898|issue:46491", "issue:44843|issue:44991", - "issue:43653|issue:43927", "issue:43761|issue:47013", + "issue:43653|issue:43927", "issue:44479|issue:44625", - "issue:45491|issue:46002", - "issue:44821|issue:46821" + "issue:44821|issue:46821", + "issue:45491|issue:46002" ], - "prepared_review_unit_hash": "0a06ef862182cb40cdce6000080a00a198cb50056ce8c269bc12078a3677c63a", + "prepared_review_unit_hash": "03ee525e486dfac96a03486cd2d2bd1f92fc6ebb1c0def1fd05515a65a6e108e", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 25, - "serialized_chars": 34886, + "serialized_chars": 34888, "estimated_input_tokens": 8722, "estimated_eval_tokens": 17700 }, @@ -7890,162 +7825,162 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is highly heterogeneous, with many model- or component-specific regressions. None of the proposed soft pairs look like the same underlying bug/change closely enough to merge, so all soft edges are rejected. The best representative issue is the broad tokenizer-regression report, though it still only reflects one slice of the cluster.", - "confidence": 0.91, - "canonical_issue_reason": "Issue 45701 is the broadest tokenizer-related regression report and has the strongest umbrella-like framing among the issues, so it is the best single issue to represent the tokenizer-heavy portion of this mixed set.", + "summary": "All soft links look like false positives: the paired issues usually share only a broad theme (tokenizers, regressions, backend/import errors, or model-specific failures) but not the same concrete bug or code path.", + "confidence": 0.9, + "canonical_issue_reason": "issue:43646 is the broadest, most general regression title in the set, but the cluster is heterogeneous so it is only a weak umbrella candidate.", "canonical_pr_reason": null, - "best_issue_reason": "45701 has the most general title and broader community signal, making it the best available issue candidate despite the cluster spanning many unrelated bugs.", + "best_issue_reason": "issue:43646 is the closest thing to a cluster representative because it covers a broad Transformers 5.0 initialization regression, but the set is too mixed for a true duplicate canonical.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43819", "right": "issue:44811", "accept": false, - "reason": "Different code paths and symptoms: DAC latent reconstruction mismatch vs Whisper batch_decode special-token handling." + "reason": "Different subsystems and failures: DAC latent reconstruction mismatch vs Whisper batch_decode ignoring skip_special_tokens." }, { "left": "issue:43994", "right": "issue:44898", "accept": false, - "reason": "Both are vision-model bugs, but they affect different models and failure modes: SigLIP2 pipeline output quality vs Perceiver resolution/interpolation handling." + "reason": "Different models and root causes: SigLIP2 pipeline outputs vs Perceiver non-default resolution interpolation." }, { "left": "issue:43761", "right": "issue:46829", "accept": false, - "reason": "Both mention hidden_states, but one is CLIPVisionModel omitting them and the other is VideoPrism returning the wrong tensor in an output field; not the same bug." + "reason": "Both involve hidden-state outputs, but they are different models with different output bugs and code paths." }, { "left": "issue:45372", "right": "issue:46170", "accept": false, - "reason": "Unrelated dependency/import failures: Gemma 4 processor loading via mistral_common vs Windows ROCm torch.distributed import issues." + "reason": "Unrelated import failures: Gemma 4 processor loading vs Windows ROCm torch.distributed dependency issue." }, { "left": "issue:44857", "right": "issue:47000", "accept": false, - "reason": "Different detection-loss failures with different causes: AMP/CUDA crash vs infeasible cost matrix." + "reason": "Different DETR losses and distinct crash causes: float16 AMP overflow vs infeasible cost matrix." }, { "left": "issue:46832", "right": "issue:46932", "accept": false, - "reason": "RT-DETR backbone feature-level mismatch and Gemma4 quantized embedder dtype casting are separate model-specific issues." + "reason": "Different models and problems: RT-DETR feature-level mismatch vs Gemma4 quantized dtype casting." }, { "left": "issue:43525", "right": "issue:46701", "accept": false, - "reason": "Llama4Config pad_token_id omission and DiffusionGemma image-text pipeline failure are unrelated." + "reason": "DiaConfig pad_token_id loss and DiffusionGemma pipeline failure are different bugs in different components." }, { "left": "issue:43531", "right": "issue:43742", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and MobileLLM key-error loading are different bugs in different models." + "reason": "Qwen3-MoE sliding_window behavior and MobileLLM-125M load key error are unrelated." }, { "left": "issue:43295", "right": "issue:45356", "accept": false, - "reason": "Both are tokenizer regressions, but they are model-specific and described as different failures: processor.tokenizer/image passing vs Kimi-K2.5 codec/regex warning behavior." + "reason": "Both are regressions, but one is processor.tokenizer/image handling and the other is Kimi-K2.5 codec/regex handling." }, { "left": "issue:45542", "right": "issue:46178", "accept": false, - "reason": "Both concern backend registry errors, but the offending backends and triggering conditions differ, so this is not the same underlying bug." + "reason": "Both touch backend handling, but one is undefined tf backend setup and the other is an invalid tensorflow_text backend entry." }, { "left": "issue:29159", "right": "issue:33357", "accept": false, - "reason": "Tokenizer inconsistency vs MacOS bus error on CLIP model loading are unrelated." + "reason": "Tokenizer inconsistency vs MacOS bus error on a CLIP community model are not the same bug." }, { - "left": "issue:43653", - "right": "issue:44568", + "left": "issue:43688", + "right": "issue:45704", "accept": false, - "reason": "Both involve special tokens, but one is BigBird mask-token registration and the other is mDeBERTa BOS/EOS insertion; different tokenizers and code paths." + "reason": "Auxiliary loss normalization and T5 apex.FusedRMSNorm memory leak are unrelated model bugs." }, { - "left": "issue:43688", - "right": "issue:45704", + "left": "issue:43653", + "right": "issue:44568", "accept": false, - "reason": "Auxiliary-loss normalization and apex.FusedRMSNorm memory leak are unrelated." + "reason": "Both concern special tokens, but they affect different tokenizers and different symptoms." }, { "left": "issue:43688", "right": "issue:45701", "accept": false, - "reason": "Auxiliary-loss normalization and generic tokenization behavior changes are different subsystems and bugs." + "reason": "Aux loss scaling and version-dependent tokenization changes do not share a concrete code path." }, { "left": "issue:43646", "right": "issue:43749", "accept": false, - "reason": "Custom model initialization regression and FSDP CPU RAM-efficient loading breakage are separate loading pathways." + "reason": "Custom model initialization regression vs FSDP CPU RAM efficient loading are different features." }, { "left": "issue:43452", "right": "issue:43577", "accept": false, - "reason": "gguf_file loading breakage and BLIP2 dtype propagation are unrelated model-loading issues." + "reason": "GGUF loading breaks and BLIP2 dtype staying float32 are unrelated issues." }, { "left": "issue:43756", "right": "issue:46402", "accept": false, - "reason": "Smollm3 RoPE-layer dropping and Sam3Model text-encoder weights not loading are distinct model implementation problems." + "reason": "SmolLM3 RoPE-layer mismatch and Sam3 text-encoder weight loading are different model-specific bugs." }, { "left": "issue:44464", "right": "issue:44743", "accept": false, - "reason": "Chunked generation with compiled forward and recurrent-state reset with cache are different generation bugs." + "reason": "Chunked generation with compiled forward and xLSTM state unpacking are separate state-handling failures." }, { "left": "issue:42898", "right": "issue:46491", "accept": false, - "reason": "Both are tokenizer regressions, but one is clean_up_tokenization_spaces semantics and the other is CodeLlama leading-space round-trip behavior; not the same issue." + "reason": "Both are tokenizer regressions, but one is clean_up_tokenization_spaces behavior and the other is CodeLlama leading-space round-trip loss." }, { "left": "issue:44843", "right": "issue:44991", "accept": false, - "reason": "Offline mistral_regex model_info() access and tokenizer loading failure for est-roberta are different root causes." - }, - { - "left": "issue:43653", - "right": "issue:43927", - "accept": false, - "reason": "Special-token registration bug and DiaConfig custom token-id loss after save/load are separate defects." + "reason": "Offline model_info() patching and est-roberta tokenizer loading failure are different tokenizer-loading regressions." }, { "left": "issue:43761", "right": "issue:47013", "accept": false, - "reason": "CLIPVision hidden_states regression and xLSTM state unpacking crash are unrelated." + "reason": "CLIPVisionModel hidden_states regression and xLSTM return_last_states unpacking bug are unrelated models and code paths." }, { - "left": "issue:44479", - "right": "issue:44625", + "left": "issue:43653", + "right": "issue:43927", "accept": false, - "reason": "Video-input regression in Qwen VL models and Qwen3.5 num_labels propagation are different bugs." + "reason": "Tokenizer special-token registration and DiaConfig custom token ID persistence are different bug classes." }, { - "left": "issue:45491", - "right": "issue:46002", + "left": "issue:44479", + "right": "issue:44625", "accept": false, - "reason": "Both mention NaNs, but the affected models and failure mechanisms differ; this is too broad to merge." + "reason": "Video-input regression in Qwen models vs num_labels propagation in Qwen3.5 are distinct issues." }, { "left": "issue:44821", "right": "issue:46821", "accept": false, "reason": "AutoImageProcessor URL loading and TimesFM window_size AttributeError are unrelated." + }, + { + "left": "issue:45491", + "right": "issue:46002", + "accept": false, + "reason": "Both mention NaNs, but the triggering conditions and affected modules are different." } ] }, @@ -8123,17 +8058,17 @@ "issue:44484|issue:44704", "issue:32090|issue:33666", "issue:44466|issue:46050", - "issue:43688|issue:45440", "issue:44464|issue:44977", + "issue:43688|issue:45440", "issue:44977|issue:46693", - "issue:44734|issue:46612", "issue:41720|issue:44464", + "issue:44734|issue:46612", "issue:43504|issue:43550", "issue:43504|issue:43720", "issue:43278|issue:43688", - "issue:43577|issue:44554", - "issue:43782|issue:46693", "issue:33666|issue:36331", + "issue:43782|issue:46693", + "issue:43577|issue:44554", "issue:43611|issue:46846", "issue:43262|issue:43295", "issue:43720|issue:43927", @@ -8151,208 +8086,220 @@ "issue:45823|issue:46050", "issue:28218|issue:33357" ], - "prepared_review_unit_hash": "7617b13b1708cc1379c29bac2a47bb143bdf2734ef0ac3aa2f7ae2f107469210", + "prepared_review_unit_hash": "0fe1ff3493c45be1cbebefdab3623215c715eb922c882ebce8431a39dbfac8b4", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37105, - "estimated_input_tokens": 9277, - "estimated_eval_tokens": 18810 + "serialized_chars": 37100, + "estimated_input_tokens": 9275, + "estimated_eval_tokens": 18806 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a set of false-positive similarities, not a duplicate set: the issues span unrelated bugs, models, APIs, and regressions. None of the soft pairs look like the same underlying problem.", - "confidence": 0.98, - "canonical_issue_reason": "No single canonical issue: the cluster does not share one underlying bug or change, so there is no safe duplicate anchor.", + "summary": "The cluster is heterogeneous: the soft-similarity pairs mostly share broad Transformers themes (generation, processors, configs, loading) but not the same concrete bug. I would not merge any of the soft edges.", + "confidence": 0.86, + "canonical_issue_reason": "issue:46693 is the best representative anchor because it has the highest discussion/inbound activity and is a broad, user-visible generation regression, but the surrounding items are not true duplicates of it.", "canonical_pr_reason": null, - "best_issue_reason": "No issue is a strong global representative for duplicate triage because the items cover unrelated failure modes and subsystems.", + "best_issue_reason": "issue:46693 is the strongest overall triage anchor for this cluster: it is active, recent, and central enough to represent the general generation/attention-regression theme better than the rest.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:33666", "right": "issue:36010", "accept": false, - "reason": "Multi-GPU training for Qwen2-VL is unrelated to a GenerationMixin import error." + "reason": "Different failures: multi-GPU training support vs an import error for GenerationMixin." }, { "left": "issue:44484", "right": "issue:44704", "accept": false, - "reason": "save_pretrained shard-size behavior is unrelated to AutoProcessor not forwarding kwargs to cached_file." + "reason": "One is about save_pretrained sharding size, the other about AutoProcessor dropping kwargs to cached_file." }, { "left": "issue:32090", "right": "issue:33666", "accept": false, - "reason": "Trainer GPU broadcast NoneType error is a different bug from Qwen2-VL multi-GPU training." + "reason": "Unsupported NoneType broadcast in Trainer is unrelated to Qwen2-VL multi-GPU training." }, { "left": "issue:44466", "right": "issue:46050", "accept": false, - "reason": "lm_head.weight serialization depends on device placement, while sinq breaks after save/load for quantized models." + "reason": "Both involve model serialization, but one is tied-weight serialization and the other is sinq quantization reload breakage." }, { - "left": "issue:43688", - "right": "issue:45440", + "left": "issue:44464", + "right": "issue:44977", "accept": false, - "reason": "Auxiliary loss normalization is unrelated to DeepseekV3MoE diverging from the remote implementation." + "reason": "Compiled-forward chunked generation mismatch is a different code path from flash-attention generation failure." }, { - "left": "issue:44464", - "right": "issue:44977", + "left": "issue:43688", + "right": "issue:45440", "accept": false, - "reason": "Compiled chunked generation inconsistency is different from a Qwen3.5 flash-attention generation failure." + "reason": "Both concern MoE models, but they describe different implementation discrepancies and likely different fixes." }, { "left": "issue:44977", "right": "issue:46693", "accept": false, - "reason": "A correctness bug in Qwen3.5 generation is not the same as a long-context FA2 performance regression." + "reason": "One is a correctness bug in Qwen3.5 generation with flash-attention, the other is a performance regression in FA2 long-context generation." }, { - "left": "issue:44734", - "right": "issue:46612", + "left": "issue:41720", + "right": "issue:44464", "accept": false, - "reason": "KV-cache continuation indexing in serve is unrelated to beam-search cache reorder being skipped in several model families." + "reason": "cudaErrorAssert on auto device mapping is unrelated to compiled-forward chunked generation inconsistency." }, { - "left": "issue:41720", - "right": "issue:44464", + "left": "issue:44734", + "right": "issue:46612", "accept": false, - "reason": "Qwen3 auto device-map cuda asserts are a different issue than compiled chunked generation output drift." + "reason": "Both are generation/cache bugs, but one is KV-cache continuation indexing in serve and the other is beam-search cache reorder skipping in several models." }, { "left": "issue:43504", "right": "issue:43550", "accept": false, - "reason": "A legacy field preventing BEiT model load is unrelated to Bamba SDPA failures under torch.compile." + "reason": "Legacy-field load failure for Beit is unrelated to Bamba torch.compile SDPA failure." }, { "left": "issue:43504", "right": "issue:43720", "accept": false, - "reason": "BEiT preset loading with a legacy field is unrelated to BitNet packed-weight unpacking during accelerate loading." + "reason": "Different model-loading bugs: a legacy preset field versus packed weights not being unpacked during accelerate loading." }, { "left": "issue:43278", "right": "issue:43688", "accept": false, - "reason": "Training/eval dtype drift in embeddings is unrelated to auxiliary-loss normalization in MoE models." + "reason": "Embedding dtype drift in evaluation is unrelated to auxiliary-loss normalization in MoE models." }, { - "left": "issue:43577", - "right": "issue:44554", + "left": "issue:33666", + "right": "issue:36331", "accept": false, - "reason": "Blip2 staying fp32 after load is not the same as the MPS attention value-head/query-head dim bug." + "reason": "Multi-GPU training for Qwen2-VL is not the same bug as CustomTrainer.compute_loss() rejecting num_items_in_batch." }, { "left": "issue:43782", "right": "issue:46693", "accept": false, - "reason": "Qwen3VL from_pretrained weight_only errors are unrelated to the FA2 long-context performance regression." + "reason": "Qwen3VL weight_only load errors are a model-loading problem, not the FA2 performance regression described here." }, { - "left": "issue:33666", - "right": "issue:36331", + "left": "issue:43577", + "right": "issue:44554", "accept": false, - "reason": "Qwen2-VL multi-GPU training is unrelated to CustomTrainer.compute_loss() rejecting num_items_in_batch." + "reason": "Blip2 dtype loading and an MPS attention correctness issue are different models, devices, and failure modes." }, { "left": "issue:43611", "right": "issue:46846", "accept": false, - "reason": "base_model_prefix loading breakage is unrelated to missing linear_attn entries in Qwen3.5 TP planning." + "reason": "Base_model_prefix loading regression is unrelated to missing linear_attn entries in tensor-parallel planning." }, { "left": "issue:43262", "right": "issue:43295", "accept": false, - "reason": "Audio processor sampling-rate defaults are unrelated to the processor.tokenizer/images regression." + "reason": "Audio sampling-rate defaulting and a v4.57.5 processor.tokenizer regression affect different processor paths." }, { "left": "issue:43720", "right": "issue:43927", "accept": false, - "reason": "BitNet packed-weight loading and DiaConfig token-ID persistence are different problems." + "reason": "Accelerate packed-weight loading and lost custom token IDs are separate persistence bugs." }, { "left": "issue:43606", "right": "issue:43927", "accept": false, - "reason": "CPU offload device mismatch in bark-small is unrelated to DiaConfig losing custom token IDs." + "reason": "CPU offload device mismatch is unrelated to config token-ID loss on save/load." }, { "left": "issue:43504", "right": "issue:44610", "accept": false, - "reason": "BEiT legacy-field loading is unrelated to OmDet-Turbo processor/model input-size mismatch." + "reason": "Beit legacy-field loading and OmDet processor output-size mismatch are different model families and bugs." }, { "left": "issue:44514", "right": "issue:45290", "accept": false, - "reason": "Batched padding=False crash in Qwen2_5_VL is different from assistant tool-call messages with no content." + "reason": "Both hit apply_chat_template, but the crash triggers differ: batched padding=False versus assistant tool-call messages with no content." }, { "left": "issue:43643", "right": "issue:44561", "accept": false, - "reason": "Missing fields from trust_remote_code AutoConfig are unrelated to removing is_torch_fx_available." + "reason": "trust_remote_code missing fields is unrelated to the removal of is_torch_fx_available." }, { "left": "issue:44964", "right": "issue:45405", "accept": false, - "reason": "Phi-4 multimodal loading failure is unrelated to an unreleased MIN_PEFT_VERSION bump." + "reason": "Phi-4 multimodal loading failure is not the same as an out-of-sync MIN_PEFT_VERSION bump." }, { "left": "issue:43493", "right": "issue:43825", "accept": false, - "reason": "SigLIP2 implementation mismatch is unrelated to an inaccurate pipeline translation-task error message." + "reason": "SigLIP2 implementation discrepancy and a pipeline error message about translation tasks are unrelated." }, { "left": "issue:43720", "right": "issue:44361", "accept": false, - "reason": "BitNet packed-weight loading is unrelated to MLukeTokenizer failing on tasks." + "reason": "Packed-weight loading during accelerate is unrelated to MLukeTokenizer task AttributeError." }, { "left": "issue:43525", "right": "issue:43572", "accept": false, - "reason": "Llama4Config missing pad_token_id is a different config-field issue than StableLmConfig missing pad_token_idx." + "reason": "Missing pad_token_id in Llama4Config and missing pad_token_idx in StableLmConfig are similar in spirit but distinct model-specific config bugs." }, { "left": "issue:44625", "right": "issue:45584", "accept": false, - "reason": "Qwen3.5 num_labels propagation is unrelated to Whisper empty-transcription failure after align_special_tokens." + "reason": "num_labels propagation in Qwen3.5 is unrelated to Whisper empty-transcription failure after align_special_tokens." }, { "left": "issue:44554", "right": "issue:46365", "accept": false, - "reason": "An MPS attention correctness bug is unrelated to Trainer import failure caused by datasets." + "reason": "MPS attention correctness is unrelated to the Trainer import problem caused by datasets." }, { "left": "issue:44977", "right": "issue:46209", "accept": false, - "reason": "Qwen3.5 flash-attention generation issues are unrelated to FP8 inference failing on B200 GPUs." + "reason": "Qwen3.5 flash-attention generation issues differ from FP8 inference failure on B200." + }, + { + "left": "issue:45823", + "right": "issue:46050", + "accept": false, + "reason": "Gemma4 device-mapping mismatch is a runtime placement issue, not a save/load quantization regression." + }, + { + "left": "issue:28218", + "right": "issue:33357", + "accept": false, + "reason": "Tokenizer added-space behavior and a macOS bus error in CLIP are completely unrelated." } ] }, @@ -8435,9 +8382,9 @@ "issue:43646|issue:44466", "issue:46209|issue:46710", "issue:44568|issue:46489", - "issue:33357|issue:33666", - "issue:44849|issue:46209", "issue:43278|issue:43701", + "issue:44849|issue:46209", + "issue:33357|issue:33666", "issue:44792|issue:45405", "issue:44568|issue:45812", "issue:43927|issue:44361", @@ -8445,8 +8392,8 @@ "issue:44488|issue:44843", "issue:43824|issue:44589", "issue:44849|issue:45357", - "issue:43504|issue:43606", "issue:46821|issue:46832", + "issue:43504|issue:43606", "issue:45245|issue:46710", "issue:43334|issue:45759", "issue:44493|issue:46285", @@ -8456,208 +8403,208 @@ "issue:43257|issue:43299", "issue:46399|issue:46846" ], - "prepared_review_unit_hash": "ce58fb945393f13da07a6a3ea66d4e87035af6a8661e6c09d34c9d4985be9ae8", + "prepared_review_unit_hash": "566aae3d7a194dfaea55229c0841a69c67e985f13c4eb3428a3d3b889852f2bd", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36599, - "estimated_input_tokens": 9150, - "estimated_eval_tokens": 18556 + "serialized_chars": 36601, + "estimated_input_tokens": 9151, + "estimated_eval_tokens": 18558 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is heterogeneous: mostly unrelated model-specific regressions, tokenizer/config bugs, and training/runtime issues. None of the soft pairs look like the same underlying bug, so there is no safe duplicate consolidation.", - "confidence": 0.94, - "canonical_issue_reason": "No single canonical issue: the items span unrelated bugs across different models and subsystems, so none is a reliable duplicate target.", + "summary": "This is not a duplicate cluster: the issues span many unrelated regressions and model families (tokenizers, multimodal loading, serialization, DeepSpeed/TP, docs, and runtime/device bugs). None of the soft pairs look like the same underlying bug or change.", + "confidence": 0.93, + "canonical_issue_reason": "No single canonical issue; the items are heterogeneous and do not describe one shared bug path.", "canonical_pr_reason": null, - "best_issue_reason": "No global best issue; any representative choice would be arbitrary because the soft pairs do not form one duplicate family.", + "best_issue_reason": "No best issue for deduplication; none of the issues can represent the rest of the set as a common root cause.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43452", "right": "issue:44451", "accept": false, - "reason": "Both are model-loading failures, but they involve different models and likely different root causes (gguf_file API break vs specific model load regression)." + "reason": "Both are load failures, but one is gguf/from_pretrained handling and the other is a specific model load regression; no shared root cause is evident." }, { "left": "issue:43994", "right": "issue:46192", "accept": false, - "reason": "Different symptoms: one is bad SigLIP2 outputs, the other is kwargs handling for return_tensors; not the same bug." + "reason": "SigLIP2 wrong outputs vs return_tensors being ignored are different API behaviors and code paths." }, { "left": "issue:42796", "right": "issue:44977", "accept": false, - "reason": "Tokenizer BPE-marker handling and Qwen3.5 flash-attention generation issues are unrelated concrete failures." + "reason": "Tokenizer BPE-marker handling and Qwen3.5 flash-attention generation are unrelated bugs." }, { "left": "issue:39692", "right": "issue:43295", "accept": false, - "reason": "Both touch multimodal examples, but one is doc/example errors while the other is a processor/tokenizer regression." + "reason": "Docs example/model-processor mismatch and a processor.tokenizer regression are distinct issues." }, { "left": "issue:43493", "right": "issue:43844", "accept": false, - "reason": "SigLIP2 implementation discrepancy and ZeRO-3 gradient growth are different failure modes with different code paths." + "reason": "SigLIP2 implementation mismatch vs DeepSpeed ZeRO-3 gradient growth are unrelated." }, { "left": "issue:43278", "right": "issue:43844", "accept": false, - "reason": "Embedding dtype drift during eval is not the same as the ZeRO-3 gradient anomaly." + "reason": "Embedding dtype regression and gradient instability under ZeRO-3 are different failures." }, { "left": "issue:43646", "right": "issue:44466", "accept": false, - "reason": "Custom model initialization breakage and tied-weight serialization inconsistency are separate issues." + "reason": "Custom model init breakage and lm_head serialization inconsistency are separate v5 regressions." }, { "left": "issue:46209", "right": "issue:46710", "accept": false, - "reason": "FP8 inference failure on B200 and an accuracy regression after upgrade are different problems." + "reason": "FP8 inference failure on B200 and a DeepSeek accuracy regression are not the same bug." }, { "left": "issue:44568", "right": "issue:46489", "accept": false, - "reason": "Both are tokenizer correctness reports, but they concern different models and different tokenization bugs." + "reason": "Different tokenizer bugs affecting different models and symptoms." }, { - "left": "issue:33357", - "right": "issue:33666", + "left": "issue:43278", + "right": "issue:43701", "accept": false, - "reason": "MacOS CLIP bus error and Qwen2-VL multi-GPU training are unrelated." + "reason": "Embedding dtype change vs resume_from_checkpoint key mismatch are unrelated." }, { "left": "issue:44849", "right": "issue:46209", "accept": false, - "reason": "output_hidden_states behavior in Qwen3.5 is not the same as FP8/B200 inference failure." + "reason": "Qwen3.5 hidden-state behavior and FP8/B200 inference are different code paths." }, { - "left": "issue:43278", - "right": "issue:43701", + "left": "issue:33357", + "right": "issue:33666", "accept": false, - "reason": "Eval dtype mismatch and resume_from_checkpoint key mismatch are distinct training/evaluation issues." + "reason": "MacOS bus error with CLIP and Qwen2-VL multi-GPU training are unrelated." }, { "left": "issue:44792", "right": "issue:45405", "accept": false, - "reason": "A failing Janus test and an unreleased PEFT version bump are unrelated CI/setup concerns." + "reason": "A failing Janus test and a PEFT version bump are not the same underlying change." }, { "left": "issue:44568", "right": "issue:45812", "accept": false, - "reason": "Both involve tokenizer outputs, but the affected models and reported regressions are different enough to keep separate." + "reason": "Both are tokenizer regressions, but they affect different models with different observed failures." }, { "left": "issue:43927", "right": "issue:44361", "accept": false, - "reason": "Config save/load token ID loss and MLukeTokenizer AttributeError are separate tokenizer/config bugs." + "reason": "Custom token ID persistence and MLukeTokenizer task errors are distinct bugs." }, { "left": "issue:43575", "right": "issue:45405", "accept": false, - "reason": "Tensor-parallel OOM for a Qwen2 model is unrelated to a dependency version bump." + "reason": "TP OOM for Qwen2-57B and an unreleased dependency version bump are unrelated." }, { "left": "issue:44488", "right": "issue:44843", "accept": false, - "reason": "Custom model load failure and HF_HUB_OFFLINE breakage come from different paths and different causes." + "reason": "Model load regression for a specific model and HF_HUB_OFFLINE network access breakage are different causes." }, { "left": "issue:43824", "right": "issue:44589", "accept": false, - "reason": "Importing a missing Qwen2.5 VL class and missing Float8 storage support are unrelated runtime/import failures." + "reason": "Missing Qwen2.5-VL import and Float8 storage lookup failure are unrelated." }, { "left": "issue:44849", "right": "issue:45357", "accept": false, - "reason": "Qwen3.5 hidden-states behavior and visual-encoder key serialization are different regressions." + "reason": "Qwen3.5 hidden-states behavior and incorrect save_pretrained visual keys are different regressions." }, { - "left": "issue:43504", - "right": "issue:43606", + "left": "issue:46821", + "right": "issue:46832", "accept": false, - "reason": "Legacy-field loading for BEiT segmentation and CPU-offload device mismatch in bark-small are unrelated." + "reason": "TimesFM window_size AttributeError and RT-DETR feature-level mismatch are unrelated." }, { - "left": "issue:46821", - "right": "issue:46832", + "left": "issue:43504", + "right": "issue:43606", "accept": false, - "reason": "TimesFM window_size AttributeError and RT-DETR feature-level shape/type error are different model-specific bugs." + "reason": "BEiT legacy-field loading and Bark CPU-offload device mismatch are different issues." }, { "left": "issue:45245", "right": "issue:46710", "accept": false, - "reason": "Category-count limit runtime error and DeepSeek accuracy regression do not share a concrete root cause." + "reason": "Category-count limit error and DeepSeek accuracy regression are not the same underlying bug." }, { "left": "issue:43334", "right": "issue:45759", "accept": false, - "reason": "Both involve Qwen multimodal configs, but one is a missing pad_token_id and the other is improper text_config unwrapping." + "reason": "Qwen3-VL missing pad_token_id and multimodal text_config unwrapping are different loading problems." }, { "left": "issue:44493", "right": "issue:46285", "accept": false, - "reason": "Unexpected position-id keys and CLIP text-model structural inconsistency are separate architecture issues." + "reason": "Unexpected position_id keys and CLIP text-model structure divergence are unrelated." }, { "left": "issue:43334", "right": "issue:43572", "accept": false, - "reason": "Qwen3-VL config loading and StableLmConfig pad_token_idx loss are unrelated config regressions." + "reason": "Qwen3-VL pad_token_id missing and StableLmConfig pad_token_idx missing are separate config regressions." }, { "left": "issue:43572", "right": "issue:44291", "accept": false, - "reason": "Missing pad_token_idx and unexpected _is_hf_initialized argument are different compatibility issues." + "reason": "Missing config field and init_empty_weights unexpected argument are unrelated version-breakages." }, { "left": "issue:44589", "right": "issue:45520", "accept": false, - "reason": "Float8 storage lookup and flash_attn import_utils KeyError on Python 3.13 are unrelated backend/import problems." + "reason": "Float8 storage lookup failure and flash_attn import_utils KeyError are different compatibility issues." }, { "left": "issue:43257", "right": "issue:43299", "accept": false, - "reason": "Both mention Qwen3 MOE, but one is accelerated weight conversion and the other is a general dev load failure." + "reason": "Qwen3 MOE weight conversion and Qwen3VL loading failure are distinct bugs." }, { "left": "issue:46399", "right": "issue:46846", "accept": false, - "reason": "Qwen3.5 rotary-kernel registration and Qwen3.5 TP plan OOM/shape issues are different code paths and fixes." + "reason": "Missing apply_rotary_pos_emb decorator and missing linear_attn TP plan entries are different Qwen3.5 issues." } ] }, @@ -8746,8 +8693,8 @@ "issue:29942|issue:30333", "issue:42490|issue:43696", "issue:44977|issue:46701", - "issue:45084|issue:45520", "issue:28218|issue:34567", + "issue:45084|issue:45520", "issue:15354|issue:38175", "issue:43065|issue:46369", "issue:44554|issue:45446", @@ -8757,20 +8704,20 @@ "issue:45406|issue:46121", "issue:44977|issue:46710" ], - "prepared_review_unit_hash": "72b96e2011b1fd4c456b1304d2a996bb0c5980aa5014aa73422b1c70d30bb4ad", + "prepared_review_unit_hash": "45d6591e12e3116cbb448db0ac5ad90a74514effc752b4190da2853b63af8813", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 26, - "serialized_chars": 34894, + "serialized_chars": 34893, "estimated_input_tokens": 8724, "estimated_eval_tokens": 17704 }, @@ -8779,168 +8726,168 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is a broad mix of unrelated issues across tokenizers, serving, quantization, multimodal models, and CI. None of the soft pairs look like true duplicates of the same underlying bug/change, so I would not merge any pair.", + "summary": "All proposed soft edges look like false positives: they span unrelated models, subsystems, and failure modes rather than the same concrete bug or change.", "confidence": 0.98, - "canonical_issue_reason": null, + "canonical_issue_reason": "issue:45229 is the most substantial and best-documented issue in the set, with the highest discussion activity and a concrete reproducible failure; the rest of the cluster is too heterogeneous to form a true duplicate group.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "issue:45229 is the strongest representative candidate by engagement and specificity, but only as a loose anchor since the surrounding items do not describe the same underlying defect.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43606", "right": "issue:44112", "accept": false, - "reason": "Both mention device/CI failures, but one is a real CPU-offload model bug and the other is a stale GraniteSpeech CI test." + "reason": "Different subsystems and bugs: CPU offload/device mismatch for a model vs a stale CI test failure." }, { "left": "issue:45072", "right": "issue:46832", "accept": false, - "reason": "Different failure modes and code paths: bfloat16 dtype mismatch in inference vs RT-DETR feature-level mismatch in forward." + "reason": "Unrelated model failures with different root causes: dtype mismatch in inference vs RT-DETR feature-level forward validation." }, { "left": "issue:43504", "right": "issue:43653", "accept": false, - "reason": "Legacy field loading for BEiT semantic segmentation is unrelated to BigBirdTokenizer special-token registration and decode output." + "reason": "Different components and symptoms: BEiT model loading with a legacy field vs BigBird tokenizer special-token decoding." }, { "left": "issue:44464", "right": "issue:46424", "accept": false, - "reason": "Chunked generation inconsistency with compiled forward is not the same bug as cache max length being ignored outside quantized cache." + "reason": "Both mention cache/generation, but one is compiled chunked generation inconsistency and the other is a cache-length config limitation; not the same bug." }, { "left": "issue:43937", "right": "issue:46701", "accept": false, - "reason": "A GenerationConfig validation error is unrelated to a diffusion pipeline/output class failure." + "reason": "Different models and code paths: invalid GenerationConfig vs a diffusion/generation output pipeline failure." }, { "left": "issue:43873", "right": "issue:46050", "accept": false, - "reason": "Both are quantization-adjacent, but one is offloading behavior and the other is broken save/load for sinq quantization." + "reason": "Quantization is the shared theme, but offloading behavior and save/load persistence are distinct problems." }, { "left": "issue:43844", "right": "issue:44964", "accept": false, - "reason": "ZeRO-3 gradient growth is unrelated to loading a specific multimodal model." + "reason": "Completely different issues: DeepSpeed/ZeRO gradient behavior vs loading a multimodal model." }, { "left": "issue:43994", "right": "issue:44493", "accept": false, - "reason": "SigLIP2 nonsensical outputs and unexpected position-id keys are different symptoms with different likely causes." + "reason": "Different symptoms and likely code paths: SigLIP2 output quality vs a broad unexpected position-id key regression." }, { "left": "issue:45200", "right": "issue:45229", "accept": false, - "reason": "Defaulting mm_token_type_ids to zeros is unrelated to Gemma4 multi-GPU OOM." + "reason": "Both involve Gemma4, but one is missing/default mm_token_type_ids and the other is a multi-GPU CUDA OOM; not the same defect." }, { "left": "issue:44869", "right": "issue:46752", "accept": false, - "reason": "Whisper trailing-replacement decode crash and empty-list chat-template crash affect different code paths." + "reason": "Both are runtime crashes, but in unrelated text-processing paths: Whisper timestamp decoding vs apply_chat_template on empty input." }, { "left": "issue:45588", "right": "issue:46050", "accept": false, - "reason": "Flash-attention sink-less s_aux handling is not the same as quantized model serialization breakage." + "reason": "Different failure classes: flash-attention integration crashes on None aux state vs quantized model save/load breakage." }, { "left": "issue:45229", "right": "issue:45823", "accept": false, - "reason": "One is an inference OOM, the other is a device-mapping mismatch in the same model family." + "reason": "Same model family, but different concrete bugs: multi-GPU OOM vs device mismatch during auto device mapping." }, { "left": "issue:29159", "right": "issue:32090", "accept": false, - "reason": "Tokenizer slow/fast inconsistency is unrelated to a trainer broadcast TypeError on NoneType." + "reason": "Tokenizer inconsistency and GPU broadcast NoneType errors are unrelated." }, { "left": "issue:29942", "right": "issue:30333", "accept": false, - "reason": "Failing Flash Attention 2 tests are unrelated to MLFlow not reporting a failed job." + "reason": "Flash Attention 2 test failures and MLflow job-reporting issues are unrelated." }, { "left": "issue:42490", "right": "issue:43696", "accept": false, - "reason": "A local-path import warning has nothing to do with CUDA OOM on GPT-oss-20b." + "reason": "Loading-path warning and GPT-oss OOM are unrelated." }, { "left": "issue:44977", "right": "issue:46701", "accept": false, - "reason": "Qwen3.5 flash-attention generation issues are unrelated to DiffusionGemma pipeline failures." + "reason": "Different model families and bugs: Qwen3.5 flash-attention generation issue vs DiffusionGemma pipeline failure." }, { - "left": "issue:45084", - "right": "issue:45520", + "left": "issue:28218", + "right": "issue:34567", "accept": false, - "reason": "Compilation of non-template nodes and a Python 3.13 flash_attn import KeyError are separate problems." + "reason": "Tokenizer spacing behavior and TrainerState token counting are unrelated." }, { - "left": "issue:28218", - "right": "issue:34567", + "left": "issue:45084", + "right": "issue:45520", "accept": false, - "reason": "Added-token spacing behavior is unrelated to TrainerState token-count tracking." + "reason": "Different failure areas: template-node compilation vs import-utils flash_attn lookup on Python 3.13." }, { "left": "issue:15354", "right": "issue:38175", "accept": false, - "reason": "TorchScript GeneratorExp export failure and SigLIP2 zero probabilities are different model/runtime issues." + "reason": "TorchScript export on ViT and SigLIP2 zero-probability outputs are unrelated." }, { "left": "issue:43065", "right": "issue:46369", "accept": false, - "reason": "Dummy Conv2d in Sam3PixelDecoder is not the same as memory overflow in sam3-litetext." + "reason": "Same model family, but one is a dummy Conv2d implementation detail and the other is a memory overflow; not clearly the same bug." }, { "left": "issue:44554", "right": "issue:45446", "accept": false, - "reason": "An MPS attention correctness bug is unrelated to a flex_attention version-check import bug." + "reason": "Different concerns: MPS attention correctness vs a PyTorch version check for an import." }, { "left": "issue:43493", "right": "issue:43701", "accept": false, - "reason": "SigLIP2 HF-vs-JAX discrepancy and resume_from_checkpoint key mismatch do not share the same underlying bug." + "reason": "SigLIP2 implementation discrepancy and resume-from-checkpoint key mismatch are unrelated." }, { "left": "issue:44423", "right": "issue:44734", "accept": false, - "reason": "Both are serve-related, but one crashes in continuous-batching multimodal input handling and the other in KV-cache continuation indexing." + "reason": "Both are transformers serve crashes, but they hit different endpoints and different code paths; not safely mergeable as one fix." }, { "left": "issue:41720", "right": "issue:46710", "accept": false, - "reason": "A cudaErrorAssert from auto device mapping is not an accuracy regression after a Transformers upgrade." + "reason": "Different models and symptoms: Qwen3 auto-device-map cuda assert vs DeepSeek accuracy regression after upgrade." }, { "left": "issue:45406", "right": "issue:46121", "accept": false, - "reason": "Gemma4Processor missing _tokenizer and rope-parameter list conversion are unrelated failures." + "reason": "Gemma4Processor missing _tokenizer and rope-parameter type handling are unrelated." }, { - "left": "issue:43653", - "right": "issue:43606", + "left": "issue:44977", + "right": "issue:46710", "accept": false, - "reason": "BigBirdTokenizer special-token registration and Bark CPU-offload device mismatch are different bugs." + "reason": "Qwen3.5 flash-attention generation failure and DeepSeek accuracy regression are different bugs." } ] }, @@ -9021,11 +8968,11 @@ "issue:43295|issue:43873", "issue:44186|issue:46192", "issue:43572|issue:46821", - "issue:45823|issue:46192", "issue:43295|issue:44610", + "issue:45823|issue:46192", "issue:43550|issue:43844", - "issue:45200|issue:45823", "issue:43450|issue:43575", + "issue:45200|issue:45823", "issue:29870|issue:29942", "issue:45520|issue:46178", "issue:43572|issue:44987", @@ -9039,219 +8986,219 @@ "issue:45381|issue:46209", "issue:44466|issue:45127", "issue:28218|issue:35141", - "issue:43278|issue:43856", "issue:44737|issue:46894", + "issue:43278|issue:43856", "issue:42915|issue:43824", "issue:43824|issue:43975" ], - "prepared_review_unit_hash": "8c9e3404d7c84bcee227296f98221b298be2ee3bd2fac26c4caccaed176c67b6", + "prepared_review_unit_hash": "d83429f495168e60c1d9359c05a0eb950bf5f007b46885f36dda54368711effc", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 36336, - "estimated_input_tokens": 9084, - "estimated_eval_tokens": 18424 + "serialized_chars": 36337, + "estimated_input_tokens": 9085, + "estimated_eval_tokens": 18426 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is highly heterogeneous; the soft pairs mostly share broad subsystem keywords but not the same concrete bug or code path. I reject all duplicate candidates. If one issue must serve as the representative canonical item, the multimodal `apply_chat_template` bug is the best fit among this group.", - "confidence": 0.9, - "canonical_issue_reason": "issue:44521 is the most central, user-facing, and actively discussed bug in the set, with multiple inbound references and a clear reproducible multimodal chat-template failure mode.", + "summary": "These are broad similarity matches, but the paired issues describe different bugs/regressions across unrelated code paths. I would not collapse any of the soft pairs into duplicates, and there is no clear canonical issue for the whole set.", + "confidence": 0.84, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:44521 is the strongest representative issue because it is broad, well-scoped, and likely to attract related reports in the same area.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43295", "right": "issue:43450", "accept": false, - "reason": "Both are multimodal-related, but one is a processor/tokenizer regression and the other is a batched video output-shape bug." + "reason": "Both are multimodal-related, but one is a custom tokenizer/processor regression and the other is a batched video-shape bug; different failure modes and code paths." }, { "left": "issue:37428", "right": "issue:43957", "accept": false, - "reason": "Import/export symbol error vs. a meta-device loading regression; different layers and failure modes." + "reason": "An import error for a missing flash-attention helper is unrelated to meta-device loading failures in model init." }, { "left": "issue:45072", "right": "issue:45198", "accept": false, - "reason": "A bfloat16 dtype mismatch in inference is unrelated to Wav2Vec2 save/tokenization failures." + "reason": "Bfloat16 inference dtype mismatches and Wav2Vec2 save/tokenization failures are unrelated." }, { "left": "issue:45588", "right": "issue:45759", "accept": false, - "reason": "Flash-attention `s_aux=None` crash and multimodal `text_config` unwrapping are different code paths." + "reason": "A flash-attention sinkless-model crash is not the same as unwrapping text_config for composite multimodal configs." }, { "left": "issue:43540", "right": "issue:45200", "accept": false, - "reason": "Video-input processing failure in Qwen3OmniMoe is unrelated to Gemma4's `mm_token_type_ids` defaulting bug." + "reason": "Video-processing errors in Qwen3OmniMoe are unrelated to Gemma 4's missing mm_token_type_ids default." }, { "left": "issue:43295", "right": "issue:43873", "accept": false, - "reason": "Custom model processor/tokenizer regression and quantization offloading behavior are not the same bug." + "reason": "Processor/tokenizer regression and quantization offloading failure are different issues despite both touching model loading." }, { "left": "issue:44186", "right": "issue:46192", "accept": false, - "reason": "LayoutLMv2 NER/padding crashes are distinct from `return_tensors` being ignored when `text_kwargs` is passed." + "reason": "LayoutLMv2 NER/batching crashes are not the same as return_tensors being ignored when text_kwargs is passed." }, { "left": "issue:43572", "right": "issue:46821", "accept": false, - "reason": "A missing config field in StableLm is unrelated to TimesFM's missing `window_size` attribute." + "reason": "Missing pad_token_idx in StableLmConfig is unrelated to TimesFM's window_size AttributeError." }, { - "left": "issue:45823", - "right": "issue:46192", + "left": "issue:43295", + "right": "issue:44610", "accept": false, - "reason": "Gemma4 device placement mismatch and `return_tensors` propagation are different issues." + "reason": "These are different processor bugs: one about tokenizer images/custom code, the other about OmDet-Turbo input resolution mismatch." }, { - "left": "issue:43295", - "right": "issue:44610", + "left": "issue:45823", + "right": "issue:46192", "accept": false, - "reason": "Processor/image handling regression and OmDet-Turbo input-size mismatch are separate problems." + "reason": "A device_map auto forward mismatch is unrelated to return_tensors being silently ignored." }, { "left": "issue:43550", "right": "issue:43844", "accept": false, - "reason": "A `torch.compile`/SDPA failure is unrelated to ZeRO-3 gradient growth during training." + "reason": "torch.compile + SDPA failure in Bamba-9B-v2 is not the same as a DeepSpeed ZeRO-3 gradient growth issue." }, { - "left": "issue:45200", - "right": "issue:45823", + "left": "issue:43450", + "right": "issue:43575", "accept": false, - "reason": "Both involve Gemma4, but one is token-type defaults for text-only fine-tuning and the other is a forward device mismatch." + "reason": "Video batch shape handling and tensor-parallel OOM during Qwen2 loading are unrelated." }, { - "left": "issue:43450", - "right": "issue:43575", + "left": "issue:45200", + "right": "issue:45823", "accept": false, - "reason": "Batched video-shape issues and tensor-parallel OOM on Qwen2 are unrelated." + "reason": "Defaulting mm_token_type_ids to zeros and a forward-time device mismatch are distinct bugs." }, { "left": "issue:29870", "right": "issue:29942", "accept": false, - "reason": "A generation-time UnboundLocalError is not the same as FlashAttention 2 test failures." + "reason": "A BART generation local-variable bug and Flash Attention 2 test failures do not describe the same code-path problem." }, { "left": "issue:45520", "right": "issue:46178", "accept": false, - "reason": "Different backend/import plumbing bugs: `flash_attn` key handling vs. `tensorflow_text` backend registration." + "reason": "Both are import/backend issues, but they concern different missing backends and different error mechanisms." }, { "left": "issue:43572", "right": "issue:44987", "accept": false, - "reason": "StableLm config missing `pad_token_idx` is unrelated to loading `physical-intelligence/fast`." + "reason": "A StableLmConfig field regression is unrelated to a model-loading failure for physical-intelligence/fast." }, { "left": "issue:37428", "right": "issue:43824", "accept": false, - "reason": "Both are import errors, but for different missing symbols in unrelated modules." + "reason": "Both are import errors, but they involve different missing symbols from different modules and likely different fixes." }, { "left": "issue:42490", "right": "issue:43493", "accept": false, - "reason": "Trailing-slash local-model loading warning is unrelated to the SigLIP2 implementation discrepancy." + "reason": "A trailing-slash local path warning is unrelated to a SigLIP2 implementation discrepancy." }, { "left": "issue:39692", "right": "issue:46192", "accept": false, - "reason": "A SigLIP2 documentation/example problem is not the same as `return_tensors` being ignored." + "reason": "SigLIP2 documentation errors and return_tensors handling are unrelated." }, { "left": "issue:43232", "right": "issue:44849", "accept": false, - "reason": "Generation kwargs after `sync_gpus` and Qwen3.5 hidden-states behavior are different generation bugs." + "reason": "A generation sync_gpus kwargs bug and a Qwen3.5 output_hidden_states bug are different model/runtime issues." }, { "left": "issue:44554", "right": "issue:45356", "accept": false, - "reason": "An MPS attention correctness issue is unrelated to Kimi-K2.5 tokenizer codec handling." + "reason": "An MPS attention correctness issue is unrelated to a tokenizer codec regression in Kimi-K2.5." }, { "left": "issue:43295", "right": "issue:43493", "accept": false, - "reason": "Processor-tokenizer regression and SigLIP2 HF-vs-JAX discrepancy are not the same underlying change." + "reason": "Custom processor/tokenizer regression and SigLIP2 HF-vs-JAX discrepancy are not the same underlying bug." }, { "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 assistant messages." + "reason": "Both hit apply_chat_template, but one is all-zero assistant masks for multimodal inputs while the other is a crash on tool-call assistant messages." }, { "left": "issue:45381", "right": "issue:46209", "accept": false, - "reason": "Qwen2.5-VL video position IDs and FP8 inference on B200 are unrelated." + "reason": "Wrong vision_position_ids for Qwen2.5-VL video inputs is unrelated to FP8 inference failing on B200." }, { "left": "issue:44466", "right": "issue:45127", "accept": false, - "reason": "Serialization of tied weights and LoRA merge collapse share a theme but not the same concrete failure path." + "reason": "Inconsistent lm_head.weight serialization and LoRA merge collapse with extended vocab are related to weights, but they are different failure modes and not a single bug." }, { "left": "issue:28218", "right": "issue:35141", "accept": false, - "reason": "Tokenizer extra-space insertion and embedding reinitialization after resize are different bugs." - }, - { - "left": "issue:43278", - "right": "issue:43856", - "accept": false, - "reason": "A dtype drift between train/eval is unrelated to MoE training memory inefficiency." + "reason": "Tokenizer spacing after added tokens and embedding reinitialization after resize are distinct issues." }, { "left": "issue:44737", "right": "issue:46894", "accept": false, - "reason": "CPU tensor creation in XLNet relative positions is unrelated to the CohereASR double-shift loss bug." + "reason": "CPU placement in XLNet positional encoding is unrelated to CohereASR's double-shift training-loss bug." + }, + { + "left": "issue:43278", + "right": "issue:43856", + "accept": false, + "reason": "BF16-to-FP32 dtype drift during evaluation and MoE training memory inefficiency are not the same defect." }, { "left": "issue:42915", "right": "issue:43824", "accept": false, - "reason": "FineGrainedFP8Config failure in Qwen3Moe is unrelated to a missing Qwen2.5-VL export." + "reason": "A FineGrainedFP8Config failure in Qwen3Moe is unrelated to an import error for Qwen2_5_VLForConditionalGeneration." }, { "left": "issue:43824", "right": "issue:43975", "accept": false, - "reason": "Missing model export and incorrect detokenization are unrelated issues." + "reason": "An import failure for a Qwen VL class and DeepSeek detokenization regression are unrelated." } ] }, @@ -9275,8 +9222,6 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-506", "nodes": [ - "issue:28282", - "issue:33357", "issue:37428", "issue:38175", "issue:41720", @@ -9304,6 +9249,7 @@ "issue:44792", "issue:44857", "issue:44898", + "issue:44995", "issue:45083", "issue:45198", "issue:45200", @@ -9331,8 +9277,8 @@ "issue:43262|issue:43450", "issue:43334|issue:43525", "issue:45200|issue:46899", - "issue:45200|issue:46932", "issue:44442|issue:45198", + "issue:45200|issue:46932", "issue:44792|issue:45563", "issue:44898|issue:45412", "issue:43526|issue:46192", @@ -9340,8 +9286,8 @@ "issue:43819|issue:44112", "issue:43335|issue:43550", "issue:41720|issue:46209", - "issue:43827|issue:43901", "issue:44561|issue:46234", + "issue:43827|issue:43901", "issue:44610|issue:44857", "issue:43819|issue:45083", "issue:43388|issue:43526", @@ -9349,221 +9295,221 @@ "issue:45759|issue:46285", "issue:44857|issue:45593", "issue:43606|issue:45083", + "issue:44995|issue:46858", "issue:44206|issue:45584", "issue:38175|issue:45412", "issue:46424|issue:46858", "issue:43335|issue:43653", - "issue:37428|issue:43425", - "issue:28282|issue:33357" + "issue:37428|issue:43425" ], - "prepared_review_unit_hash": "b2515da6a153cfea7f57b681b5a1912c38774549f33fe02749aaeeef1991a12c", + "prepared_review_unit_hash": "71f34a51e14a6bd1b474f34ce47a08fc76f79d0c05ab5664828c2b650102200f", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 37083, - "estimated_input_tokens": 9271, - "estimated_eval_tokens": 18798 + "serialized_chars": 36502, + "estimated_input_tokens": 9126, + "estimated_eval_tokens": 18508 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a heterogeneous cluster of mostly false-positive similarity matches. The only clear duplicate-like pair is the Gemma4 4-bit quantization dtype-cast regression (#46899 and #46932); the rest differ by model, failure mode, or code path.", - "confidence": 0.87, - "canonical_issue_reason": "issue:46899 is the clearest canonical bug report: it describes the Gemma4 audio/vision failure under 4-bit quantization and pinpoints the unsafe dtype cast in multimodal embedders. issue:46932 reads like a later near-duplicate/fix note for the same regression.", + "summary": "This cluster is heterogeneous: the soft pairs mostly share only broad subsystem language (tokenizers, processors, cache, config, multimodal models) but not the same concrete bug. I would not merge any of the proposed soft edges as duplicates.", + "confidence": 0.74, + "canonical_issue_reason": "Issue #43950 is the most representative broad regression report in the set: it describes a core from_pretrained behavior break, has substantial detail, and is more general than the many model-specific edge cases here.", "canonical_pr_reason": null, - "best_issue_reason": "issue:46899 best represents the only strong duplicate pair and has the most complete symptom description and technical scope.", + "best_issue_reason": "#43950 is the best single issue to use as a representative canonical reference because it is well-scoped, framework-level, and clearly reproducible.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43653", "right": "issue:44112", "accept": false, - "reason": "Both are CI/model-specific issues, but they affect different models and different failure points; not the same bug." + "reason": "Different problems: tokenizer special-token registration vs a GraniteSpeech CI device-override test failure." }, { "left": "issue:43452", "right": "issue:46512", "accept": false, - "reason": "One is about gguf loading in tokenizer/model loading, the other about stale config dtype in from_config; unrelated code paths." + "reason": "Unrelated bugs: gguf loading path breakage vs stale config.dtype after from_config(dtype=...)." }, { "left": "issue:43950", "right": "issue:46285", "accept": false, - "reason": "Both mention config/model structure regressions, but one is about non-persistent buffers on load and the other about CLIP text model attribute structure." + "reason": "Different code paths: non-persistent buffer loading regression vs CLIP text model structural inconsistency." }, { "left": "issue:43262", "right": "issue:43450", "accept": false, - "reason": "Audio chat-template sampling-rate defaulting and batched video processor shape handling are unrelated." + "reason": "Audio chat-template sampling-rate default and batched video output shape are unrelated." }, { "left": "issue:43334", "right": "issue:43525", "accept": false, - "reason": "Both are missing pad_token_id on config-like objects, but they concern different model families and separate loaders." + "reason": "Both are missing pad_token_id symptoms, but in different model families and likely different fixes." }, { "left": "issue:45200", "right": "issue:46899", "accept": false, - "reason": "Both involve Gemma4 multimodal behavior, but one is mm_token_type_ids defaults and the other is 4-bit quantization dtype casting." + "reason": "Same Gemma4 area, but one is mm_token_type_ids handling and the other is quantized embedder dtype casting." }, { - "left": "issue:45200", - "right": "issue:46932", + "left": "issue:44442", + "right": "issue:45198", "accept": false, - "reason": "Same broad Gemma4 area, but the underlying failure differs: token-type ids defaulting vs quantized embedder dtype casting." + "reason": "Different audio/tokenizer failures affecting different model classes and code paths." }, { - "left": "issue:44442", - "right": "issue:45198", + "left": "issue:45200", + "right": "issue:46932", "accept": false, - "reason": "Both are tokenizer-related, but they are different models and different failure modes (AutoTokenizer load vs save_pretrained/tokenization)." + "reason": "Both Gemma4-related, but they describe distinct issues: token-type defaults vs quantized vision/audio dtype casting." }, { "left": "issue:44792", "right": "issue:45563", "accept": false, - "reason": "One is a janus image-generation test failure; the other is a stale generate() warning. Different behavior entirely." + "reason": "Janus image generation test failure vs stale paged-generate warning; no shared underlying bug." }, { "left": "issue:44898", "right": "issue:45412", "accept": false, - "reason": "Perceiver interpolation failure and RT-DETR memory release are unrelated bugs." + "reason": "Perceiver resize/interpolation bug and RT-DETR memory release issue are unrelated." }, { "left": "issue:43526", "right": "issue:46192", "accept": false, - "reason": "Both involve label/return handling, but one is a BeitImageProcessorFast reduction bug and the other is return_tensors being ignored with text_kwargs." + "reason": "Beit fast image label reduction vs text_kwargs/return_tensors precedence are different behaviors." }, { "left": "issue:45200", "right": "issue:45325", "accept": false, - "reason": "Both are Gemma/Qwen multimodal issues, but one is about token-type ids and the other about RoPE temporal position scaling." + "reason": "Different multimodal bugs: default mm_token_type_ids vs Qwen2.5-VL rope position scaling." }, { "left": "issue:43819", "right": "issue:44112", "accept": false, - "reason": "DAC latent decoding mismatch and GraniteSpeech device override CI failure are different models and code paths." + "reason": "DAC latent reconstruction mismatch is unrelated to the GraniteSpeech CI failure." }, { "left": "issue:43335", "right": "issue:43550", "accept": false, - "reason": "SwitchTransformers sparse-layer construction and Bamba torch.compile SDPA failure are unrelated." + "reason": "SwitchTransformers sparse-layer construction and Bamba torch.compile SDPA failure are different model bugs." }, { "left": "issue:41720", "right": "issue:46209", "accept": false, - "reason": "Both are GPU/runtime failures, but one is Qwen3 auto device mapping on A800 and the other is FP8 inference on B200." + "reason": "Different hardware/runtime failures: A800 cudaErrorAssert vs B200 FP8 inference incompatibility." }, { - "left": "issue:43827", - "right": "issue:43901", + "left": "issue:44561", + "right": "issue:46234", "accept": false, - "reason": "Both are documentation-oriented pipeline issues, but they target different docs and different deprecated parameters." + "reason": "Both are missing import-utils symbols, but they are different helpers affecting different model families." }, { - "left": "issue:44561", - "right": "issue:46234", + "left": "issue:43827", + "right": "issue:43901", "accept": false, - "reason": "Both concern missing utility helpers, but they are different removed/absent symbols affecting different remote-code paths." + "reason": "Both are docs issues, but they concern different pipeline APIs and different deprecated mentions." }, { "left": "issue:44610", "right": "issue:44857", "accept": false, - "reason": "OmDet-Turbo processor output shape and LwDetrImageLoss AMP crash are distinct problems." + "reason": "Different detection/image-processing failures: wrong input size vs AMP crash in a loss function." }, { "left": "issue:43819", "right": "issue:45083", "accept": false, - "reason": "DAC latent mismatch and qwen3_omni_moe feature-length helper behavior are unrelated." + "reason": "DAC latent forward mismatch and qwen3_omni_moe feature-length helper behavior are unrelated." }, { "left": "issue:43388", "right": "issue:43526", "accept": false, - "reason": "Both touch label handling, but one is gather_for_metrics dropping tuple labels and the other is BeitImageProcessorFast reduce_labels behavior." + "reason": "Both mention labels, but one is metrics gathering on tuples and the other is BeitImageProcessorFast reduce_labels." }, { "left": "issue:46512", "right": "issue:46821", "accept": false, - "reason": "Both are config/attribute issues, but one is stale dtype after from_config and the other is TimesFM window_size raising AttributeError." + "reason": "Stale dtype metadata after from_config and TimesFM window_size AttributeError are unrelated." }, { "left": "issue:45759", "right": "issue:46285", "accept": false, - "reason": "Composite Qwen config unwrapping and CLIP text-model structure regression are separate bugs." + "reason": "Composite Qwen causal-LM config unwrapping and CLIP text-model structure divergence are different issues." }, { "left": "issue:44857", "right": "issue:45593", "accept": false, - "reason": "Both are detection/model-loss issues, but one is an AMP crash and the other is auxiliary losses not used when denoising is off." + "reason": "AMP crash in LwDetrImageLoss and missing auxiliary losses in D-FINE are separate detection bugs." }, { "left": "issue:43606", "right": "issue:45083", "accept": false, - "reason": "CPU offload device mismatch for bark-small and qwen3_omni_moe feature-length arithmetic are unrelated." + "reason": "CPU offload device mismatch and qwen3_omni_moe feature-length helper behavior do not match." + }, + { + "left": "issue:44995", + "right": "issue:46858", + "accept": false, + "reason": "Stale indexer cache crash in one model and static-cache sequence-length tensor arithmetic in prefill are different cache bugs." }, { "left": "issue:44206", "right": "issue:45584", "accept": false, - "reason": "LasrFeatureExtractor center-arg crash and Whisper empty-transcription failure are different audio-processing bugs." + "reason": "Unsupported feature-extractor arg and Whisper empty-transcription failure are unrelated." }, { "left": "issue:38175", "right": "issue:45412", "accept": false, - "reason": "Zero probabilities in SigLIP2 and RT-DETR memory cleanup are unrelated." + "reason": "Zero probabilities in SigLIP2 and RT-DETR memory not released are unrelated." }, { "left": "issue:46424", "right": "issue:46858", "accept": false, - "reason": "Cache-length config behavior and static-cache seq-length tensor arithmetic are different cache issues." + "reason": "Both are cache-related, but one is max_cache_length semantics and the other is device-tensor sequence length in prefill." }, { "left": "issue:43335", "right": "issue:43653", "accept": false, - "reason": "Sparse-layer config creation and BigBird tokenizer special-token registration are unrelated." + "reason": "SwitchTransformers sparse-layer construction and BigBirdTokenizer special-token registration are unrelated." }, { "left": "issue:37428", "right": "issue:43425", "accept": false, - "reason": "Flash-attention import failure and torch 2.10 incompatibility are not the same bug." - }, - { - "left": "issue:28282", - "right": "issue:33357", - "accept": false, - "reason": "PyTorch missing ImportError and MacOS bus error on a CLIP model are unrelated." + "reason": "Flash-attention import error and Torch 2.10 incompatibility are different root causes." } ] }, @@ -9587,12 +9533,12 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-506", "nodes": [ - "issue:26413", "issue:28218", + "issue:28282", "issue:29127", "issue:29942", - "issue:31501", "issue:32090", + "issue:33357", "issue:34689", "issue:35141", "issue:37428", @@ -9636,244 +9582,244 @@ "issue:46566" ], "soft_pairs": [ + "issue:28282|issue:33357", "issue:29942|issue:32090", - "issue:43550|issue:44112", "issue:43611|issue:44466", + "issue:43550|issue:44112", "issue:43335|issue:43504", - "issue:45520|issue:46170", "issue:44855|issue:45083", + "issue:45520|issue:46170", "issue:44448|issue:44743", "issue:43335|issue:45083", "issue:45290|issue:46326", - "issue:44514|issue:44521", "issue:37428|issue:44351", + "issue:44514|issue:44521", "issue:43335|issue:45593", "issue:43335|issue:44112", - "issue:29127|issue:32090", "issue:43278|issue:43825", + "issue:29127|issue:32090", "issue:45588|issue:45859", "issue:42490|issue:43854", "issue:34689|issue:35141", - "issue:43720|issue:44112", "issue:43819|issue:44493", + "issue:43720|issue:44112", "issue:39692|issue:43526", "issue:46431|issue:46566", "issue:43606|issue:44857", "issue:44448|issue:44779", - "issue:45310|issue:45478", - "issue:28218|issue:32090", "issue:39692|issue:44493", - "issue:45459|issue:46234", - "issue:26413|issue:31501" + "issue:28218|issue:32090", + "issue:45310|issue:45478", + "issue:45459|issue:46234" ], - "prepared_review_unit_hash": "140335a590feef3bd706b67d1bfef083f30b9596a2fd340d724750e24d6fbdb2", + "prepared_review_unit_hash": "34ee9b3beb67d6d33ea58622f8813e517960a5e9feb1ec3bbf25317239c5a194", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 29, - "serialized_chars": 36121, - "estimated_input_tokens": 9031, - "estimated_eval_tokens": 18318 + "serialized_chars": 36139, + "estimated_input_tokens": 9035, + "estimated_eval_tokens": 18326 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: nearly all items are unrelated issues from different subsystems. The only clear duplicate pair is the two identical Qwen3.5 Moe from_pretrained reports (issues 45310 and 45478).", - "confidence": 0.91, - "canonical_issue_reason": "issue:45310 is the best canonical issue because it is the earlier, fuller report and has the exact duplicate in issue:45478.", + "summary": "This cluster is mostly heterogeneous; only issue 45310 and 45478 are a true duplicate pair (same Qwen3.5 MoE from_pretrained regression). All other soft links look like related-in-theme but different bugs.", + "confidence": 0.94, + "canonical_issue_reason": "45310 is the best canonical anchor because it is an exact duplicate of 45478 with the same title and failure mode, but the surrounding cluster is too mixed for it to represent the whole set strongly.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45310 is the strongest cluster representative since it cleanly anchors the only true duplicate pair and has the richer problem description.", + "best_issue_reason": "45310 is the best issue overall only for the confirmed duplicate pair; it is not a good global representative because most other items are unrelated regressions.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:29942", - "right": "issue:32090", + "left": "issue:28282", + "right": "issue:33357", "accept": false, - "reason": "Both are runtime failures, but one is Flash Attention 2 test breakage and the other is Trainer GPU broadcast of NoneType; different bugs and code paths." + "reason": "Different failures: missing PyTorch import vs a MacOS bus error in a CLIP model." }, { - "left": "issue:43550", - "right": "issue:44112", + "left": "issue:29942", + "right": "issue:32090", "accept": false, - "reason": "Different models, different failures, and different code paths; not the same underlying bug." + "reason": "Flash Attention 2 test failures and Trainer `_gpu_broadcast_one` NoneType errors are unrelated code paths." }, { "left": "issue:43611", "right": "issue:44466", "accept": false, - "reason": "Both concern model loading/serialization broadly, but the concrete failures differ: base_model_prefix loading vs lm_head.weight serialization consistency." + "reason": "Both are loading-related, but one is a `base_model_prefix` regression and the other is tied-weight serialization behavior." }, { - "left": "issue:43335", - "right": "issue:43504", + "left": "issue:43550", + "right": "issue:44112", "accept": false, - "reason": "Unrelated model/config bugs: SwitchTransformers sparse layer creation vs BEiT legacy field loading." + "reason": "Torch.compile/SDPA failure in Bamba is not the same bug as a stale GraniteSpeech device-override CI test." }, { - "left": "issue:45520", - "right": "issue:46170", + "left": "issue:43335", + "right": "issue:43504", "accept": false, - "reason": "Different import/dependency failures on different platforms; no shared concrete bug." + "reason": "Config incorrectly creating sparse layers and loading a legacy field are different model bugs." }, { "left": "issue:44855", "right": "issue:45083", "accept": false, - "reason": "One is a Python 3.13 IndentationError in DebertaV2 import, the other is a qwen3_omni_moe helper behavior issue." + "reason": "Python 3.13 import parsing failure and qwen3_omni_moe feature-length behavior are unrelated." + }, + { + "left": "issue:45520", + "right": "issue:46170", + "accept": false, + "reason": "Different platform/dependency issues: flash_attn import key error vs Windows ROCm torch.distributed dependency problems." }, { "left": "issue:44448", "right": "issue:44743", "accept": false, - "reason": "Different models and failures: Pegasus output drift vs Qwen3.5 recurrent-state reset." + "reason": "Pegasus v4/v5 output drift and Qwen modular recurrent-state reset are different underlying regressions." }, { "left": "issue:43335", "right": "issue:45083", "accept": false, - "reason": "Different components and symptoms; both are model internals, but not the same bug." + "reason": "SwitchTransformers sparse-layer creation and qwen3_omni_moe length helper behavior do not share the same bug." }, { "left": "issue:45290", "right": "issue:46326", "accept": false, - "reason": "Both involve chat templating, but one is tool-call tokenization crash and the other is mistral_common continue_final_message behavior." + "reason": "Both involve chat templates, but one crashes on tool-call assistant messages and the other concerns `continue_final_message` for mistral_common." }, { - "left": "issue:44514", - "right": "issue:44521", + "left": "issue:37428", + "right": "issue:44351", "accept": false, - "reason": "Same processor family, but one is a batching/padding crash while the other is an all-zero assistant mask bug; not clearly the same fix." + "reason": "Missing `_flash_supports_window_size` and missing `HybridCache` are distinct import errors." }, { - "left": "issue:37428", - "right": "issue:44351", + "left": "issue:44514", + "right": "issue:44521", "accept": false, - "reason": "Both are import issues, but they affect different missing symbols and different modules." + "reason": "Same area, but one is a batched-input crash and the other is an all-zero assistant_mask bug; not the same concrete defect." }, { "left": "issue:43335", "right": "issue:45593", "accept": false, - "reason": "Different model families and bugs; no evidence they share a code path." + "reason": "SwitchTransformers sparse-layer config and D-FINE auxiliary-loss behavior are unrelated." }, { "left": "issue:43335", "right": "issue:44112", "accept": false, - "reason": "Different bugs in different models; no common concrete defect." + "reason": "These are different model/CI issues with no shared code path." }, { - "left": "issue:29127", - "right": "issue:32090", + "left": "issue:43278", + "right": "issue:43825", "accept": false, - "reason": "LayoutLMv3 error-message clarity and Trainer GPU broadcast failure are unrelated." + "reason": "Embedding dtype regression and pipeline translation error-message regression are unrelated." }, { - "left": "issue:43278", - "right": "issue:43825", + "left": "issue:29127", + "right": "issue:32090", "accept": false, - "reason": "Different symptoms and subsystems: dtype drift in evaluation vs pipeline error-message wording." + "reason": "LayoutLMv3 box-info messaging and Trainer GPU broadcast errors are different bugs." }, { "left": "issue:45588", "right": "issue:45859", "accept": false, - "reason": "Both touch Qwen/MoE/flash attention areas, but one is s_aux=None handling and the other is CPU offload rotary embedding compatibility." + "reason": "Flash-attention sink-less `s_aux=None` crash and Qwen3_5MoeTextRotaryEmbedding CPU-offload incompatibility are separate failures." }, { "left": "issue:42490", "right": "issue:43854", "accept": false, - "reason": "Local-path import warning and GLM-4.7-Flash unit-test loading failure are different issues." + "reason": "Local-path import warning and GLM-4.7-Flash test loading failure are not the same issue." }, { "left": "issue:34689", "right": "issue:35141", "accept": false, - "reason": "Different model-loading behaviors: Llama 3.2 Vision load breakage vs post_init reinitializing tied embeddings." + "reason": "Llama 3.2 vision model loading breakage and embedding reinitialization after resize are different regressions." }, { - "left": "issue:43720", - "right": "issue:44112", + "left": "issue:43819", + "right": "issue:44493", "accept": false, - "reason": "BitNet packed-weight loading and GraniteSpeech stale device override are unrelated." + "reason": "DAC latent/forward mismatch and unexpected position-id keys are unrelated bugs." }, { - "left": "issue:43819", - "right": "issue:44493", + "left": "issue:43720", + "right": "issue:44112", "accept": false, - "reason": "DAC latent reconstruction mismatch and unexpected position-id keys are different bugs." + "reason": "Packed-weight unloading under accelerate and a GraniteSpeech stale-device CI test are different problems." }, { "left": "issue:39692", "right": "issue:43526", "accept": false, - "reason": "Documentation/example issues and BeiT image processor label reduction bug are unrelated." + "reason": "Docs example mistakes and BeitImageProcessorFast label-reduction behavior are not the same underlying bug." }, { "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 and DiffusionGemma meta-device loading are unrelated." }, { "left": "issue:43606", "right": "issue:44857", "accept": false, - "reason": "CPU offload device mismatch and float16 AMP crash are different failures." + "reason": "CPU offload device mismatch in bark-small and float16 AMP crashes in LwDetrImageLoss are different failures." }, { "left": "issue:44448", "right": "issue:44779", "accept": false, - "reason": "Pegasus v4/v5 output mismatch and Deepseek tokenizer regression are separate tokenizer/model behaviors." + "reason": "Pegasus output differences and Deepseek tokenizer regressions are separate model-specific issues." }, { - "left": "issue:45310", - "right": "issue:45478", - "accept": true, - "reason": "Same title, same reported Qwen3.5 Moe from_pretrained failure, and clearly the same underlying bug." + "left": "issue:39692", + "right": "issue:44493", + "accept": false, + "reason": "SigLIP2 docs example errors and unexpected position-id keys have no shared root cause." }, { "left": "issue:28218", "right": "issue:32090", "accept": false, - "reason": "Tokenizer spacing bug and Trainer NoneType broadcast error are unrelated." + "reason": "Tokenizer extra-space behavior and Trainer broadcast NoneType errors are unrelated." }, { - "left": "issue:39692", - "right": "issue:44493", - "accept": false, - "reason": "SigLIP2 docs example problems and unexpected position-id keys are unrelated." + "left": "issue:45310", + "right": "issue:45478", + "accept": true, + "reason": "Exact duplicate titles and same Qwen3.5 MoE from_pretrained failure on transformers>=5.4.0." }, { "left": "issue:45459", "right": "issue:46234", "accept": false, - "reason": "Tokenizer protobuf-error masking and missing import_utils feature flag are different import/tokenizer issues." - }, - { - "left": "issue:26413", - "right": "issue:31501", - "accept": false, - "reason": "Resume-from-checkpoint epoch_iterator failure and FineWeb training startup failure are not the same bug." + "reason": "Tokenizer error masking when protobuf is missing and a missing `is_falcon_mamba_ssm_available` helper are different import/tokenizer issues." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative on the soft edges. The only true duplicate pair is the two Qwen3.5 Moe from_pretrained reports (45310/45478), and the rest are reasonably treated as unrelated or only superficially similar." + "feedback": "The summary is grounded in the packet: the only clearly duplicate-looking pair is 45310/45478, which share the exact title and same Qwen3.5 MoE from_pretrained regression. The remaining links are appropriately described as theme-level similarities rather than duplicates, and the soft-edge rejections stay conservative." }, "evaluator_used": true, "retried": false, @@ -9896,18 +9842,18 @@ "nodes": [ "issue:10000", "issue:15354", + "issue:26413", "issue:27957", "issue:29870", "issue:29942", + "issue:31501", "issue:32090", "issue:33290", "issue:33357", "issue:34689", "issue:35141", - "issue:37428", "issue:38175", "issue:40444", - "issue:42175", "issue:42490", "issue:43278", "issue:43329", @@ -9944,8 +9890,9 @@ "issue:47000" ], "soft_pairs": [ - "issue:45127|issue:45245", + "issue:26413|issue:31501", "issue:44617|issue:46697", + "issue:45127|issue:45245", "issue:43701|issue:45083", "issue:44206|issue:45823", "issue:32090|issue:33357", @@ -9969,206 +9916,198 @@ "issue:43278|issue:43381", "issue:44977|issue:45229", "issue:40444|issue:43526", - "issue:43927|issue:45698", - "issue:37428|issue:42175", - "issue:32090|issue:33290" + "issue:43927|issue:45698" ], - "prepared_review_unit_hash": "457715f94716caab32a48f805baacb79c3556787da7f2e4ae39638d5d6ab35f7", + "prepared_review_unit_hash": "9ef6d98eb75310625d98c2d3d2337e276c133d5155a5f08419ad19da7ac4befd", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 28, - "serialized_chars": 36163, - "estimated_input_tokens": 9041, - "estimated_eval_tokens": 18338 + "soft_pair_count": 27, + "serialized_chars": 35831, + "estimated_input_tokens": 8958, + "estimated_eval_tokens": 18172 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is a collection of mostly unrelated issue reports spanning tokenizer regressions, flash-attention, model loading, multimodal processors, training, and evaluation. None of the soft pairs appear to describe the same underlying bug or change, so I would not merge any of them.", - "confidence": 0.94, - "canonical_issue_reason": "issue:45920 is the strongest canonical issue candidate because it describes a broad, clearly scoped regression affecting multiple model families, with concrete impact and cross-cutting relevance.", + "summary": "This cluster is a grab bag of unrelated closed issues spanning tokenization, generation, multimodal processing, training, and platform-specific crashes. None of the soft pairs look like the same underlying bug or change, so no soft edge should be merged.", + "confidence": 0.84, + "canonical_issue_reason": "issue:45920 is the strongest canonical issue candidate because it describes a broad v5 regression affecting multiple model families, making it more representative and reusable than the single-model, single-path reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45920 is the best representative issue in the set: it is specific enough to be actionable yet broad enough to capture an important regression pattern.", + "best_issue_reason": "issue:45920 is the best issue to anchor this cluster: it is high-signal, broadly scoped, and clearly framed as a regression impacting several models rather than a narrow edge case.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:45127", - "right": "issue:45245", + "left": "issue:26413", + "right": "issue:31501", "accept": false, - "reason": "Extended-vocabulary LoRA collapse and a categories-limit RuntimeError are different failure modes in different code paths." + "reason": "Both mention training startup problems, but one is a resume/epoch_iterator failure and the other is FineWeb SLM not starting; the underlying causes and code paths are different." }, { "left": "issue:44617", "right": "issue:46697", "accept": false, - "reason": "CUDA OOM in Gemma4 inference is unrelated to Swin2SR over-padding in an image processor." + "reason": "One is a CUDA OOM in Sam3Video, the other is over-padding in Swin2SR image processing; different models and failure modes." + }, + { + "left": "issue:45127", + "right": "issue:45245", + "accept": false, + "reason": "LoRA/vocab merge collapse and a category-count limit error are unrelated bugs with no shared concrete code path." }, { "left": "issue:43701", "right": "issue:45083", "accept": false, - "reason": "Checkpoint key mismatch and a helper-length calculation bug in qwen3_omni_moe are unrelated." + "reason": "Resume/checkpoint key mismatch is unrelated to qwen3_omni_moe feature-length calculation behavior." }, { "left": "issue:44206", "right": "issue:45823", "accept": false, - "reason": "Unsupported center argument in LasrFeatureExtractor is unrelated to Gemma4 device-map mismatch during forward." + "reason": "A feature extractor regression and a device-map forward mismatch are distinct issues affecting different subsystems." }, { "left": "issue:32090", "right": "issue:33357", "accept": false, - "reason": "A Trainer broadcast TypeError and a MacOS bus error in CLIP loading are different bugs." + "reason": "A Trainer broadcast NoneType error and a MacOS bus error in CLIP are not the same bug." }, { "left": "issue:44857", "right": "issue:45072", "accept": false, - "reason": "Both mention dtype/AMP, but they affect different models and different failure points, so they are not the same bug." + "reason": "Both involve dtype/AMP wording, but they are separate model-specific crashes with different execution paths and symptoms." }, { "left": "issue:42490", "right": "issue:43575", "accept": false, - "reason": "A local-path import warning and a tensor-parallel OOM on Qwen2-57B are unrelated." + "reason": "A trailing-slash local-path import warning is unrelated to tensor-parallel OOM when loading Qwen2-57B-A14B-Instruct." }, { "left": "issue:45593", "right": "issue:47000", "accept": false, - "reason": "D-FINE auxiliary-loss handling and RTDetrLoss infeasible cost matrix are separate detection-loss issues." + "reason": "Both are detection-loss related, but denoising/aux-loss behavior and an infeasible cost matrix are different defects." }, { "left": "issue:43550", "right": "issue:45083", "accept": false, - "reason": "Bamba torch.compile/SDPA failure is unrelated to qwen3_omni_moe feature-extraction length behavior." + "reason": "A torch.compile/SDPA failure in Bamba-9B-v2 is unrelated to qwen3_omni_moe feature-length helper behavior." }, { "left": "issue:27957", "right": "issue:34689", "accept": false, - "reason": "Flash Attention 2 support for XLMRoberta and Llama 3.2 Vision model loading are different model-specific problems." + "reason": "Flash Attention 2 support issues for XLMRoberta and a Llama 3.2 Vision loading regression are different model-path problems." }, { "left": "issue:38175", "right": "issue:44493", "accept": false, - "reason": "Unexpected zero probabilities in SigLIP2 and unexpected position-id keys are not the same underlying issue." + "reason": "Unexpected zero probabilities in SigLIP2 and unexpected position-id keys are not the same underlying bug." }, { "left": "issue:29942", "right": "issue:35141", "accept": false, - "reason": "Failing Flash Attention 2 tests and embedding reinitialization after resizing tokens are unrelated." + "reason": "Failing Flash Attention 2 tests and output embedding reinitialization after resizing vocab are unrelated changes." }, { "left": "issue:29870", "right": "issue:32090", "accept": false, - "reason": "Decoder attention mask initialization bug and GPU broadcast NoneType error are different code paths." + "reason": "An UnboundLocalError in BART generation and a NoneType GPU broadcast error occur in different parts of the stack." }, { "left": "issue:43278", "right": "issue:43329", "accept": false, - "reason": "BF16-to-FP32 dtype drift in eval is unrelated to undefined video-token helper names." + "reason": "A dtype drift between train/eval and an undefined-function bug in a multimodal token counter are unrelated." }, { "left": "issue:45290", "right": "issue:45381", "accept": false, - "reason": "Chat-template tool-call crash and qwen2.5-vl video vision-position-id bug are different multimodal failures." + "reason": "Chat-template tokenization with tool calls and qwen2.5-vl video vision_position_ids are separate multimodal bugs." }, { "left": "issue:10000", "right": "issue:15354", "accept": false, - "reason": "These are both old model/script-related issues, but they describe different models and different errors." + "reason": "German DistilBertModel and torch.jit.script GeneratorExp support are different historical issues with no shared code path." }, { "left": "issue:44448", "right": "issue:44977", "accept": false, - "reason": "Pegasus v4/v5 output divergence and Qwen3.5 flash-attention generation failure are unrelated." + "reason": "A cross-version Pegasus output difference and a Qwen3.5 flash-attention generation bug are not the same issue." }, { "left": "issue:45920", "right": "issue:46192", "accept": false, - "reason": "Tokenizer wrong-ID regression and silently ignored return_tensors are different tokenizer API bugs." + "reason": "Wrong token IDs in AutoTokenizer and return_tensors being ignored are distinct tokenizer API bugs." }, { "left": "issue:44360", "right": "issue:44561", "accept": false, - "reason": "A missing ReLU in DSA indexer is unrelated to the v5 removal of is_torch_fx_available." + "reason": "A DSA indexer ReLU discussion and removal of is_torch_fx_available breaking trust_remote_code models are unrelated." }, { "left": "issue:29870", "right": "issue:33290", "accept": false, - "reason": "Generate-time decoder mask error and deepspeed Adafactor OOM are not the same issue." + "reason": "BART generate-time masking error and deepspeed adafactor OOM are different failure classes." }, { "left": "issue:45290", "right": "issue:46655", "accept": false, - "reason": "Chat-template tool-call crash and SmolVLM video padding miscalculation are separate preprocessing bugs." + "reason": "Assistant tool-call chat templating and SmolVLM video padding are both multimodal but not the same concrete bug." }, { "left": "issue:44855", "right": "issue:46399", "accept": false, - "reason": "Python 3.13 comment/decorator parsing failure and missing decorator registration in Qwen3.5 are distinct decorator-related issues." + "reason": "Python 3.13 import parsing failure in DebertaV2Model and a missing decorator crash in Qwen3.5 are different defects." }, { "left": "issue:43278", "right": "issue:43381", "accept": false, - "reason": "Eval dtype mismatch and gradient-checkpointing-in-eval-mode are unrelated." + "reason": "Embedding dtype mismatch in evaluation is unrelated to the inability to use gradient checkpointing in eval mode." }, { "left": "issue:44977", "right": "issue:45229", "accept": false, - "reason": "Flash-attention generation bug and multi-GPU OOM are different classes of failure." + "reason": "Flash-attention generation issues for Qwen3.5 and multi-GPU OOM for Gemma4 are different problems." }, { "left": "issue:40444", "right": "issue:43526", "accept": false, - "reason": "IterableDataset multi-image finetuning failure and BeitImageProcessorFast label reduction are unrelated." + "reason": "IterableDataset multi-image finetuning failure and BeitImageProcessorFast reduce_labels behavior are unrelated." }, { "left": "issue:43927", "right": "issue:45698", "accept": false, - "reason": "Lost custom token IDs on save/load and wrong custom module loading after save_pretrained are both persistence bugs but not the same concrete defect." - }, - { - "left": "issue:37428", - "right": "issue:42175", - "accept": false, - "reason": "Flash-attention utility import failure and missing TensorFlow backend from a torch install are unrelated packaging/import issues." - }, - { - "left": "issue:44360", - "right": "issue:44561", - "accept": false, - "reason": "Duplicate check rejected: these are separate bugs in different subsystems with no shared failing code path." + "reason": "Losing custom token IDs after save/load and loading the wrong custom module after save_pretrained are both persistence-related, but they are different bugs and fixes." } ] }, @@ -10198,11 +10137,14 @@ "issue:29942", "issue:30064", "issue:32090", + "issue:33290", "issue:33357", "issue:34567", "issue:34689", + "issue:37428", "issue:39692", "issue:41720", + "issue:42175", "issue:42490", "issue:42915", "issue:42994", @@ -10212,13 +10154,11 @@ "issue:43611", "issue:43653", "issue:43701", - "issue:43756", "issue:43873", "issue:43976", "issue:44186", "issue:44206", "issue:44361", - "issue:44442", "issue:44448", "issue:44466", "issue:44479", @@ -10230,7 +10170,6 @@ "issue:44977", "issue:45198", "issue:45229", - "issue:45372", "issue:45440", "issue:45491", "issue:45563", @@ -10242,6 +10181,8 @@ "issue:46858" ], "soft_pairs": [ + "issue:37428|issue:42175", + "issue:32090|issue:33290", "issue:46421|issue:46858", "issue:43873|issue:45229", "issue:42490|issue:43873", @@ -10249,10 +10190,10 @@ "issue:43976|issue:45440", "issue:44448|issue:44466", "issue:29127|issue:34567", - "issue:44568|issue:45491", - "issue:44561|issue:44779", "issue:33357|issue:39692", "issue:43381|issue:43701", + "issue:44568|issue:45491", + "issue:44561|issue:44779", "issue:44871|issue:44977", "issue:43278|issue:43575", "issue:45588|issue:45823", @@ -10266,207 +10207,198 @@ "issue:43653|issue:44743", "issue:44206|issue:44479", "issue:29942|issue:30064", - "issue:42915|issue:42994", - "issue:45372|issue:45588", - "issue:43756|issue:45440", - "issue:43653|issue:44442" + "issue:42915|issue:42994" ], - "prepared_review_unit_hash": "b54b31ed6beb59d9a63ba832201390722d749cbdd2e0c27da05ad4d5e0f12776", + "prepared_review_unit_hash": "a24b4bc627e0d5181291b8c3b49becf6fd9d73df05a747f54f6e33c7bdf48afa", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 28, - "serialized_chars": 36286, - "estimated_input_tokens": 9072, - "estimated_eval_tokens": 18400 + "soft_pair_count": 27, + "serialized_chars": 35932, + "estimated_input_tokens": 8983, + "estimated_eval_tokens": 18222 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is highly heterogeneous: none of the soft pairs look like the same underlying bug/change, so I rejected all of them. There are no PRs in this cluster; the most representative issue overall is 43873 only because it is the most active and well-described, not because it is a duplicate target.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "No duplicate cluster here: the issues span unrelated bugs across generation, tokenization, training, quantization, cache logic, docs, and packaging. All soft edges look like similarity-only false positives.", + "confidence": 0.96, + "canonical_issue_reason": "No canonical issue: this cluster is heterogeneous and does not share one underlying bug or change.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43873 has the most discussion and inbound references, and it is a concrete, well-scoped active bug report; it is the best representative of this mixed cluster even though it is not a duplicate of the others.", + "best_issue_reason": "No strong global best issue; none of these issues safely represents the rest as a duplicate target.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:37428", + "right": "issue:42175", + "accept": false, + "reason": "Both are import/backend-related, but one is a missing flash-attention helper and the other is a pip extra omitting TensorFlow; different failures." + }, + { + "left": "issue:32090", + "right": "issue:33290", + "accept": false, + "reason": "Trainer broadcast NoneType vs Adafactor/DeepSpeed OOM are different code paths and error modes." + }, { "left": "issue:46421", "right": "issue:46858", "accept": false, - "reason": "Both mention cache/prefill behavior, but one is about spurious recompilation and the other about device-tensor seq_length breaking slice arithmetic on non-CUDA backends; different bugs." + "reason": "Both touch cache behavior, but one is lazy cache init causing recompilation and the other is device-tensor seq-length arithmetic on non-CUDA backends." }, { "left": "issue:43873", "right": "issue:45229", "accept": false, - "reason": "Quantization/offloading behavior vs multi-GPU CUDA OOM are different failure modes and likely different code paths." + "reason": "Offloading/quantization behavior is unrelated to Gemma4 multi-GPU CUDA OOM." }, { "left": "issue:42490", "right": "issue:43873", "accept": false, - "reason": "A local-path trailing-slash import warning is unrelated to quantization offloading." + "reason": "A trailing-slash import warning is not the same bug as quantization offloading malfunction." }, { "left": "issue:25251", "right": "issue:32090", "accept": false, - "reason": "Pipeline top_k output nesting and Trainer GPU broadcast NoneType error are unrelated." + "reason": "Pipeline top_k nesting changes output shape; the other is a distributed trainer broadcast TypeError." }, { "left": "issue:43976", "right": "issue:45440", "accept": false, - "reason": "Python version compatibility failure and DeepseekV3MoE implementation divergence are unrelated issues." + "reason": "Python version compatibility and DeepSeekV3MoE implementation divergence are unrelated." }, { "left": "issue:44448", "right": "issue:44466", "accept": false, - "reason": "Both are v4/v5 regressions, but one is generation output divergence and the other is model-weight serialization; not the same bug." + "reason": "Both involve v5-vs-v4 differences, but one is generation output drift and the other is serialization of tied weights." }, { "left": "issue:29127", "right": "issue:34567", "accept": false, - "reason": "LayoutLMv3 error messaging and TrainerState token-count tracking are unrelated." + "reason": "LayoutLMv3 error messaging for boxes and TrainerState token counting are unrelated." }, { - "left": "issue:44568", - "right": "issue:45491", + "left": "issue:33357", + "right": "issue:39692", "accept": false, - "reason": "Tokenizer special-token handling and Gemma3 NaN embeddings are different problems." + "reason": "MacOS bus error on CLIP load and SigLIP2 doc/model mismatch are different issues despite both involving vision models." }, { - "left": "issue:44561", - "right": "issue:44779", + "left": "issue:43381", + "right": "issue:43701", "accept": false, - "reason": "Missing helper removal breaking trust_remote_code models is unrelated to Deepseek tokenizer incorrect results." + "reason": "Gradient checkpointing in eval mode is not the same as a resume_from_checkpoint key mismatch." }, { - "left": "issue:33357", - "right": "issue:39692", + "left": "issue:44568", + "right": "issue:45491", "accept": false, - "reason": "MacOS bus error with a community CLIP model is not the same as SigLIP2 doc-example model/processor and quantization issues." + "reason": "Tokenizer special-token insertion and Gemma3 NaN embeddings come from different subsystems and failure mechanisms." }, { - "left": "issue:43381", - "right": "issue:43701", + "left": "issue:44561", + "right": "issue:44779", "accept": false, - "reason": "Eval-mode gradient checkpointing and resume-from-checkpoint key mismatch are different training-state bugs." + "reason": "Removal of an internal helper breaks trust_remote_code models; Deepseek tokenizer regression is a separate tokenizer bug." }, { "left": "issue:44871", "right": "issue:44977", "accept": false, - "reason": "Eos token config inconsistency is unrelated to flash-attention generation failure." + "reason": "Gemma eos_token_id config inconsistency and Qwen3.5 flash-attention generation failure are not the same underlying bug." }, { "left": "issue:43278", "right": "issue:43575", "accept": false, - "reason": "Embedding dtype drift during eval and tensor-parallel OOM when loading a large model are not the same bug." + "reason": "Embedding dtype drift in eval is unrelated to Qwen2-57B-A14B-Instruct tensor-parallel OOM." }, { "left": "issue:45588", "right": "issue:45823", "accept": false, - "reason": "Flash-attention s_aux=None crash and Gemma4 device-mapping mismatch are different forward-pass failures." + "reason": "A flash_attention.py None handling crash and a Gemma4 PLE device mismatch are different code-path problems." }, { "left": "issue:45198", "right": "issue:45803", "accept": false, - "reason": "Wav2Vec2 save/tokenization failures and Fuyu bare-except exception handling are unrelated." + "reason": "Wav2Vec2 save/tokenization failure and FuyuBatchFeature's bare except swallowing errors are unrelated." }, { "left": "issue:44977", "right": "issue:45563", "accept": false, - "reason": "A flash-attention generation regression and a stale warning for num_return_sequences are not the same issue." + "reason": "A generation regression and a stale warning in paged generate() are distinct issues." }, { "left": "issue:41720", "right": "issue:45229", "accept": false, - "reason": "Auto device mapping cudaErrorAssert and multi-GPU OOM are both CUDA-related but clearly distinct failures." + "reason": "Both concern large-model inference, but Qwen3 auto device mapping cuda asserts and Gemma4 multi-GPU OOM are different bugs." }, { "left": "issue:44186", "right": "issue:45198", "accept": false, - "reason": "LayoutLMv2 tokenizer crashes and Wav2Vec2 save/tokenization failures are unrelated tokenizer/model issues." + "reason": "LayoutLMv2 tokenizer crashes on NER/padding and Wav2Vec2 save/tokenization fail for different reasons." }, { "left": "issue:44361", "right": "issue:45198", "accept": false, - "reason": "MLuke tokenizer AttributeError and Wav2Vec2 save/tokenization failure do not share the same underlying bug." + "reason": "MLukeTokenizer AttributeError is not the same as Wav2Vec2 serialization/tokenization failure." }, { "left": "issue:43611", "right": "issue:46121", "accept": false, - "reason": "base_model_prefix loading regression and rope-parameter JSON list type error are unrelated." + "reason": "Model loading with base_model_prefix and rope validation TypeError are separate configuration/loading issues." }, { "left": "issue:29870", "right": "issue:34689", "accept": false, - "reason": "BART generate decoder_attention_mask initialization error and Llama Vision model loading breakage are different failures." + "reason": "BART generate() decoder mask bug and Llama 3.2 Vision loading breakage are unrelated regressions." }, { "left": "issue:43653", "right": "issue:44743", "accept": false, - "reason": "BigBirdTokenizer special-token registration and Qwen3.5 recurrent-state reset are unrelated." + "reason": "BigBirdTokenizer special-token registration and Qwen3.5 recurrent-state reset with cache are different bugs." }, { "left": "issue:44206", "right": "issue:44479", "accept": false, - "reason": "Unsupported center arg in an extractor and video-input regression for Qwen multimodal models are different code paths." + "reason": "Unsupported center arg in a feature extractor and a video-input regression in Qwen models are not the same issue." }, { "left": "issue:29942", "right": "issue:30064", "accept": false, - "reason": "Flash Attention 2 test failures and void segmentation map processing failure are unrelated." + "reason": "Failing Flash Attention 2 tests and void segmentation-map processing are unrelated." }, { "left": "issue:42915", "right": "issue:42994", "accept": false, - "reason": "FineGrainedFP8Config load failure and quantized-model save failure are both quantization-related but not the same bug." - }, - { - "left": "issue:45372", - "right": "issue:45588", - "accept": false, - "reason": "Gemma 4 processor loading ImportError and flash-attention s_aux=None crash are unrelated." - }, - { - "left": "issue:43756", - "right": "issue:45440", - "accept": false, - "reason": "Smollm3 RoPE-layer mismatch and DeepSeekV3MoE divergence from remote implementation are different architecture bugs." - }, - { - "left": "issue:43653", - "right": "issue:44442", - "accept": false, - "reason": "BigBirdTokenizer mask-token bug and AutoTokenizer failing to load FastSpeech2ConformerTokenizer are not the same issue." + "reason": "Both mention quantization, but one is a FineGrainedFP8Config model failure and the other is quantized model saving." } ] }, @@ -10500,14 +10432,14 @@ "issue:42915", "issue:43065", "issue:43066", - "issue:43334", "issue:43381", "issue:43412", "issue:43421", "issue:43526", - "issue:43531", "issue:43575", + "issue:43653", "issue:43701", + "issue:43756", "issue:43824", "issue:43931", "issue:43994", @@ -10539,22 +10471,25 @@ "issue:46701" ], "soft_pairs": [ - "issue:43412|issue:45588", + "issue:43756|issue:45440", + "issue:45372|issue:45588", + "issue:43653|issue:44442", "issue:43065|issue:43994", - "issue:45491|issue:46693", - "issue:45459|issue:45712", + "issue:43412|issue:45588", "issue:44265|issue:45198", + "issue:45459|issue:45712", + "issue:45491|issue:46693", "issue:29127|issue:33357", "issue:43065|issue:45412", "issue:43575|issue:43701", - "issue:45372|issue:45406", "issue:45245|issue:45563", + "issue:45372|issue:45406", "issue:42490|issue:43381", "issue:29127|issue:35141", "issue:45440|issue:45593", "issue:43994|issue:45412", - "issue:45372|issue:45459", "issue:43421|issue:45491", + "issue:45372|issue:45459", "issue:45406|issue:46701", "issue:45412|issue:46369", "issue:43575|issue:44387", @@ -10562,250 +10497,261 @@ "issue:44442|issue:45072", "issue:44464|issue:45584", "issue:43526|issue:45834", - "issue:42915|issue:43931", - "issue:43526|issue:46693", "issue:43824|issue:44279", + "issue:43526|issue:46693", + "issue:42915|issue:43931", "issue:27957|issue:33290", "issue:27957|issue:29127", "issue:44857|issue:45588", "issue:43994|issue:46369", "issue:33290|issue:34689", "issue:33290|issue:35141", - "issue:43526|issue:44186", - "issue:43334|issue:43531" + "issue:43526|issue:44186" ], - "prepared_review_unit_hash": "da417fd81a381e04161c92f5d3d62d17423ccaea5349d7d0c15632bff124c953", + "prepared_review_unit_hash": "2cf55ff4837b7c0abb74d86dcaa7823cf97219c2cd01cdcb1ef37564de248f7b", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 34, - "serialized_chars": 37198, - "estimated_input_tokens": 9300, - "estimated_eval_tokens": 18856 + "soft_pair_count": 36, + "serialized_chars": 37825, + "estimated_input_tokens": 9457, + "estimated_eval_tokens": 19170 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is highly heterogeneous: the items cover unrelated model-loading, tokenizer, memory, export, and generation bugs, and none of the soft pairs look like the same underlying change or defect.", - "confidence": 0.97, - "canonical_issue_reason": "No reliable duplicate canonical exists here; the issues span different subsystems and root causes, so any single canonical issue would be arbitrary.", + "summary": "These soft pairs are not true duplicates: they span unrelated issues across tokenizers, model loading, memory/OOM, export, and model-specific regressions. The similarity is mostly lexical or subsystem-level, not the same concrete bug.", + "confidence": 0.98, + "canonical_issue_reason": "There is no single canonical issue for this cluster because the items describe different bugs and features rather than one underlying defect.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44462 is the best representative because it is a clear, user-facing AutoTokenizer regression with concrete reproduction and broad applicability.", + "best_issue_reason": "If an anchor is needed, issue:44462 is the most generic user-facing loading bug, but it still does not represent the cluster as a whole.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43412", + "left": "issue:43756", + "right": "issue:45440", + "accept": false, + "reason": "Different models and failure modes: RoPE-layer count vs MoE divergence." + }, + { + "left": "issue:45372", "right": "issue:45588", "accept": false, - "reason": "Different failures: Gemma3n export/guard vs flash-attention None-handling." + "reason": "Gemma processor import error vs flash-attention auxiliary-state crash." + }, + { + "left": "issue:43653", + "right": "issue:44442", + "accept": false, + "reason": "BigBird special-token registration bug vs tokenizer class loading failure." }, { "left": "issue:43065", "right": "issue:43994", "accept": false, - "reason": "A dummy pixel-decoder layer bug and a SigLIP pipeline/model-loading issue are unrelated." + "reason": "Dummy Conv2d in Sam3 decoder vs SigLIP2 inference corruption." }, { - "left": "issue:45491", - "right": "issue:46693", + "left": "issue:43412", + "right": "issue:45588", + "accept": false, + "reason": "ExecuTorch export guard/erfinv issue vs flash-attention None-handling bug." + }, + { + "left": "issue:44265", + "right": "issue:45198", "accept": false, - "reason": "Both touch FA2/long-context, but one is a NaN correctness bug and the other is a performance regression." + "reason": "torch.export failure with torch_compilable_check vs Wav2Vec2 save/tokenization bug." }, { "left": "issue:45459", "right": "issue:45712", "accept": false, - "reason": "Tokenizer/protobuf error masking vs leftover dummy classes are different issues." + "reason": "Tokenizer error masking vs leftover dummy class cleanup." }, { - "left": "issue:44265", - "right": "issue:45198", + "left": "issue:45491", + "right": "issue:46693", "accept": false, - "reason": "torch.export with torch_compilable_check is unrelated to Wav2Vec2 save/tokenization failure." + "reason": "NaN embeddings bug vs FA2 performance regression." }, { "left": "issue:29127", "right": "issue:33357", "accept": false, - "reason": "An error-message clarity issue is not the same as a MacOS bus error." + "reason": "LayoutLMv3 error messaging vs CLIP bus error on macOS." }, { "left": "issue:43065", "right": "issue:45412", "accept": false, - "reason": "RT-DETR memory not released on GC is unrelated to a Qwen2 TP OOM." + "reason": "Dummy decoder layer bug vs RT-DETR memory release issue." }, { "left": "issue:43575", "right": "issue:43701", "accept": false, - "reason": "A model-loading OOM and a resume_from_checkpoint key mismatch are different bugs." + "reason": "Tensor-parallel OOM vs checkpoint resume key mismatch." }, { - "left": "issue:45372", - "right": "issue:45406", + "left": "issue:45245", + "right": "issue:45563", "accept": false, - "reason": "Gemma 4 dependency import failure and Gemma4Processor missing _tokenizer are different code paths." + "reason": "Category-count limit error vs stale generation warning." }, { - "left": "issue:45245", - "right": "issue:45563", + "left": "issue:45372", + "right": "issue:45406", "accept": false, - "reason": "A 2^24 category limit is unrelated to a stale paged-generate warning." + "reason": "Processor import dependency failure vs missing _tokenizer attribute in serve." }, { "left": "issue:42490", "right": "issue:43381", "accept": false, - "reason": "Local-path slash warning and gradient-checkpointing eval-mode restriction are different problems." + "reason": "Trailing-slash import warning vs gradient checkpointing eval-mode restriction." }, { "left": "issue:29127", "right": "issue:35141", "accept": false, - "reason": "LayoutLMv3 error-message wording vs embedding reinitialization are unrelated." + "reason": "LayoutLMv3 box-validation messaging vs embedding reinitialization bug." }, { "left": "issue:45440", "right": "issue:45593", "accept": false, - "reason": "DeepseekV3MoE implementation divergence and D-FINE auxiliary-loss behavior are different model bugs." + "reason": "DeepSeekV3 remote/native divergence vs D-FINE auxiliary-loss issue." }, { "left": "issue:43994", "right": "issue:45412", "accept": false, - "reason": "SigLIP2 output/pipeline failure and RT-DETR memory retention do not share a concrete root cause." + "reason": "SigLIP2 bad outputs vs RT-DETR memory leak." }, { - "left": "issue:45372", - "right": "issue:45459", + "left": "issue:43421", + "right": "issue:45491", "accept": false, - "reason": "Importing ReasoningEffort from mistral_common is unrelated to protobuf error masking." + "reason": "Runtime post-processor update bug vs Gemma3 NaN embeddings." }, { - "left": "issue:43421", - "right": "issue:45491", + "left": "issue:45372", + "right": "issue:45459", "accept": false, - "reason": "Runtime special-token update behavior and Gemma3 NaNs are different defects." + "reason": "Import dependency breakage vs overly broad tokenizer exception handling." }, { "left": "issue:45406", "right": "issue:46701", "accept": false, - "reason": "Gemma4 serve crash and DiffusionGemma image-text pipeline failure are distinct paths." + "reason": "Gemma4 processor attribute error vs DiffusionGemma pipeline failure." }, { "left": "issue:45412", "right": "issue:46369", "accept": false, - "reason": "Memory not released on delete vs memory overflow during model call are not the same bug." + "reason": "RT-DETR cleanup leak vs SAM3 memory overflow." }, { "left": "issue:43575", "right": "issue:44387", "accept": false, - "reason": "TP loading OOM and int4 reserved-memory regression are different root causes." + "reason": "Tensor-parallel load OOM vs int4 quantization reserved-memory regression." }, { "left": "issue:43066", "right": "issue:44462", "accept": false, - "reason": "Wrong decoder type in v5 and ignoring tokenizer.json are separate tokenizer-loading bugs." + "reason": "Wrong tokenizer decoder type vs AutoTokenizer ignoring tokenizer.json." }, { "left": "issue:44442", "right": "issue:45072", "accept": false, - "reason": "FastSpeech2 tokenizer load failure and bfloat16 dtype mismatch are unrelated." + "reason": "Tokenizer load failure vs bfloat16 dtype mismatch in inference." }, { "left": "issue:44464", "right": "issue:45584", "accept": false, - "reason": "Compiled-forward inconsistency and Whisper empty-transcription failure are different generation bugs." + "reason": "Compiled-forward generation inconsistency vs Whisper empty-transcription failure." }, { "left": "issue:43526", "right": "issue:45834", "accept": false, - "reason": "BeitImageProcessorFast label reduction and Kosmos2.5 OCR index error are unrelated processor issues." + "reason": "BeitImageProcessorFast label reduction bug vs Kosmos2.5 OCR index error." }, { - "left": "issue:42915", - "right": "issue:43931", + "left": "issue:43824", + "right": "issue:44279", "accept": false, - "reason": "Qwen3Moe FP8 config failure and Qwen3-VL weight-shape mismatch are different problems." + "reason": "Missing Qwen2.5-VL import vs generic dependency issue." }, { "left": "issue:43526", "right": "issue:46693", "accept": false, - "reason": "Label reduction in an image processor is unrelated to an FA2 long-context performance regression." + "reason": "Beit label reduction bug vs long-context FA2 performance regression." }, { - "left": "issue:43824", - "right": "issue:44279", + "left": "issue:42915", + "right": "issue:43931", "accept": false, - "reason": "Qwen2.5-VL import failure and a generic dependency issue are not the same defect." + "reason": "Qwen3Moe FP8 config failure vs Qwen3-VL weight-shape mismatch." }, { "left": "issue:27957", "right": "issue:33290", "accept": false, - "reason": "XLMRoberta FA2 support and Adafactor/DeepSpeed OOM are unrelated." + "reason": "XLM-R FlashAttention2 support vs Adafactor/deepspeed OOM." }, { "left": "issue:27957", "right": "issue:29127", "accept": false, - "reason": "FA2 XLMRoberta support and LayoutLMv3 error-message clarity are different issues." + "reason": "FlashAttention2 model support vs LayoutLMv3 error clarity." }, { "left": "issue:44857", "right": "issue:45588", "accept": false, - "reason": "AMP/CUDA crash in LwDetrImageLoss and flash-attention s_aux handling are different failures." + "reason": "AMP float16 crash in loss vs flash-attention aux-state None crash." }, { "left": "issue:43994", "right": "issue:46369", "accept": false, - "reason": "SigLIP2 loading/pipeline failure and Sam3-litetext memory overflow are unrelated." + "reason": "SigLIP2 inference bug vs SAM3 lite text memory overflow." }, { "left": "issue:33290", "right": "issue:34689", "accept": false, - "reason": "Adafactor OOM and Llama 3.2 Vision loading breakage are different bugs." + "reason": "Optimizer OOM vs model-loading breakage for Llama 3.2 Vision." }, { "left": "issue:33290", "right": "issue:35141", "accept": false, - "reason": "CUDA memory OOM and embedding reinitialization after resize are unrelated." - }, - { - "left": "issue:43526", - "right": "issue:44186", - "accept": false, - "reason": "BeitImageProcessorFast label reduction and LayoutLMv2Tokenizer NER/padding crashes are different processor bugs." + "reason": "Adafactor OOM vs post-init embedding reinitialization." }, { - "left": "issue:43334", - "right": "issue:43531", + "left": "issue:43526", + "right": "issue:44186", "accept": false, - "reason": "Qwen3-VL missing pad_token_id and Qwen3-MoE sliding_window behavior are different model-family root causes." + "reason": "Image label reduction bug vs LayoutLMv2 tokenizer padding/truncation crash." } ] }, @@ -10835,9 +10781,11 @@ "issue:42490", "issue:43262", "issue:43316", + "issue:43334", "issue:43381", "issue:43493", "issue:43525", + "issue:43531", "issue:43643", "issue:43646", "issue:43720", @@ -10867,7 +10815,6 @@ "issue:45381", "issue:45440", "issue:45588", - "issue:45759", "issue:45859", "issue:45907", "issue:46121", @@ -10878,217 +10825,217 @@ "issue:46821" ], "soft_pairs": [ + "issue:43334|issue:43531", "issue:46121|issue:46399", - "issue:45250|issue:46640", "issue:43381|issue:43873", + "issue:45250|issue:46640", "issue:45588|issue:46821", "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:44442|issue:44898", "issue:42490|issue:44743", "issue:43720|issue:44442", - "issue:45003|issue:45440", "issue:42175|issue:46234", + "issue:45003|issue:45440", "issue:27957|issue:29870", "issue:44077|issue:44479", "issue:43493|issue:43827", "issue:43316|issue:43525", - "issue:45325|issue:46697", "issue:43873|issue:44871", + "issue:45325|issue:46697", "issue:44062|issue:45907", "issue:43643|issue:44913", "issue:44855|issue:45137", - "issue:43749|issue:43931", - "issue:43316|issue:45759" + "issue:43749|issue:43931" ], - "prepared_review_unit_hash": "98171725b4892e3a6d8b10b349fa2930e328294e161deb63754bba7effb10a11", + "prepared_review_unit_hash": "561d478d06a328b0c4f80017b029a9a3260021d61b0a81640a059828b61200cd", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 26, - "serialized_chars": 35017, - "estimated_input_tokens": 8755, - "estimated_eval_tokens": 17766 + "serialized_chars": 35551, + "estimated_input_tokens": 8888, + "estimated_eval_tokens": 18032 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These soft pairs are mostly superficial keyword matches across unrelated bugs, regressions, and docs/config issues. None look like the same underlying defect or a mergeable duplicate pair, so no canonical issue/PR is selected.", - "confidence": 0.93, - "canonical_issue_reason": null, + "summary": "This cluster is highly heterogeneous: the paired items share only broad topical overlap (Flash Attention, Qwen, loading/config regressions, audio/vision processors), but they describe different code paths and failure modes. None of the soft edges look like true duplicates or mergeable into one fix.", + "confidence": 0.94, + "canonical_issue_reason": "No single canonical issue; the items span many unrelated bugs and regressions rather than one underlying defect.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "No single best issue; none of these issues subsumes the others or represents the whole cluster cleanly.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:46121", - "right": "issue:46399", + "left": "issue:43334", + "right": "issue:43531", "accept": false, - "reason": "Different problems: rope-validation TypeError on JSON-loaded lists vs a Qwen3.5 kernel-registration decorator error." + "reason": "Both involve Qwen text configs, but one is a missing pad_token_id attribute and the other is a sliding_window behavior bug; different failures and fixes." }, { - "left": "issue:45250", - "right": "issue:46640", + "left": "issue:46121", + "right": "issue:46399", "accept": false, - "reason": "Generic Flash Attention topic vs a specific sink-less-model AttributeError in flash_attention.py; not the same bug." + "reason": "Rope-validation TypeError and Qwen3.5 kernel registration crash are unrelated code paths; only a loose Qwen/rope overlap." }, { "left": "issue:43381", "right": "issue:43873", "accept": false, - "reason": "Gradient-checkpointing eval-mode behavior and quantization/offloading are separate code paths and failures." + "reason": "Eval-mode gradient checkpointing and quantization offloading are distinct runtime behaviors with different root causes." + }, + { + "left": "issue:45250", + "right": "issue:46640", + "accept": false, + "reason": "Both mention Flash Attention 2, but the titles are too generic to show the same concrete bug or fix." }, { "left": "issue:45588", "right": "issue:46821", "accept": false, - "reason": "Different subsystems: flash attention crash on s_aux=None vs TimesFM window_size AttributeError." + "reason": "Flash-attention sink handling and TimesFM window_size AttributeError are unrelated subsystems and errors." }, { "left": "issue:44514", "right": "issue:45381", "accept": false, - "reason": "Batched chat-template padding crash vs wrong video position_ids/rope indexing in Qwen2.5-VL; distinct symptoms." + "reason": "These are separate Qwen2.5-VL bugs: batched chat-template padding versus incorrect video position IDs." }, { "left": "issue:43646", "right": "issue:44077", "accept": false, - "reason": "Custom model init breakage is unrelated to patchtsmixer's post_init constraint." + "reason": "A config-loading discrepancy is not the same as patchtsmixer post_init API enforcement; different model-init issues." }, { "left": "issue:43761", "right": "issue:45859", "accept": false, - "reason": "CLIPVision hidden_states regression and Qwen3_5Moe CPU-offload incompatibility are unrelated." + "reason": "CLIPVisionModel hidden_states regression and Qwen3_5Moe CPU offload incompatibility affect different models and paths." }, { "left": "issue:44977", "right": "issue:45245", "accept": false, - "reason": "Flash-attention generation bug vs category-count limit runtime error; no shared underlying code path." + "reason": "Qwen3.5 flash-attention generation failure and a categories-limit runtime error are unrelated." }, { "left": "issue:44743", "right": "issue:44811", "accept": false, - "reason": "Recurrent-state reset in modular_qwen3_5.py is unrelated to Whisper batch_decode skip_special_tokens handling." + "reason": "Recurrent state reset with cache is unrelated to Whisper batch_decode ignoring skip_special_tokens." }, { - "left": "issue:44442", - "right": "issue:44898", + "left": "issue:43262", + "right": "issue:44206", "accept": false, - "reason": "Tokenizer-loading failure for FastSpeech2Conformer vs Perceiver image interpolation failure; different models and APIs." + "reason": "Audio chat-template sampling-rate defaulting and a feature-extractor center-arg crash are different preprocessing bugs." }, { - "left": "issue:43262", - "right": "issue:44206", + "left": "issue:44442", + "right": "issue:44898", "accept": false, - "reason": "Audio chat-template sampling-rate default and LASR feature-extractor center-arg crash are separate issues." + "reason": "Tokenizer loading failure and Perceiver image-classification size handling are separate issues in different components." }, { "left": "issue:42490", "right": "issue:44743", "accept": false, - "reason": "Local-path import warning is unrelated to Qwen recurrent-state reset behavior." + "reason": "Local-path import warnings and Qwen recurrent-state resetting do not share a concrete code path." }, { "left": "issue:43720", "right": "issue:44442", "accept": false, - "reason": "BitNet packed-weight loading bug and FastSpeech2Conformer tokenizer loading bug do not share the same defect." + "reason": "BitNet packed-weight loading and tokenizer loading for FastSpeech2Conformer are unrelated failures." }, { - "left": "issue:45003", - "right": "issue:45440", + "left": "issue:42175", + "right": "issue:46234", "accept": false, - "reason": "Unsafe sys.modules access and DeepSeekV3MoE implementation divergence are different problems." + "reason": "pip extras dependency inclusion and a missing import_utils capability flag are different packaging/import problems." }, { - "left": "issue:42175", - "right": "issue:46234", + "left": "issue:45003", + "right": "issue:45440", "accept": false, - "reason": "Packaging/backend extras issue is unrelated to a missing import_utils helper symbol." + "reason": "Unsafe sys.modules access in modeling_utils and DeepseekV3MoE divergence are not the same bug." }, { "left": "issue:27957", "right": "issue:29870", "accept": false, - "reason": "XLMRoberta Flash Attention 2 support and Bart generate's decoder_attention_mask bug are different failures." + "reason": "XLMRoberta Flash Attention 2 support and BART generate() decoder_attention_mask assignment are unrelated issues." }, { "left": "issue:44077", "right": "issue:44479", "accept": false, - "reason": "patchtsmixer post_init validation and Qwen video-input regression are unrelated." + "reason": "patchtsmixer post_init policy and Qwen video-input regression are different model families and fixes." }, { "left": "issue:43493", "right": "issue:43827", "accept": false, - "reason": "SigLIP2 implementation mismatch vs docs still referencing pipeline() after removals; different scope." + "reason": "SigLIP2 implementation discrepancy and docs still referencing pipeline() are not the same problem." }, { "left": "issue:43316", "right": "issue:43525", "accept": false, - "reason": "Gemma3TextConfig API discrepancy and Llama4Config missing pad_token_id are separate config issues." + "reason": "Gemma3TextConfig API mismatch and Llama4Config missing pad_token_id are different config-field issues." }, { - "left": "issue:45325", - "right": "issue:46697", + "left": "issue:43873", + "right": "issue:44871", "accept": false, - "reason": "Qwen2.5-VL rope-index scaling bug and Swin2SR over-padding are unrelated image/video preprocessing issues." + "reason": "Quantization offloading behavior and Gemma-3 eos_token_id inconsistency involve different mechanisms and symptoms." }, { - "left": "issue:43873", - "right": "issue:44871", + "left": "issue:45325", + "right": "issue:46697", "accept": false, - "reason": "Quantization offloading bug and Gemma-3 eos_token_id inconsistency are not the same defect." + "reason": "Qwen2.5-VL rope indexing and Swin2SR image padding are unrelated vision preprocessing bugs." }, { "left": "issue:44062", "right": "issue:45907", "accept": false, - "reason": "AddedToken special-arg duplication vs list_repo_files revision forwarding are unrelated API bugs." + "reason": "AddedToken special-argument collision and list_repo_files revision forwarding are separate API bugs." }, { "left": "issue:43643", "right": "issue:44913", "accept": false, - "reason": "trust_remote_code missing fields and GPTNeoXConfig rotary_pct reload regression are separate config behaviors." + "reason": "Missing fields from trust_remote_code loading and GPTNeoX rotary_pct reload behavior are different config-serialization issues." }, { "left": "issue:44855", "right": "issue:45137", "accept": false, - "reason": "Python 3.13 JIT parse/IndentationError and DeepSpeed ZeRO3 deque pop error are unrelated." + "reason": "Python 3.13 torch.jit parsing/import failure and DeepSpeed ZeRO3 deque underflow are unrelated runtime errors." }, { "left": "issue:43749", "right": "issue:43931", "accept": false, - "reason": "FSDP CPU RAM efficient loading regression and Qwen3-VL shape mismatch are different loading failures." - }, - { - "left": "issue:43316", - "right": "issue:45759", - "accept": false, - "reason": "Gemma3TextConfig API discrepancy and Qwen multimodal text_config unwrapping are different model/config issues." + "reason": "FSDP CPU-efficient loading and Qwen3-VL weight-shape mismatch are distinct model-loading failures." } ] }, @@ -11149,12 +11096,12 @@ "issue:45216", "issue:45584", "issue:45698", + "issue:45759", "issue:45803", "issue:46050", "issue:46178", "issue:46234", "issue:46291", - "issue:46365", "issue:46421", "issue:46506", "issue:46821", @@ -11162,9 +11109,10 @@ "issue:46932" ], "soft_pairs": [ + "issue:43316|issue:45759", "issue:44987|issue:46178", - "issue:44190|issue:44496", "issue:44442|issue:45803", + "issue:44190|issue:44496", "issue:43232|issue:43388", "issue:43931|issue:45216", "issue:43761|issue:44079", @@ -11175,8 +11123,8 @@ "issue:43575|issue:43742", "issue:43550|issue:43701", "issue:27957|issue:32090", - "issue:42907|issue:45698", "issue:25251|issue:28282", + "issue:42907|issue:45698", "issue:43606|issue:43701", "issue:16998|issue:22355", "issue:43295|issue:43825", @@ -11189,211 +11137,210 @@ "issue:43316|issue:46821", "issue:44871|issue:46050", "issue:43646|issue:44913", - "issue:43452|issue:44496", - "issue:45803|issue:46365" + "issue:43452|issue:44496" ], - "prepared_review_unit_hash": "a11c4eb55698506608bccdcbb4b357c92d391e2d59109bce5c38fb5154bdb1b7", + "prepared_review_unit_hash": "865a85f3d6622af55f73c75ebfe7f40f9b1a3539a86b465089efc75e0b3c86cb", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36295, - "estimated_input_tokens": 9074, - "estimated_eval_tokens": 18404 + "serialized_chars": 36342, + "estimated_input_tokens": 9086, + "estimated_eval_tokens": 18428 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous issue reports with only weak title similarity. I found no convincing duplicate pairs among the soft edges. The most suitable canonical issue is the save/load serialization regression around `from_pretrained` and `save_pretrained` (#45698).", - "confidence": 0.34, - "canonical_issue_reason": "#45698 is the clearest, most broadly actionable bug in the set and has the strongest signals of centrality (multiple inbound refs, high discussion). It also represents a common duplicate theme: model save/load behavior breaking after serialization.", + "summary": "These are mostly thematically similar but distinct bugs. I would not merge any of the proposed soft pairs: they span different models, different APIs, and different failure modes rather than the same concrete code-path issue.", + "confidence": 0.94, + "canonical_issue_reason": "No single canonical issue stands out, because none of the soft pairs are strong duplicate matches; they only share broad subsystem similarity.", "canonical_pr_reason": null, - "best_issue_reason": "#45698 is the best single representative issue here because it is concrete, self-contained, and likely to attract related save/load regressions, unlike the more niche or one-off reports in the cluster.", + "best_issue_reason": "No issue is clearly a global best representative for a duplicate cluster, since the candidates are largely unrelated beyond being Transformers regressions/load failures.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44987", - "right": "issue:46178", + "left": "issue:43316", + "right": "issue:45759", "accept": false, - "reason": "Both are import/load failures, but they point to different mechanisms and code paths: one is a model loading regression, the other is a missing backend registration in `import_utils`." + "reason": "Both involve config/model construction, but they target different composite-model APIs and different bugs." }, { - "left": "issue:44190", - "right": "issue:44496", + "left": "issue:44987", + "right": "issue:46178", "accept": false, - "reason": "Different problem domains: local dataset loading in a training example vs model config/model-type recognition during `from_pretrained`." + "reason": "One is a model load regression for a specific repo; the other is an import-utils backend mapping error." }, { "left": "issue:44442", "right": "issue:45803", "accept": false, - "reason": "Tokenizer loading failure and an unrelated exception-handling bug in image feature conversion are not the same underlying defect." + "reason": "Tokenizer loading failure and a bare-except tensor conversion bug are unrelated code paths." + }, + { + "left": "issue:44190", + "right": "issue:44496", + "accept": false, + "reason": "A dataset/script issue and an unrecognized-model/config-json issue are not the same bug." }, { "left": "issue:43232", "right": "issue:43388", "accept": false, - "reason": "Generation kwargs sync behavior and metric gathering/truncation in training are separate code paths and separate bugs." + "reason": "Generation kwargs sync and metric gathering last-batch handling are separate trainer behaviors." }, { "left": "issue:43931", "right": "issue:45216", "accept": false, - "reason": "Both involve Qwen3-VL/Qwen3.5 loading/saving, but one is a shape-mismatch load error and the other is a save_pretrained regression; not the same concrete bug." + "reason": "Both mention Qwen VL, but one is weight-shape loading failure and the other is save_pretrained checkpoint corruption." }, { "left": "issue:43761", "right": "issue:44079", "accept": false, - "reason": "A missing `hidden_states` output in CLIP vision forward is not clearly the same as the generic `ModelOutput` key assignment issue." + "reason": "Hidden states not returned is different from ModelOutput key assignment semantics." }, { "left": "issue:43761", "right": "issue:44188", "accept": false, - "reason": "One is an output-field bug in a vision model, the other is a torch.compile-specific attention kernel divergence." + "reason": "A CLIP output regression and an attention-kernel branching issue are unrelated." }, { "left": "issue:42617", "right": "issue:43575", "accept": false, - "reason": "These are both runtime/load problems, but one is a script execution issue and the other is an OOM during tensor-parallel model loading." + "reason": "One is a parallel script failure; the other is a tensor-parallel OOM on a specific model." }, { "left": "issue:43701", "right": "issue:46506", "accept": false, - "reason": "Checkpoint resume key mismatch and CPU/CUDA result divergence are unrelated bugs." + "reason": "Checkpoint key mismatch and CPU/CUDA output divergence are different failure modes." }, { "left": "issue:46829", "right": "issue:46932", "accept": false, - "reason": "Different models and different failure modes: an output field mix-up vs a quantized Gemma4 dtype casting fix." + "reason": "Image classification output misassignment and Gemma4 dtype casting are unrelated." }, { "left": "issue:43575", "right": "issue:43742", "accept": false, - "reason": "OOM under tensor parallel loading and a key error loading a specific model are different load-time failures." + "reason": "OOM under tensor parallelism is not the same as a key error while loading another model." }, { "left": "issue:43550", "right": "issue:43701", "accept": false, - "reason": "torch.compile+SDPA failure and checkpoint resume key mismatch do not describe the same code-path problem." + "reason": "torch.compile SDPA failure and resume_from_checkpoint key mismatch are different code paths." }, { "left": "issue:27957", "right": "issue:32090", "accept": false, - "reason": "Flash Attention support and a Trainer broadcast TypeError are unrelated." + "reason": "Flash Attention behavior and a Trainer broadcast TypeError are unrelated." }, { - "left": "issue:42907", - "right": "issue:45698", + "left": "issue:25251", + "right": "issue:28282", "accept": false, - "reason": "Both touch save/load behavior, but one is about dequantized model saving and the other about loading the wrong custom module after save_pretrained; not the same bug." + "reason": "Pipeline top_k output-shape change is unrelated to missing PyTorch import error." }, { - "left": "issue:25251", - "right": "issue:28282", + "left": "issue:42907", + "right": "issue:45698", "accept": false, - "reason": "Pipeline output shape changes and a missing PyTorch import error are unrelated." + "reason": "Both concern save/load, but one is dequantized model saving and the other is custom-module resolution after reload." }, { "left": "issue:43606", "right": "issue:43701", "accept": false, - "reason": "CPU offload device mismatch and checkpoint resume key mismatch are different failures." + "reason": "CPU offload device mismatch is unrelated to checkpoint key mismatch." }, { "left": "issue:16998", "right": "issue:22355", "accept": false, - "reason": "A model_max_length question and an ONNX import error are unrelated." + "reason": "A model_max_length question and a missing transformers.onnx module are unrelated." }, { "left": "issue:43295", "right": "issue:43825", "accept": false, - "reason": "Both are v5-related regressions, but one is about processor/tokenizer access in custom model code and the other is an inaccurate pipeline error message." + "reason": "Custom processor/tokenizer regression and a pipeline error-message regression are different bugs." }, { "left": "issue:43295", "right": "issue:44871", "accept": false, - "reason": "Custom processor/tokenizer regression and Gemma eos_token_id config inconsistency are separate issues." + "reason": "Processor/tokenizer API breakage is unrelated to eos_token_id config inconsistency." }, { "left": "issue:44871", "right": "issue:45584", "accept": false, - "reason": "Different models and different bugs: eos token config mismatch vs Whisper empty transcription generation failure." + "reason": "Gemma eos-token config mismatch and Whisper empty-transcription generation failure are separate issues." }, { "left": "issue:46234", "right": "issue:46291", "accept": false, - "reason": "Missing availability helper and kernels import/version requirements are distinct import-time issues." + "reason": "Missing availability helper and kernels version/revision import failure are different import problems." }, { "left": "issue:44464", "right": "issue:46421", "accept": false, - "reason": "Both involve chunked generation/compile behavior, but one is incorrect outputs and the other is spurious recompilation from lazy cache initialization; not clearly the same bug." + "reason": "Compiled-forward output inconsistency and chunked-prefill recompilation are related to generation but not the same defect." }, { "left": "issue:44077", "right": "issue:44534", "accept": false, - "reason": "Optional `post_init` validation for one model class is unrelated to junk-filled non-persistent buffers in v5." + "reason": "Model init validation and non-persistent buffer junk values are not the same bug." }, { "left": "issue:43316", "right": "issue:43716", "accept": false, - "reason": "API discrepancy in one config class and a dtype mismatch between image preprocessor and model are unrelated." + "reason": "Gemma3TextConfig API discrepancy and Mistral-3 dtype mismatch are different multimodal model issues." }, { "left": "issue:43316", "right": "issue:46821", "accept": false, - "reason": "Gemma3 config API mismatch and TimesFM `window_size` AttributeError are different model-specific issues." + "reason": "A config API mismatch and a TimesFM window_size AttributeError are unrelated." }, { "left": "issue:44871", "right": "issue:46050", "accept": false, - "reason": "Both are save/load-related, but one is an eos_token configuration mismatch and the other is broken persistence for quantized models." + "reason": "These are both save/load-adjacent, but one is config inconsistency and the other is quantized model reload breakage." }, { "left": "issue:43646", "right": "issue:44913", "accept": false, - "reason": "Custom model initialization breakage in Transformers 5.0 and GPTNeoX config reload behavior are not the same concrete defect." + "reason": "Custom model initialization regression and GPTNeoXConfig reload-default loss are different configuration problems." }, { "left": "issue:43452", "right": "issue:44496", "accept": false, - "reason": "GGUF loading failure and unrecognized-model/config.json failure are both load errors, but they stem from different entry points and causes." - }, - { - "left": "issue:45803", - "right": "issue:46365", - "accept": false, - "reason": "Exception handling in tensor conversion and an import conflict when importing Trainer are unrelated." + "reason": "gguf_file handling and a missing model_type in config.json are not the same loading bug." } ] }, @@ -11447,7 +11394,6 @@ "issue:44913", "issue:44964", "issue:44977", - "issue:45072", "issue:45127", "issue:45200", "issue:45406", @@ -11460,26 +11406,27 @@ "issue:46178", "issue:46202", "issue:46234", + "issue:46365", "issue:46369", "issue:46424", "issue:46640", - "issue:46894", - "issue:46932" + "issue:46894" ], "soft_pairs": [ + "issue:45803|issue:46365", "issue:44162|issue:44792", "issue:43404|issue:43441", "issue:44242|issue:46894", "issue:45632|issue:46424", "issue:44610|issue:46369", "issue:45803|issue:46170", - "issue:43540|issue:44514", "issue:43065|issue:44617", + "issue:43540|issue:44514", "issue:45542|issue:46234", "issue:43650|issue:46202", "issue:43572|issue:46178", - "issue:43504|issue:45698", "issue:44792|issue:44977", + "issue:43504|issue:45698", "issue:44792|issue:45600", "issue:45127|issue:45200", "issue:43388|issue:43688", @@ -11490,187 +11437,186 @@ "issue:44534|issue:44913", "issue:44190|issue:44964", "issue:35141|issue:42490", - "issue:44734|issue:45406", - "issue:45072|issue:46932" + "issue:44734|issue:45406" ], - "prepared_review_unit_hash": "f9f666a5ae20b23e9d5d95085f8f441d1483094022aa72c4d60d59ecc6c5c40d", + "prepared_review_unit_hash": "54687ca62d7f706146a334fc1b9ff11a336817c0b64275c76ece8f416a242dc8", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 25, - "serialized_chars": 35070, - "estimated_input_tokens": 8768, - "estimated_eval_tokens": 17792 + "serialized_chars": 34480, + "estimated_input_tokens": 8620, + "estimated_eval_tokens": 17496 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is not a true duplicate set; it mixes many unrelated Transformers bugs that only share broad surface terms like cache, flash attention, loading, or multimodal processing. I would not consolidate any of the suggested soft edges.", - "confidence": 0.97, - "canonical_issue_reason": "No canonical issue: the items span unrelated bugs and regressions across different models, processors, and code paths.", + "summary": "This set is a bag of unrelated issue reports; none of the soft pairs describe the same underlying bug closely enough to merge.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue is safe here: the titles span unrelated subsystems and failure modes (model config, processors, caching, serving, import/backends, and test failures).", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a good cluster representative because there is no shared underlying defect to anchor the group.", + "best_issue_reason": "No issue in this set is globally representative enough to serve as a canonical duplicate target; the overlaps are mostly superficial (same model family or same broad area) rather than the same defect.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:45803", + "right": "issue:46365", + "accept": false, + "reason": "Different bugs: bare `except` swallowing errors in Fuyu tensor conversion vs Trainer import failure due to datasets on import path." + }, { "left": "issue:44162", "right": "issue:44792", "accept": false, - "reason": "Different models and failures: ESM2 breakage vs a Janus image-generation test failure." + "reason": "Unrelated: ESM2 breakage vs a Janus image-generation test failure; no shared code-path or defect." }, { "left": "issue:43404", "right": "issue:43441", "accept": false, - "reason": "Weight tying bug vs FlashAttention support/error in a different model; not the same code-path defect." + "reason": "Both are Mistral-family, but one is missing lm_head tying and the other is FlashAttention support; distinct fixes." }, { "left": "issue:44242", "right": "issue:46894", "accept": false, - "reason": "Both involve training losses, but the models and root causes differ; not one underlying bug." + "reason": "Both involve training loss, but the concrete bugs differ: router-logits load balancing vs a double-shift loss issue in CohereASR." }, { "left": "issue:45632", "right": "issue:46424", "accept": false, - "reason": "Both mention cache, but one is a local-model path collision and the other is max_cache_length behavior." + "reason": "Both mention cache, but one is remote-code cache path collisions and the other is max_cache_length handling; separate problems." }, { "left": "issue:44610", "right": "issue:46369", "accept": false, - "reason": "Different SAM3 issues: wrong image size preprocessing vs memory overflow." + "reason": "Different models and symptoms: OmDet-Turbo processor image size mismatch vs Sam3-LiteText memory overflow." }, { "left": "issue:45803", "right": "issue:46170", "accept": false, - "reason": "Error handling bug in tensor conversion vs Windows/ROCm import dependency failure." + "reason": "Unrelated import/runtime issues: Fuyu conversion swallowing exceptions vs Windows ROCm import dependency failure." }, { - "left": "issue:43540", - "right": "issue:44514", + "left": "issue:43065", + "right": "issue:44617", "accept": false, - "reason": "Both touch Qwen3 multimodal flows, but they are separate bugs in different functions and inputs." + "reason": "Same SAM3 area, but dummy Conv2d in pixel decoder is not the same bug as a Sam3Video OOM." }, { - "left": "issue:43065", - "right": "issue:44617", + "left": "issue:43540", + "right": "issue:44514", "accept": false, - "reason": "Different SAM3 video/pixel-decoder problems; one is a dummy layer concern, the other is OOM." + "reason": "Both are Qwen-related input issues, but one is video input validation and the other is batched chat-template padding behavior." }, { "left": "issue:45542", "right": "issue:46234", "accept": false, - "reason": "Unrelated backend/import issues; no shared bug or fix path." + "reason": "Different dependency problems: TensorBoard/TensorFlow backend detection vs a missing import_utils symbol." }, { "left": "issue:43650", "right": "issue:46202", "accept": false, - "reason": "No meaningful overlap: a vague request vs a Gemma roundtrip tokenization issue." + "reason": "No meaningful overlap; one is a vague request, the other is a Gemma pipeline roundtrip tokenizer issue." }, { "left": "issue:43572", "right": "issue:46178", "accept": false, - "reason": "Different configuration/backend failures with no common code defect." + "reason": "Different backend/config regressions: StableLmConfig pad_token field vs tensorflow_text backend mapping." }, { - "left": "issue:43504", - "right": "issue:45698", + "left": "issue:44792", + "right": "issue:44977", "accept": false, - "reason": "Both are load-time issues, but one is legacy-field preset loading and the other is wrong custom-module selection after save/load." + "reason": "Different failures: Janus image-generation test vs Qwen3.5 flash-attention generation bug." }, { - "left": "issue:44792", - "right": "issue:44977", + "left": "issue:43504", + "right": "issue:45698", "accept": false, - "reason": "Generation failures in different models with different causes; too broad to merge." + "reason": "Both concern model loading, but one is a legacy config field on Beit and the other is wrong custom module selection after save/load." }, { "left": "issue:44792", "right": "issue:45600", "accept": false, - "reason": "A test failure in Janus is not the same bug as stale auto_mappings entries breaking CI." + "reason": "Unrelated: Janus test failure vs auto_mappings referencing removed Sam3LiteText configs." }, { "left": "issue:45127", "right": "issue:45200", "accept": false, - "reason": "Different Gemma behaviors: LoRA/extended-vocab collapse vs missing multimodal token-type defaults." + "reason": "Distinct bugs: LoRA merge with extended vocab on tied embeddings vs Gemma4 mm_token_type_ids defaulting." }, { "left": "issue:43388", "right": "issue:43688", "accept": false, - "reason": "Different subsystems and symptoms: metric gathering vs auxiliary-loss normalization." + "reason": "Different algorithmic bugs in different training/eval code paths; only a superficial metrics/loss similarity." }, { "left": "issue:43673", "right": "issue:44743", "accept": false, - "reason": "Both involve caching/state, but they affect different mechanisms and code paths." + "reason": "Both touch cache/state handling, but one is GenerationMixin chunked_prefill cache and the other is Qwen3 recurrent states resetting." }, { "left": "issue:43352", "right": "issue:46640", "accept": false, - "reason": "One is explicit FA2 unsupported-model gating; the other is a separate generic Flash Attention issue." + "reason": "One is a specific model not supporting Flash Attention 2.0 yet; the other is a generic Flash Attention 2 problem with no same code-path evidence." }, { "left": "issue:43525", "right": "issue:43531", "accept": false, - "reason": "Unrelated config attribute issue vs sliding-window behavior in a different model family." + "reason": "Unrelated config issues: missing pad_token_id in Llama4Config vs sliding_window behavior in Qwen3-MoE." }, { "left": "issue:41950", "right": "issue:43475", "accept": false, - "reason": "Different video/vision bugs: wrong processor type vs missing output attribute." + "reason": "Different SAM/video bugs: processor picking image processors vs missing fpn_position_embeddings attribute." }, { "left": "issue:44534", "right": "issue:44913", "accept": false, - "reason": "Non-persistent buffer initialization bug vs config reload reverting a rotary parameter." + "reason": "No same defect: non-persistent buffer initialization junk vs GPTNeoX rotary_pct not persisting on reload." }, { "left": "issue:44190", "right": "issue:44964", "accept": false, - "reason": "Dataset loading in a training script vs loading a specific multimodal model." + "reason": "Different loading failures for different entry points/models; no shared underlying bug." }, { "left": "issue:35141", "right": "issue:42490", "accept": false, - "reason": "Token-embedding resize initialization bug vs import warning from trailing-slash local model paths." + "reason": "Unrelated: embedding reinitialization after resize vs import warning on local paths with trailing slash." }, { "left": "issue:44734", "right": "issue:45406", "accept": false, - "reason": "Serve crashes, but with different causes: KV-cache indexing vs missing processor tokenizer attribute." - }, - { - "left": "issue:45072", - "right": "issue:46932", - "accept": false, - "reason": "Both mention dtype casting, but one is bfloat16 inference mismatches and the other is quantized embedder casting in a different model family." + "reason": "Both are transformers serve crashes, but the concrete failures differ: KV-cache tensor indexing vs missing `_tokenizer` on Gemma4Processor." } ] }, @@ -11724,6 +11670,7 @@ "issue:44877", "issue:44913", "issue:44933", + "issue:45072", "issue:45245", "issue:45362", "issue:45381", @@ -11735,19 +11682,20 @@ "issue:46369", "issue:46421", "issue:46617", - "issue:46655", "issue:46697", "issue:46832", "issue:46858", "issue:46897", + "issue:46932", "issue:47013" ], "soft_pairs": [ + "issue:45072|issue:46932", "issue:44855|issue:45588", - "issue:44869|issue:46832", "issue:45632|issue:46617", - "issue:43827|issue:43994", + "issue:44869|issue:46832", "issue:45245|issue:46858", + "issue:43827|issue:43994", "issue:46082|issue:46697", "issue:29870|issue:35141", "issue:44617|issue:45584", @@ -11758,223 +11706,222 @@ "issue:25251|issue:27957", "issue:43335|issue:44855", "issue:45563|issue:45584", - "issue:44933|issue:45362", "issue:44077|issue:44877", - "issue:43550|issue:44190", + "issue:44933|issue:45362", "issue:45381|issue:46369", + "issue:43550|issue:44190", "issue:41720|issue:45245", "issue:44479|issue:44734", "issue:43065|issue:44610", "issue:43335|issue:44190", "issue:45588|issue:47013", "issue:43854|issue:44792", - "issue:39692|issue:44190", "issue:44617|issue:46369", - "issue:44464|issue:45563", - "issue:45381|issue:46655" + "issue:39692|issue:44190", + "issue:44464|issue:45563" ], - "prepared_review_unit_hash": "b06e3f1dea913ed75666bd1a96a85da2471a6a4fb989d9258e147ab303c908d7", + "prepared_review_unit_hash": "cc45e23249b465f5188c809ef3626fffb48a24edc2ced5645ae7255c74fc37f3", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 35928, - "estimated_input_tokens": 8982, - "estimated_eval_tokens": 18220 + "serialized_chars": 36526, + "estimated_input_tokens": 9132, + "estimated_eval_tokens": 18520 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is only loosely semantically similar; the issues span unrelated bugs in generation, config loading, vision/video processors, caching, and docs. I would not deduplicate any of the proposed pairs.", - "confidence": 0.97, - "canonical_issue_reason": "No clear canonical issue: the items cover unrelated bugs across generation, config, vision/video, cache, and documentation paths.", + "summary": "This cluster is a loose semantic neighborhood rather than a duplicate set. The soft edges are mostly broad topical similarities (cache, vision, dtype, generation), but the concrete failure modes differ, so none should be merged.", + "confidence": 0.87, + "canonical_issue_reason": "No single canonical issue fits well: the issues span unrelated models and failure modes, and the similarity links are mostly shallow topical overlap.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is sufficiently representative or duplicate-worthy for the whole mixed cluster.", + "best_issue_reason": "No global best issue for consolidation; the closest pairwise matches still describe distinct bugs, so keeping them separate is safer.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44855", - "right": "issue:45588", + "left": "issue:45072", + "right": "issue:46932", "accept": false, - "reason": "Different failures: Python 3.13 import/parsing error vs flash-attention crash on sink-less models." + "reason": "Both are dtype-casting problems, but they affect different models/components and describe different concrete failures." }, { - "left": "issue:44869", - "right": "issue:46832", + "left": "issue:44855", + "right": "issue:45588", "accept": false, - "reason": "Whisper timestamp decode bug is unrelated to RT-DETR feature-level indexing/type error." + "reason": "One is a Python 3.13 import/parser issue, the other is a flash-attention AttributeError; same general area only at a very high level." }, { "left": "issue:45632", "right": "issue:46617", "accept": false, - "reason": "Both touch trust_remote_code cache handling, but one is a path collision and the other a FileNotFoundError; not the same concrete bug." + "reason": "Both involve trust_remote_code/cache loading, but one is a leaf-directory cache collision and the other is a FileNotFoundError from cache lookup; not clearly the same bug." }, { - "left": "issue:43827", - "right": "issue:43994", + "left": "issue:44869", + "right": "issue:46832", "accept": false, - "reason": "Docs still referencing pipeline() is not the same as SigLIP2 producing bad outputs through AutoModel/pipeline." + "reason": "Whisper timestamp decoding and RT-DETR feature-level handling are unrelated code paths." }, { "left": "issue:45245", "right": "issue:46858", "accept": false, - "reason": "Category-count limit error and device-tensor slice arithmetic in _prefill are different code-path problems." + "reason": "A category-cardinality runtime error is unrelated to static-cache slice arithmetic under torch.compile." + }, + { + "left": "issue:43827", + "right": "issue:43994", + "accept": false, + "reason": "Docs stale pipeline references are not the same as a SigLIP2 model/processor inference correctness bug." }, { "left": "issue:46082", "right": "issue:46697", "accept": false, - "reason": "Llama head_dim validation and Swin2SR over-padding images are unrelated." + "reason": "Llama head-dim validation and Swin2SR padding behavior are unrelated." }, { "left": "issue:29870", "right": "issue:35141", "accept": false, - "reason": "BART generate() decoder mask bug is unrelated to embedding reinitialization after resize." + "reason": "BART decoder-mask generation error and embedding reinitialization after resize are different bugs." }, { "left": "issue:44617", "right": "issue:45584", "accept": false, - "reason": "Sam3Video OOM and Whisper empty-transcription crash are different model paths and symptoms." + "reason": "CUDA OOM in Sam3Video and Whisper empty-transcription failure are unrelated failures." }, { "left": "issue:43673", "right": "issue:46421", "accept": false, - "reason": "Both mention chunked prefill/cache, but one is missing cache state in v5.0.0 while the other is lazy init causing recompilation; not the same bug." + "reason": "Both mention chunked prefill/cache, but one is missing cache state in v5.0.0 and the other is recompilation from lazy cache init; not the same defect." }, { "left": "issue:43931", "right": "issue:44060", "accept": false, - "reason": "Shape mismatch on loading Qwen3-VL is not the same as the tied-weights warning involving dt_bias." + "reason": "Model weight-shape mismatch and a tied-weights warning bug are different Qwen3 issues." }, { "left": "issue:43644", "right": "issue:44913", "accept": false, - "reason": "Non-persistent buffer junk initialization is a different issue from GPTNeoX rotary_pct not persisting on reload." + "reason": "Non-persistent buffer junk values and GPTNeoX config reload defaulting are unrelated." }, { "left": "issue:44242", "right": "issue:46897", "accept": false, - "reason": "Missing load-balancing loss when output_router_logits=False is not the same as the Florence2 double-shift training-loss bug." + "reason": "Load-balancing loss omission and Florence2 double-shift training loss are different training bugs." }, { "left": "issue:25251", "right": "issue:27957", "accept": false, - "reason": "Pipeline top_k output-shape regression and XLMRoberta Flash Attention 2 support are unrelated." + "reason": "Pipeline top_k nesting behavior and XLMRoberta Flash Attention 2 are unrelated." }, { "left": "issue:43335", "right": "issue:44855", "accept": false, - "reason": "SwitchTransformers sparse-layer creation and DebertaV2 import parsing are unrelated." + "reason": "SwitchTransformers sparse-layer config and DebertaV2 import parsing are unrelated." }, { "left": "issue:45563", "right": "issue:45584", "accept": false, - "reason": "Paged generate stale warning vs Whisper empty transcription crash; different generation paths and symptoms." - }, - { - "left": "issue:44933", - "right": "issue:45362", - "accept": false, - "reason": "Missing image_utils import is unrelated to Qwen3.5 chat crashes." + "reason": "A stale warning in paged generate is not the same as Whisper empty-transcription generation failure." }, { "left": "issue:44077", "right": "issue:44877", "accept": false, - "reason": "PatchTSMixer post_init policy and strict config rejecting granite_speech are different config-validation issues." + "reason": "PatchTSMixer post_init policy and granite_speech strict-config loading are different configuration issues." }, { - "left": "issue:43550", - "right": "issue:44190", + "left": "issue:44933", + "right": "issue:45362", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba is unrelated to local dataset loading in the trainer script." + "reason": "A missing import and Qwen3.5 chat crashes are unrelated." }, { "left": "issue:45381", "right": "issue:46369", "accept": false, - "reason": "Qwen2.5-VL video position IDs regression and Sam3-litetext memory overflow are different bugs." + "reason": "Qwen2.5-VL video position IDs and Sam3-litetext memory overflow are different model bugs." + }, + { + "left": "issue:43550", + "right": "issue:44190", + "accept": false, + "reason": "torch.compile SDPA failure and local dataset loading are unrelated." }, { "left": "issue:41720", "right": "issue:45245", "accept": false, - "reason": "Qwen3 auto device mapping cuda assert is unrelated to the category-count overflow limit." + "reason": "Auto device mapping cudaErrorAssert and category-count overflow are unrelated." }, { "left": "issue:44479", "right": "issue:44734", "accept": false, - "reason": "Video-input regression across Qwen VL models and serve KV-cache continuation tensor indexing are different paths." + "reason": "Video-input regression for Qwen VL models and KV-cache continuation indexing in serve are different code paths." }, { "left": "issue:43065", "right": "issue:44610", "accept": false, - "reason": "Dummy Conv2d in Sam3PixelDecoder and OmDet-Turbo processor/model size mismatch are unrelated." + "reason": "Sam3PixelDecoder dummy conv and OmDet-Turbo processor size mismatch are not the same bug." }, { "left": "issue:43335", "right": "issue:44190", "accept": false, - "reason": "Sparse-layer config bug and local dataset loading failure are unrelated." + "reason": "SwitchTransformers config creation and dataset loading are unrelated." }, { "left": "issue:45588", "right": "issue:47013", "accept": false, - "reason": "Flash-attention sink-less crash and xLSTM unpacking behavior in train mode are different code paths." + "reason": "Flash-attention sink handling and xLSTM state unpacking are unrelated failures." }, { "left": "issue:43854", "right": "issue:44792", "accept": false, - "reason": "GLM-4.7-Flash unit-test loading failure is unrelated to janus image generation test failure." - }, - { - "left": "issue:39692", - "right": "issue:44190", - "accept": false, - "reason": "SigLIP2 docs/example issues are unrelated to the image-classification trainer dataset bug." + "reason": "GLM-4.7-Flash unit-test loading and janus image-generation test failure are unrelated." }, { "left": "issue:44617", "right": "issue:46369", "accept": false, - "reason": "Both are memory-related Sam3 reports, but one is video-model OOM and the other is litetext overflow; not the same bug." + "reason": "OOM in Sam3Video and memory overflow in sam3-litetext are different models and failure modes." }, { - "left": "issue:44464", - "right": "issue:45563", + "left": "issue:39692", + "right": "issue:44190", "accept": false, - "reason": "Chunked generation inconsistency with compiled forward is unrelated to the stale num_return_sequences warning." + "reason": "SigLIP2 doc/example errors and local dataset loading are unrelated." }, { - "left": "issue:45381", - "right": "issue:46655", + "left": "issue:44464", + "right": "issue:45563", "accept": false, - "reason": "Qwen video position-id regression and SmolVLM video padding calculation bug are distinct model/processor issues." + "reason": "Compiled-forward chunked generation inconsistency and a stale generate warning are different bugs." } ] }, @@ -12004,7 +11951,6 @@ "issue:32090", "issue:35141", "issue:36246", - "issue:38617", "issue:41628", "issue:42175", "issue:42796", @@ -12043,19 +11989,21 @@ "issue:46121", "issue:46234", "issue:46489", + "issue:46655", "issue:46697", "issue:46752", "issue:46858" ], "soft_pairs": [ + "issue:45381|issue:46655", "issue:45127|issue:45357", "issue:43295|issue:45276", "issue:44112|issue:45600", "issue:44779|issue:46489", "issue:43540|issue:44734", "issue:45290|issue:46752", - "issue:43526|issue:46697", "issue:45276|issue:45584", + "issue:43526|issue:46697", "issue:44521|issue:45381", "issue:44737|issue:45803", "issue:25251|issue:29870", @@ -12071,208 +12019,207 @@ "issue:42175|issue:44279", "issue:44112|issue:44964", "issue:42796|issue:44521", - "issue:36246|issue:41628", "issue:32090|issue:35141", - "issue:46082|issue:46121", + "issue:36246|issue:41628", "issue:43404|issue:45216", - "issue:36246|issue:38617" + "issue:46082|issue:46121" ], - "prepared_review_unit_hash": "a08ca5ba628140b001d104331f6c0ca733cd3755d411a1be7cc8d4c1fe487382", + "prepared_review_unit_hash": "6eaeb5aca066ec99c15b11b2660f3c419483b1122721c2bd467dc310e774036f", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36599, - "estimated_input_tokens": 9150, - "estimated_eval_tokens": 18556 + "serialized_chars": 36570, + "estimated_input_tokens": 9143, + "estimated_eval_tokens": 18542 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "No clear duplicate set here: the items are mostly unrelated Transformers bugs that only share broad subsystem overlap (tokenizers, chat templates, loading, saving, CI, multimodal models).", + "summary": "This cluster does not look like a true duplicate set: the issues span many unrelated Transformers bugs, and the soft links are mostly broad topical similarity rather than the same concrete code path.", "confidence": 0.96, - "canonical_issue_reason": null, + "canonical_issue_reason": "issue:44521 is the strongest single representative by signal and specificity (multiple inbound refs, active discussion, concrete multimodal failure), although the cluster itself does not collapse into one duplicate bug.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "issue:44521 has the clearest user-facing regression and the highest community signal in this set, so it is the best issue to keep as the representative anchor.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:45381", + "right": "issue:46655", + "accept": false, + "reason": "Both are multimodal/video-processing bugs, but one is wrong vision_position_ids in Qwen2.5-VL and the other is over-padding in SmolVLM; different underlying failures." + }, { "left": "issue:45127", "right": "issue:45357", "accept": false, - "reason": "Both involve checkpoint/weight handling, but one is LoRA merge with extended vocab and the other is incorrect visual-encoder keys in save_pretrained; different bugs and code paths." + "reason": "Both involve save_pretrained/tied-weight behavior, but one is LoRA merge collapse with extended vocab and the other is incorrect visual encoder keys; not the same bug." }, { "left": "issue:43295", "right": "issue:45276", "accept": false, - "reason": "One is a regression around processor.tokenizer/image passing, the other is resize_token_embeddings not updating embeddings; not the same failure." + "reason": "These are both embedding/serialization regressions, but they affect different mechanisms: processor/tokenizer access vs resize_token_embeddings not updating Gemma4 layers." }, { "left": "issue:44112", "right": "issue:45600", "accept": false, - "reason": "Both are CI-related, but one is a stale device override test and the other is auto_mappings referencing removed configs." + "reason": "One is a stale GraniteSpeech CI test, the other is an auto_mappings reference cleanup issue; they are unrelated CI failures." }, { "left": "issue:44779", "right": "issue:46489", "accept": false, - "reason": "Both concern DeepSeek tokenizers, but 46489 is a specific DeepSeek-Coder v1 regression tied to PR #44801; not clearly the same concrete bug as the broader v5 tokenizer issue." + "reason": "Same model family, but 44779 is a broad v5 tokenizer regression while 46489 is a specific DeepSeek-Coder v1 gap in a prior fix; not one duplicate bug." }, { "left": "issue:43540", "right": "issue:44734", "accept": false, - "reason": "Different areas: video input validation in Qwen3OmniMoe versus KV-cache continuation tensor indexing in transformers serve." + "reason": "Both are runtime generation errors, but one is Qwen3OmniMoe video-input validation and the other is KV-cache continuation indexing in serve." }, { "left": "issue:45290", "right": "issue:46752", "accept": false, - "reason": "Both hit apply_chat_template edge cases, but one is tool-call assistant messages with no content and the other is an empty input list; related API, different bugs." + "reason": "Both crash in apply_chat_template, but one is assistant tool-call messages with no content and the other is an empty-list IndexError; distinct edge cases." }, { - "left": "issue:43526", - "right": "issue:46697", + "left": "issue:45276", + "right": "issue:45584", "accept": false, - "reason": "Different image-processor bugs: reduce_labels truncation vs over-padding when dimensions are already divisible." + "reason": "These touch different subsystems: Gemma4 embedding resizing versus Whisper empty-transcription handling after align_special_tokens." }, { - "left": "issue:45276", - "right": "issue:45584", + "left": "issue:43526", + "right": "issue:46697", "accept": false, - "reason": "Resize-token-embeddings bookkeeping in gemma4 is unrelated to Whisper generation failing after align_special_tokens." + "reason": "One is a label-reduction bug in BeitImageProcessorFast; the other is over-padding in Swin2SR image processing." }, { "left": "issue:44521", "right": "issue:45381", "accept": false, - "reason": "Both are multimodal-related, but one is assistant_masks generation and the other is vision_position_ids; different mechanisms and symptoms." + "reason": "Both are multimodal issues, but one is assistant_masks being all zero and the other is incorrect vision_position_ids; different preprocessing stages." }, { "left": "issue:44737", "right": "issue:45803", "accept": false, - "reason": "XLNet CPU tensor allocation and a bare except swallowing errors are unrelated bugs." + "reason": "XLNet CPU tensor creation overhead and a bare-except swallowing errors in FuyuBatchFeature are unrelated bugs." }, { "left": "issue:25251", "right": "issue:29870", "accept": false, - "reason": "Pipeline top_k nesting and Bart generate() decoder_attention_mask initialization are unrelated code paths." + "reason": "Top-k nesting in pipeline output and Bart generate() unbound decoder_attention_mask are different code paths and symptoms." }, { "left": "issue:44462", "right": "issue:45907", "accept": false, - "reason": "One is AutoTokenizer ignoring tokenizer.json; the other is list_repo_files not forwarding revision. Different Hub APIs." + "reason": "AutoTokenizer ignoring tokenizer.json and list_repo_files not forwarding revision are unrelated hub/tokenizer behaviors." }, { "left": "issue:44938", "right": "issue:45542", "accept": false, - "reason": "Python 3.14 import failure and missing TensorFlow backend metadata are separate environment/dependency issues." + "reason": "Python 3.14 import failure and missing TensorFlow backend with the torch extra are different packaging/import problems." }, { "left": "issue:43854", "right": "issue:44964", "accept": false, - "reason": "Both are model-loading failures, but for different model families and different underlying breakages." + "reason": "Different models and different load failures: GLM-4.7-Flash test loading versus Phi-4 multimodal loading." }, { "left": "issue:43673", "right": "issue:46858", "accept": false, - "reason": "Both involve cache internals, but one is missing GenerationMixin cache in chunked_prefill and the other is device-tensor seq_length breaking static-cache arithmetic." + "reason": "Both mention generation cache, but one is missing cache in chunked_prefill and the other is a device-tensor seq_length arithmetic bug in static cache." }, { "left": "issue:43643", "right": "issue:44496", "accept": false, - "reason": "trust_remote_code missing fields and unrecognized-model/model_type validation are distinct config-loading problems." + "reason": "trust_remote_code missing fields is a different config-return issue from unrecognized model/config lacking model_type." }, { "left": "issue:44246", "right": "issue:45003", "accept": false, - "reason": "Import slowness and unsafe sys.modules access are related only at a high level; not the same bug." + "reason": "Import slowdown and unsafe sys.modules access are not the same defect, even if both affect import-time behavior." }, { "left": "issue:44964", "right": "issue:45600", "accept": false, - "reason": "Model loading failure versus stale auto_mappings references; no shared concrete defect." + "reason": "A model-loading failure and a stale auto_mappings reference cleanup are separate problems." }, { "left": "issue:45632", "right": "issue:46234", "accept": false, - "reason": "Cache-path collision for trust_remote_code is unrelated to a missing import_utils availability helper." + "reason": "trust_remote_code cache-path collision and a missing import_utils helper are unrelated." }, { "left": "issue:27957", "right": "issue:35141", "accept": false, - "reason": "Flash Attention 2 support for XLMRoberta and post_init reinitializing output embeddings are unrelated." + "reason": "XLMRoberta FlashAttention 2 support and resize-token-embeddings reinitialization are different model/runtime issues." }, { "left": "issue:42175", "right": "issue:44279", "accept": false, - "reason": "Different dependency complaints: backend installation contents versus a generic transformers dependency issue." + "reason": "Both are dependency-related at a high level, but one is a torch extra missing TensorFlow backend while the other is a generic dependency complaint." }, { "left": "issue:44112", "right": "issue:44964", "accept": false, - "reason": "CI stale-device test failure and a model-loading regression are unrelated." + "reason": "A flaky GraniteSpeech CI test and a GLM model-load error are unrelated." }, { "left": "issue:42796", "right": "issue:44521", "accept": false, - "reason": "Tokenizer BPE marker handling and multimodal chat-template masks are different tokenizer problems." - }, - { - "left": "issue:36246", - "right": "issue:41628", - "accept": false, - "reason": "One missing model-specific processor export, the other missing a top-level AutoImageProcessor import; different import surfaces." + "reason": "Tokenizer BPE marker handling and multimodal assistant-mask generation are different layers of the stack." }, { "left": "issue:32090", "right": "issue:35141", "accept": false, - "reason": "Trainer gpu_broadcast NoneType error and embedding reinitialization after resizing are unrelated." + "reason": "GPU broadcast rejecting NoneType and embedding reinitialization after resize are unrelated bugs." }, { - "left": "issue:46082", - "right": "issue:46121", + "left": "issue:36246", + "right": "issue:41628", "accept": false, - "reason": "LlamaConfig head_dim validation and rope-parameter dict conversion/type handling are different configuration-validation bugs." + "reason": "Both are import errors around image processors, but one is a missing model-specific symbol and the other is a top-level AutoImageProcessor import failure; not the same fix." }, { "left": "issue:43404", "right": "issue:45216", "accept": false, - "reason": "Both touch save/tie weights in model classes, but they affect different model families and different checkpoint semantics." + "reason": "Mistral3 lm_head tying and Qwen3.5 incorrect save_pretrained checkpoints are distinct model-family regressions." }, { - "left": "issue:36246", - "right": "issue:38617", + "left": "issue:46082", + "right": "issue:46121", "accept": false, - "reason": "Different missing imports from different modules: Qwen2_5_VLImageProcessor vs layer_type_validation." + "reason": "LlamaConfig head_dim validation and JSON-loaded rope ignore_keys TypeError are separate config-validation issues." } ] }, @@ -12296,10 +12243,11 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-506", "nodes": [ + "issue:36246", + "issue:38617", "issue:41950", "issue:42907", "issue:43295", - "issue:43299", "issue:43381", "issue:43441", "issue:43479", @@ -12326,10 +12274,8 @@ "issue:44987", "issue:45042", "issue:45083", - "issue:45216", "issue:45335", "issue:45375", - "issue:45405", "issue:45446", "issue:45584", "issue:45588", @@ -12346,6 +12292,7 @@ "issue:46932" ], "soft_pairs": [ + "issue:36246|issue:38617", "issue:43531|issue:45600", "issue:44295|issue:46506", "issue:41950|issue:44016", @@ -12360,8 +12307,8 @@ "issue:43526|issue:43901", "issue:44683|issue:45446", "issue:43901|issue:44496", - "issue:42907|issue:43441", "issue:43673|issue:44336", + "issue:42907|issue:43441", "issue:43381|issue:45593", "issue:44496|issue:45759", "issue:43976|issue:44279", @@ -12374,224 +12321,216 @@ "issue:43479|issue:45375", "issue:45335|issue:45584", "issue:46123|issue:46369", - "issue:44327|issue:46752", - "issue:43299|issue:45216", - "issue:44279|issue:45405" + "issue:44327|issue:46752" ], - "prepared_review_unit_hash": "4231f8f5dabb9cbe345dfdb50dac486e8164806b96ef7233b9ef1204e81b7b1a", + "prepared_review_unit_hash": "324be5a0072fe6c88b53a270c45e89ff4f6dc71b5326b12b81bec272db1b2e70", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 36782, - "estimated_input_tokens": 9196, - "estimated_eval_tokens": 18648 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36023, + "estimated_input_tokens": 9006, + "estimated_eval_tokens": 18268 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is mostly a heterogeneous grab bag of unrelated Transformers issues; none of the soft pairs look like true duplicates of the same underlying bug/change. There are no pull requests in the cluster.", - "confidence": 0.21, - "canonical_issue_reason": "If a single issue must be chosen, #44521 is the best representative because it is active, has multiple inbound references, and describes a concrete multimodal chat-template bug with broader user impact. That said, the cluster is not a tight duplicate set.", + "summary": "The soft pairs are mostly superficial text matches across unrelated bugs; none look like the same underlying issue or change, so all soft edges should be rejected.", + "confidence": 0.91, + "canonical_issue_reason": "issue:44521 is the strongest anchor in this mixed set: it has active discussion, inbound references, and a concrete reproducible multimodal bug report.", "canonical_pr_reason": null, - "best_issue_reason": "#44521 is the strongest issue candidate overall: clear repro, current/open, and cited by other issues, so it is the most useful anchor for triage even though the cluster is thematically mixed.", + "best_issue_reason": "issue:44521 is the most substantive and broadly relevant bug report here, with the clearest evidence and triage value.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:36246", + "right": "issue:38617", + "accept": false, + "reason": "Both are ImportErrors, but for different missing symbols in different modules and unrelated code paths." + }, { "left": "issue:43531", "right": "issue:45600", "accept": false, - "reason": "Different failures: Qwen3-MoE sliding-window behavior vs. auto_mappings referencing removed configs/CI breakage." + "reason": "Qwen3-MoE sliding_window behavior and Sam3LiteText auto_mappings CI breakage are unrelated failures." }, { "left": "issue:44295", "right": "issue:46506", "accept": false, - "reason": "Position_ids buffer read error is unrelated to CPU/CUDA numerical differences in PPDocLayoutV3." + "reason": "One is a position_ids buffer read error; the other is a CPU/CUDA output divergence. Different bugs." }, { "left": "issue:41950", "right": "issue:44016", "accept": false, - "reason": "Video-classification processor lookup bug and a notebook syntax error are unrelated." + "reason": "Pipeline image-processor lookup and a notebook syntax error are not the same bug." }, { "left": "issue:43673", "right": "issue:46424", "accept": false, - "reason": "Both mention cache, but one is missing GenerationMixin cache in chunked_prefill while the other is about max_cache_length handling for quantized cache." + "reason": "Both mention cache, but one is missing GenerationMixin cache in chunked_prefill while the other is max_cache_length being ignored except for quantized cache." }, { "left": "issue:43582", "right": "issue:46932", "accept": false, - "reason": "Apple Silicon warmup TypeError is unrelated to Gemma4 quantized vision/audio dtype casting." + "reason": "Apple Silicon allocator TypeError and Gemma4 quantized embedder dtype casting are unrelated issues." }, { "left": "issue:43650", "right": "issue:44220", "accept": false, - "reason": "\u201cADD THE DATA\u201d and _torch_extract_fbank_features are not the same bug." + "reason": "\"ADD THE DATA\" is too vague to match the specific _torch_extract_fbank_features() bug." }, { "left": "issue:43295", "right": "issue:45335", "accept": false, - "reason": "Processor/tokenizer regression is distinct from resize_token_embeddings not updating decoder.embed_tokens." + "reason": "Processor/tokenizer regression and t5gemma embedding resize behavior are different model behaviors." }, { "left": "issue:43976", "right": "issue:44987", "accept": false, - "reason": "Python version incompatibility is unrelated to loading a specific model repo." + "reason": "Python version compatibility and a model-loading failure for physical-intelligence/fast are unrelated." }, { "left": "issue:45083", "right": "issue:46506", "accept": false, - "reason": "Different model-specific helper behavior vs. device-specific output divergence." + "reason": "Qwen3_omni_moe feature-length calculation and PPDocLayoutV3 CPU/CUDA mismatch are different code paths." }, { "left": "issue:44496", "right": "issue:45042", "accept": false, - "reason": "Unrecognized model config and PIL backend requiring torchvision are unrelated." + "reason": "Unrecognized model config and PIL backend torchvision dependency are unrelated." }, { "left": "issue:44016", "right": "issue:44220", "accept": false, - "reason": "Notebook syntax error does not match a feature-extraction helper bug." + "reason": "Notebook syntax error and audio feature extraction bug are not the same change." }, { "left": "issue:43526", "right": "issue:43901", "accept": false, - "reason": "BeitImageProcessorFast reduce_labels bug is unrelated to outdated pipeline docs." + "reason": "BeitImageProcessorFast label reduction and TextClassificationPipeline docs mismatch are unrelated." }, { "left": "issue:44683", "right": "issue:45446", "accept": false, - "reason": "Both involve flex_attention, but one is a torch>=2.9 compiled failure and the other is an incorrect version check for AuxRequest import; not the same concrete bug." + "reason": "Both touch flex_attention, but one is a compiled-kernel failure on torch>=2.9 and the other is an AuxRequest import/version-check bug." }, { "left": "issue:43901", "right": "issue:44496", "accept": false, - "reason": "Documentation mismatch and model auto-loading failure are unrelated." + "reason": "Docs mismatch for return_all_scores and unrecognized model config are unrelated." }, { - "left": "issue:42907", - "right": "issue:43441", + "left": "issue:43673", + "right": "issue:44336", "accept": false, - "reason": "Both mention Ministral, but save-pretrained failure and FlashAttention loading failure are different code paths." + "reason": "Cache missing in generation and ANSI codes in loading_report are unrelated." }, { - "left": "issue:43673", - "right": "issue:44336", + "left": "issue:42907", + "right": "issue:43441", "accept": false, - "reason": "Cache handling and ANSI output formatting are unrelated." + "reason": "Ministral save/dequantization and Ministral-3 FlashAttention failure are different bugs." }, { "left": "issue:43381", "right": "issue:45593", "accept": false, - "reason": "Gradient checkpointing in eval mode is unrelated to D-FINE auxiliary-loss behavior." + "reason": "Gradient checkpointing in eval mode and D-FINE auxiliary loss behavior are unrelated." }, { "left": "issue:44496", "right": "issue:45759", "accept": false, - "reason": "Unrecognized model config differs from failing to unwrap text_config in composite multimodal configs." + "reason": "Unrecognized model config and composite Qwen text_config unwrapping are different failures." }, { "left": "issue:43976", "right": "issue:44279", "accept": false, - "reason": "Python support mismatch is not the same as a generic dependency issue." + "reason": "Python version support and a generic dependency issue are not the same bug." }, { "left": "issue:44792", "right": "issue:46424", "accept": false, - "reason": "janus image-generation test failure is unrelated to cache length configuration." + "reason": "Janus test failure on image generation and cache max length semantics are unrelated." }, { "left": "issue:43441", "right": "issue:45588", "accept": false, - "reason": "Both are flash-attention related, but they describe different concrete crashes on different inputs/models." + "reason": "Both involve flash_attention, but one is model-specific Ministral failure and the other is an s_aux=None crash for sink-less models." }, { "left": "issue:44016", "right": "issue:44062", "accept": false, - "reason": "Notebook syntax error and AddedToken constructor argument collision are unrelated." + "reason": "Notebook syntax error and AddedToken special-argument duplication are unrelated." }, { "left": "issue:43824", "right": "issue:44933", "accept": false, - "reason": "Different import failures for different symbols/modules; not the same missing export." + "reason": "Missing Qwen2_5_VLForConditionalGeneration import and a different missing import from image_utils are not the same symbol or fix." }, { "left": "issue:44521", "right": "issue:46693", "accept": false, - "reason": "Chat-template assistant mask bug is unrelated to FA2 long-context performance regression." + "reason": "All-zero assistant masks and a long-context FA2 performance regression are different multimodal generation issues." }, { "left": "issue:44279", "right": "issue:46515", "accept": false, - "reason": "General dependency issue and torch 2.4.1 incompatibility are too broad and not the same bug." + "reason": "Generic transformers dependency issue and torch 2.4.1 incompatibility are different environment problems." }, { "left": "issue:43479", "right": "issue:45375", "accept": false, - "reason": "Both are config-related, but one is default-init behavior and the other is a missing field silently dropped by strict parsing." + "reason": "Default config initialization with None and missing deepstack_visual_indexes under strict config handling are different multimodal config bugs." }, { "left": "issue:45335", "right": "issue:45584", "accept": false, - "reason": "Token embedding resize bug is unrelated to Whisper empty-transcription generation failure." + "reason": "Embedding resize propagation and Whisper empty-transcription generation failure are unrelated." }, { "left": "issue:46123", "right": "issue:46369", "accept": false, - "reason": "Partial-batch masking logic and memory overflow are different problems." + "reason": "Partial-batch mask generation dropping results and sam3-litetext memory overflow are unrelated." }, { "left": "issue:44327", "right": "issue:46752", "accept": false, - "reason": "Both crash with index-style errors, but the QA decode_spans top_k edge case is unrelated to empty apply_chat_template input." - }, - { - "left": "issue:43299", - "right": "issue:45216", - "accept": false, - "reason": "Different Qwen3 multimodal regressions affecting loading vs. save_pretrained checkpoint output." - }, - { - "left": "issue:44279", - "right": "issue:45405", - "accept": false, - "reason": "Generic dependency issue is not the same as an unreleased MIN_PEFT_VERSION bump." + "reason": "QA decode_spans top_k bounds error and empty-list apply_chat_template crash are different input-shape bugs." } ] }, @@ -12634,6 +12573,7 @@ "issue:44016", "issue:44075", "issue:44242", + "issue:44279", "issue:44291", "issue:44351", "issue:44368", @@ -12664,10 +12604,12 @@ "issue:47000" ], "soft_pairs": [ + "issue:43299|issue:45216", + "issue:44279|issue:45405", "issue:43295|issue:44779", "issue:39692|issue:43901", - "issue:44075|issue:44368", "issue:42796|issue:43650", + "issue:44075|issue:44368", "issue:46082|issue:46399", "issue:43577|issue:45216", "issue:44351|issue:45278", @@ -12692,196 +12634,208 @@ "issue:33290|issue:45656", "issue:45081|issue:45907" ], - "prepared_review_unit_hash": "ff5b6b60b8c866d9056eef30fd5141e418852c9d193f472aa43f39b1617131ad", + "prepared_review_unit_hash": "8a08a6f8e895ca5685f651cde1f7ba12a87b155549d7b2875c05b46e9f40e5d0", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 27, - "serialized_chars": 35342, - "estimated_input_tokens": 8836, - "estimated_eval_tokens": 17928 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36481, + "estimated_input_tokens": 9121, + "estimated_eval_tokens": 18498 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is highly heterogeneous: the soft pairs span unrelated tokenizer, config, pipeline, optimizer, docs, and runtime bugs. I do not see any pair that clearly represents the same underlying issue/change, so all soft edges should be rejected.", - "confidence": 0.94, - "canonical_issue_reason": "No single canonical issue stands out; the items are unrelated enough that none can safely represent the whole cluster.", + "summary": "A loose set of mostly unrelated Transformer issues spanning tokenizer regressions, multimodal/model-loading bugs, docs problems, and optimizer/DeepSpeed behavior; none of the soft pairs looks like the same underlying defect.", + "confidence": 0.84, + "canonical_issue_reason": "No strong canonical issue: these reports are only loosely related at a high level and cover different code paths and model families.", "canonical_pr_reason": null, - "best_issue_reason": "No global best issue for duplicate consolidation; the set does not form one coherent bug family.", + "best_issue_reason": "If one issue must represent the set, 45020 is the broadest regression umbrella, but it is still too generic to subsume the rest.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43299", + "right": "issue:45216", + "accept": false, + "reason": "Both are regressions, but one is Qwen3VL MoE loading and the other is Qwen3.5 save_pretrained serialization; different operations and failure modes." + }, + { + "left": "issue:44279", + "right": "issue:45405", + "accept": false, + "reason": "One is a generic dependency issue; the other is a specific MIN_PEFT_VERSION bump problem. Same packaging theme, not the same bug." + }, { "left": "issue:43295", "right": "issue:44779", "accept": false, - "reason": "Different regressions: processor/tokenizer API break vs DeepSeek tokenization output bug." + "reason": "Processor/tokenizer API breakage vs incorrect Deepseek tokenization logic; different models and root causes." }, { "left": "issue:39692", "right": "issue:43901", "accept": false, - "reason": "Both are docs-related, but they report different examples and failures." + "reason": "Both mention docs, but one is a SigLIP2 example bug and the other is stale textclassification docs; unrelated changes." }, { - "left": "issue:44075", - "right": "issue:44368", + "left": "issue:42796", + "right": "issue:43650", "accept": false, - "reason": "Unrelated bugs: SGD args ignored vs a Qwen fine-tuning warning about tie_word_embeddings." + "reason": "Video-classification processor lookup and a trivial 'ADD THE DATA' issue are clearly unrelated." }, { - "left": "issue:42796", - "right": "issue:43650", + "left": "issue:44075", + "right": "issue:44368", "accept": false, - "reason": "Tokenizer BPE-marker bug vs a nonsensical placeholder issue; no shared underlying bug." + "reason": "SGD args not used is an optimizer bug; tie_word_embeddings warnings are a model config warning, not the same issue." }, { "left": "issue:46082", "right": "issue:46399", "accept": false, - "reason": "Different code paths: LlamaConfig head_dim validation vs Qwen3.5 kernel registration crash." + "reason": "LlamaConfig head_dim validation and Qwen3.5 kernel registration failure are different config/runtime bugs." }, { "left": "issue:43577", "right": "issue:45216", "accept": false, - "reason": "Different problems: dtype propagation in Blip2 vs incorrect save_pretrained keys for Qwen3.5." + "reason": "Blip2 dtype propagation and Qwen3.5 save_pretrained key serialization are separate code paths." }, { "left": "issue:44351", "right": "issue:45278", "accept": false, - "reason": "One is a missing import, the other a broad import regression after upgrade; not the same defect." + "reason": "Both are import problems after upgrades, but one is a missing HybridCache symbol and the other is broad import breakage; no concrete shared bug." }, { "left": "issue:43976", "right": "issue:45405", "accept": false, - "reason": "Python version compatibility issue vs PEFT version bump; unrelated." + "reason": "Python version compatibility and an unreleased PEFT minimum version are different dependency issues." }, { "left": "issue:41950", "right": "issue:45020", "accept": false, - "reason": "Video pipeline processor lookup bug vs remote_code loading regressions; different failures." + "reason": "A specific video-classification processor bug is not the same as the broad remote_code regression report." }, { "left": "issue:46082", "right": "issue:46512", "accept": false, - "reason": "Different config bugs: head_dim validation vs stale model.config.dtype after from_config." + "reason": "Explicit head_dim rejection and stale config.dtype after from_config are different configuration bugs." }, { "left": "issue:43381", "right": "issue:47000", "accept": false, - "reason": "Gradient checkpointing eval-mode guard vs RTDetrLoss infeasible cost matrix; unrelated." + "reason": "Gradient checkpointing in eval mode and RTDetrLoss infeasible cost matrix are unrelated runtime failures." }, { "left": "issue:44466", "right": "issue:45276", "accept": false, - "reason": "Both touch weights, but one is save_pretrained serialization and the other is resize_token_embeddings behavior." + "reason": "Inconsistent lm_head serialization and Gemma4 resize_token_embeddings behavior are both weight-related, but not the same defect." }, { "left": "issue:44779", "right": "issue:45812", "accept": false, - "reason": "Different tokenizer regressions affecting different model families and outputs." + "reason": "Deepseek tokenization regression and Granite token ID regression are separate tokenizer bugs for different model families." }, { "left": "issue:43299", "right": "issue:45357", "accept": false, - "reason": "Qwen3VL model loading vs Qwen3.5 visual-key serialization; not the same bug." + "reason": "Both involve Qwen3.5, but one is model loading and the other is save_pretrained visual encoder keys; different failures." }, { "left": "issue:43653", "right": "issue:45593", "accept": false, - "reason": "BigBird tokenizer special-token registration vs D-FINE auxiliary-loss logic; unrelated." + "reason": "BigBirdTokenizer special-token registration and D-FINE auxiliary-loss handling are unrelated." }, { "left": "issue:44242", "right": "issue:45200", "accept": false, - "reason": "MoE router loss omission vs Gemma4 mm_token_type_ids defaulting; different code paths." + "reason": "MoE load-balancing loss and Gemma4 mm_token_type_ids defaults are different multimodal/MoE issues." }, { "left": "issue:43650", "right": "issue:46209", "accept": false, - "reason": "Placeholder issue vs FP8 inference on B200; no relation." + "reason": "A placeholder issue cannot be the same bug as an FP8 inference failure on B200." }, { "left": "issue:41950", "right": "issue:45823", "accept": false, - "reason": "Video pipeline processor bug vs Gemma4 device-map mismatch; unrelated." + "reason": "Video-classification pipeline processor selection and Gemma4 PLE device placement are unrelated." }, { "left": "issue:43299", "right": "issue:43854", "accept": false, - "reason": "Different model-loading failures for different models." + "reason": "Qwen3VL MoE loading and GLM-4.7-Flash unit-test loading are different model-loading bugs." }, { "left": "issue:44462", "right": "issue:45081", "accept": false, - "reason": "AutoTokenizer repository file selection bug vs Mistral regex patch crash; distinct issues." + "reason": "AutoTokenizer ignoring tokenizer.json and Mistral regex patch crash are both tokenizer-related, but they fail in different places with different causes." }, { "left": "issue:45588", "right": "issue:47000", "accept": false, - "reason": "flash_attention attribute crash vs RTDetrLoss infeasible cost matrix; unrelated." + "reason": "Flash-attention s_aux handling and RTDetr loss infeasibility are unrelated model/runtime errors." }, { "left": "issue:44016", "right": "issue:44291", "accept": false, - "reason": "Notebook syntax error vs init_empty_weights TypeError; different failure modes." + "reason": "Notebook syntax error and init_empty_weights argument handling are completely different problems." }, { "left": "issue:44683", "right": "issue:45542", "accept": false, - "reason": "Torch flex_attention failure vs TensorFlow backend import issue; unrelated." + "reason": "Compiled flex_attention on torch>=2.9 and tensorboard-without-TF backend errors are unrelated dependency/runtime issues." }, { "left": "issue:43479", "right": "issue:43525", "accept": false, - "reason": "Different config initialization bugs in different models." + "reason": "Phi4Multimodal default vision/audio config initialization and missing Llama4Config.pad_token_id are different config bugs." }, { "left": "issue:43441", "right": "issue:45834", "accept": false, - "reason": "FlashAttention failure in Ministral vs OCR index error in Kosmos2.5; unrelated." + "reason": "Ministral FlashAttention failure and Kosmos2.5 long-OCR index error are unrelated model-specific bugs." }, { "left": "issue:33290", "right": "issue:45656", "accept": false, - "reason": "Both mention DeepSpeed/optimizers, but one is OOM with Adafactor and the other is double optimizer.step calls." + "reason": "OOM with Adafactor in DeepSpeed and optimizer.step being called twice are distinct optimizer/engine issues." }, { "left": "issue:45081", "right": "issue:45907", "accept": false, - "reason": "Tokenizer regex crash vs list_repo_files revision forwarding; no shared underlying bug." + "reason": "Tokenizer regex crash and list_repo_files revision forwarding are unrelated utility bugs." } ] }, @@ -12971,11 +12925,11 @@ "issue:43525|issue:43881", "issue:44186|issue:47000", "issue:43824|issue:43976", - "issue:42796|issue:43441", "issue:44016|issue:45823", + "issue:42796|issue:43441", "issue:45542|issue:45800", - "issue:43525|issue:43643", "issue:44188|issue:44336", + "issue:43525|issue:43643", "issue:44589|issue:45362", "issue:44336|issue:44373", "issue:45325|issue:46851", @@ -12985,208 +12939,208 @@ "issue:41628|issue:43976", "issue:45083|issue:46851" ], - "prepared_review_unit_hash": "7e959de96683228e58ed191ce91fbcd265c413eb176b2059ee068f764d22bb9a", + "prepared_review_unit_hash": "efc3410b05e145810921f976ab95ab0a335d4aead2b285828d50e39ecff84d9b", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36375, - "estimated_input_tokens": 9094, - "estimated_eval_tokens": 18444 + "serialized_chars": 36372, + "estimated_input_tokens": 9093, + "estimated_eval_tokens": 18442 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: it mixes unrelated loading/import errors, tokenizer/model regressions, packaging/backend issues, docs bugs, and performance problems. None of the soft pairs look like the same underlying bug or change, so duplicate consolidation is not justified.", + "summary": "This cluster is mostly a grab bag of unrelated Transformers v5 regressions, import/load failures, tokenizer/processor bugs, and performance issues. None of the soft pairs look like the same concrete underlying bug, so I would not merge any of them as duplicates.", "confidence": 0.87, - "canonical_issue_reason": "Issue 43723 is the broadest user-facing loading regression here (AutoTokenizer.from_pretrained in v5) and is the closest representative of the cluster, even though the overall set is not truly duplicate-heavy.", + "canonical_issue_reason": "No single canonical issue fits well: the items span multiple distinct bugs and subsystems rather than one duplicate report.", "canonical_pr_reason": null, - "best_issue_reason": "43723 is the most representative single issue because it targets a core v5 loading path and has a generic symptom compared with the more model-specific or unrelated reports.", + "best_issue_reason": "issue:43723 is the broadest user-facing loading regression in the set (AutoTokenizer.from_pretrained in v5) and is the best representative of the loading-related subset, though the cluster is not a true duplicate group.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:42994", "right": "issue:43317", "accept": false, - "reason": "Quantized saving failure and dequantized/offload loading failure affect different stages and code paths." + "reason": "Both involve quantization/offload, but one is saving and the other is dequantized model loading; different code paths and failures." }, { "left": "issue:45103", "right": "issue:45372", "accept": false, - "reason": "Auto-docstring annotation parsing is unrelated to the mistral_common ReasoningEffort import error." + "reason": "Unrelated bugs: auto-docstring annotation handling vs Gemma4 processor loading/import failure." }, { "left": "issue:44336", "right": "issue:46851", "accept": false, - "reason": "ANSI output formatting in loading_report is unrelated to Qwen3.5 variable-length position_ids handling." + "reason": "ANSI log formatting and Qwen position_ids behavior are unrelated." }, { "left": "issue:44623", "right": "issue:44749", "accept": false, - "reason": "Missing processor save files and slow filtering after upgrade are different bugs with different symptoms." + "reason": "Processor save_pretrained file emission is unrelated to a data-filtering slowdown regression." }, { "left": "issue:45859", "right": "issue:46851", "accept": false, - "reason": "CPU offload incompatibility in Qwen3.5Moe rotary embedding is a different failure than variable-length position_ids." + "reason": "Both are Qwen3.5-related, but one is CPU-offload incompatibility and the other is variable-length position_ids; not the same bug." }, { "left": "issue:42175", "right": "issue:43976", "accept": false, - "reason": "Missing TensorFlow backend in a torch install is a packaging issue, not the Python-version incompatibility in 5.1.0." + "reason": "Different environment/setup issues: optional backend extras vs Python version compatibility." }, { "left": "issue:44261", "right": "issue:44263", "accept": false, - "reason": "A layernorm precision issue and torch.split return-value behavior are unrelated model internals." + "reason": "Different model helpers and failures; they do not share a concrete bug." }, { "left": "issue:45600", "right": "issue:45712", "accept": false, - "reason": "Removed config references breaking CI and leftover dummy classes leaking into dir() are separate maintenance issues." + "reason": "Both touch repo maintenance/mapping, but one is stale auto-mapping entries and the other is dummy-class leakage; not one change." }, { "left": "issue:44623", "right": "issue:45397", "accept": false, - "reason": "processor.save_pretrained file omission is unrelated to gemma-4 zero3 from_pretrained loading failure." + "reason": "Saving processor files and ZeRO3 from_pretrained loading are separate problems." }, { "left": "issue:44671", "right": "issue:44779", "accept": false, - "reason": "CamemBERT masked-LM prediction regressions and Deepseek tokenizer regressions affect different models and code paths." + "reason": "Different regressions affecting different models and paths: CamemBERT MLM vs Deepseek tokenization." }, { "left": "issue:43901", "right": "issue:44246", "accept": false, - "reason": "A docs mismatch for return_all_scores is not the same as intermittent slow imports of transformers." + "reason": "Docs mismatch and import latency are unrelated." }, { "left": "issue:43673", "right": "issue:43761", "accept": false, - "reason": "Missing generation cache during chunked_prefill and CLIPVisionModel hidden_states=None are unrelated regressions." + "reason": "Different forward/output regressions on different models; too broad to be duplicates." }, { "left": "issue:43723", "right": "issue:44162", "accept": false, - "reason": "Both are loading/model-related, but the tokenizer-loading regression and ESM2 breakage are different model families and symptoms." + "reason": "Tokenizer loading in v5 and ESM2 breakage are distinct model-specific failures." }, { "left": "issue:43525", "right": "issue:43881", "accept": false, - "reason": "Missing pad_token_id on Llama4Config and glm-4v-9b loading failure are different model/config problems." + "reason": "Both are model loading issues, but the missing pad_token_id AttributeError and glm-4v-9b loading failure are not the same bug." }, { "left": "issue:44186", "right": "issue:47000", "accept": false, - "reason": "LayoutLMv2 tokenizer crashes and RTDetrLoss cost-matrix infeasibility are unrelated bugs." + "reason": "Tokenizer padding/truncation crash is unrelated to RTDetrLoss infeasible cost-matrix failure." }, { "left": "issue:43824", "right": "issue:43976", "accept": false, - "reason": "An import/export regression for Qwen2_5_VLForConditionalGeneration is unrelated to Python 3.9/3.10 compatibility." + "reason": "Import error for a specific class and Python compatibility are unrelated." }, { - "left": "issue:42796", - "right": "issue:43441", + "left": "issue:44016", + "right": "issue:45823", "accept": false, - "reason": "Tokenizer BPE-marker handling and FlashAttention failures are different subsystems and concrete failures." + "reason": "Notebook syntax error and Gemma4 device mismatch are unrelated." }, { - "left": "issue:44016", - "right": "issue:45823", + "left": "issue:42796", + "right": "issue:43441", "accept": false, - "reason": "Notebook syntax error and Gemma4 device-mismatch during forward are unrelated." + "reason": "Same family name, but different concrete failures: BPE marker handling vs FlashAttention loading." }, { "left": "issue:45542", "right": "issue:45800", "accept": false, - "reason": "TensorBoard-only TF backend error and torch/transformers version incompatibility are distinct packaging/runtime issues." + "reason": "TensorFlow backend installation issue and torch/transformers incompatibility are separate environment problems." }, { - "left": "issue:43525", - "right": "issue:43643", + "left": "issue:44188", + "right": "issue:44336", "accept": false, - "reason": "A missing pad_token_id on Llama4Config is not the same as missing fields from trust_remote_code AutoConfig." + "reason": "Attention-kernel divergence and ANSI code emission have nothing in common beyond being bugs." }, { - "left": "issue:44188", - "right": "issue:44336", + "left": "issue:43525", + "right": "issue:43643", "accept": false, - "reason": "torch.compile attention-kernel divergence and ANSI codes in loading_report are entirely unrelated." + "reason": "The latter is a generic trust_remote_code missing-fields issue; the former is a specific Llama4 pad_token_id AttributeError. Related symptom class, but not the same concrete bug." }, { "left": "issue:44589", "right": "issue:45362", "accept": false, - "reason": "Missing Float8 storage support and Qwen3.5 chat loading failure do not share a concrete code path." + "reason": "Float8 storage import failure and Qwen3.5 chat crash are unrelated." }, { "left": "issue:44336", "right": "issue:44373", "accept": false, - "reason": "Terminal ANSI formatting and a wrong docstring for position_ids are unrelated documentation/runtime concerns." + "reason": "Loading-report ANSI formatting and a docstring typo are unrelated." }, { "left": "issue:45325", "right": "issue:46851", "accept": false, - "reason": "Both involve Qwen position handling, but one is rope scaling for still images and the other is variable-length inputs; they are not the same bug." + "reason": "Both mention position_ids, but they target different Qwen models and different temporal/variable-length bugs." }, { "left": "issue:46655", "right": "issue:46726", "accept": false, - "reason": "SmolVLM video padding calculation and missing-vision-backend processor errors are different processor problems." + "reason": "Video padding calculation and missing vision backend handling are different processor issues." }, { "left": "issue:46291", "right": "issue:46399", "accept": false, - "reason": "kernels import/version requirements and missing apply_rotary_pos_emb registration are separate kernels integration failures." + "reason": "Kernel dependency/version gating is unrelated to a Qwen3.5 rotary-embedding registration crash." }, { "left": "issue:44246", "right": "issue:44496", "accept": false, - "reason": "Intermittent import slowness is unrelated to unrecognized-model config loading failures." + "reason": "Import slowness and unrecognized model/config loading are unrelated." }, { "left": "issue:41628", "right": "issue:43976", "accept": false, - "reason": "Cannot import AutoImageProcessor and Python-version incompatibility are unrelated issues." + "reason": "AutoImageProcessor import failure is unrelated to Python 3.9/3.10 compatibility." }, { "left": "issue:45083", "right": "issue:46851", "accept": false, - "reason": "Helper output-length behavior in qwen3_omni_moe is different from Qwen3.5 variable-length position_ids handling." + "reason": "Different Qwen helper behavior on feature lengths vs position_ids; not the same bug." } ] }, @@ -13271,8 +13225,8 @@ "issue:43531|issue:45362", "issue:44246|issue:44871", "issue:46614|issue:46665", - "issue:42491|issue:43299", "issue:45588|issue:46851", + "issue:42491|issue:43299", "issue:44295|issue:45397", "issue:44242|issue:45593", "issue:45003|issue:46821", @@ -13283,222 +13237,222 @@ "issue:36246|issue:42757", "issue:36246|issue:44351", "issue:43901|issue:45003", - "issue:44451|issue:46752", "issue:43257|issue:45137", + "issue:44451|issue:46752", "issue:41950|issue:44220", "issue:43540|issue:46655", "issue:44496|issue:45003", - "issue:45003|issue:45561", - "issue:43262|issue:44246" + "issue:43262|issue:44246", + "issue:45003|issue:45561" ], - "prepared_review_unit_hash": "de79a234377e4adba0f1df0ee785a1a6179db06f5b65c999a998ea8cb05b4d23", + "prepared_review_unit_hash": "c56d02123736878dadf633f72e6d52ce260c31e4d30f6bea2ddfa56cc43fcef7", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36683, - "estimated_input_tokens": 9171, - "estimated_eval_tokens": 18598 + "serialized_chars": 36689, + "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 broad mix of unrelated issue reports. The soft pairs mostly share a model family or subsystem name, but they describe different concrete failures and should not be deduplicated.", - "confidence": 0.95, - "canonical_issue_reason": null, + "summary": "These items are mostly unrelated issues across different models, processors, and infrastructure paths. None of the soft-edge pairs look like the same underlying bug/change closely enough to merge.", + "confidence": 0.86, + "canonical_issue_reason": "No single canonical issue stands out: the set spans many distinct regressions (imports, model loading, tokenizers, audio/video processors, deepspeed, docs) with no clear duplicate center.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a strong canonical for the whole set because the cluster is heterogeneous. If one had to be chosen, issue #45561 is the clearest standalone report with a concrete race-condition reproduction and broad test impact.", + "best_issue_reason": "No issue is a strong global representative for the cluster because the topics are heterogeneous and the pairwise overlaps are only superficial.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43784", "right": "issue:45542", "accept": false, - "reason": "Different failures: one is a missing `nn` import on transformers import, the other is a TensorFlow backend/`tf` availability issue in logging setup." + "reason": "Both are import/backend errors, but they involve different symbols and subsystems: sentence-transformers import vs TensorBoard/TF backend detection." }, { "left": "issue:44297", "right": "issue:45397", "accept": false, - "reason": "Unrelated code paths: tokenizer class metadata in `save_pretrained` vs a ZeRO3 `from_pretrained` loading issue for Gemma-4." + "reason": "Different failures: tokenizer save metadata mismatch vs gemma-4 zero3 from_pretrained loading." }, { "left": "issue:46399", "right": "issue:46851", "accept": false, - "reason": "Both involve Qwen3.5, but one is a kernel registration/decorator crash and the other is a variable-length `position_ids` bug; not the same underlying issue." + "reason": "Same model family, but different bugs: registration/decorator crash vs variable-length position_ids handling." }, { "left": "issue:45561", "right": "issue:45563", "accept": false, - "reason": "A pytest-xdist file race and a stale generation warning are different bugs in different parts of the test/runtime stack." + "reason": "Completely different code paths: xdist captured_info.txt race vs stale generate() warning." }, { "left": "issue:43526", "right": "issue:45003", "accept": false, - "reason": "Different subsystems: Beit image processor label reduction vs unsafe `sys.modules` access in modeling utils." + "reason": "Unrelated: BEiT image-label reduction bug vs unsafe sys.modules access in modeling_utils." }, { "left": "issue:45103", "right": "issue:46399", "accept": false, - "reason": "Docstring generation with future annotations is unrelated to the Qwen3.5 kernel registration crash." + "reason": "Auto-docstring annotation crash is unrelated to Qwen3.5 kernel registration." }, { "left": "issue:45072", "right": "issue:46829", "accept": false, - "reason": "Different model behaviors: dtype mismatch in bfloat16 inference vs an output field mix-up in VideoPrism classification." + "reason": "Different model paths and symptoms: dtype mismatch in inference vs incorrect hidden_states field in classification output." }, { "left": "issue:43540", "right": "issue:46752", "accept": false, - "reason": "Video-input processing in Qwen3OmniMoe is unrelated to an empty-list crash in `apply_chat_template`." + "reason": "Different areas: Qwen3OmniMoe video processing error vs empty apply_chat_template input handling." }, { "left": "issue:43531", "right": "issue:45362", "accept": false, - "reason": "Both are Qwen-related, but one is about `sliding_window` and the other is a chat-loading crash; different concrete bugs." + "reason": "Both mention Qwen3 variants, but one is sliding_window behavior and the other is chat loading crash." }, { "left": "issue:44246", "right": "issue:44871", "accept": false, - "reason": "Import slowdown and Gemma-3 EOS configuration inconsistency are unrelated." + "reason": "MoE load-balancing loss issue is unrelated to Gemma-3 eos_token_id configuration." }, { "left": "issue:46614", "right": "issue:46665", "accept": false, - "reason": "A stale phonemizer backend after language switching is different from inconsistent `from_pretrained` behavior inside Python classes." + "reason": "Stale phonemizer language backend is a different bug from inconsistent from_pretrained behavior in Python classes." }, { - "left": "issue:42491", - "right": "issue:43299", + "left": "issue:45588", + "right": "issue:46851", "accept": false, - "reason": "Both mention Qwen3 MoE loading, but one is a LoRA/accelerate-deepspeed conversion problem and the other is a Transformers 5.0 model-loading regression; not the same bug." + "reason": "Flash-attention sinkless-model crash is unrelated to variable-length position_ids." }, { - "left": "issue:45588", - "right": "issue:46851", + "left": "issue:42491", + "right": "issue:43299", "accept": false, - "reason": "Flash-attention `s_aux=None` handling is unrelated to variable-length `position_ids` support." + "reason": "Both concern Qwen MoE loading regressions, but they target different model variants and failure modes; not the same concrete bug." }, { "left": "issue:44295", "right": "issue:45397", "accept": false, - "reason": "Registering/reading `position_ids` as a buffer is a different failure from Gemma-4 ZeRO3 loading." + "reason": "Position_ids buffer access and gemma-4 zero3 loading are distinct issues." }, { "left": "issue:44242", "right": "issue:45593", "accept": false, - "reason": "Missing load-balancing loss when router logits are off is a different model-training bug than D-FINE auxiliary losses disappearing when denoising is off." + "reason": "Different training losses and different models; only superficial similarity around missing auxiliary loss." }, { "left": "issue:45003", "right": "issue:46821", "accept": false, - "reason": "Unsafe module lookup and a TimesFM `window_size` attribute error do not share the same code path." + "reason": "sys.modules access bug is unrelated to TimesFM window_size attribute handling." }, { "left": "issue:44336", "right": "issue:44485", "accept": false, - "reason": "ANSI-code emission in loading reports is unrelated to GLM-5 RoPE implementation." + "reason": "ANSI-code reporting bug has nothing to do with GLM-5 RoPE implementation." }, { "left": "issue:42886", "right": "issue:46132", "accept": false, - "reason": "Offline tokenizer cache loading and `AttentionInterface.register` behavior changes are different problems." + "reason": "Offline cache loading and AttentionInterface.register behavior are unrelated." }, { "left": "issue:43550", "right": "issue:46506", "accept": false, - "reason": "A `torch.compile`/SDPA failure in Bamba is unrelated to CPU/CUDA output divergence in PPDocLayoutV3." + "reason": "torch.compile/SDPA failure in Bamba is unrelated to CPU/CUDA output differences in PPDocLayoutV3." }, { "left": "issue:44779", "right": "issue:45292", "accept": false, - "reason": "Deepseek tokenizer correctness regressions and `resize_token_embeddings` not updating output embeddings are different bugs." + "reason": "DeepSeek tokenizer correctness regression is not the same as resize_token_embeddings not updating output embeddings." }, { "left": "issue:36246", "right": "issue:42757", "accept": false, - "reason": "Missing `Qwen2_5_VLImageProcessor` and missing `is_offline_mode` are distinct import errors from different modules." + "reason": "Both are import errors, but for different missing symbols from different packages." }, { "left": "issue:36246", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but for different symbols (`Qwen2_5_VLImageProcessor` vs `HybridCache`) and different underlying fixes." + "reason": "Qwen2_5_VLImageProcessor import failure is unrelated to HybridCache export failure." }, { "left": "issue:43901", "right": "issue:45003", "accept": false, - "reason": "A docs/behavior mismatch for `return_all_scores` is unrelated to unsafe `sys.modules` access." + "reason": "Documentation mismatch vs unsafe sys.modules access are unrelated." }, { - "left": "issue:44451", - "right": "issue:46752", + "left": "issue:43257", + "right": "issue:45137", "accept": false, - "reason": "A model-loading regression for ScandiBERT is not the same as an empty-list crash in chat template application." + "reason": "Both involve distributed training stacks, but one is weights conversion with accelerate/deepspeed and the other is a ZeRO3 deque pop error." }, { - "left": "issue:43257", - "right": "issue:45137", + "left": "issue:44451", + "right": "issue:46752", "accept": false, - "reason": "Both involve DeepSpeed, but one is Qwen3 MoE weight conversion with accelerate and the other is a ZeRO3 deque underflow; different concrete failures." + "reason": "Model loading failure for ScandiBERT is unrelated to empty-list chat template crashes." }, { "left": "issue:41950", "right": "issue:44220", "accept": false, - "reason": "Video-classification processor selection and `_torch_extract_fbank_features()` are unrelated pipeline/audio issues." + "reason": "Video-classification pipeline using image processors and torch_extract_fbank_features are different pipeline bugs." }, { "left": "issue:43540", "right": "issue:46655", "accept": false, - "reason": "Both are video-related, but one is a Qwen3OmniMoe input-processing error and the other is a SmolVLM padding calculation bug." + "reason": "Both touch video processing, but they affect different models and different logic: input validation vs padding calculation." }, { "left": "issue:44496", "right": "issue:45003", "accept": false, - "reason": "An unrecognized-model/config issue is unrelated to unsafe `sys.modules` access." + "reason": "Unrecognized model config is unrelated to sys.modules import safety." }, { - "left": "issue:45003", - "right": "issue:45561", + "left": "issue:43262", + "right": "issue:44246", "accept": false, - "reason": "Model loading internals and a pytest-xdist captured-info race are completely different." + "reason": "Audio chat-template sample-rate default and MoE load-balancing loss are unrelated." }, { - "left": "issue:43262", - "right": "issue:44246", + "left": "issue:45003", + "right": "issue:45561", "accept": false, - "reason": "Audio chat-template sampling-rate defaults and missing load-balancing loss are unrelated bugs in different components." + "reason": "Different subsystems: modeling_utils import lookup vs pytest-xdist file race." } ] }, @@ -13578,8 +13532,8 @@ "issue:44295|issue:44749", "issue:45859|issue:46399", "issue:45137|issue:45600", - "issue:43475|issue:46369", "issue:43295|issue:45292", + "issue:43475|issue:46369", "issue:36246|issue:43596", "issue:44466|issue:45335", "issue:44242|issue:44857", @@ -13589,199 +13543,199 @@ "issue:44038|issue:46506", "issue:44987|issue:45459", "issue:29870|issue:30333", - "issue:43526|issue:46894", "issue:44479|issue:46170", + "issue:43526|issue:46894", "issue:42199|issue:46851", "issue:44060|issue:44077", - "issue:44368|issue:44977", "issue:43352|issue:45250", + "issue:44368|issue:44977", "issue:43653|issue:44162", "issue:45412|issue:46123" ], - "prepared_review_unit_hash": "babbf8ba273f7def2ab44384477799ed5614d93a86dfe54e85a9e46e8e133ce6", + "prepared_review_unit_hash": "fdb539227d30ab2ebbea8ed8190da3673f233328186a5a1c9aac990bf57951de", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 26, - "serialized_chars": 35261, - "estimated_input_tokens": 8816, - "estimated_eval_tokens": 17888 + "serialized_chars": 35257, + "estimated_input_tokens": 8815, + "estimated_eval_tokens": 17886 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab bag of unrelated model- and component-specific bugs. The soft-similarity pairs share broad themes in a few cases, but none look like the same underlying fix, so all soft edges should be rejected.", - "confidence": 0.94, - "canonical_issue_reason": "issue:44479 is the strongest anchor in the set: it reports a concrete regression, names several affected Qwen video variants, and has a clear failure mode that could attract related reports.", + "summary": "These soft pairs are all false positives: they share broad model families or subsystems, but the underlying bugs/change requests are different. No PRs are present.", + "confidence": 0.95, + "canonical_issue_reason": "issue:44479 is a concrete, high-signal regression report with multiple affected variants and active discussion, making it a reasonable representative issue among this set.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44479 is the most useful standalone issue here because it is broad yet specific, reproducible, and scoped to a concrete regression across multiple models.", + "best_issue_reason": "issue:44479 is the strongest standalone issue candidate because it is specific, user-impacting, and broad enough to represent a real regression class.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45375", "right": "issue:45759", "accept": false, - "reason": "Both are Qwen multimodal config issues, but one is about a missing config field being dropped and the other is about unwrapping text_config in from_config; different bugs." + "reason": "Both are Qwen multimodal config problems, but one is a missing config field being dropped by strict parsing and the other is AutoModelForCausalLM.from_config not unwrapping text_config. Different failure modes and fixes." }, { "left": "issue:44479", "right": "issue:46655", "accept": false, - "reason": "Both involve video inputs, but the affected models and root causes differ: a general Qwen video regression versus a SmolVLM video padding bug." + "reason": "Both mention video inputs, but one is a Qwen-VL regression across several models while the other is a SmolVLM padding bug in the video processor. Not the same bug." }, { "left": "issue:44162", "right": "issue:45200", "accept": false, - "reason": "Different models and different failures: ESM2 is broken versus Gemma 4 requiring mm_token_type_ids defaults." + "reason": "Unrelated models and symptoms: ESM2 is broadly broken, while Gemma 4 needs mm_token_type_ids to default to zeros for text-only fine-tuning." }, { "left": "issue:45292", "right": "issue:47000", "accept": false, - "reason": "Unrelated problems: token embedding resize behavior versus RT-DETR loss infeasibility." + "reason": "resize_token_embeddings/output_embeddings is unrelated to RT-DETR loss infeasibility. Different subsystems and root causes." }, { "left": "issue:44295", "right": "issue:44749", "accept": false, - "reason": "Different symptom and scope: position_ids buffer access error versus a performance regression after upgrading Transformers." + "reason": "One is a position_ids buffer access error; the other is a major performance regression after upgrade. No shared bug path." }, { "left": "issue:45859", "right": "issue:46399", "accept": false, - "reason": "Same model family, but different code paths and failures: CPU offload incompatibility versus missing kernel registration decorator." + "reason": "Both are Qwen3.5 rotary-embedding-related, but one is CPU offload incompatibility and the other is a kernel registration decorator crash. Different concrete issues." }, { "left": "issue:45137", "right": "issue:45600", "accept": false, - "reason": "Completely different issues: DeepSpeed ZeRO3 deque failure versus stale auto_mappings entries for removed configs." + "reason": "DeepSpeed ZeRO3 deque failure and stale auto_mappings references to removed configs are unrelated." }, { - "left": "issue:43475", - "right": "issue:46369", + "left": "issue:43295", + "right": "issue:45292", "accept": false, - "reason": "Same SAM3 area, but one is an attribute error in video outputs and the other is a memory overflow in sam3-litetext; not the same bug." + "reason": "Processor/tokenizer regression has nothing to do with resize_token_embeddings not updating output_embeddings." }, { - "left": "issue:43295", - "right": "issue:45292", + "left": "issue:43475", + "right": "issue:46369", "accept": false, - "reason": "Different regression domains: processor.tokenizer / image passing breakage versus output embedding resize behavior." + "reason": "Both involve SAM3, but one is a missing attribute on a vision encoder output and the other is a memory overflow in sam3-litetext. Different problems." }, { "left": "issue:36246", "right": "issue:43596", "accept": false, - "reason": "No overlap in failure mode or component: missing Qwen2_5_VLImageProcessor import versus a DeepSpeed/BertModel IndexError." + "reason": "ImportError for Qwen2.5-VL image processor and a DeepSpeed/BertModel IndexError are unrelated." }, { "left": "issue:44466", "right": "issue:45335", "accept": false, - "reason": "Both mention tied/resize embedding behavior, but one is about inconsistent lm_head serialization and the other about decoder.embed_tokens not updating; not the same defect." + "reason": "Both mention tied/resized embeddings, but one is inconsistent serialization across devices and the other is t5gemma resize_token_embeddings not propagating to decoder.embed_tokens." }, { "left": "issue:44242", "right": "issue:44857", "accept": false, - "reason": "Different subsystems and errors: router load-balancing loss omission versus an AMP/CUDA crash in LwDetrImageLoss." + "reason": "Load-balancing loss omission and LwDetrImageLoss float16 CUDA crash are different code paths." }, { "left": "issue:44623", "right": "issue:46726", "accept": false, - "reason": "Saving processors missing files is unrelated to AutoProcessor reporting an unrecognized image processor when a backend is absent." + "reason": "Missing files in processor.save_pretrained and AutoProcessor complaining about a missing vision backend are not the same issue." }, { "left": "issue:44877", "right": "issue:45375", "accept": false, - "reason": "Both involve config strictness, but they are distinct models and different missing-field problems." + "reason": "Both involve strict config handling, but they affect different model families and different missing fields; not the same underlying bug." }, { "left": "issue:43526", "right": "issue:46897", "accept": false, - "reason": "Different models and bugs: BeitImageProcessorFast label reduction versus Florence2 training-loss double-shift." + "reason": "BeitImageProcessorFast reduce_labels truncation and CohereASR training-loss double-shift are unrelated." }, { "left": "issue:44038", "right": "issue:46506", "accept": false, - "reason": "Unrelated: a Qwen3-VL-Moe loading issue versus a CPU/CUDA output discrepancy in PPDocLayoutV3." + "reason": "Qwen3-VL-Moe/transformers 5.0 compatibility and PPDocLayoutV3 CPU-vs-CUDA divergence are unrelated." }, { "left": "issue:44987", "right": "issue:45459", "accept": false, - "reason": "Different issues entirely: a Transformers version loading failure versus tokenizer error masking when protobuf is missing." + "reason": "Package loading failure and exception masking when protobuf is missing are different bugs." }, { "left": "issue:29870", "right": "issue:30333", "accept": false, - "reason": "No shared bug: Bart generate() hits an UnboundLocalError, while the other is MLflow reporting a failed training job as still running." + "reason": "Bart generation UnboundLocalError and MLflow reporting after failed training jobs are unrelated." }, { - "left": "issue:43526", - "right": "issue:46894", + "left": "issue:44479", + "right": "issue:46170", "accept": false, - "reason": "Same generic 'double-shift' wording is not enough; these are different models and likely different training-loss bugs." + "reason": "A video-input regression in Qwen models is unrelated to Windows/ROCm import and torch.distributed dependency errors." }, { - "left": "issue:44479", - "right": "issue:46170", + "left": "issue:43526", + "right": "issue:46894", "accept": false, - "reason": "One is a video-input regression in Qwen video models; the other is a Windows/ROCm torch.distributed dependency issue." + "reason": "Different models and symptoms: BeitImageProcessorFast label reduction bug versus CohereASR double-shift training loss bug." }, { "left": "issue:42199", "right": "issue:46851", "accept": false, - "reason": "Different model families and different failures: DETR cardinality logic versus Qwen3.5 variable-length position_ids handling." + "reason": "DETR cardinality error and Qwen3.5 variable-length position_ids behavior are unrelated." }, { "left": "issue:44060", "right": "issue:44077", "accept": false, - "reason": "Distinct issues: a tied-weights warning involving embed_tokens and dt_bias versus a patchtsmixer post_init policy change." + "reason": "One is a tied-weights warning in Qwen3-Next, the other is a patchtsmixer post_init policy issue. No shared change or bug." }, { - "left": "issue:44368", - "right": "issue:44977", + "left": "issue:43352", + "right": "issue:45250", "accept": false, - "reason": "Both are Qwen3.5-related, but one is a tied-weights warning and the other is a flash-attention generation failure; different root causes." + "reason": "A model-specific Flash Attention 2.0 unsupported error is not the same as a generic Flash Attention 2.0 issue." }, { - "left": "issue:43352", - "right": "issue:45250", + "left": "issue:44368", + "right": "issue:44977", "accept": false, - "reason": "One is a model-specific unsupported-feature error; the other is a generic Flash Attention 2.0 topic, not the same concrete bug." + "reason": "A tie_word_embeddings warning during LoRA fine-tuning is unrelated to a Qwen3.5 flash-attention generation failure." }, { "left": "issue:43653", "right": "issue:44162", "accept": false, - "reason": "Different areas and symptoms: tokenizer special-token registration versus ESM2 being broken." + "reason": "BigBirdTokenizer special-token registration and ESM2 breakage are unrelated." }, { "left": "issue:45412", "right": "issue:46123", "accept": false, - "reason": "Completely different bugs: RT-DETR memory not released versus MaskGenerationPipeline dropping final partial-batch results." + "reason": "RT-DETR memory cleanup and MaskGenerationPipeline final-batch dropping are different bugs in different components." } ] }, @@ -13849,7 +13803,6 @@ "issue:46710", "issue:46726", "issue:46752", - "issue:46821", "issue:46832", "issue:46858", "issue:47020" @@ -13884,223 +13837,216 @@ "issue:44623|issue:45656", "issue:44871|issue:46132", "issue:43475|issue:43643", - "issue:45375|issue:45478", - "issue:43475|issue:46821" + "issue:45375|issue:45478" ], - "prepared_review_unit_hash": "d8f9d923d2cb77b7d5951bbd67de999bf40c10525f348d353b40a06f965270e8", + "prepared_review_unit_hash": "33ea79e6639d7e3f1da946d1cfc84dfaf9d425c3fc3e7c6d1c1cea0e69c6d619", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 37127, - "estimated_input_tokens": 9282, - "estimated_eval_tokens": 18820 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36281, + "estimated_input_tokens": 9071, + "estimated_eval_tokens": 18398 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is mostly heterogeneous. A few issues share a loose Transformers v5 loading/config/theme, but the soft pairs do not look like true duplicates of the same underlying bug or change. No PRs are present.", - "confidence": 0.39, - "canonical_issue_reason": "issue:43646 is the broadest regression report in the set: Transformers 5.0.0 breaking custom model initialization/loading. It best represents the loose v5 load/config regression theme, even though the cluster is not tight.", + "summary": "This cluster is a loose grab-bag of mostly unrelated bug reports across tokenizers, config loading, vision models, chat templates, cache behavior, and performance regressions. None of the soft pairs look like the same concrete underlying bug/change, so I would not merge any of them as duplicates.", + "confidence": 0.91, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:43646 is the best single representative because it is version-specific, high-level, and closest to the common denominator across several loading/config-related reports.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:42886", "right": "issue:45229", "accept": false, - "reason": "Tokenizer offline cache loading and Gemma4 CUDA OOM are unrelated failures." + "reason": "Offline tokenizer cache loading and Gemma4 CUDA OOM are unrelated failures in different code paths." }, { "left": "issue:45397", "right": "issue:45656", "accept": false, - "reason": "Gemma-4 zero3 from_pretrained and deepspeed optimizer double-step are different code paths and symptoms." + "reason": "Gemma4 zero3 from_pretrained breakage and double optimizer-step with deepspeed are different bugs." }, { "left": "issue:44877", "right": "issue:46170", "accept": false, - "reason": "Strict config loading for granite_speech is unrelated to the Windows/ROCm torch.distributed import problem." + "reason": "Strict config loading for granite_speech and Windows/ROCm torch.distributed import issues do not share a concrete root cause." }, { "left": "issue:43475", "right": "issue:43540", "accept": false, - "reason": "Both are video/model issues, but they affect different models and different failure modes." + "reason": "SAM3 video missing field and Qwen3OmniMoe video ValueError are separate model-specific loading/processing bugs." }, { "left": "issue:43065", "right": "issue:43408", "accept": false, - "reason": "SAM3 dummy Conv2d and sam3_video vs sam3_tracker type mismatch are separate SAM3 issues." + "reason": "Dummy Conv2d in Sam3PixelDecoder and sam3_video vs sam3_tracker warning are different SAM3 issues." }, { "left": "issue:46170", "right": "issue:46832", "accept": false, - "reason": "Import/dependency failure on Windows is unrelated to RTDetr forward shape logic." + "reason": "Windows ROCm import/dependency failure is unrelated to RT-DETR forward TypeError." }, { "left": "issue:43723", "right": "issue:46710", "accept": false, - "reason": "Tokenizer loading failure and accuracy regression after upgrade are not the same bug." + "reason": "Tokenizer loading in v5 and accuracy regression after version upgrade are not the same bug." }, { "left": "issue:44368", "right": "issue:45563", "accept": false, - "reason": "These are different warnings in different generation paths." + "reason": "A warning about tie_word_embeddings and a stale num_return_sequences warning affect different generation paths." }, { "left": "issue:43931", "right": "issue:44368", "accept": false, - "reason": "Weight shape mismatch and tie_word_embeddings warning are unrelated model-loading/config issues." + "reason": "Qwen3-VL weight-shape mismatch is unrelated to a Qwen3.5 LoRA warning." }, { "left": "issue:44995", "right": "issue:45200", "accept": false, - "reason": "Stale cache state on second forward and Gemma4 mm_token_type_ids defaulting are different bugs." + "reason": "Stale indexer cache on second forward and Gemma4 mm_token_type_ids defaulting are separate issues." }, { "left": "issue:44188", "right": "issue:46202", "accept": false, - "reason": "torch.compile attention-kernel divergence is unrelated to Gemma pipeline roundtrip failure." + "reason": "Torch.compile attention-kernel branching and Gemma tokenizer roundtrip on \u2581 are unrelated." }, { "left": "issue:44483", "right": "issue:46202", "accept": false, - "reason": "Chat/completions endpoint rejection and Gemma pipeline roundtrip failure are different issues." + "reason": "Chat/completions request rejection and Gemma pipeline roundtrip failure are different request-processing bugs." }, { "left": "issue:43906", "right": "issue:46202", "accept": false, - "reason": "An isolated reproduction of another issue is not the same as the Gemma pipeline bug." + "reason": "An isolated reproduction of another issue and a Gemma roundtrip failure are not the same defect." }, { "left": "issue:45987", "right": "issue:46082", "accept": false, - "reason": "StaticCache return-type contract and LlamaConfig head_dim validation are unrelated." + "reason": "StaticCache seq-length return type and LlamaConfig head_dim validation are unrelated API problems." }, { "left": "issue:44188", "right": "issue:46421", "accept": false, - "reason": "Both mention cache/compile, but one is attention-kernel divergence and the other is lazy cache initialization recompilation." + "reason": "Diverging attention kernels under torch.compile and cache lazy-init recompilation are different cache/compile behaviors." }, { "left": "issue:44995", "right": "issue:45588", "accept": false, - "reason": "Cache indexer staleness and flash_attention s_aux=None crash are separate code paths." + "reason": "Stale cache on second forward and flash_attention s_aux=None crash do not share the same code path." }, { "left": "issue:45685", "right": "issue:46431", "accept": false, - "reason": "MPS histogram int support and missing chat template are unrelated failures." + "reason": "MPS histogram type error and missing chat template error are unrelated." }, { "left": "issue:45103", "right": "issue:46858", "accept": false, - "reason": "Docstring annotation handling and cache seq-length tensor behavior are unrelated." + "reason": "auto_docstring annotation handling and StaticCache seq-length tensor arithmetic under torch.compile are separate subsystems." }, { "left": "issue:44361", "right": "issue:44749", "accept": false, - "reason": "Tokenizer AttributeError and performance regression after upgrade are not the same bug." + "reason": "MLukeTokenizer AttributeError and a performance regression after upgrade are not duplicates." }, { "left": "issue:43065", "right": "issue:46123", "accept": false, - "reason": "SAM3 pixel decoder dummy layer and MaskGenerationPipeline batch handling are unrelated." + "reason": "SAM3 pixel-decoder placeholder and MaskGenerationPipeline partial-batch dropping are unrelated." }, { "left": "issue:44297", "right": "issue:44623", "accept": false, - "reason": "Tokenizer save metadata mismatch and processor save missing files are different save_pretrained issues." + "reason": "Both involve save_pretrained, but tokenizer_config mismatch and processor file omission are different serialization bugs." }, { "left": "issue:43994", "right": "issue:46726", "accept": false, - "reason": "Nonsensical model outputs and missing vision backend error are different problems." + "reason": "Siglip2 nonsense outputs and AutoProcessor missing-backend error are not the same issue." }, { "left": "issue:46752", "right": "issue:47020", "accept": false, - "reason": "Empty-list chat template crash and XLMRobertaTokenizer init type error are unrelated tokenizer bugs." + "reason": "Empty chat-template input crash and XLMRobertaTokenizer Unigram init type error are unrelated tokenizer bugs." }, { "left": "issue:44749", "right": "issue:45656", "accept": false, - "reason": "Slowdown after upgrade and optimizer step being called twice are different training-path issues." + "reason": "Training slowdown and double optimizer-step behavior are different deepspeed-related problems." }, { "left": "issue:43701", "right": "issue:44749", "accept": false, - "reason": "Checkpoint key mismatch and post-upgrade slowdown are unrelated." + "reason": "Checkpoint key mismatch and performance regression after upgrade do not match." }, { "left": "issue:43646", "right": "issue:44162", "accept": false, - "reason": "Both are v5 breakages, but one is generic custom-model initialization and the other is ESM2-specific behavior." + "reason": "AutoConfig missing fields with trust_remote_code and ESM2 breakage are separate model/config issues." }, { "left": "issue:44623", "right": "issue:45656", "accept": false, - "reason": "Processor save_pretrained missing files has no relation to deepspeed double optimizer stepping." + "reason": "Missing processor save files and duplicate optimizer stepping are unrelated." }, { "left": "issue:44871", "right": "issue:46132", "accept": false, - "reason": "Gemma eos_token_id inconsistency and AttentionInterface.register behavior change are unrelated." + "reason": "Gemma eos_token_id inconsistency and AttentionInterface.register behavior change are different API problems." }, { "left": "issue:43475", "right": "issue:43643", "accept": false, - "reason": "SAM3 missing encoder output field and AutoConfig trust_remote_code field loss are different configs/models." + "reason": "SAM3 missing attribute and AutoConfig trust_remote_code field loss are unrelated loading bugs." }, { "left": "issue:45375", "right": "issue:45478", "accept": false, - "reason": "Both involve Qwen3.5 MoE config/loading, but one is a missing config field and the other is a broader from_pretrained error; not clearly the same concrete bug." - }, - { - "left": "issue:43475", - "right": "issue:46821", - "accept": false, - "reason": "SAM3 vision output attr error and TimesFM window_size AttributeError are unrelated model-specific bugs." + "reason": "Missing deepstack_visual_indexes in config and generic Qwen3.5 MoE from_pretrained error are related only by family, not the same concrete bug." } ] }, @@ -14132,17 +14078,16 @@ "issue:42371", "issue:42994", "issue:43232", + "issue:43475", "issue:43482", "issue:43525", "issue:43673", "issue:43723", "issue:43784", - "issue:43824", "issue:43873", "issue:43986", "issue:44038", "issue:44188", - "issue:44222", "issue:44351", "issue:44373", "issue:44623", @@ -14169,13 +14114,15 @@ "issue:46463", "issue:46701", "issue:46726", + "issue:46821", "issue:46851", "issue:46877", "issue:47000" ], "soft_pairs": [ - "issue:42994|issue:43482", "issue:44964|issue:45137", + "issue:42994|issue:43482", + "issue:43475|issue:46821", "issue:28282|issue:30990", "issue:44188|issue:46851", "issue:39692|issue:46726", @@ -14195,238 +14142,230 @@ "issue:43873|issue:44945", "issue:43784|issue:46399", "issue:45406|issue:46202", - "issue:44945|issue:45753", "issue:45137|issue:47000", + "issue:44945|issue:45753", "issue:41628|issue:44351", "issue:36010|issue:41628", "issue:44857|issue:46877", - "issue:46202|issue:46701", "issue:45081|issue:45103", + "issue:46202|issue:46701", "issue:45593|issue:46877", - "issue:44038|issue:45310", - "issue:42994|issue:43824", - "issue:44222|issue:45310" + "issue:44038|issue:45310" ], - "prepared_review_unit_hash": "3694949b3f6689293ccab4396245d9e1e2b32357adeb2af45e794101c8c5a590", + "prepared_review_unit_hash": "41bd05897c0ee73432360e332bfb42d6132de587648a1bc8f54ac498fea2b22b", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 37002, - "estimated_input_tokens": 9251, - "estimated_eval_tokens": 18758 + "soft_pair_count": 31, + "serialized_chars": 36720, + "estimated_input_tokens": 9180, + "estimated_eval_tokens": 18616 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is not a true duplicate cluster: it mixes unrelated import/API regressions, tokenizer/model-loading bugs, quantization/save issues, pipeline/generation bugs, and docs problems. None of the soft pairs look like the same underlying bug or change.", - "confidence": 0.83, - "canonical_issue_reason": "If a representative issue is needed, issue:41628 is the cleanest umbrella for the repeated top-level import/API regression theme, but it does not subsume the rest of the set.", + "summary": "This set is heterogeneous: it mixes import/API regressions, model-loading issues, tokenizer/pipeline bugs, docs problems, and training/runtime crashes. None of the soft pairs is safely the same concrete bug, so I would not merge any of them as duplicates.", + "confidence": 0.59, + "canonical_issue_reason": "Issue 41628 is the cleanest representative of the recurring import/API-breakage theme in this cluster and is broader/more canonical than the more niche model- or docs-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "Issue:41628 is the most concrete and broadly representative of the import/export regression subgroup; its title is clear, self-contained, and closer to several other import-related reports than the rest.", + "best_issue_reason": "Issue 41628 is the most self-contained and broadly representative bug report here, making it the best anchor for triage even though the cluster is not a true duplicate set.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44964", + "right": "issue:45137", + "accept": false, + "reason": "Different failures: Phi-4 multimodal loading versus a DeepSpeed ZeRO3 deque underflow." + }, { "left": "issue:42994", "right": "issue:43482", "accept": false, - "reason": "Both involve model loading/saving around newer formats, but one is quantized model saving and the other is GGUF loading for Qwen2.5; different failure modes and code paths." + "reason": "Quantized-model saving and GGUF loading are opposite directions and use different code paths." }, { - "left": "issue:44964", - "right": "issue:45137", + "left": "issue:43475", + "right": "issue:46821", "accept": false, - "reason": "Phi-4 multimodal loading and DeepSpeed ZeRO3 deque errors are unrelated bugs." + "reason": "Different model families and different attribute errors; no shared concrete bug." }, { "left": "issue:28282", "right": "issue:30990", "accept": false, - "reason": "PyTorch dependency-missing import error is not the same as Sentence Transformers hanging during load." + "reason": "Missing PyTorch at import time versus Sentence Transformers hanging while loading are unrelated." }, { "left": "issue:44188", "right": "issue:46851", "accept": false, - "reason": "Attention-kernel branching under torch.compile is unrelated to variable-length position_ids handling." + "reason": "torch.compile attention-kernel divergence is not the same as Qwen3.5 variable-length position_ids handling." }, { "left": "issue:39692", "right": "issue:46726", "accept": false, - "reason": "A broken SigLIP2 doc example is not the same as AutoProcessor failing because a vision backend is missing." + "reason": "Both involve vision processors, but one is a broken docs example and the other is missing-backend handling; not the same bug." }, { "left": "issue:43232", "right": "issue:43673", "accept": false, - "reason": "Both are generation-related, but one is a sync_gpus bookkeeping bug and the other is a cache-missing issue in chunked_prefill; not the same fix." + "reason": "Generation kwargs/update logic and missing GenerationMixin cache in chunked prefill are distinct generation internals." }, { "left": "issue:43986", "right": "issue:46726", "accept": false, - "reason": "Both mention missing vision dependencies, but one is an AutoProcessor crash without torchvision and the other is an unrecognized image processor error; different root causes." + "reason": "Both mention missing vision dependencies, but one is a video-model AutoProcessor/torchvision crash and the other is an unrecognized image-processor error." }, { "left": "issue:41628", "right": "issue:44933", "accept": false, - "reason": "Both are missing-export/import complaints, but they target different symbols and modules, so they are not the same bug." + "reason": "Top-level AutoImageProcessor export failure and a missing import from image_utils are different missing symbols." }, { "left": "issue:46877", "right": "issue:47000", "accept": false, - "reason": "Doc-lint stdout noise at import time is unrelated to RTDetr loss infeasible-cost crashes." + "reason": "Doc-lint stdout noise at import time is unrelated to an RTDetrLoss infeasible-cost-matrix crash." }, { "left": "issue:45310", "right": "issue:45375", "accept": false, - "reason": "Both touch Qwen3.5-MoE, but one is a from_pretrained error and the other is a config field silently dropped; not the same underlying defect." + "reason": "Same model family, but one is a broad from_pretrained failure and the other is a specific missing config field; not enough evidence of one bug." }, { "left": "issue:46463", "right": "issue:46877", "accept": false, - "reason": "T5 decoder-input validation messaging has nothing to do with auto_docstring printing diagnostics." + "reason": "T5 decoder-input validation messaging and auto_docstring stdout diagnostics are unrelated." }, { "left": "issue:45230", "right": "issue:45397", "accept": false, - "reason": "A placeholder bug report is not the same as a gemma-4 ZeRO3 from_pretrained failure." + "reason": "Generic bug report versus a Gemma-4 ZeRO3 loading issue; no overlap in concrete behavior." }, { "left": "issue:44869", "right": "issue:45356", "accept": false, - "reason": "Whisper timestamp decoding on trailing replacement characters and Gemma tokenizer codec handling are different tokenizer bugs." + "reason": "Whisper timestamp decoding and Kimi tokenizer codec/regression handling are different tokenizer bugs." }, { "left": "issue:44373", "right": "issue:46851", "accept": false, - "reason": "A wrong docstring for position_ids is not the same as a runtime failure with variable-length position_ids." + "reason": "A wrong docstring is not the same as a runtime position_ids bug." }, { "left": "issue:44038", "right": "issue:44623", "accept": false, - "reason": "Qwen3-VL-Moe loading issues and processor.save_pretrained missing files are different components and failure modes." + "reason": "Qwen3-VL-Moe loading and processor.save_pretrained missing files are separate issues." }, { "left": "issue:42994", "right": "issue:43525", "accept": false, - "reason": "Quantized saving failure and missing pad_token_id on Llama4Config are unrelated." + "reason": "Quantized saving and a missing pad_token_id on Llama4Config are unrelated." }, { "left": "issue:43723", "right": "issue:46326", "accept": false, - "reason": "Tokenizer loading in v5 and continue_final_message on mistral_common tokenizers are different tokenizer code paths." + "reason": "AutoTokenizer loading in v5 and continue_final_message for mistral_common tokenizers are different tokenizer paths." }, { "left": "issue:42371", "right": "issue:45092", "accept": false, - "reason": "TF32 settings deprecation is unrelated to old InternVL2 checkpoints breaking under meta initialization." + "reason": "TF32 API guidance and InternVL2 meta-initialization incompatibility are unrelated." }, { "left": "issue:43873", "right": "issue:44945", "accept": false, - "reason": "Quantization offloading behavior and pipeline-parallel output correctness are different runtime bugs." + "reason": "Quantization offloading and pipeline-parallel output corruption are different execution bugs." }, { "left": "issue:43784", "right": "issue:46399", "accept": false, - "reason": "Importing sentence-transformers with a missing nn symbol and Qwen3.5 kernel registration errors are unrelated." + "reason": "sentence-transformers import failure and Qwen3.5 kernel-registration errors are unrelated." }, { "left": "issue:45406", "right": "issue:46202", "accept": false, - "reason": "Gemma4 processor missing _tokenizer and Gemma pipeline roundtrip on \u2581 are Gemma-related but not the same underlying bug." + "reason": "Gemma4Processor missing _tokenizer and Gemma pipeline roundtrip on '\u2581' are different bugs." }, { - "left": "issue:44945", - "right": "issue:45753", + "left": "issue:45137", + "right": "issue:47000", "accept": false, - "reason": "Pipeline-parallel output corruption and Qwen3_5 looping on a specific image are different classes of model inference bugs." + "reason": "DeepSpeed ZeRO3 deque underflow and RTDetrLoss infeasible-cost-matrix crash are unrelated." }, { - "left": "issue:45137", - "right": "issue:47000", + "left": "issue:44945", + "right": "issue:45753", "accept": false, - "reason": "ZeRO3 deque underflow and RTDetr cost-matrix infeasibility are unrelated training/runtime failures." + "reason": "Pipeline-parallel output corruption is not the same as an image-triggered infinite loop." }, { "left": "issue:41628", "right": "issue:44351", "accept": false, - "reason": "Both are import-regression reports, but they concern different exported symbols and different subsystems." + "reason": "Both are import/export regressions, but they involve different missing APIs and different fixes." }, { "left": "issue:36010", "right": "issue:41628", "accept": false, - "reason": "Top-level import errors, but one is dependency gating for AutoModel/PyTorch and the other is a missing AutoImageProcessor export." + "reason": "Both are import errors, but GenerationMixin import-path breakage and AutoImageProcessor export failure are different symbols/code paths." }, { "left": "issue:44857", "right": "issue:46877", "accept": false, - "reason": "LwDetr float16 AMP crash and auto_docstring stdout diagnostics are unrelated." + "reason": "AMP/CUDA loss crashing and doc-lint stdout noise are unrelated." }, { - "left": "issue:46202", - "right": "issue:46701", + "left": "issue:45081", + "right": "issue:45103", "accept": false, - "reason": "Both are Gemma pipeline issues, but one is tokenizer roundtrip behavior and the other is an image-text-to-text pipeline/output type problem." + "reason": "Mistral regex patch backend-tokenizer access and auto_docstring annotation handling are unrelated AttributeErrors." }, { - "left": "issue:45081", - "right": "issue:45103", + "left": "issue:46202", + "right": "issue:46701", "accept": false, - "reason": "Mistral regex patch crashing on tokenizer internals and auto_docstring failing with future annotations are unrelated." + "reason": "Gemma tokenizer roundtrip failure and DiffusionGemma image-text pipeline failure are different bugs." }, { "left": "issue:45593", "right": "issue:46877", "accept": false, - "reason": "Missing auxiliary losses in D-FINE and docstring lint output at import time are not the same bug." + "reason": "D-FINE auxiliary-loss wiring and auto_docstring logging are unrelated." }, { "left": "issue:44038", "right": "issue:45310", "accept": false, - "reason": "Both mention Qwen3 MoE models, but they are distinct loading failures on different model variants and are not clearly the same concrete fix." - }, - { - "left": "issue:42994", - "right": "issue:43824", - "accept": false, - "reason": "Quantized model saving and missing Qwen2_5_VLForConditionalGeneration export are unrelated." - }, - { - "left": "issue:44222", - "right": "issue:45310", - "accept": false, - "reason": "FP8 save_pretrained for MoE and Qwen3.5 MoE from_pretrained errors are separate save-vs-load issues." + "reason": "Related model family, but Qwen3-VL-Moe loading and Qwen3.5 MoE from_pretrained errors are not the same concrete regression." } ] }, @@ -14462,6 +14401,7 @@ "issue:42673", "issue:42831", "issue:42898", + "issue:42994", "issue:43232", "issue:43295", "issue:43479", @@ -14485,7 +14425,6 @@ "issue:45103", "issue:45200", "issue:45310", - "issue:45356", "issue:45405", "issue:45561", "issue:45600", @@ -14500,6 +14439,8 @@ "issue:46909" ], "soft_pairs": [ + "issue:42994|issue:43824", + "issue:44222|issue:45310", "issue:42673|issue:43704", "issue:43232|issue:44792", "issue:42831|issue:44222", @@ -14529,233 +14470,235 @@ "issue:30333|issue:34689", "issue:42898|issue:43295", "issue:44734|issue:45103", - "issue:42222|issue:45200", - "issue:42898|issue:45356" + "issue:42222|issue:45200" ], - "prepared_review_unit_hash": "7d135d37e0ec8ae3a2ade4c3ac894f55b544bbd51bf3c97ab623bbb556641c3d", + "prepared_review_unit_hash": "6abad843d8304235e82a1b679f017968d80e717f0c023fc60ac760e6f44cfbca", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 37103, - "estimated_input_tokens": 9276, - "estimated_eval_tokens": 18808 + "soft_pair_count": 32, + "serialized_chars": 37305, + "estimated_input_tokens": 9327, + "estimated_eval_tokens": 18910 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly noisy: almost all pairs are unrelated and only share broad terminology. The one clear duplicate is the exact-title Qwen3ForCausalLM VRAM leak report pair, where issue:43704 is the duplicate of issue:42673.", - "confidence": 0.88, - "canonical_issue_reason": "issue:42673 is the best canonical issue because it is the earlier report with the exact same title and concrete VRAM-leak symptom as issue:43704, which appears to be a duplicate.", + "summary": "This set is mostly a grab bag of unrelated closed issues across different bugs, regressions, docs, and model-specific failures. None of the soft-similarity pairs are close enough to be treated as the same underlying issue or change.", + "confidence": 0.92, + "canonical_issue_reason": "Issue 42898 is the clearest self-contained report: it has a specific regression boundary (v5), an explicit behavior change, and a focused symptom around tokenization cleanup. If a single representative issue must be chosen, this one is the most canonical-style bug report in the set.", "canonical_pr_reason": null, - "best_issue_reason": "issue:42673 is the strongest issue representative for the cluster because it captures the same Qwen3ForCausalLM multi-dataloader-thread VRAM leak and has the clearest duplicate relationship in the packet.", + "best_issue_reason": "Issue 42898 is the strongest representative issue overall because it is concrete, reproducible, and narrowly scoped to one API behavior change. Several others are also well-written, but this one best fits a canonical regression report.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:42994", + "right": "issue:43824", + "accept": false, + "reason": "Different problem classes: quantized model saving failure vs missing Qwen2.5-VL import. No shared code-path." + }, + { + "left": "issue:44222", + "right": "issue:45310", + "accept": false, + "reason": "Both mention FP8/MoE/Qwen3.5, but one is a save-pretrained issue and the other is a from_pretrained load regression. Not the same bug." + }, { "left": "issue:42673", "right": "issue:43704", - "accept": true, - "reason": "Exact same title and same concrete bug: Qwen3ForCausalLM leaking VRAM with multiple dataloader threads." + "accept": false, + "reason": "These are the same title about VRAM leaks, but the second appears to be a duplicate report, not evidence of a mergeable distinct same-change pair; still, as soft-duplicate triage they look like the same underlying bug and would normally be rejected as separate items." }, { "left": "issue:43232", "right": "issue:44792", "accept": false, - "reason": "Different failures: generation kwargs sync_gpus behavior versus a janus image-generation test failure." + "reason": "Generation kwargs update after sync_gpus is unrelated to a janus image generation test failure. Different code paths." }, { "left": "issue:42831", "right": "issue:44222", "accept": false, - "reason": "Both mention FP8, but one is an accuracy issue and the other is save_pretrained for MoE; not the same bug." + "reason": "FineGrainedFP8 accuracy regression is distinct from FP8 save_pretrained/moe serialization. Same theme, different failure mode." }, { "left": "issue:42175", "right": "issue:44933", "accept": false, - "reason": "Unrelated: backend dependency installation versus a missing import from image_utils." + "reason": "TensorFlow backend packaging issue vs a nonexistent image_utils import. No overlap beyond packaging/import context." }, { "left": "issue:43504", "right": "issue:45561", "accept": false, - "reason": "Different areas and symptoms: pretrained preset loading versus a pytest-xdist file race in testing utils." + "reason": "Beit pretrained preset loading bug is unrelated to the pytest-xdist captured_info.txt race. Different subsystem and symptom." }, { "left": "issue:45561", "right": "issue:45600", "accept": false, - "reason": "A test-file race and a stale auto_mappings config reference are unrelated." + "reason": "Both touch testing infrastructure, but one is a file race in patched testing utils and the other is auto_mappings referencing removed configs. Not the same bug." }, { "left": "issue:44496", "right": "issue:45561", "accept": false, - "reason": "Model config loading error versus a test utility race; not the same underlying problem." + "reason": "Unrecognized model/config loading failure is unrelated to a test-worker file race. Different paths and fixes." }, { "left": "issue:45405", "right": "issue:46877", "accept": false, - "reason": "PEFT version pinning is unrelated to auto_docstring printing diagnostics at import time." + "reason": "PEFT version pin issue is a dependency/versioning problem; auto_docstring stdout diagnostics is import-time logging noise. Not the same change." }, { "left": "issue:43976", "right": "issue:44933", "accept": false, - "reason": "Python version compatibility issue versus missing import; no shared code-path bug." + "reason": "Python version compatibility regression is unrelated to a missing import from image_utils. No shared code path." }, { "left": "issue:30333", "right": "issue:33290", "accept": false, - "reason": "MLflow reporting on failed jobs is unrelated to an optimizer OOM in deepspeed." + "reason": "MLflow reporting on training failure and Adafactor OOM in DeepSpeed are different failure modes with different fixes." }, { "left": "issue:30990", "right": "issue:35141", "accept": false, - "reason": "Loading hang in sentence-transformers versus embedding resize/post_init reinitialization are different bugs." + "reason": "Sentence-transformers loading stall is unrelated to embedding resize/post_init reinitialization. Different subsystem." }, { "left": "issue:42898", "right": "issue:44568", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different behaviors and code paths." + "reason": "Tokenization cleanup behavior change and add_special_tokens/BOS-EOS regression both concern tokenizers, but they affect different APIs and behaviors." }, { "left": "issue:42175", "right": "issue:45712", "accept": false, - "reason": "Different packaging/check-repo problems; one is missing TensorFlow backend, the other dummy PT leftovers." + "reason": "Backend dependency omission is unrelated to leftover dummy classes and dir(transformers) leakage. Different maintenance areas." }, { "left": "issue:43479", "right": "issue:46829", "accept": false, - "reason": "Multimodal config initialization versus incorrect hidden_states return value are unrelated." + "reason": "Phi4Multimodal config initialization bug is unrelated to VideoPrism output field misassignment. Different model families and code paths." }, { "left": "issue:46077", "right": "issue:46123", "accept": false, - "reason": "Attention-mask construction and partial-batch pipeline completion are different issues." + "reason": "Gemma4 attention-mask construction and mask generation pipeline batch handling are separate generation bugs with different symptoms." }, { "left": "issue:44623", "right": "issue:45310", "accept": false, - "reason": "Missing files in processor.save_pretrained and a Qwen3.5 MoE from_pretrained error are separate failures." + "reason": "Processor.save_pretrained missing files is a serialization bug, while Qwen3.5 MoE from_pretrained error is a loading/config issue. Not the same bug." }, { "left": "issue:42673", "right": "issue:46693", "accept": false, - "reason": "Both touch generation/performance, but a VRAM leak in threads is not the same bug as FA2 long-context regression." + "reason": "Both involve VRAM/performance around FA2-like workloads, but one is a thread-related leak in Qwen3ForCausalLM and the other is a long-context performance regression after 5.2.0. Too broad to merge." }, { "left": "issue:43479", "right": "issue:44869", "accept": false, - "reason": "Multimodal config defaults versus Whisper timestamp decoding crash are unrelated." + "reason": "Phi4Multimodal config defaults are unrelated to Whisper timestamp decoding on trailing replacement characters." }, { "left": "issue:42898", "right": "issue:45812", "accept": false, - "reason": "Different tokenizer regressions: clean_up_tokenization_spaces behavior versus wrong Granite token IDs." + "reason": "Both are tokenizer regressions, but Granite token IDs and clean_up_tokenization_spaces are distinct behaviors with different fixes." }, { "left": "issue:43723", "right": "issue:44568", "accept": false, - "reason": "Tokenizer loading regression versus add_special_tokens BOS/EOS regression; not the same concrete defect." + "reason": "AutoTokenizer.from_pretrained loading issue in v5 is not the same as add_special_tokens not inserting BOS/EOS for mdeberta-v3-base." }, { "left": "issue:44016", "right": "issue:46231", "accept": false, - "reason": "Notebook syntax error and doc parameter-name mismatch are unrelated documentation problems." + "reason": "Notebook syntax error and documentation parameter-name mismatch are unrelated doc issues." }, { "left": "issue:39692", "right": "issue:46909", "accept": false, - "reason": "Two broken docs examples, but for different models and failures; not a single duplicate issue." + "reason": "SigLIP2 doc example errors and DIA doc example issues are different examples with different failure modes." }, { "left": "issue:43479", "right": "issue:44079", "accept": false, - "reason": "Config defaulting bug versus ModelOutput key assignment bug are separate code paths." + "reason": "Phi4Multimodal config initialization and ModelOutput key assignment are unrelated framework behaviors." }, { "left": "issue:42898", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer regressions, but they concern different models and likely different fixes." + "reason": "Clean-up-tokenization behavior regression and DeepSeek tokenizer incorrect results are separate tokenizer bugs, not one underlying change." }, { "left": "issue:43784", "right": "issue:43824", "accept": false, - "reason": "Sentence-transformers import NameError versus missing Qwen2.5-VL export are unrelated import failures." + "reason": "Sentence-transformers import NameError and Qwen2.5-VL import error are different missing-symbol/import problems." }, { "left": "issue:27957", "right": "issue:30333", "accept": false, - "reason": "Flash Attention 2 support for XLMRoberta is unrelated to MLflow reporting on failed jobs." + "reason": "Flash Attention 2 support for XLM-Roberta is unrelated to MLflow not reporting failed jobs." }, { "left": "issue:30333", "right": "issue:34689", "accept": false, - "reason": "Failed-job tracking versus Llama 3.2 Vision model loading regression are different bugs." + "reason": "MLflow job-status reporting and Llama 3.2 vision model loading regression are different bugs." }, { "left": "issue:42898", "right": "issue:43295", "accept": false, - "reason": "Tokenizer cleanup behavior and custom processor.tokenizer access regression are not the same issue." + "reason": "Tokenization cleanup behavior change and processor.tokenizer/image-passing regression are separate v5 issues with different APIs." }, { "left": "issue:44734", "right": "issue:45103", "accept": false, - "reason": "KV-cache indexing crash in serve is unrelated to auto_docstring failing under future annotations." + "reason": "KV-cache continuation crash in serve is unrelated to auto_docstring crashing on future annotations." }, { "left": "issue:42222", "right": "issue:45200", "accept": false, - "reason": "Broken ViTPose models versus Gemma4 mm_token_type_ids defaulting are different model-specific problems." - }, - { - "left": "issue:42898", - "right": "issue:45356", - "accept": false, - "reason": "Both are tokenizer regressions in nearby versions, but they affect different models and failure modes." + "reason": "ViTPose breakage and Gemma4 mm_token_type_ids defaulting are unrelated model-specific bugs." } ] }, - "evaluator_result": { - "accept": false, - "feedback": "The core duplicate call on issue:42673 vs issue:43704 is supported because the titles match exactly, but the analyst overstates the evidence by calling issue:42673 the \"earlier open report\" and the \"original tracker\". In the packet, issue:42673 is closed, not open, and there is no explicit evidence that it is the original tracker rather than just the earlier same-title report. The rest of the soft-edge rejects are conservative, but the canonical/best-issue rationale is not fully grounded." - }, - "evaluator_used": true, - "retried": true, - "accepted_nontrivial_soft_edge": true, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -14780,6 +14723,7 @@ "issue:42757", "issue:42796", "issue:42831", + "issue:42898", "issue:43116", "issue:43122", "issue:43232", @@ -14791,7 +14735,6 @@ "issue:44075", "issue:44190", "issue:44222", - "issue:44263", "issue:44295", "issue:44297", "issue:44360", @@ -14807,10 +14750,10 @@ "issue:45125", "issue:45245", "issue:45310", + "issue:45356", "issue:45405", "issue:45656", "issue:45712", - "issue:45834", "issue:46082", "issue:46132", "issue:46209", @@ -14822,6 +14765,7 @@ "issue:46932" ], "soft_pairs": [ + "issue:42898|issue:45356", "issue:43874|issue:45092", "issue:44295|issue:45656", "issue:30333|issue:36331", @@ -14850,182 +14794,186 @@ "issue:44222|issue:44484", "issue:43122|issue:46209", "issue:44190|issue:46082", - "issue:43232|issue:46424", - "issue:43232|issue:45834", - "issue:44263|issue:45092" + "issue:43232|issue:46424" ], - "prepared_review_unit_hash": "a0f3987a142eb9a390a1119c287d8c4551a09d33075af78be254cbc38c14abe6", + "prepared_review_unit_hash": "7b90f4c9f99d4d7e278ece44d4b648b12c4fcda308437ddc7e484eaa07cbad9f", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 36674, - "estimated_input_tokens": 9169, - "estimated_eval_tokens": 18594 + "soft_pair_count": 30, + "serialized_chars": 36462, + "estimated_input_tokens": 9116, + "estimated_eval_tokens": 18488 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is over-broad: the soft-paired issues span unrelated modules, models, and bug classes, so none of the candidate edges should be merged as duplicates.", - "confidence": 0.98, - "canonical_issue_reason": "No single canonical issue stands out because the items are not a duplicate set; they describe distinct bugs across different subsystems.", + "summary": "The cluster is broadly heterogeneous; the high-similarity pairs mostly share only surface wording (tokenizer, save_pretrained, import, config, or regression language) and do not look like the same underlying bug or change. I reject all soft edges. If forced to name a representative issue, issue:44222 is the densest concrete save_pretrained-related bug, but the cluster is not a true duplicate set.", + "confidence": 0.73, + "canonical_issue_reason": "issue:44222 is the most central concrete bug report in the set, with the densest save_pretrained-adjacent similarity, but this is only a weak representative because the cluster is heterogeneous.", "canonical_pr_reason": null, - "best_issue_reason": "No global best issue for deduplication here; the cluster appears heterogeneous rather than centered on one underlying defect.", + "best_issue_reason": "issue:44222 is the best single representative if one must be chosen: it is a concrete implementation bug and sits in the most connected part of the similarity graph.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:42898", + "right": "issue:45356", + "accept": false, + "reason": "Both are tokenizer regressions, but one is about clean_up_tokenization_spaces behavior and the other about codec handling/warning text; different bugs." + }, { "left": "issue:43874", "right": "issue:45092", "accept": false, - "reason": "Different problems: one is a missing image-processor method, the other is a multimodal checkpoint/meta-initialization incompatibility." + "reason": "Both concern multimodal/v5 breakage, but one is a missing image-patch API and the other is remote-code/meta-init compatibility; not the same issue." }, { "left": "issue:44295", "right": "issue:45656", "accept": false, - "reason": "Unrelated failures: buffer registration/readback error versus DeepSpeed causing optimizer.step to run twice." + "reason": "Position_ids buffer access and double optimizer stepping with deepspeed are unrelated failures." }, { "left": "issue:30333", "right": "issue:36331", "accept": false, - "reason": "MLflow reporting on failed training and CustomTrainer compute_loss signature mismatch are separate training issues." + "reason": "MLflow reporting on failed jobs and a CustomTrainer compute_loss signature error are different training-path bugs." }, { "left": "issue:22355", "right": "issue:28282", "accept": false, - "reason": "Both are import errors, but for different symbols and causes; they do not indicate the same underlying bug." + "reason": "These are different import errors: missing transformers.onnx versus missing PyTorch dependency for AutoModel." }, { "left": "issue:45405", "right": "issue:45712", "accept": false, - "reason": "One is a dependency version bump concern, the other is leftover dummy classes leaking into the package API." + "reason": "PEFT version pinning and leftover dummy classes in dummy_pt_objects.py are unrelated maintenance issues." }, { "left": "issue:43116", "right": "issue:44190", "accept": false, - "reason": "Different example-script failures: multi-label classification returns empty results vs local dataset loading failure." + "reason": "Different example scripts and different failure modes; one is empty multi-label outputs, the other is local dataset loading." }, { "left": "issue:44222", "right": "issue:44297", "accept": false, - "reason": "Both mention save_pretrained, but one is FP8 MoE serialization and the other is tokenizer config metadata mismatch." + "reason": "Both mention save_pretrained, but one is FP8 MoE serialization and the other is tokenizer_config class mismatch; different code paths." }, { "left": "issue:44075", "right": "issue:46231", "accept": false, - "reason": "Optimizer argument handling and a documentation parameter-name mismatch are not the same bug." + "reason": "Optimizer SGD arguments not being used is unrelated to a documentation parameter-name mismatch." }, { "left": "issue:44060", "right": "issue:45125", "accept": false, - "reason": "Different Qwen model issues: a bad tied-weights warning versus a missing tensor-parallel plan." + "reason": "Same model family only; one is a tied-weights warning, the other is a missing tensor-parallel plan." }, { "left": "issue:44933", "right": "issue:45712", "accept": false, - "reason": "One concerns a nonexistent import from image_utils; the other concerns dummy PT objects leaking into dir(transformers)." + "reason": "Different import/dummy-class problems with different root causes." }, { "left": "issue:42831", "right": "issue:44623", "accept": false, - "reason": "FineGrainedFP8 accuracy regression and processor.save_pretrained missing files are unrelated." + "reason": "FP8 accuracy regression and processor.save_pretrained missing files are distinct bugs." }, { "left": "issue:44448", "right": "issue:46132", "accept": false, - "reason": "Different code paths: Pegasus v4/v5 output regression versus AttentionInterface.register changing callable behavior." + "reason": "Pegasus output differences and AttentionInterface.register behavior changes are different code paths." }, { "left": "issue:42796", "right": "issue:43723", "accept": false, - "reason": "Tokenizer BPE-marker handling and AutoTokenizer.from_pretrained loading failure are separate tokenizer problems." + "reason": "Both are tokenizer-related, but BPE marker handling and AutoTokenizer loading are distinct bugs." }, { "left": "issue:44297", "right": "issue:45656", "accept": false, - "reason": "Tokenizer save metadata mismatch has nothing to do with optimizer stepping twice." + "reason": "Tokenizer serialization mismatch and deepspeed optimizer-step behavior are unrelated." }, { "left": "issue:43479", "right": "issue:43644", "accept": false, - "reason": "Multimodal config default initialization and non-persistent buffer junk values are different defects." + "reason": "Config default initialization for multimodal subconfigs and non-persistent buffer junk are different issues." }, { "left": "issue:45245", "right": "issue:46132", "accept": false, - "reason": "Category-cardinality runtime error and attention registration behavior are unrelated." + "reason": "Category-count runtime error and AttentionInterface.register behavior change are unrelated." }, { "left": "issue:44222", "right": "issue:44623", "accept": false, - "reason": "FP8 MoE save_pretrained serialization and processor file emission are different saving bugs." + "reason": "Both involve save_pretrained, but FP8 MoE saving and processor file emission are different failures." }, { "left": "issue:44360", "right": "issue:46762", "accept": false, - "reason": "A missing ReLU in a DSA indexer and numerical differences in Minimax MSA attention are not the same issue." + "reason": "DSA indexer missing ReLU and Minimax attention numerical drift are not the same bug." }, { "left": "issue:46132", "right": "issue:46710", "accept": false, - "reason": "AttentionInterface registration semantics and a DeepSeek accuracy regression are different underlying problems." + "reason": "AttentionInterface.register behavior changes and a DeepSeek accuracy regression are different code-path problems." }, { "left": "issue:41628", "right": "issue:42757", "accept": false, - "reason": "Import errors, but for different packages and symbols: transformers.AutoImageProcessor vs huggingface_hub.is_offline_mode." + "reason": "Both are import errors, but they arise from different packages and missing symbols." }, { "left": "issue:44373", "right": "issue:44485", "accept": false, - "reason": "A docstring error for position_ids and a GLM-5 RoPE implementation discussion are not duplicates." + "reason": "A wrong docstring and a GLM-5 RoPE implementation issue are unrelated." }, { "left": "issue:44297", "right": "issue:45310", "accept": false, - "reason": "Tokenizer save_pretrained metadata mismatch does not match a Qwen3.5 MoE from_pretrained error." + "reason": "Tokenizer class metadata on save is unrelated to Qwen3.5 Moe from_pretrained failures." }, { "left": "issue:43723", "right": "issue:44977", "accept": false, - "reason": "Generic tokenizer-loading failure and Qwen3.5 flash-attention generation regression are unrelated." + "reason": "Tokenizer loading regression and flash-attention generation failure are different bugs." }, { "left": "issue:45071", "right": "issue:46082", "accept": false, - "reason": "Both involve configs, but one is type checking in v5.4.0 and the other is a head_dim validation rule." + "reason": "PretrainedConfig typing regression and LlamaConfig head_dim validation are distinct config issues." }, { "left": "issue:46614", @@ -15037,37 +14985,25 @@ "left": "issue:44222", "right": "issue:44484", "accept": false, - "reason": "Serialization bug in FP8 MoE save_pretrained is unrelated to the max_shard_size default question." + "reason": "Both mention save_pretrained, but shard-size defaults and FP8 MoE saving are not the same problem." }, { "left": "issue:43122", "right": "issue:46209", "accept": false, - "reason": "Tokenizer behavior change across versions and FP8 inference failure on B200 are distinct issues." + "reason": "Tokenization regression across versions and B200 FP8 inference failure are unrelated." }, { "left": "issue:44190", "right": "issue:46082", "accept": false, - "reason": "A local dataset loading problem and a LlamaConfig head_dim validation error are unrelated." - }, - { - "left": "issue:43232", - "right": "issue:46424", - "accept": false, - "reason": "Generation kwargs update timing and cache max length handling are different code paths." - }, - { - "left": "issue:43232", - "right": "issue:45834", - "accept": false, - "reason": "Generation sync_gpus behavior and Kosmos2.5 long OCR input indexing are unrelated." + "reason": "Local dataset loading in an example script and a LlamaConfig validation bug are different issues." }, { - "left": "issue:44263", - "right": "issue:45092", + "left": "issue:43232", + "right": "issue:46424", "accept": false, - "reason": "A torch.split return-value issue in a DSA indexer does not match a remote-code checkpoint/meta-init incompatibility." + "reason": "Generation sync_gpus handling and cache max_length enforcement are different generation/cache paths." } ] }, @@ -15099,13 +15035,12 @@ "issue:39401", "issue:41628", "issue:42757", - "issue:42796", "issue:42831", "issue:42886", "issue:42907", "issue:42994", "issue:43066", - "issue:43122", + "issue:43232", "issue:43408", "issue:43475", "issue:43644", @@ -15129,6 +15064,7 @@ "issue:45381", "issue:45406", "issue:45753", + "issue:45834", "issue:46002", "issue:46050", "issue:46209", @@ -15141,6 +15077,8 @@ "issue:46909" ], "soft_pairs": [ + "issue:43232|issue:45834", + "issue:44263|issue:45092", "issue:46209|issue:46431", "issue:30990|issue:38175", "issue:44295|issue:44623", @@ -15157,8 +15095,8 @@ "issue:45381|issue:46305", "issue:44483|issue:45406", "issue:44829|issue:46002", - "issue:41628|issue:43824", "issue:42831|issue:44038", + "issue:41628|issue:43824", "issue:43644|issue:46305", "issue:43408|issue:46402", "issue:44261|issue:45092", @@ -15168,225 +15106,230 @@ "issue:42886|issue:42994", "issue:42886|issue:43066", "issue:42757|issue:45278", - "issue:27957|issue:30990", "issue:44188|issue:44373", - "issue:43408|issue:46369", - "issue:42796|issue:43122" + "issue:27957|issue:30990", + "issue:43408|issue:46369" ], - "prepared_review_unit_hash": "019bdba910415f9a056d2c61db3d9f5cff290149278b340b717aa88d7690dd2a", + "prepared_review_unit_hash": "916310409de5ecc25f7c086d2b904c65ff04314ed47030ab2f9ba41a015f292d", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 36512, - "estimated_input_tokens": 9128, - "estimated_eval_tokens": 18512 + "soft_pair_count": 32, + "serialized_chars": 36792, + "estimated_input_tokens": 9198, + "estimated_eval_tokens": 18652 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is not a duplicate cluster: the issues span unrelated import errors, tokenizer regressions, SAM3/Qwen/SmolVLM-specific bugs, quantization/save-load problems, and docs/perf issues. None of the soft pairs look like the same underlying bug or change.", - "confidence": 0.97, - "canonical_issue_reason": "No single issue cleanly represents a duplicate set; the items are heterogeneous and do not share one underlying bug.", + "summary": "This set is mostly a grab-bag of unrelated Transformers issues. None of the soft edges look like true duplicates or the same concrete fix, so all candidate links are rejected.", + "confidence": 0.86, + "canonical_issue_reason": "issue:45278 is the broadest umbrella-style report here: a version-upgrade import regression with multiple missing symbols, so it is the least specific and most plausible anchor if one issue must represent the set.", "canonical_pr_reason": null, - "best_issue_reason": "No best issue candidate: the cluster does not cohere around one fixable problem.", + "best_issue_reason": "It is the most general, user-facing regression report among the items and could plausibly subsume some narrower import-error reports, but it is still not a duplicate of them.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43232", + "right": "issue:45834", + "accept": false, + "reason": "Different bug classes: generation kwargs handling vs OCR long-input indexing." + }, + { + "left": "issue:44263", + "right": "issue:45092", + "accept": false, + "reason": "Unrelated code paths: torch.split/indexer behavior vs remote-code checkpoint meta-init compatibility." + }, { "left": "issue:46209", "right": "issue:46431", "accept": false, - "reason": "FP8 inference failure on B200 and missing chat template are unrelated problems." + "reason": "FP8 inference failure and chat-template absence are unrelated issues." }, { "left": "issue:30990", "right": "issue:38175", "accept": false, - "reason": "Loading hang in Sentence Transformers and zero-probability outputs in SigLIP2 are different bugs." + "reason": "Different symptoms and models: loading stall vs zero-probability outputs." }, { "left": "issue:44295", "right": "issue:44623", "accept": false, - "reason": "Buffer read error for position_ids and missing files on processor save are separate code paths." + "reason": "Buffer registration/read error is not the same as missing files in processor.save_pretrained." }, { "left": "issue:43475", "right": "issue:46305", "accept": false, - "reason": "SAM3 attribute error and MoonshineStreaming off-by-one masking are unrelated." + "reason": "Different models and bugs: missing attribute in SAM3 vs sliding-window off-by-one." }, { "left": "issue:46305", "right": "issue:46655", "accept": false, - "reason": "Both concern sequence/video processing, but they are different models and different defects." + "reason": "Both concern video processing, but the off-by-one mask bug is not the same as SmolVLM padding calculation." }, { "left": "issue:44263", "right": "issue:44373", "accept": false, - "reason": "DSA indexer split behavior and a wrong docstring for position_ids are not the same bug." + "reason": "A code bug in DSA indexing is not the same as a docstring issue." }, { "left": "issue:46369", "right": "issue:46909", "accept": false, - "reason": "SAM3-litetext memory overflow and a broken documentation example for DIA are unrelated." + "reason": "Memory overflow in SAM3-lite-text is unrelated to a broken documentation example." }, { "left": "issue:44263", "right": "issue:44360", "accept": false, - "reason": "Same subsystem, but one is torch.split return values and the other is a missing ReLU; not the same fix." + "reason": "Same subsystem, but torch.split return handling and missing ReLU are different fixes." }, { "left": "issue:42907", "right": "issue:46050", "accept": false, - "reason": "Different quantization/save-load failures for different methods and models." + "reason": "Both touch quantized saving/loading, but they describe different quantization paths and failures." }, { "left": "issue:42831", "right": "issue:45310", "accept": false, - "reason": "FineGrainedFP8 accuracy regression and Qwen3.5 from_pretrained error are unrelated." + "reason": "FineGrainedFP8 accuracy problems and Qwen3.5-MoE from_pretrained errors are different issues." }, { "left": "issue:44279", "right": "issue:46328", "accept": false, - "reason": "Generic dependency issue and a padding performance regression are not duplicates." + "reason": "A dependency problem is unrelated to a SequenceFeatureExtractor.pad performance regression." }, { "left": "issue:22355", "right": "issue:27957", "accept": false, - "reason": "Missing transformers.onnx module and XLMRoberta Flash Attention 2 are unrelated." + "reason": "Importing transformers.onnx and XLMRoberta with Flash Attention 2 are unrelated." }, { "left": "issue:45092", "right": "issue:45357", "accept": false, - "reason": "Both involve v5-era model handling, but one is meta-init incompatibility and the other is incorrect visual encoder keys." + "reason": "Both are multimodal model regressions, but the checkpoint/meta-init issue is not the same as incorrect visual-encoder keys on save_pretrained." }, { "left": "issue:45381", "right": "issue:46305", "accept": false, - "reason": "Qwen2.5-VL vision_position_ids bug and MoonshineStreaming masking bug are different models and different code paths." + "reason": "Different models and code paths: Qwen2.5-VL vision_position_ids vs MoonshineStreaming mask off-by-one." }, { "left": "issue:44483", "right": "issue:45406", "accept": false, - "reason": "Chat/completions request rejection and Gemma4Processor missing _tokenizer are unrelated." + "reason": "Serve API request rejection and Gemma4Processor _tokenizer crash are different failures." }, { "left": "issue:44829", "right": "issue:46002", "accept": false, - "reason": "Degenerate training with flash_attention_3 and NaNs after checkout/init are different failures." + "reason": "Training divergence with flash_attention_3 and NaNs after checkout/init are not the same bug." }, { - "left": "issue:41628", - "right": "issue:43824", + "left": "issue:42831", + "right": "issue:44038", "accept": false, - "reason": "Both are import errors, but they concern different missing symbols and different model families." + "reason": "Generic FP8 accuracy issue does not match a Qwen3-VL-Moe loading bug." }, { - "left": "issue:42831", - "right": "issue:44038", + "left": "issue:41628", + "right": "issue:43824", "accept": false, - "reason": "FP8 accuracy issue and Qwen3-VL-Moe from_pretrained failure are unrelated." + "reason": "Both are import errors, but they involve different missing symbols and likely different causes." }, { "left": "issue:43644", "right": "issue:46305", "accept": false, - "reason": "Non-persistent buffer junk and MoonshineStreaming off-by-one masking are different bugs." + "reason": "Non-persistent buffer junk in v5.0 is unrelated to the MoonshineStreaming window bug." }, { "left": "issue:43408", "right": "issue:46402", "accept": false, - "reason": "SAM3 model-type warning and text-encoder weight loading failure are separate issues." + "reason": "Both involve SAM3, but the model-type warning is not the same as missing text encoder weights." }, { "left": "issue:44261", "right": "issue:45092", "accept": false, - "reason": "MLA layernorm epsilon precision issue and InternVL2 meta-init incompatibility are unrelated." + "reason": "Layernorm epsilon precision mismatch and checkpoint meta-init incompatibility are unrelated." }, { "left": "issue:22355", "right": "issue:28218", "accept": false, - "reason": "Module import failure and tokenizer extra-space behavior are different problems." + "reason": "No module named transformers.onnx is unrelated to tokenizer spacing after added tokens." }, { "left": "issue:42831", "right": "issue:45753", "accept": false, - "reason": "FP8 accuracy regression and Qwen3_5 image loop are unrelated." + "reason": "FP8 accuracy issues are not the same as a Qwen3_5 infinite-loop bug on a specific image." }, { "left": "issue:39401", "right": "issue:44568", "accept": false, - "reason": "Wrong offset_mapping and missing BOS/EOS with add_special_tokens are different tokenizer bugs." + "reason": "Different tokenizer bugs on different models: offset mapping vs BOS/EOS not added." }, { "left": "issue:42886", "right": "issue:42994", "accept": false, - "reason": "Offline tokenizer cache loading and quantized model saving are unrelated." + "reason": "Offline cache loading failure is unrelated to quantized model save failures." }, { "left": "issue:42886", "right": "issue:43066", "accept": false, - "reason": "Both involve tokenizers, but one is offline cache loading and the other is a decoder-type regression." + "reason": "Cache/offline loading and decoder-type serialization are different tokenizer problems." }, { "left": "issue:42757", "right": "issue:45278", "accept": false, - "reason": "A missing huggingface_hub symbol and broad import errors after upgrade are not the same concrete defect." - }, - { - "left": "issue:27957", - "right": "issue:30990", - "accept": false, - "reason": "Flash Attention 2 behavior for XLMRoberta and Sentence Transformers loading hang are unrelated." + "reason": "Missing huggingface_hub import is a dependency-layer issue, not the same as Transformers import regressions after upgrade." }, { "left": "issue:44188", "right": "issue:44373", "accept": false, - "reason": "torch.compile attention-kernel divergence and a wrong docstring are unrelated." + "reason": "torch.compile attention-kernel divergence is unrelated to a docstring correction." }, { - "left": "issue:43408", - "right": "issue:46369", + "left": "issue:27957", + "right": "issue:30990", "accept": false, - "reason": "Same SAM3 family, but one is a type warning and the other is a memory overflow in a different model." + "reason": "Flash Attention 2 on XLMRoberta and Sentence Transformers loading stall are unrelated." }, { - "left": "issue:42796", - "right": "issue:43122", + "left": "issue:43408", + "right": "issue:46369", "accept": false, - "reason": "Specific Ministral BPE marker handling and a broad tokenizer difference across versions are not the same bug." + "reason": "Both mention SAM3, but a config/model-type mismatch is not the same as memory overflow in sam3-litetext." } ] }, @@ -15417,6 +15360,7 @@ "issue:39401", "issue:41628", "issue:42371", + "issue:42796", "issue:42907", "issue:43122", "issue:43232", @@ -15455,11 +15399,11 @@ "issue:46002", "issue:46172", "issue:46326", - "issue:46614", "issue:46762", "issue:46909" ], "soft_pairs": [ + "issue:42796|issue:43122", "issue:43232|issue:45593", "issue:45381|issue:46909", "issue:43540|issue:46909", @@ -15480,8 +15424,8 @@ "issue:46172|issue:46762", "issue:39401|issue:44521", "issue:44625|issue:44829", - "issue:43408|issue:43475", "issue:44704|issue:45290", + "issue:43408|issue:43475", "issue:43122|issue:44521", "issue:44360|issue:44485", "issue:43122|issue:46326", @@ -15489,223 +15433,222 @@ "issue:44038|issue:44297", "issue:22355|issue:29159", "issue:44038|issue:44295", - "issue:45216|issue:45478", - "issue:39401|issue:46614" + "issue:45216|issue:45478" ], - "prepared_review_unit_hash": "81758b88ace2de5868b52e3a85310454c19d62b879ea2418111342ed214e4143", + "prepared_review_unit_hash": "5a94b7d6f128609db70b27d0a115a2b208a307cb672e127d4d1cf3836439163e", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 36578, - "estimated_input_tokens": 9145, - "estimated_eval_tokens": 18546 + "serialized_chars": 36520, + "estimated_input_tokens": 9130, + "estimated_eval_tokens": 18516 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a loose lexical match set: tokenizer/import/regression/model-family issues, but the concrete failure modes differ. I don\u2019t see any pair that clearly represents the same underlying bug or change.", - "confidence": 0.37, - "canonical_issue_reason": "issue:45278 is the broadest, most representative regression report here ('many import errors after update'), so it is the best umbrella issue for this heterogeneous cluster.", + "summary": "These issues are mostly unrelated standalone bugs/regressions. A few are in adjacent areas (tokenizers, imports, Qwen/SAM families), but none of the soft pairs look like the same concrete underlying defect.", + "confidence": 0.91, + "canonical_issue_reason": "Issue 45278 is the broadest representative regression report in the set (many import errors after upgrading), so it is the best anchor if one issue must serve as the cluster representative.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the most general and cluster-like issue title, and it is closest to the few import/regression-themed items; the rest are mostly narrow, unrelated model-specific bugs.", + "best_issue_reason": "Issue 45278 is the most general, user-facing catch-all among these reports and the likeliest hub for related import-breakage complaints.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:42796", + "right": "issue:43122", + "accept": false, + "reason": "Both are tokenizer-related, but one is a model-specific BPE-marker problem and the other is a version-wide tokenization regression." + }, { "left": "issue:43232", "right": "issue:45593", "accept": false, - "reason": "Different subsystems and failures: generation kwargs after sync_gpus vs auxiliary loss handling in D-FINE." + "reason": "Different subsystems and failure modes: generation kwargs after sync_gpus versus missing auxiliary losses in D-FINE." }, { "left": "issue:45381", "right": "issue:46909", "accept": false, - "reason": "Qwen2.5-VL video position IDs bug vs a broken DIA documentation example; unrelated." + "reason": "Qwen2.5-VL video position IDs and a broken DIA docs example are unrelated." }, { "left": "issue:43540", "right": "issue:46909", "accept": false, - "reason": "Qwen3OmniMoe video-input ValueError vs documentation example; different bug." + "reason": "A video-input processing ValueError and a documentation example bug are not the same issue." }, { "left": "issue:43646", "right": "issue:46002", "accept": false, - "reason": "Both are regressions, but custom model initialization breakage and NaNs in classification heads are different concrete problems." + "reason": "Both mention initialization, but one is custom model init breakage and the other is NaNs during classification-head init." }, { "left": "issue:30990", "right": "issue:33357", "accept": false, - "reason": "Both involve loading, but one is Sentence Transformers hanging and the other is a MacOS bus error with a CLIP model; not the same bug." + "reason": "Different failures: SentenceTransformers hanging on load versus a Mac bus error loading a CLIP model." }, { "left": "issue:42907", "right": "issue:44945", "accept": false, - "reason": "Saving dequantized Ministrals/Devstrals and incorrect pipeline-parallel LLM output are unrelated code paths." + "reason": "Saving dequantized Ministrals/Devstrals and incorrect pipeline-parallel LLM output are unrelated." }, { "left": "issue:42907", "right": "issue:44964", "accept": false, - "reason": "Dequantized model saving vs loading Phi-4 multimodal are different failures." + "reason": "A dequantized-save regression and a multimodal model load failure are different bugs." }, { "left": "issue:43976", "right": "issue:44945", "accept": false, - "reason": "Python version support regression vs pipeline-parallel generation output bug; not the same issue." + "reason": "Python-version incompatibility and pipeline-parallel output corruption are unrelated." }, { "left": "issue:44295", "right": "issue:44297", "accept": false, - "reason": "One is about reading a buffered position_ids tensor, the other about tokenizer_class metadata on save; different bugs." + "reason": "One is a model buffer access problem; the other is tokenizer metadata written by save_pretrained." }, { "left": "issue:43756", "right": "issue:44060", "accept": false, - "reason": "Different model implementations and failures: missing RoPE layers vs incorrect tied-weights warning." + "reason": "Different models and different bugs: RoPE layer dropping versus tied-weights warnings." }, { "left": "issue:43576", "right": "issue:43784", "accept": false, - "reason": "CLI env command breakage and a NameError during sentence-transformers import are unrelated." + "reason": "A broken CLI env command and a NameError on importing sentence-transformers are not the same code path." }, { "left": "issue:39401", "right": "issue:44488", "accept": false, - "reason": "Tokenizer offset_mapping bug vs model load failure for cjvt/sleng-bert; not the same underlying problem." + "reason": "Wrong offset_mapping in Qwen3 tokenizer versus failure to load cjvt/sleng-bert are unrelated." }, { "left": "issue:41628", "right": "issue:45278", "accept": false, - "reason": "45278 is a broad import-regression report, but 41628 is one specific missing import; too general to call the same bug." + "reason": "Both are import regressions, but one is a specific missing symbol and the other is a broad upgrade-wide import breakage; not enough evidence of the same code path." }, { "left": "issue:42907", "right": "issue:43937", "accept": false, - "reason": "Model saving issue vs invalid GenerationConfig for GLM-5; unrelated." + "reason": "Different model families and failure modes: save/dequantization versus invalid GenerationConfig." }, { "left": "issue:42371", "right": "issue:44263", "accept": false, - "reason": "TF32 API guidance and a torch.split return-value issue are unrelated." + "reason": "TF32 API guidance and a torch.split/indexer bug are unrelated." }, { "left": "issue:43784", "right": "issue:45600", "accept": false, - "reason": "Import NameError in sentence-transformers vs stale auto_mappings references to removed configs; different code paths." + "reason": "One is an import-time NameError in sentence-transformers; the other is a CI break from removed Sam3LiteText configs." }, { "left": "issue:44117", "right": "issue:44704", "accept": false, - "reason": "Tokenizer mapping None-handling and AutoProcessor kwargs forwarding are different plumbing bugs." + "reason": "Both touch from_pretrained helpers, but they fail in different places: TOKENIZER_MAPPING_NAMES versus cached_file kwargs propagation." }, { "left": "issue:46172", "right": "issue:46762", "accept": false, - "reason": "Different numerical issues in different models; no shared concrete bug." + "reason": "Different model implementations and different numerical discrepancies; not the same bug." }, { "left": "issue:39401", "right": "issue:44521", "accept": false, - "reason": "Qwen3 tokenizer offset mapping and multimodal assistant_mask generation are distinct tokenizer/template bugs." + "reason": "Tokenizer offset mapping and multimodal assistant masks are separate issues." }, { "left": "issue:44625", "right": "issue:44829", "accept": false, - "reason": "num_labels propagation in Qwen3.5 config and flash_attention_3 training degeneration are unrelated." + "reason": "Qwen3.5 num_labels propagation and flash_attention_3 training degeneration are unrelated." }, { - "left": "issue:43408", - "right": "issue:43475", + "left": "issue:44704", + "right": "issue:45290", "accept": false, - "reason": "Same SAM3 area, but one is a model-type warning and the other a missing attribute; different failures." + "reason": "Both involve processing helpers, but AutoProcessor kwargs forwarding and chat-template tool-call crashes are different code paths." }, { - "left": "issue:44704", - "right": "issue:45290", + "left": "issue:43408", + "right": "issue:43475", "accept": false, - "reason": "AutoProcessor cached_file kwargs forwarding vs chat_template crashing on tool-call assistant messages are unrelated." + "reason": "Both are SAM3 video-related, but one is a model-type warning and the other is a missing attribute in vision encoder output." }, { "left": "issue:43122", "right": "issue:44521", "accept": false, - "reason": "Different tokenization regressions: version-to-version tokenization drift vs multimodal assistant mask construction." + "reason": "A tokenizer version regression and all-zero assistant masks in multimodal chat templating are different defects." }, { "left": "issue:44360", "right": "issue:44485", "accept": false, - "reason": "DSA ReLU discussion and GLM-5 RoPE implementation are unrelated model internals." + "reason": "A DSA indexer activation question and a GLM-5 RoPE implementation issue are unrelated." }, { "left": "issue:43122", "right": "issue:46326", "accept": false, - "reason": "Both tokenizer-related, but one is tokenization drift across versions and the other is continue_final_message handling for mistral_common tokenizers." + "reason": "Both involve tokenization, but one is a cross-version regression and the other is continue_final_message behavior for mistral_common tokenizers." }, { "left": "issue:42907", "right": "issue:43452", "accept": false, - "reason": "Model save/dequantization issue vs gguf_file breakage in AutoTokenizer/AutoModel loading; different problems." + "reason": "Dequantized-saving failures and gguf_file loading failures are different pathways." }, { "left": "issue:44038", "right": "issue:44297", "accept": false, - "reason": "Qwen3-VL-Moe regression and tokenizer save metadata mismatch are unrelated." + "reason": "Qwen3-VL-Moe loading issues and tokenizer save_pretrained metadata mismatches are unrelated." }, { "left": "issue:22355", "right": "issue:29159", "accept": false, - "reason": "Missing transformers.onnx module and slow/fast tokenizer inconsistency are different issues." + "reason": "Missing transformers.onnx and slow/fast tokenizer inconsistency are different problems." }, { "left": "issue:44038", "right": "issue:44295", "accept": false, - "reason": "Qwen3-VL-Moe regression vs buffered position_ids error; no shared bug." + "reason": "Qwen3-VL-Moe loading and position_ids buffer access are unrelated." }, { "left": "issue:45216", "right": "issue:45478", "accept": false, - "reason": "Same model family, but save_pretrained checkpoint corruption and from_pretrained loading error are different code paths and not clearly the same bug." - }, - { - "left": "issue:39401", - "right": "issue:46614", - "accept": false, - "reason": "Qwen3 tokenizer offset mapping vs stale phonemizer backend state in Wav2Vec2PhonemeCTCTokenizer are unrelated." + "reason": "Same model family, but one is save_pretrained checkpoint correctness and the other is from_pretrained loading; too different to treat as one duplicate." } ] }, @@ -15733,8 +15676,8 @@ "issue:29127", "issue:36296", "issue:38617", + "issue:39401", "issue:42175", - "issue:42371", "issue:42673", "issue:42757", "issue:42898", @@ -15745,7 +15688,6 @@ "issue:43493", "issue:43611", "issue:43824", - "issue:43874", "issue:43937", "issue:43976", "issue:44038", @@ -15779,6 +15721,7 @@ "issue:46981" ], "soft_pairs": [ + "issue:39401|issue:46614", "issue:43824|issue:44351", "issue:44038|issue:44222", "issue:43116|issue:45200", @@ -15789,15 +15732,15 @@ "issue:22355|issue:29127", "issue:42175|issue:44938", "issue:43493|issue:45753", - "issue:46231|issue:46877", "issue:44188|issue:46328", + "issue:46231|issue:46877", "issue:42898|issue:46614", "issue:38617|issue:45278", "issue:43937|issue:44485", "issue:43976|issue:45446", "issue:43066|issue:44779", - "issue:43611|issue:46002", "issue:42673|issue:45072", + "issue:43611|issue:46002", "issue:44038|issue:45478", "issue:43116|issue:43493", "issue:44355|issue:46231", @@ -15806,211 +15749,210 @@ "issue:44610|issue:46909", "issue:45083|issue:45446", "issue:46726|issue:46981", - "issue:44075|issue:44936", - "issue:42371|issue:43874" + "issue:44075|issue:44936" ], - "prepared_review_unit_hash": "708153a4bbe67f8730fa732a45a52c3b326b48e795cd03fac17726acb915cf4d", + "prepared_review_unit_hash": "88edf5d5cd6860171c000b9dbff6e97fd5975e23387fb428553ad8b49fd446bb", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 29, - "serialized_chars": 35995, - "estimated_input_tokens": 8999, - "estimated_eval_tokens": 18254 + "serialized_chars": 35355, + "estimated_input_tokens": 8839, + "estimated_eval_tokens": 17934 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab-bag of unrelated Transformers issues; the soft pairs are thematically similar in a few cases but not the same underlying bug/change. No PRs are present. If a single issue must anchor the cluster, the broadest umbrella is the version-upgrade import-error report.", - "confidence": 0.81, - "canonical_issue_reason": "Issue 45278 is the broadest umbrella-style report in the set: it aggregates many import failures after upgrading Transformers, making it the closest thing to a cluster sink among these mostly unrelated issues.", + "summary": "The cluster is highly heterogeneous: it mixes unrelated tokenizer, import, training, model, documentation, and environment issues. None of the soft pairs look like the same underlying bug/change, so I would not merge any of them.", + "confidence": 0.93, + "canonical_issue_reason": "No single issue cleanly explains the cluster; issue:45278 is only the broadest umbrella for import-regression reports, but the rest of the items are largely unrelated.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45278 is the best representative only in a loose sense because it is broad and can subsume adjacent import-error reports; however, the overall cluster is too heterogeneous for a strong canonical duplicate.", + "best_issue_reason": "issue:45278 is the least-specific, broadest report in the set, but it is still not a good global canonical because the cluster does not share one root cause.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:39401", + "right": "issue:46614", + "accept": false, + "reason": "Both are tokenizer-related, but one is a Qwen3 offset_mapping regression and the other is a Wav2Vec2 phonemizer backend state bug; different code paths." + }, { "left": "issue:43824", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but for different symbols and likely different missing exports; not the same concrete bug." + "reason": "Both are import errors, but they involve different missing symbols from different modules and likely unrelated fixes." }, { "left": "issue:44038", "right": "issue:44222", "accept": false, - "reason": "Both mention Qwen/MoE, but one is a model bug and the other is FP8 save_pretrained behavior; too different to merge." + "reason": "Same model family only at a very high level; one is a Qwen3-VL-Moe bug and the other is FP8 save_pretrained behavior, so they are not the same change." }, { "left": "issue:43116", "right": "issue:45200", "accept": false, - "reason": "Different subsystems and symptoms: example script classification output vs Gemma 4 multimodal token ids propagation." + "reason": "Different problems: example script multi-label output handling versus Gemma 4 mm_token_type_ids defaults." }, { "left": "issue:43122", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer regressions in v5, but they affect different tokenizers/models and likely different code paths." + "reason": "Both mention tokenizer regressions in v5, but they affect different models and likely different implementation bugs." }, { "left": "issue:42757", "right": "issue:44351", "accept": false, - "reason": "Separate import failures from different packages/modules; same symptom class, not the same bug." + "reason": "Two separate import failures from different packages/modules; no shared underlying code path." }, { "left": "issue:43278", "right": "issue:44945", "accept": false, - "reason": "Training/eval dtype drift is unrelated to pipeline-parallel wrong outputs." + "reason": "A dtype mismatch during evaluation is unrelated to incorrect outputs under pipeline parallelism." }, { "left": "issue:36296", "right": "issue:43116", "accept": false, - "reason": "Tensor parallel training bug and multi-label example-script failure are unrelated." + "reason": "Tensor parallel training bug and example-script classification output bug are different failures." }, { "left": "issue:22355", "right": "issue:29127", "accept": false, - "reason": "Unrelated issues: module import failure vs LayoutLMv3 error-message clarity." + "reason": "One is a missing module import and the other is an error-message clarity issue in LayoutLMv3; not duplicates." }, { "left": "issue:42175", "right": "issue:44938", "accept": false, - "reason": "Different release/build problems: missing TensorFlow backend in extra install vs Python 3.14 load failure." + "reason": "Dependency/backend packaging issue versus a Python 3.14 loading failure; different root causes." }, { "left": "issue:43493", "right": "issue:45753", "accept": false, - "reason": "Different model families and failure modes; not the same underlying change." + "reason": "Different models and symptoms: SigLIP2 HF/JAX discrepancy versus Qwen3_5 infinite-loop behavior." }, { - "left": "issue:46231", - "right": "issue:46877", + "left": "issue:44188", + "right": "issue:46328", "accept": false, - "reason": "Documentation typo vs import-time doc-lint stdout noise; unrelated." + "reason": "One is a torch.compile attention-kernel branching bug; the other is a padding performance issue." }, { - "left": "issue:44188", - "right": "issue:46328", + "left": "issue:46231", + "right": "issue:46877", "accept": false, - "reason": "Attention-kernel divergence under torch.compile is unrelated to a padding performance issue." + "reason": "Documentation parameter mismatch versus import-time diagnostics printing; not the same bug." }, { "left": "issue:42898", "right": "issue:46614", "accept": false, - "reason": "Tokenizer cleanup behavior change and phonemizer backend caching are different tokenizer-stack bugs." + "reason": "Tokenizer cleanup behavior change and stale phonemizer language state are unrelated." }, { "left": "issue:38617", "right": "issue:45278", "accept": false, - "reason": "45278 is a broad umbrella of many import errors, but 38617 is one specific missing symbol; not enough evidence of the same exact bug." + "reason": "General import-regression reports, but the concrete missing symbol and the broader update breakage are not the same underlying fix." }, { "left": "issue:43937", "right": "issue:44485", "accept": false, - "reason": "GenerationConfig validation and GLM-5 RoPE implementation are different issues." + "reason": "Both are GLM-5 related, but one concerns GenerationConfig validation and the other RoPE implementation; distinct code paths." }, { "left": "issue:43976", "right": "issue:45446", "accept": false, - "reason": "Python version compatibility and a PyTorch version check bug are unrelated." + "reason": "Python-version compatibility issue and a PyTorch version check bug are different environment guards." }, { "left": "issue:43066", "right": "issue:44779", "accept": false, - "reason": "Both concern tokenization regressions, but they target different model/tokenizer behaviors and code paths." + "reason": "Both are tokenizer regressions, but for different models and with different failure modes." }, { - "left": "issue:43611", - "right": "issue:46002", + "left": "issue:42673", + "right": "issue:45072", "accept": false, - "reason": "Base-model-prefix loading regression and NaNs in classification heads are unrelated." + "reason": "VRAM leak in dataloader threads is unrelated to bfloat16 dtype mismatch in inference." }, { - "left": "issue:42673", - "right": "issue:45072", + "left": "issue:43611", + "right": "issue:46002", "accept": false, - "reason": "VRAM leak in multithreaded causal LM use is unrelated to dtype mismatches in inference." + "reason": "Model loading via base_model_prefix and NaNs in classification heads are separate issues." }, { "left": "issue:44038", "right": "issue:45478", "accept": false, - "reason": "Both are Qwen3.5/MoE-related, but one is a broad bug report and the other is a specific from_pretrained error; not clearly the same concrete bug." + "reason": "Both touch Qwen3.5/Qwen3-VL-Moe, but one is a 5.0-era bug and the other is a from_pretrained regression in newer versions; not one concrete fix." }, { "left": "issue:43116", "right": "issue:43493", "accept": false, - "reason": "Example-script classification failure and SigLIP2 HF-vs-JAX discrepancy are unrelated." + "reason": "Example-script classification behavior and SigLIP2 implementation discrepancy are unrelated." }, { "left": "issue:44355", "right": "issue:46231", "accept": false, - "reason": "Compiled Python file errors and a docs parameter-name mismatch are unrelated." + "reason": "Compiled Python file loading errors are unrelated to a documentation typo." }, { "left": "issue:45357", "right": "issue:45478", "accept": false, - "reason": "Both involve Qwen3.5 and save/load paths, but one is a save_pretrained key regression while the other is a from_pretrained load error; not enough to treat as one bug." + "reason": "Both involve Qwen3.5 Moe, but one is save_pretrained key handling and the other is from_pretrained loading; different bugs." }, { "left": "issue:44625", "right": "issue:46002", "accept": false, - "reason": "num_labels propagation in Qwen3.5 is unrelated to NaNs in classification heads." + "reason": "Config propagation of num_labels and NaNs on init/checkout are different failures." }, { "left": "issue:44610", "right": "issue:46909", "accept": false, - "reason": "Processor output-size bug and broken docs example are unrelated." + "reason": "Processor input-size mismatch and a broken documentation example are not the same issue." }, { "left": "issue:45083", "right": "issue:45446", "accept": false, - "reason": "Feature-extractor length calculation and a flex_attention version check are unrelated." + "reason": "Helper-function length calculation in qwen3_omni_moe is unrelated to the flex_attention version-check bug." }, { "left": "issue:46726", "right": "issue:46981", "accept": false, - "reason": "Both are AutoProcessor-related, but one is missing-vision-backend handling and the other is HTTP 429 retry behavior; different code paths." + "reason": "Both are AutoProcessor-related, but one is missing-vision-backend handling and the other is retry behavior on HTTP 429; different code paths." }, { "left": "issue:44075", "right": "issue:44936", "accept": false, - "reason": "Unused SGD args and trainer.evaluate failing after train are not the same failure mode." - }, - { - "left": "issue:42371", - "right": "issue:43874", - "accept": false, - "reason": "TF32 API guidance and a missing image-patches method are unrelated." + "reason": "Optimizer argument handling and train-then-evaluate failure are separate trainer/optimizer bugs." } ] }, @@ -16038,6 +15980,7 @@ "issue:33453", "issue:39401", "issue:42222", + "issue:42371", "issue:42673", "issue:42796", "issue:43064", @@ -16049,13 +15992,13 @@ "issue:43824", "issue:43825", "issue:43827", + "issue:43874", "issue:43937", "issue:43992", "issue:44117", "issue:44263", "issue:44291", "issue:44380", - "issue:44410", "issue:44466", "issue:44704", "issue:44779", @@ -16063,7 +16006,6 @@ "issue:44995", "issue:45071", "issue:45092", - "issue:45310", "issue:45405", "issue:45446", "issue:45563", @@ -16083,13 +16025,14 @@ "issue:47000" ], "soft_pairs": [ + "issue:42371|issue:43874", "issue:43618|issue:46829", "issue:43116|issue:43825", "issue:43992|issue:44704", "issue:43066|issue:46710", "issue:46614|issue:46752", - "issue:44995|issue:46697", "issue:28348|issue:46640", + "issue:44995|issue:46697", "issue:42673|issue:42796", "issue:44263|issue:46762", "issue:43827|issue:45446", @@ -16099,8 +16042,8 @@ "issue:43937|issue:45563", "issue:42222|issue:43441", "issue:43824|issue:46981", - "issue:42222|issue:47000", "issue:43673|issue:46328", + "issue:42222|issue:47000", "issue:42673|issue:46655", "issue:45405|issue:45446", "issue:42222|issue:46752", @@ -16109,205 +16052,204 @@ "issue:44380|issue:46077", "issue:43937|issue:44977", "issue:44466|issue:46002", - "issue:39401|issue:44779", - "issue:44410|issue:45310" + "issue:39401|issue:44779" ], - "prepared_review_unit_hash": "39f40cde5a71742978380290cfcfee4d8654a59b551ec681e5b1b66d537f08f0", + "prepared_review_unit_hash": "e40515dcd41bf390f90aef3ba9047115e31fb9e8601fb706ae3953b05bf8ef81", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 28, - "serialized_chars": 35394, - "estimated_input_tokens": 8849, - "estimated_eval_tokens": 17954 + "serialized_chars": 35473, + "estimated_input_tokens": 8869, + "estimated_eval_tokens": 17994 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a heterogeneous set of unrelated bugs/features. The soft pairs share broad themes like tokenizers, flash attention, loading, or generation, but they describe different concrete failure modes or code paths, so none are strong duplicates.", - "confidence": 0.84, - "canonical_issue_reason": "issue:44995 is the clearest self-contained bug report in the set: it has a concrete repro ('second forward pass') and a specific suspected cause (stale indexer cache). The rest of the items are too scattered to serve as a better cluster anchor.", + "summary": "This cluster is very heterogeneous: the soft pairs span unrelated tokenizer, generation, attention, training, processor, and model-export bugs. None of the candidate edges look like true duplicates of the same concrete fix path, so I would not collapse the cluster into a single canonical duplicate.", + "confidence": 0.97, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:44995 is the most actionable standalone issue: narrow scope, explicit failure mode, and an obvious code-path to inspect. If one issue had to be kept as the representative report, this is the strongest candidate.", + "best_issue_reason": "issue:46640 is the broadest umbrella item in the set (Flash Attention 2), but it is still too vague to serve as a true duplicate canonical for this mostly unrelated cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:42371", + "right": "issue:43874", + "accept": false, + "reason": "TF32 control behavior and a missing image-processor method are different bugs in different subsystems." + }, { "left": "issue:43618", "right": "issue:46829", "accept": false, - "reason": "Both are output-field regressions, but they affect different models and different tensors (`attentions` vs `hidden_states`); not the same bug." + "reason": "Both are output-field issues, but they affect different models and different attributes, so they are not the same bug." }, { "left": "issue:43116", "right": "issue:43825", "accept": false, - "reason": "One is a multi-label example script bug, the other is a pipeline task-support message regression; different code paths and symptoms." + "reason": "Example-script multi-label behavior and a pipeline error-message regression are unrelated." }, { "left": "issue:43992", "right": "issue:44704", "accept": false, - "reason": "`from_pretrained` missing a weight vs `AutoProcessor` not forwarding kwargs to `cached_file`; unrelated loading issues." + "reason": "Both involve from_pretrained flows, but one misses model weights and the other forwards kwargs to cached_file; different code paths." }, { "left": "issue:43066", "right": "issue:46710", "accept": false, - "reason": "Tokenizer decoder-type regression vs accuracy regression in a DeepSeek model; different components and failure modes." + "reason": "Tokenizer decoder typing and a model accuracy regression are not the same underlying issue." }, { "left": "issue:46614", "right": "issue:46752", "accept": false, - "reason": "Stale phonemizer language state vs empty-list handling in `apply_chat_template`; no shared underlying bug." + "reason": "A stale phonemizer backend state bug is unrelated to apply_chat_template crashing on an empty list." }, { - "left": "issue:44995", - "right": "issue:46697", + "left": "issue:28348", + "right": "issue:46640", "accept": false, - "reason": "A stale cache crash on second forward vs image over-padding in a processor; unrelated implementation areas." + "reason": "One is a GPT-2 flash-attention support request; the other is only a vague flash-attention issue, not the same concrete change." }, { - "left": "issue:28348", - "right": "issue:46640", + "left": "issue:44995", + "right": "issue:46697", "accept": false, - "reason": "A feature request for GPT-2 flash-attn2 support vs a generic flash-attention bug report; not the same change or defect." + "reason": "Stale indexer cache on second forward pass and image over-padding are unrelated." }, { "left": "issue:42673", "right": "issue:42796", "accept": false, - "reason": "VRAM leak in threaded Qwen3 causal LM use vs tokenizer BPE-marker handling; completely different subsystems." + "reason": "VRAM leakage in threaded loading and tokenizer BPE-marker handling are different bugs." }, { "left": "issue:44263", "right": "issue:46762", "accept": false, - "reason": "`torch.split` return-value handling in one GLM MoE path vs numerical differences in Minimax attention; different concrete problems." + "reason": "A torch.split return-value issue in one indexer is not the same as Minimax attention numerical drift." }, { "left": "issue:43827", "right": "issue:45446", "accept": false, - "reason": "Docs still referencing removed `pipeline()` vs a PyTorch version-check bug in flex attention; unrelated." + "reason": "Docs still referencing pipeline() and a bad PyTorch version check in flex_attention are unrelated." }, { "left": "issue:45092", "right": "issue:46762", "accept": false, - "reason": "Meta-init incompatibility for old InternVL2 checkpoints vs attention numerical drift; different code paths." + "reason": "Meta-initialization incompatibility for old checkpoints is unrelated to attention numerical differences." }, { "left": "issue:33453", "right": "issue:39401", "accept": false, - "reason": "Tokenizer loading regression vs wrong `offset_mapping` for Qwen3; related theme but not the same bug." + "reason": "Both are tokenizer problems, but one is a loading regression and the other is wrong offset mappings for a specific model." }, { "left": "issue:43064", "right": "issue:43673", "accept": false, - "reason": "FSDP2/PEFT optimizer-state mismatch vs missing generation cache during chunked prefill; separate training vs generation failures." + "reason": "Training-time optimizer state corruption and generation cache loss are distinct failures." }, { "left": "issue:43937", "right": "issue:45563", "accept": false, - "reason": "Invalid `GenerationConfig` vs a stale warning for `num_return_sequences`; different symptoms and fixes." + "reason": "Invalid GenerationConfig and a stale warning in paged generate() are different issues." }, { "left": "issue:42222", "right": "issue:43441", "accept": false, - "reason": "VitPose models broken vs Ministral FlashAttention failure; both model-specific breakages, but not the same defect." + "reason": "VitPose model breakage and Ministral flash-attention failure do not share the same bug." }, { "left": "issue:43824", "right": "issue:46981", "accept": false, - "reason": "ImportError for a missing model class vs repeated 429 retries in `AutoProcessor.from_pretrained`; unrelated processor/loading behavior." + "reason": "An import error for a missing class and retry behavior on HTTP 429 are unrelated." }, { - "left": "issue:42222", - "right": "issue:47000", + "left": "issue:43673", + "right": "issue:46328", "accept": false, - "reason": "VitPose breakage vs RTDetr loss infeasible-cost-matrix crash; different model families and code paths." + "reason": "Generation cache handling and an inefficient pad conversion path are different problems." }, { - "left": "issue:43673", - "right": "issue:46328", + "left": "issue:42222", + "right": "issue:47000", "accept": false, - "reason": "Generation cache missing vs `SequenceFeatureExtractor.pad` inefficiency; one is correctness, the other performance." + "reason": "VitPose breakage and RTDetrLoss infeasible cost matrix are unrelated." }, { "left": "issue:42673", "right": "issue:46655", "accept": false, - "reason": "VRAM leak in multi-threaded causal LM use vs incorrect video padding calculation; no shared bug." + "reason": "A VRAM leak in threaded loading and a video-processor padding bug are not the same bug." }, { "left": "issue:45405", "right": "issue:45446", "accept": false, - "reason": "A dependency version bump to unreleased PEFT vs a PyTorch version-check bug; unrelated release/config issues." + "reason": "A dependency version bump to unreleased PEFT and an incorrect version check in flex_attention are unrelated." }, { "left": "issue:42222", "right": "issue:46752", "accept": false, - "reason": "Broken VitPose model loading vs empty-list chat template crash; unrelated failures." + "reason": "VitPose model breakage and apply_chat_template empty-list crash are unrelated." }, { "left": "issue:44291", "right": "issue:45071", "accept": false, - "reason": "Unexpected `_is_hf_initialized` argument during init_empty_weights vs PretrainedConfig type-checking regression; different API breakages." + "reason": "Both are v5 compatibility problems, but one is an unexpected init argument and the other is a PretrainedConfig type-checking break; not the same fix." }, { "left": "issue:44117", "right": "issue:45588", "accept": false, - "reason": "Tokenizer mapping returning None vs flash-attention `s_aux=None` crash; unrelated defensive-check bugs." + "reason": "TOKENIZER_MAPPING_NAMES returning None and a flash_attention AttributeError on sink-less models are unrelated." }, { "left": "issue:44380", "right": "issue:46077", "accept": false, - "reason": "GPT-2 attention scaling ignored under SDPA/FlashAttention vs Gemma4 bidirectional mask construction; different attention implementations." + "reason": "Both touch attention behavior, but one is backend scaling config being ignored and the other is causal-mask construction for a different model." }, { "left": "issue:43937", "right": "issue:44977", "accept": false, - "reason": "GenerationConfig validation vs DeepSeek tokenizer incorrect results; different components and symptoms." + "reason": "Invalid generation config and DeepSeek tokenizer regression are different code paths." }, { "left": "issue:44466", "right": "issue:46002", "accept": false, - "reason": "State serialization inconsistency for `lm_head.weight` vs NaNs in classification heads after checkout/init; no common underlying bug." + "reason": "Serialization of tied lm_head weights and NaNs after checkout/init are unrelated." }, { "left": "issue:39401", "right": "issue:44779", "accept": false, - "reason": "Qwen3 `offset_mapping` bug vs DeepSeek tokenizer incorrect outputs; both tokenizer regressions, but not the same model or failure." - }, - { - "left": "issue:44410", - "right": "issue:45310", - "accept": false, - "reason": "Missing QKV/gate projections in qwen3next layer 0 vs Qwen3.5 MoE `from_pretrained` error; distinct model issues." + "reason": "Both are tokenizer correctness bugs, but they affect different models and failure modes." } ] }, @@ -16380,6 +16322,7 @@ "issue:46981" ], "soft_pairs": [ + "issue:44410|issue:45310", "issue:43450|issue:43986", "issue:45071|issue:46512", "issue:45341|issue:46328", @@ -16404,200 +16347,206 @@ "issue:42371|issue:46617", "issue:44075|issue:45656", "issue:38617|issue:44220", - "issue:43066|issue:43122", "issue:43208|issue:46894", + "issue:43066|issue:43122", "issue:44263|issue:44485" ], - "prepared_review_unit_hash": "73b8dad0a240badd8f7ea4da1646d8e30e4843806c459dc1c9b843873ff1d734", + "prepared_review_unit_hash": "29ad13aee0b056c59646c7d25edd2efe17524df10026f11db3558b863f554e10", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 27, - "serialized_chars": 34863, - "estimated_input_tokens": 8716, - "estimated_eval_tokens": 17688 + "soft_pair_count": 28, + "serialized_chars": 35151, + "estimated_input_tokens": 8788, + "estimated_eval_tokens": 17832 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is highly heterogeneous: most items are superficially similar only at the subsystem level (imports, tokenizers, model loading, configs, optimizers, training, video processors) but do not look like the same underlying bug. No soft-edge duplicate pairs are strong enough to merge.", - "confidence": 0.59, - "canonical_issue_reason": "issue:45278 is the broadest umbrella report in the set, covering multiple import failures after a version update and best matches the cluster\u2019s import/API-breakage theme.", + "summary": "This cluster is a loose mix of unrelated Transformers bugs across tokenizers, model loading, config validation, training, testing utilities, and docs. None of the soft pairs look like the same underlying fix, so all are rejected.", + "confidence": 0.97, + "canonical_issue_reason": "No canonical issue: the set is too heterogeneous, with no clear duplicate chain across the items.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the most representative issue overall because it frames a broad regression rather than a single model-specific or file-specific failure.", + "best_issue_reason": "issue:45278 is the broadest user-facing regression in the set (many import errors after upgrading), so it is the closest thing to a representative issue, though it is still not a true duplicate anchor.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44410", + "right": "issue:45310", + "accept": false, + "reason": "Different Qwen3 bugs: missing projections in qwen3next vs a Qwen3.5 MoE from_pretrained failure." + }, { "left": "issue:43450", "right": "issue:43986", "accept": false, - "reason": "Both involve video-related loading/processing, but one is an output-shape bug and the other is a missing-torchvision crash; different failure modes." + "reason": "One is a batched video processor shape bug; the other is an AutoProcessor crash without torchvision." }, { "left": "issue:45071", "right": "issue:46512", "accept": false, - "reason": "Both mention config/dtype behavior, but one is a type-checking regression and the other is stale config.dtype after from_config; not the same bug." + "reason": "Type-checking regression in PreTrainedConfig is unrelated to stale config.dtype after from_config()." }, { "left": "issue:45341", "right": "issue:46328", "accept": false, - "reason": "A generic testing_utils bug and a SequenceFeatureExtractor.pad performance issue are unrelated." + "reason": "testing_utils bug vs SequenceFeatureExtractor.pad performance issue; different code paths and symptoms." }, { "left": "issue:45341", "right": "issue:45561", "accept": false, - "reason": "Both touch testing_utils, but one is unspecified and the other is a pytest-xdist race on captured_info.txt; too different to merge." + "reason": "Both touch testing utils, but one is a small bug and the other is an xdist file race." }, { "left": "issue:45632", "right": "issue:46981", "accept": false, - "reason": "One is a trust_remote_code cache-path collision, the other is repeated HTTP-429 retries in AutoProcessor; different code paths." + "reason": "Cache path collision for trust_remote_code models is unrelated to repeated 429 retries in AutoProcessor." }, { "left": "issue:44829", "right": "issue:45216", "accept": false, - "reason": "Different model families and different problems: flash_attention_3 training degeneration vs Qwen3.5 save_pretrained checkpoint corruption." + "reason": "Flash-attention training degeneration and a Qwen3.5 save_pretrained regression are different problems." }, { "left": "issue:44117", "right": "issue:45081", "accept": false, - "reason": "Both concern tokenizer loading, but one is a mapping-table None handling bug and the other is a Mistral regex patch crash." + "reason": "Tokenizer mapping None handling is unrelated to the Mistral regex patch crash." }, { "left": "issue:42617", "right": "issue:45341", "accept": false, - "reason": "A 3d_parallel.py execution problem is not the same as a testing_utils bug." + "reason": "3d_parallel.py execution failure is unrelated to the testing_utils bug." }, { "left": "issue:43408", "right": "issue:43742", "accept": false, - "reason": "Different models and different failure types: sam3 model-type mismatch warning vs MobileLLM loading KeyError." + "reason": "Model-type warning for sam3 is unrelated to the MobileLLM-125M loading KeyError." }, { "left": "issue:38617", "right": "issue:41628", "accept": false, - "reason": "Both are import errors, but they refer to different missing symbols and likely different breakages." + "reason": "Both are import errors, but they concern different symbols and different underlying causes." }, { "left": "issue:43122", "right": "issue:46298", "accept": false, - "reason": "Different tokenization regression vs a noisy logging message; no common underlying bug." + "reason": "Tokenizer behavior regression is unrelated to a noisy logging message." }, { "left": "issue:45362", "right": "issue:46328", "accept": false, - "reason": "Qwen3.5 chat crash and a padding-performance issue are unrelated." + "reason": "Qwen chat crash and padding inefficiency are different issues in different subsystems." }, { "left": "issue:42222", "right": "issue:46178", "accept": false, - "reason": "VitPose model breakage and an unsupported backend error are separate issues." + "reason": "Vitpose breakage and BACKENDS_MAPPING validation are unrelated backend/model-loading failures." }, { "left": "issue:39401", "right": "issue:46209", "accept": false, - "reason": "Tokenizer offset_mapping bug and FP8 inference failure are unrelated." + "reason": "Qwen3 offset_mapping bug and FP8 inference failure on B200 are unrelated." }, { "left": "issue:44877", "right": "issue:46231", "accept": false, - "reason": "Strict config loading and documentation parameter mismatch are not the same bug." + "reason": "Strict config loading for granite_speech is unrelated to a docs parameter-name mismatch." }, { "left": "issue:44484", "right": "issue:45310", "accept": false, - "reason": "max_shard_size behavior and Qwen3.5 MoE from_pretrained failure are unrelated." + "reason": "save_pretrained shard-size behavior is unrelated to the Qwen3.5 MoE from_pretrained error." }, { "left": "issue:43937", "right": "issue:44496", "accept": false, - "reason": "Both are model-loading/validation errors, but on different models and different validation paths." + "reason": "Invalid GenerationConfig in GLM-5 and unrecognized Olmo-Hybrid model config are distinct failures." }, { "left": "issue:42371", "right": "issue:44261", "accept": false, - "reason": "TF32 API guidance and MLA epsilon precision error are unrelated." + "reason": "TF32 API guidance and MLA layernorm precision error do not share the same bug." }, { "left": "issue:44261", "right": "issue:46762", "accept": false, - "reason": "Both are numerical issues, but they affect different models and different computations." + "reason": "Both are numerical issues, but they affect different model components and likely different fixes." }, { "left": "issue:44933", "right": "issue:45278", "accept": false, - "reason": "Both are import-related, but one is a specific missing import and the other is a broad post-upgrade import-failure report; not clearly the same bug." + "reason": "A specific image_utils import issue is not the same as a broad multi-import regression." }, { "left": "issue:44410", "right": "issue:45478", "accept": false, - "reason": "Different Qwen model variants and different loading/projection problems." + "reason": "Missing qwen3next projections is unrelated to the Qwen3.5 MoE from_pretrained error." }, { "left": "issue:42371", "right": "issue:46617", "accept": false, - "reason": "TF32 settings guidance is unrelated to trust_remote_code cache FileNotFoundError." + "reason": "TF32 behavior settings and trust_remote_code cache FileNotFoundError are unrelated." }, { "left": "issue:44075", "right": "issue:45656", "accept": false, - "reason": "Both involve optimizers, but one is unused SGD args and the other is double optimizer stepping under deepspeed." + "reason": "SGD args ignored and optimizer stepping twice with deepspeed are different optimizer bugs." }, { "left": "issue:38617", "right": "issue:44220", "accept": false, - "reason": "An import error and an audio feature-extraction issue are unrelated." + "reason": "ImportError for layer_type_validation is unrelated to _torch_extract_fbank_features()." }, { - "left": "issue:43066", - "right": "issue:43122", + "left": "issue:43208", + "right": "issue:46894", "accept": false, - "reason": "Related tokenizer regressions, but one is about decoder type and the other is changed tokenization output; not enough evidence of the same concrete bug." + "reason": "xLSTM training blockers and CohereASR double-shift loss bug are different model-specific issues." }, { - "left": "issue:43208", - "right": "issue:46894", + "left": "issue:43066", + "right": "issue:43122", "accept": false, - "reason": "Different model families and distinct training bugs." + "reason": "Tokenizer decoder-type regression and version-to-version tokenization differences are related broadly, but not the same concrete bug." }, { "left": "issue:44263", "right": "issue:44485", "accept": false, - "reason": "A torch.split indexing issue and a RoPE implementation discussion are not the same underlying problem." + "reason": "torch.split() return handling in GlmMoeDsaIndexer is unrelated to GLM-5 RoPE implementation." } ] }, @@ -16704,20 +16653,20 @@ "issue:38617|issue:42757", "issue:43064|issue:44315" ], - "prepared_review_unit_hash": "90ee0ba24e3aee2a26e59a43d6e7dc37b20ac570f1178d9466c0c71b7383bfd7", + "prepared_review_unit_hash": "d8628484731d552c21eb48707477ba52927b86a67303beb06fc78d2b48708e40", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 33, - "serialized_chars": 36860, + "serialized_chars": 36857, "estimated_input_tokens": 9215, "estimated_eval_tokens": 18686 }, @@ -16726,210 +16675,210 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a loose collection of mostly unrelated Transformers bugs and regressions. A few pairs share a broad surface area (imports, model loading, FlashAttention, Qwen/GLM families), but none look like the same underlying issue closely enough to merge; all soft edges should be rejected.", - "confidence": 0.91, - "canonical_issue_reason": "issue:43097 is the broadest, most central regression-style report in the set, with inbound references and a generic API-breaking symptom that could plausibly anchor a noisy cluster better than the more model-specific issues.", + "summary": "This cluster is highly heterogeneous: it mixes unrelated loading, import, FlashAttention, tokenizer, config, and training bugs. All soft duplicate suggestions are false positives.", + "confidence": 0.98, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:43097 is the best representative because it concerns a widely visible v5 compatibility break rather than a narrow model or test-only defect, so it is the most plausible duplicate target among these issues.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43097", "right": "issue:44493", "accept": false, - "reason": "Removal of tie_embeddings_and_encoder_decoder is a config/API compatibility change; unexpected position-id keys is a different model-loading/state-dict issue." + "reason": "Different bugs: API/removal of tie_embeddings vs position-id key warnings." }, { "left": "issue:44077", "right": "issue:46981", "accept": false, - "reason": "patchtsmixer post_init validation and AutoProcessor retry behavior on 429s are unrelated code paths and failures." + "reason": "Unrelated areas: patchtsmixer post_init validation vs AutoProcessor 429 retry behavior." }, { "left": "issue:43673", "right": "issue:45341", "accept": false, - "reason": "GenerationMixin cache missing in chunked_prefill is unrelated to a bug in testing_utils.py." + "reason": "Generation cache/chunked_prefill bug vs a testing_utils bug." }, { "left": "issue:44483", "right": "issue:45464", "accept": false, - "reason": "One is a chat/completions API acceptance issue; the other is a Qwen3.5 streaming inference failure. Different layers and symptoms." + "reason": "Both mention chat/completions, but one is general request rejection and the other is a Qwen3.5 streaming-specific failure." }, { "left": "issue:43931", "right": "issue:44410", "accept": false, - "reason": "Qwen3-VL weight-shape mismatch and qwen3next missing projections are both model-specific, but they are distinct configuration/architecture bugs." + "reason": "Different Qwen3 problems: weight-shape mismatch vs missing attention projections." }, { "left": "issue:41628", "right": "issue:45071", "accept": false, - "reason": "AutoImageProcessor import failure and PretrainedConfig type-checking breakage are unrelated import/API regressions." + "reason": "AutoImageProcessor import failure is unrelated to PretrainedConfig type checking." }, { "left": "issue:45071", "right": "issue:45542", "accept": false, - "reason": "PretrainedConfig typing regression and missing TensorFlow backend with only tensorboard installed are separate issues." + "reason": "Type-checking regression is unrelated to the TensorFlow/tensorboard backend error." }, { "left": "issue:44484", "right": "issue:45478", "accept": false, - "reason": "save_pretrained shard-size behavior is unrelated to Qwen3.5-Moe from_pretrained load errors." + "reason": "max_shard_size question is unrelated to a Qwen3.5-Moe from_pretrained error." }, { "left": "issue:27957", "right": "issue:43352", "accept": false, - "reason": "Both mention Flash Attention 2, but they are different model support questions for different architectures and code paths." + "reason": "Different models and failures: XLM-R flash attention support vs NemotronH FA2 support." }, { "left": "issue:43441", "right": "issue:44829", "accept": false, - "reason": "Ministral-3 FlashAttention failure in v5 RC and FA3 degenerate training in sequence classification are different bugs, not one root cause." + "reason": "Both involve flash attention, but one is model loading and the other is degenerate training with FA3." }, { "left": "issue:43012", "right": "issue:43577", "accept": false, - "reason": "A PyTorch warning during bfloat16 compilation is unrelated to Blip2 dtype propagation staying at float32." + "reason": "Precision warning during compile is not the same as Blip2 submodules staying float32 on load." }, { "left": "issue:44263", "right": "issue:46172", "accept": false, - "reason": "torch.split return values in GlmMoeDsaIndexer and a DeepSeek-v4 tensor transpose/cast issue are distinct implementation bugs." + "reason": "Distinct code paths: torch.split behavior in a DSA indexer vs DeepSeek-V4 comb dtype/transpose." }, { "left": "issue:44360", "right": "issue:46172", "accept": false, - "reason": "A missing ReLU in the DSA indexer is not the same defect as the DeepSeek-v4 comb.to(...).transpose(...) issue." + "reason": "ReLU omission in the DSA indexer is unrelated to DeepSeek-V4 tensor conversion." }, { "left": "issue:43874", "right": "issue:44261", "accept": false, - "reason": "Missing get_number_of_image_patches in GLM46V image processor and missing rms_norm_eps in MLA q_a_layernorm are unrelated model-component bugs." + "reason": "Image-patch count method missing vs rms_norm_eps/config precision issue." }, { "left": "issue:43931", "right": "issue:44038", "accept": false, - "reason": "Both involve Qwen-VL variants, but one is a weight-shape mismatch while the other is a Transformers v5 compatibility bug for Qwen3-VL-Moe." + "reason": "Related model family, but different concrete failures and no shared code-path evidence." }, { "left": "issue:43097", "right": "issue:43452", "accept": false, - "reason": "A removed config helper and a gguf_file/from_pretrained regression are separate load-path problems." + "reason": "Removed tie_embeddings API is unrelated to gguf_file loading breakage." }, { "left": "issue:43986", "right": "issue:45341", "accept": false, - "reason": "Crash without torchvision for video models is unrelated to the testing_utils bug." + "reason": "Video AutoProcessor crash without torchvision is unrelated to testing_utils." }, { "left": "issue:45003", "right": "issue:45341", "accept": false, - "reason": "Unsafe sys.modules access in modeling_utils and a testing_utils defect are different modules and failure modes." + "reason": "sys.modules access bug in modeling_utils is unrelated to testing_utils." }, { "left": "issue:42886", "right": "issue:46981", "accept": false, - "reason": "Offline tokenizer cache loading and repeated 429 retries via HF_ENDPOINT are both network/cache-related, but not the same bug." + "reason": "Both are loading/caching/network-related, but offline cache failure and 429 retry behavior are different bugs." }, { "left": "issue:44261", "right": "issue:44485", "accept": false, - "reason": "Missing rms_norm_eps in MLA layernorm is unrelated to the GLM-5 RoPE implementation discussion." + "reason": "rms_norm_eps precision problem is not the same as GLM-5 RoPE implementation." }, { "left": "issue:45753", "right": "issue:46369", "accept": false, - "reason": "Qwen3_5 looping on a specific image and sam3-litetext memory overflow are different model/runtime failures." + "reason": "Infinite loop on a specific image vs memory overflow in sam3-litetext." }, { "left": "issue:43531", "right": "issue:45341", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior is unrelated to the testing_utils bug." + "reason": "Qwen3-MoE sliding_window issue is unrelated to testing_utils." }, { "left": "issue:44117", "right": "issue:44987", "accept": false, - "reason": "TOKENIZER_MAPPING_NAMES returning None and a physical-intelligence/fast loading failure do not indicate the same underlying defect." + "reason": "Tokenizer mapping None handling is not the same as physical-intelligence/fast loading failure." }, { "left": "issue:43097", "right": "issue:45588", "accept": false, - "reason": "Config removal without indication and a flash_attention.py s_aux=None crash are unrelated regressions." + "reason": "tie_embeddings removal is unrelated to flash_attention.py crashing on s_aux=None." }, { "left": "issue:43673", "right": "issue:46123", "accept": false, - "reason": "Generation cache missing in chunked_prefill and is_last handling in MaskGenerationPipeline are separate generation bugs." + "reason": "Generation cache missing in chunked_prefill is unrelated to partial-batch MaskGenerationPipeline dropping results." }, { "left": "issue:45440", "right": "issue:46172", "accept": false, - "reason": "Both touch DeepSeek-family models, but the native-vs-remote divergence is broader than the specific DeepSeek-v4 tensor operation bug." + "reason": "Different model families and failures: DeepseekV3MoE divergence vs DeepSeek-V4 tensor handling." }, { "left": "issue:43874", "right": "issue:44263", "accept": false, - "reason": "Missing image patch count method and torch.split return values are different GLM46V/GlmMoe implementation problems." + "reason": "ImageProcessor patch counting bug is unrelated to torch.split return handling." }, { "left": "issue:44410", "right": "issue:45375", "accept": false, - "reason": "Missing attn_qkv/attn_gate projections in qwen3next layer 0 and a missing deepstack_visual_indexes config field are separate Qwen3.5 config/model issues." + "reason": "Qwen3next missing projections and Qwen3_5MoeVisionConfig field dropping are different config/model issues." }, { "left": "issue:43986", "right": "issue:45372", "accept": false, - "reason": "AutoProcessor crash without torchvision and a mistral_common ReasoningEffort import error are unrelated dependency issues." + "reason": "Missing torchvision crash is unrelated to mistral_common ReasoningEffort import failure." }, { "left": "issue:43097", "right": "issue:46050", "accept": false, - "reason": "A removed embedding-tying helper and a sinq quantized save/load regression are different model serialization issues." + "reason": "API removal in config handling is unrelated to sinq quantization save/load breakage." }, { "left": "issue:44038", "right": "issue:44410", "accept": false, - "reason": "Qwen3-VL-Moe v5 bug and qwen3next missing projections are related only by model family name, not by the same concrete root cause." + "reason": "Both are Qwen3-family, but they describe different missing-parameter bugs with no shared concrete code path." }, { "left": "issue:38617", "right": "issue:42757", "accept": false, - "reason": "Both are ImportErrors, but they concern different missing symbols from different packages." + "reason": "Separate import errors from different modules: transformers.configuration_utils vs huggingface_hub." }, { "left": "issue:43064", "right": "issue:44315", "accept": false, - "reason": "FSDP2/PEFT optimizer-state corruption under cpu_ram_efficient_loading and Liger kernel not applied with model_init are separate training-path bugs." + "reason": "FSDP2/PEFT optimizer-state corruption is unrelated to Liger Kernel not being applied with model_init." } ] }, @@ -17038,238 +16987,238 @@ "issue:44077|issue:46365", "issue:44222|issue:45478" ], - "prepared_review_unit_hash": "98980799cbf0a89a6b06ced649771f2445566eaa56d3add15cc704021c5b6e82", + "prepared_review_unit_hash": "12dc6ef4826ecc4fc3a1fa92286324f5b163c2e03645fedca2efa5228085569e", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 37628, - "estimated_input_tokens": 9407, - "estimated_eval_tokens": 19070 + "serialized_chars": 37640, + "estimated_input_tokens": 9410, + "estimated_eval_tokens": 19076 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is highly heterogeneous: the soft pairs are mostly topical similarities within Transformers, not true duplicate-level matches. No pull requests are present.", - "confidence": 0.93, - "canonical_issue_reason": "issue:43502 is the most concrete, user-facing runtime bug in the set, with a clear repro around `local_files_only=True` still allowing API requests; it is a strong anchor issue even though the cluster itself is not coherent.", + "summary": "This is a heterogeneous set of mostly unrelated issues; none of the soft pairs look like the same underlying bug or change. I\u2019d treat the open quantization/offloading report as the best canonical issue in the set, but only as a representative, not a duplicate anchor.", + "confidence": 0.63, + "canonical_issue_reason": "issue:43873 is the most substantial and active report here (highest discussion/inbound refs) and is a clear, actionable bug report.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43502 is the best standalone issue here because it is narrowly scoped, reproducible, and points to a specific code path that can be fixed and verified.", + "best_issue_reason": "issue:43873 is the strongest representative issue overall because it has the broadest engagement and a well-scoped problem statement.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43208", "right": "issue:46897", "accept": false, - "reason": "Both mention training/data-shift style failures, but they are different models and different bug shapes; not the same underlying defect." + "reason": "Different models and failures: xLSTM training bugs vs Florence2 training-loss shifting." }, { "left": "issue:43986", "right": "issue:44484", "accept": false, - "reason": "One is an AutoProcessor/torchvision import crash, the other is a `save_pretrained` shard-size question; unrelated." + "reason": "Video-model/AutoProcessor import failure is unrelated to a save_pretrained shard-size question." }, { "left": "issue:46655", "right": "issue:46664", "accept": false, - "reason": "Both are SmolVLM video-processor issues, but one is padding calculation and the other is `num_frames` semantics; distinct fixes." + "reason": "Same component, but padding calculation and num_frames semantics are distinct bugs." }, { "left": "issue:46298", "right": "issue:46752", "accept": false, - "reason": "Generic logging noise versus an `apply_chat_template` empty-list crash; not the same bug." + "reason": "Noisy logging and empty-list chat-template crash are unrelated." }, { "left": "issue:42890", "right": "issue:46298", "accept": false, - "reason": "A flaky model integration test due to missing `set_seed()` is unrelated to a noisy logging message." + "reason": "Missing set_seed in a test is unrelated to a logging message issue." }, { "left": "issue:43097", "right": "issue:43873", "accept": false, - "reason": "Removed config flag documentation/behavior and quantization offloading are different problems, with different code paths." + "reason": "Removed config field guidance and quantization/offloading behavior are different problems." }, { "left": "issue:38617", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but they involve different missing symbols and likely separate API breakages." + "reason": "Both are import errors, but they involve different missing symbols and code paths." }, { "left": "issue:44062", "right": "issue:44117", "accept": false, - "reason": "`AddedToken()` argument collision and tokenizer mapping returning `None` are unrelated tokenizer-layer bugs." + "reason": "Tokenizer AddedToken construction bug is separate from TOKENIZER_MAPPING_NAMES returning None." }, { "left": "issue:43012", "right": "issue:46821", "accept": false, - "reason": "A PyTorch precision warning when compiling bf16 models is unrelated to a TimesFM `window_size` attribute error." + "reason": "bfloat16 compile warning is unrelated to a TimesFM window_size AttributeError." }, { "left": "issue:44117", "right": "issue:46701", "accept": false, - "reason": "A tokenizer mapping issue and a diffusion pipeline output issue are different subsystems and failure modes." + "reason": "Tokenizer mapping fallback bug does not match the pipeline failure described here." }, { "left": "issue:46172", "right": "issue:46710", "accept": false, - "reason": "One is a tensor transpose/dtype bug in deepseek-v4; the other is an accuracy regression after upgrade, not the same concrete defect." + "reason": "A DeepSeek tensor operation bug is not the same as an accuracy regression after upgrade." }, { "left": "issue:43502", "right": "issue:43992", "accept": false, - "reason": "`local_files_only` network leakage and a missing `embed_tokens.weight` in `UMT5Encoder.from_pretrained` are different load-path failures." + "reason": "Both involve loading, but one is network access control and the other is missing encoder weights." }, { "left": "issue:44936", "right": "issue:46365", "accept": false, - "reason": "Trainer evaluation failure after training is not the same as an import conflict caused by `datasets`." + "reason": "Trainer evaluate failure after train is unrelated to importing Trainer with datasets installed." }, { "left": "issue:44704", "right": "issue:45103", "accept": false, - "reason": "Passing kwargs to `cached_file` and the auto-docstring crash under future annotations are unrelated code paths." + "reason": "AutoProcessor kwargs forwarding and auto_docstring annotation parsing are unrelated." }, { "left": "issue:44380", "right": "issue:46421", "accept": false, - "reason": "Both touch backend/performance behavior, but one is attention scaling being ignored and the other is cache lazy initialization recompiling; different bugs." + "reason": "GPT2 attention scaling under SDPA/FlashAttention is different from cache lazy-init recompilation." }, { "left": "issue:44938", "right": "issue:46178", "accept": false, - "reason": "Python 3.14 import failure and a backend mapping error for `tensorflow_text` are unrelated compatibility issues." + "reason": "Python 3.14 import failure is not the same as an invalid BACKENDS_MAPPING backend." }, { "left": "issue:43986", "right": "issue:44016", "accept": false, - "reason": "AutoProcessor/torchvision import handling is unrelated to a notebook syntax error." + "reason": "Video-model import handling and a notebook syntax error are unrelated." }, { "left": "issue:42907", "right": "issue:45753", "accept": false, - "reason": "Saving dequantized Ministrals/Devstrals and a Qwen3_5 image infinite loop are different model-specific bugs." + "reason": "Saving dequantized models and an infinite-loop image bug are different failures." }, { "left": "issue:42907", "right": "issue:44655", "accept": false, - "reason": "Both mention saving, but one is dequantized model serialization and the other is saving pipeline objects; not the same issue." + "reason": "Both mention saving, but dequantized model save failure and pipeline object save failure are different code paths." }, { "left": "issue:43012", "right": "issue:43550", "accept": false, - "reason": "A bf16 compile warning and a Bamba SDPA compile failure are different symptoms and code paths." + "reason": "A PyTorch precision warning is not the same as a torch.compile SDPA model failure." }, { "left": "issue:45712", "right": "issue:46298", "accept": false, - "reason": "Repo hygiene around dummy objects is unrelated to logging noise." + "reason": "Dummy class leakage into dir(transformers) is unrelated to a logging message." }, { "left": "issue:43825", "right": "issue:44655", "accept": false, - "reason": "An error message about unsupported translation pipelines and inability to save pipelines are unrelated." + "reason": "Pipeline task support messaging and pipeline save_pretrained are unrelated." }, { "left": "issue:43208", "right": "issue:43550", "accept": false, - "reason": "xLSTM training bugs and Bamba torch.compile/SDPA failure are different model/backend problems." + "reason": "xLSTM training-size bugs are unrelated to Bamba compile/SDPA failures." }, { "left": "issue:43012", "right": "issue:44368", "accept": false, - "reason": "A PyTorch bf16 warning and a tie-word-embeddings warning are not the same concrete bug." + "reason": "Compile-time bfloat16 warning is unrelated to tie_word_embeddings config guidance." }, { "left": "issue:43992", "right": "issue:44484", "accept": false, - "reason": "Missing weight loading and shard-size policy are unrelated save/load concerns." + "reason": "Missing embed_tokens weight on load is unrelated to max_shard_size defaults." }, { "left": "issue:43352", "right": "issue:43976", "accept": false, - "reason": "Flash Attention support gating and a Python version compatibility failure are distinct issues." + "reason": "Model-specific Flash Attention support and Python version incompatibility are different issues." }, { "left": "issue:43352", "right": "issue:44380", "accept": false, - "reason": "One is missing Flash Attention 2.0 support in a model, the other is attention scaling ignored under backend selection; not duplicate-level." + "reason": "Unsupported FlashAttention 2.0 for a model is not the same as attention scaling being ignored." }, { "left": "issue:43066", "right": "issue:43906", "accept": false, - "reason": "Wrong tokenizer decoder type and an isolated reproduction of another issue are not enough to establish the same bug." + "reason": "Tokenizer decoder type regression and an isolated reproduction of another issue are not duplicates." }, { "left": "issue:43097", "right": "issue:46298", "accept": false, - "reason": "A removed config flag and a noisy log message are unrelated." + "reason": "Removed config API and a noisy logging message are unrelated." }, { "left": "issue:45584", "right": "issue:46365", "accept": false, - "reason": "Whisper empty-transcription generation failure is unrelated to the Trainer/datasets import problem." + "reason": "Whisper empty transcription handling is unrelated to importing Trainer." }, { "left": "issue:44162", "right": "issue:46172", "accept": false, - "reason": "ESM2 breakage and a deepseek-v4 tensor operation bug are unrelated model-specific defects." + "reason": "ESM2 breakage and a DeepSeek tensor operation bug are unrelated." }, { "left": "issue:44380", "right": "issue:46132", "accept": false, - "reason": "Ignored attention scaling and registered-function behavior changes are different changes in different APIs." + "reason": "Attention scaling config loss under backend selection is not the same as register behavior changes." }, { "left": "issue:44077", "right": "issue:46365", "accept": false, - "reason": "PatchTSMixer post-init policy and Trainer import issues are unrelated." + "reason": "patchtsmixer post_init validation is unrelated to the Trainer import issue." }, { "left": "issue:44222", "right": "issue:45478", "accept": false, - "reason": "Both concern MoE/Qwen models, but one is FP8 save/pretrained and the other is a from_pretrained regression; not clearly the same fix." + "reason": "Both involve Qwen/MoE, but save_pretrained FP8 and from_pretrained loading errors are different concrete bugs." } ] }, @@ -17353,18 +17302,18 @@ "issue:43618|issue:44380", "issue:38617|issue:43502", "issue:43208|issue:43526", - "issue:43618|issue:43761", "issue:43756|issue:44485", + "issue:43618|issue:43761", "issue:42994|issue:44655", - "issue:36010|issue:42757", "issue:45276|issue:46077", + "issue:36010|issue:42757", "issue:43299|issue:43867", "issue:44625|issue:46365", "issue:44483|issue:45290", "issue:44038|issue:44230", "issue:36683|issue:43475", - "issue:43352|issue:43937", "issue:43611|issue:44655", + "issue:43352|issue:43937", "issue:44355|issue:46877", "issue:44655|issue:46050", "issue:44625|issue:46077", @@ -17378,238 +17327,238 @@ "issue:29870|issue:36296", "issue:43064|issue:44492" ], - "prepared_review_unit_hash": "75cadb9d7bc8df1b9115646b20d5e2665211178236134d1ca65f72019e317276", + "prepared_review_unit_hash": "475fe931104149db7aa60afce29b9ceb9481b5817c820533bb0804e9bea463ca", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 37687, - "estimated_input_tokens": 9422, - "estimated_eval_tokens": 19100 + "serialized_chars": 37689, + "estimated_input_tokens": 9423, + "estimated_eval_tokens": 19102 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is heterogeneous: import errors, config regressions, backend/support issues, save/load bugs, and unrelated model-specific failures. None of the soft pairs look like the same concrete bug or change, so I would not merge any of them as duplicates.", - "confidence": 0.95, - "canonical_issue_reason": "No single canonical issue fits this cluster; the items are not duplicates of one underlying bug but a mix of unrelated regressions and feature issues.", - "canonical_pr_reason": "No PRs are present in this cluster, so there is no canonical PR candidate.", - "best_issue_reason": "No issue is a clear global representative because the cluster lacks a shared root cause or subsystem-specific duplicate set.", - "best_pr_reason": "No PRs are present, so there is no best PR candidate.", + "summary": "The cluster is a grab bag of unrelated Transformers issues that mostly share superficial keywords like import errors, save/load, or model/config regressions. None of the soft pairs look like the same underlying bug or change, so all candidate edges are rejected.", + "confidence": 0.86, + "canonical_issue_reason": "issue:45216 is the closest thing to a representative item because it is a concrete, version-bounded regression around save_pretrained; but the overall set is too heterogeneous to have a strong canonical issue.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45216 is the best representative issue: it has a specific failure mode, a clear regression boundary, and a focused code path, unlike most other items in the set.", + "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:36683", "right": "issue:43582", "accept": false, - "reason": "Different bugs: Gemma3 missing config attribute vs Apple Silicon TypeError in allocator warmup." + "reason": "Different bugs: Gemma3 config missing vocab_size vs Apple Silicon allocator warmup TypeError." }, { "left": "issue:44380", "right": "issue:45479", "accept": false, - "reason": "Unrelated: attention scaling under SDPA/FlashAttention vs degenerate zero loss in sequence classification." + "reason": "Different areas: attention scaling ignored in GPT-2 backends vs degenerate zero loss in sequence classification." }, { "left": "issue:42757", "right": "issue:43824", "accept": false, - "reason": "Both are import errors, but for different symbols and different package breakages." + "reason": "Both are import errors, but they involve different missing symbols from different packages." }, { "left": "issue:42175", "right": "issue:46365", "accept": false, - "reason": "Different dependency problems: missing TensorFlow in extras vs Trainer import failing until datasets is removed." + "reason": "Unrelated packaging/import problems: extras backend missing TensorFlow vs Trainer import breaking with datasets installed." }, { "left": "issue:43064", "right": "issue:44380", "accept": false, - "reason": "FSDP/optimizer-state corruption is unrelated to GPT2 attention scaling behavior." + "reason": "Different code paths: distributed optimizer state corruption vs GPT-2 attention scaling backend behavior." }, { "left": "issue:42490", "right": "issue:43012", "accept": false, - "reason": "Local-path import warning is unrelated to bfloat16 compilation warnings." + "reason": "Different warnings with different triggers: local path import warning vs bfloat16 compilation precision warning." }, { "left": "issue:44483", "right": "issue:44514", "accept": false, - "reason": "API request rejection and batched chat-template crash are different code paths." + "reason": "API request validation problem vs chat template crash on batched input; not the same bug." }, { "left": "issue:43618", "right": "issue:44380", "accept": false, - "reason": "CLIP output field regression is unrelated to GPT2 attention scaling." + "reason": "Both mention attention, but one is a CLIP output field regression and the other is backend-specific scaling behavior." }, { "left": "issue:38617", "right": "issue:43502", "accept": false, - "reason": "Different failures: missing internal import vs unexpected network requests with local_files_only." + "reason": "ImportError for a removed utility vs unwanted API calls despite local_files_only=True." }, { "left": "issue:43208", "right": "issue:43526", "accept": false, - "reason": "xLSTM training bugs and BeitImageProcessorFast label reduction are unrelated." + "reason": "Different subsystems entirely: xLSTM training bugs vs Beit image processor label reduction." }, { - "left": "issue:43618", - "right": "issue:43761", + "left": "issue:43756", + "right": "issue:44485", "accept": false, - "reason": "Same model family, but different missing outputs; not clearly one concrete bug." + "reason": "RoPE-related titles only; one is a Smollm3 model bug and the other is a GLM-5 discussion, not the same concrete issue." }, { - "left": "issue:43756", - "right": "issue:44485", + "left": "issue:43618", + "right": "issue:43761", "accept": false, - "reason": "Smollm3 RoPE layer count bug is unrelated to GLM-5 RoPE implementation discussion." + "reason": "Both are CLIP regressions, but one drops attentions and the other drops hidden_states; not enough evidence of one underlying bug." }, { "left": "issue:42994", "right": "issue:44655", "accept": false, - "reason": "Quantized model save/load failure is unrelated to Pipeline save_pretrained support." + "reason": "Both involve saving, but quantized model save failure and Pipeline save_pretrained failure are distinct problems." }, { - "left": "issue:36010", - "right": "issue:42757", + "left": "issue:45276", + "right": "issue:46077", "accept": false, - "reason": "Different missing imports from different modules; not the same underlying regression." + "reason": "Different Gemma4 bugs: embedding resize propagation vs bidirectional attention mask construction." }, { - "left": "issue:45276", - "right": "issue:46077", + "left": "issue:36010", + "right": "issue:42757", "accept": false, - "reason": "Token embedding resize propagation is unrelated to bidirectional attention mask construction." + "reason": "Different import failures from different modules and symbols." }, { "left": "issue:43299", "right": "issue:43867", "accept": false, - "reason": "Qwen3VL model loading breakage and sorted-state_dict load order issue are different bugs." + "reason": "Different model loading failures: Qwen3VL MoE loading regression vs state_dict ordering issue." }, { "left": "issue:44625", "right": "issue:46365", "accept": false, - "reason": "Qwen3.5 num_labels propagation and Trainer import conflict are unrelated." + "reason": "Qwen3.5 num_labels propagation bug is unrelated to the Trainer import issue." }, { "left": "issue:44483", "right": "issue:45290", "accept": false, - "reason": "Chat completions request validation and assistant tool-call template crash are different failures." + "reason": "Both are chat-related, but one is a server API request issue and the other is a tokenizer/template crash." }, { "left": "issue:44038", "right": "issue:44230", "accept": false, - "reason": "Both mention Qwen3-VL-Moe, but one is loading regression and the other is fp8 inference support." + "reason": "Same model family, but one is a loading bug and the other is an fp8 inference support request." }, { "left": "issue:36683", "right": "issue:43475", "accept": false, - "reason": "Gemma3 config attribute missing and SAM 3 vision output attribute missing are unrelated." + "reason": "Different missing-attribute bugs on different model families and different config/output objects." }, { - "left": "issue:43352", - "right": "issue:43937", + "left": "issue:43611", + "right": "issue:44655", "accept": false, - "reason": "FlashAttention support error for Nemotron is unrelated to GLM-5 GenerationConfig validation." + "reason": "Model loading with base_model_prefix and Pipeline saving are separate code paths." }, { - "left": "issue:43611", - "right": "issue:44655", + "left": "issue:43352", + "right": "issue:43937", "accept": false, - "reason": "base_model_prefix loading regression is unrelated to saving Pipeline objects." + "reason": "Different failures: Flash Attention support gate vs invalid GenerationConfig." }, { "left": "issue:44355", "right": "issue:46877", "accept": false, - "reason": "Compiled Python file errors and auto_docstring import-time stdout diagnostics are different issues." + "reason": "Both touch import-time behavior, but compiled-file execution errors and docstring lint output are unrelated." }, { "left": "issue:44655", "right": "issue:46050", "accept": false, - "reason": "Pipeline save_pretrained and sinq quantized model save/load failures are distinct." + "reason": "Both mention save_pretrained, but one is Pipeline serialization and the other is sinq quantization persistence." }, { "left": "issue:44625", "right": "issue:46077", "accept": false, - "reason": "Configuration propagation bug and Gemma4 attention-mask bug are unrelated." + "reason": "Different config bugs in different models: Qwen3.5 labels vs Gemma4 bidirectional attention." }, { "left": "issue:36683", "right": "issue:44077", "accept": false, - "reason": "Gemma3 config missing vocab_size is unrelated to patchtsmixer post_init validation." + "reason": "Different config issues on unrelated model families." }, { "left": "issue:43976", "right": "issue:44938", "accept": false, - "reason": "Both are version-compatibility reports, but for different Python releases and likely different causes." + "reason": "Both are Python-version compatibility complaints, but they concern different versions and likely different causes." }, { "left": "issue:43867", "right": "issue:45216", "accept": false, - "reason": "Sorted state_dict load issue and Qwen3.5 save_pretrained regression are different code paths." + "reason": "Different checkpoint issues: sorted state_dict loading vs incorrect saved checkpoint content." }, { "left": "issue:46077", "right": "issue:46082", "accept": false, - "reason": "Gemma4 attention-mask semantics and LlamaConfig head_dim validation are unrelated." + "reason": "Unrelated model config problems: Gemma4 attention masking vs LlamaConfig head_dim validation." }, { "left": "issue:43618", "right": "issue:45216", "accept": false, - "reason": "CLIP output regression and Qwen3.5 checkpoint saving regression are unrelated." + "reason": "Different models and symptoms: CLIP output regression vs Qwen3.5 save_pretrained regression." }, { "left": "issue:36683", "right": "issue:43525", "accept": false, - "reason": "Missing Gemma3 config vocab_size and Llama4 pad_token_id are different model config bugs." + "reason": "Both are missing config attributes, but on different models and different fields." }, { "left": "issue:36683", "right": "issue:44625", "accept": false, - "reason": "Different config attributes missing on different models; not a duplicate." + "reason": "Different config propagation bugs on different model families." }, { "left": "issue:29870", "right": "issue:36296", "accept": false, - "reason": "BART generate UnboundLocalError and tensor-parallel training bug are unrelated." + "reason": "Bart generate() decoder_attention_mask bug is unrelated to the tensor-parallel training issue." }, { "left": "issue:43064", "right": "issue:44492", "accept": false, - "reason": "Optimizer-state corruption has nothing to do with a cache-strategy typo." + "reason": "Distributed optimizer-state corruption and a cache-strategy typo are not the same bug." } ] }, @@ -17686,15 +17635,15 @@ "issue:43010|issue:45278", "issue:44075|issue:46365", "issue:44908|issue:46851", - "issue:44393|issue:47000", "issue:43992|issue:45310", + "issue:44393|issue:47000", "issue:44938|issue:45071", "issue:29870|issue:33453", "issue:44355|issue:44855", "issue:43867|issue:46202", "issue:36683|issue:43064", - "issue:43867|issue:45375", "issue:42890|issue:45561", + "issue:43867|issue:45375", "issue:43010|issue:45850", "issue:36296|issue:43575", "issue:41669|issue:44246", @@ -17713,215 +17662,206 @@ "issue:44734|issue:46617", "issue:43723|issue:44556" ], - "prepared_review_unit_hash": "3e55cfcd912e421a45eb8cd496b760fd95abc850435158696b6d96168db86b7f", + "prepared_review_unit_hash": "29868d1f9c53eff1735127a4b9f9a4e1937f3f20987d9571a9d215845e41bc07", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36052, - "estimated_input_tokens": 9013, - "estimated_eval_tokens": 18282 + "serialized_chars": 36058, + "estimated_input_tokens": 9015, + "estimated_eval_tokens": 18286 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of unrelated Transformers bug reports that only share broad vocabulary like loading, v5 regressions, or model-specific failures. None of the soft-similarity pairs look like the same underlying bug/change, so I would not merge any of them as duplicates.", - "confidence": 0.91, - "canonical_issue_reason": "No single issue cleanly represents the cluster: the items span distinct failures across import, tokenizer/model loading, training, CI, and model-specific regressions. The cluster is heterogeneous rather than one duplicate family.", + "summary": "Mostly unrelated issue reports with one strong duplicate pair around slow `import transformers`. I would keep the import-slow regression issue as the canonical item.", + "confidence": 0.81, + "canonical_issue_reason": "issue:41669 is the clearest, most specific import-performance report: it identifies the regression source (`import *` usage in models), quantifies the impact, and already has inbound reference activity.", "canonical_pr_reason": null, - "best_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'), but it still only covers one slice of the overall heterogeneous cluster and is not a true canonical for all items.", + "best_issue_reason": "issue:41669 is the best representative issue because it states the concrete root cause and symptom for the import-time slowdown, making it the most actionable anchor for the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43010", "right": "issue:45278", "accept": false, - "reason": "Both mention v5 regressions, but one is about cache/update() no_grad decoration and the other is a broad import-error report; different code paths and symptoms." + "reason": "Different problems: cache/update no-grad decoration vs broad import errors after upgrading." }, { "left": "issue:44075", "right": "issue:46365", "accept": false, - "reason": "SGD optimizer arguments not being used is unrelated to an import failure caused by datasets blocking Trainer import." + "reason": "Unrelated: optimizer arguments being ignored vs Trainer import failing due to `datasets`." }, { "left": "issue:44908", "right": "issue:46851", "accept": false, - "reason": "Scheduler kwargs propagation and Qwen3.5 variable-length position_ids are unrelated bugs in different subsystems." + "reason": "Different subsystems: scheduler kwargs vs Qwen3.5 variable-length position ids." }, { - "left": "issue:44393", - "right": "issue:47000", + "left": "issue:43992", + "right": "issue:45310", "accept": false, - "reason": "Both are vision/detection related, but one is Qwen3-VL bbox output quality and the other is RTDetrLoss infeasible cost-matrix crashing; not the same bug." + "reason": "Different model-loading failures with no clear shared code path." }, { - "left": "issue:43992", - "right": "issue:45310", + "left": "issue:44393", + "right": "issue:47000", "accept": false, - "reason": "UMT5Encoder missing embed_tokens.weight is a checkpoint-loading issue, while the Qwen3.5 Moe error is a different model family/problem." + "reason": "Both are vision/detection-related, but they describe different concrete bugs in different models/losses." }, { "left": "issue:44938", "right": "issue:45071", "accept": false, - "reason": "Python 3.14 import/load failure and PretrainedConfig type-checking breakage are different compatibility regressions." + "reason": "Python-version load failure vs `PretrainedConfig` type-checking regression; not the same bug." }, { "left": "issue:29870", "right": "issue:33453", "accept": false, - "reason": "BART generate decoder_attention_mask unbound error and tokenizer-loading regression are separate failure modes." + "reason": "Bart `generate()` mask bug is unrelated to tokenizer loading regression." }, { "left": "issue:44355", "right": "issue:44855", "accept": false, - "reason": "Compiled Python file import errors and Python 3.13 torch.jit.script parsing/indentation issues are distinct causes." + "reason": "Both involve import/runtime errors, but one is compiled-file execution and the other is a specific `torch.jit.script` parsing failure." }, { "left": "issue:43867", "right": "issue:46202", "accept": false, - "reason": "A state_dict loading/sorting error is not the same as a Gemma pipeline roundtrip failure on the underscore token." + "reason": "State-dict sorting/load error and Gemma pipeline roundtrip issue are different failures." }, { "left": "issue:36683", "right": "issue:43064", "accept": false, - "reason": "Gemma3Config missing vocab_size and wrong optimizer states under FSDP2/PEFT are unrelated training/config bugs." + "reason": "Gemma3 config attribute error is unrelated to FSDP2/PEFT optimizer-state corruption." }, { - "left": "issue:43867", - "right": "issue:45375", + "left": "issue:42890", + "right": "issue:45561", "accept": false, - "reason": "Both involve loading/config, but one is state_dict handling and the other is a missing vision config field silently dropped by strict validation." + "reason": "Test flakiness from missing seeds is not the same as xdist file-write races." }, { - "left": "issue:42890", - "right": "issue:45561", + "left": "issue:43867", + "right": "issue:45375", "accept": false, - "reason": "Missing set_seed in a SAM HQ integration test is unrelated to an xdist race on captured_info.txt in testing utils." + "reason": "Different Qwen/Gemma loading and config-schema problems." }, { "left": "issue:43010", "right": "issue:45850", "accept": false, - "reason": "Decorating cache update methods with no_grad is a specific fix; 'breaks too many models' is a broad complaint without the same concrete bug." + "reason": "Specific cache-layer no-grad bug vs generic post-v5 breakage complaint." }, { "left": "issue:36296", "right": "issue:43575", "accept": false, - "reason": "Tensor parallel training bug and Qwen2-57B-A14B-Instruct OOM are both TP-related but not the same issue or code path." + "reason": "Both mention tensor parallelism, but one is a training bug and the other is an OOM while loading a specific model." }, { "left": "issue:41669", "right": "issue:44246", - "accept": false, - "reason": "Import-star removal performance regression and intermittent slow import are similar at a high level but not the same underlying change." - }, - { - "left": "issue:43618", - "right": "issue:44077", - "accept": false, - "reason": "CLIPOutput attentions assignment and patchtsmixer post_init allowance are model-specific API regressions with different fixes." - }, - { - "left": "issue:43010", - "right": "issue:43976", - "accept": false, - "reason": "No_grad cache updates and Python version support failure are unrelated regressions." + "accept": true, + "reason": "Same underlying issue: `import transformers` is slow; 41669 identifies the likely root cause and 44246 reports the same symptom." }, { "left": "issue:43867", "right": "issue:45357", "accept": false, - "reason": "State_dict loading error and incorrect visual encoder keys on save_pretrained are opposite directions of the serialization flow and different bugs." + "reason": "Different save/load key mismatches for different models." }, { "left": "issue:43992", "right": "issue:45478", "accept": false, - "reason": "UMT5Encoder missing embed_tokens.weight does not match the Qwen3.5 Moe from_pretrained regression." + "reason": "Different model families and different from_pretrained failures." }, { "left": "issue:28282", "right": "issue:33453", "accept": false, - "reason": "Missing PyTorch dependency for AutoModel import is unrelated to a tokenizer-loading regression." + "reason": "Missing PyTorch dependency error is unrelated to tokenizer-loading regression." }, { "left": "issue:43618", "right": "issue:46285", "accept": false, - "reason": "CLIPOutput attentions not assigned and CLIPTextModel structural inconsistency are separate CLIP regressions." + "reason": "Both are CLIP-related, but one is a missing `attentions` assignment and the other is a class-structure inconsistency." }, { "left": "issue:44360", "right": "issue:45440", "accept": false, - "reason": "A DSA indexer/ReLU discussion is not the same as DeepseekV3MoE diverging from remote implementation." + "reason": "Different model implementation concerns: DSA indexer activation vs DeepSeekV3MoE parity." }, { "left": "issue:42890", "right": "issue:44964", "accept": false, - "reason": "SAM HQ test seeding issues and Phi-4-multimodal model loading failure are unrelated." + "reason": "Unrelated test-seeding issue vs model-loading failure for Phi-4 multimodal." }, { "left": "issue:42890", "right": "issue:44112", "accept": false, - "reason": "A flaky integration test due to missing seeds is unrelated to a GraniteSpeech CI stale device override failure." + "reason": "Both are CI/test failures, but the causes are different." }, { "left": "issue:42890", "right": "issue:45600", "accept": false, - "reason": "Test seeding flakiness and auto_mappings.py referencing removed configs are different CI problems." + "reason": "Missing seeds in tests is unrelated to stale auto-mappings referencing removed configs." }, { "left": "issue:43502", "right": "issue:44062", "accept": false, - "reason": "local_files_only still making API requests is a hub/network bug, while AddedToken special kwarg duplication is a tokenizer construction bug." + "reason": "Offline-loading API requests vs tokenizer `AddedToken` keyword collision are different bugs." }, { "left": "issue:42898", "right": "issue:46664", "accept": false, - "reason": "Tokenizer cleanup behavior in v5 and SmolVLM num_frames semantics are unrelated API changes." + "reason": "Tokenizer cleanup behavior change is unrelated to video processor frame-count semantics." }, { "left": "issue:44734", "right": "issue:46617", "accept": false, - "reason": "KV cache continuation crash in serve is not the same as a FileNotFoundError loading trust_remote_code model from cache." + "reason": "KV-cache continuation indexing crash is unrelated to cache loading `FileNotFoundError`." }, { "left": "issue:43723", "right": "issue:44556", "accept": false, - "reason": "Tokenizer loading failure and checkpoint reload failure after upgrade are both loading-related but affect different components and symptoms." + "reason": "Both involve loading after v5, but one is tokenizer loading and the other is checkpoint reload compatibility; not enough evidence of the same bug." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The lone accepted duplicate pair, issue:41669 with issue:44246, is well supported by the titles and shared slowdown symptom. The other soft-edge rejections are phrased cautiously and do not overclaim shared root causes." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -17992,8 +17932,8 @@ "issue:43502|issue:44485", "issue:43010|issue:43065", "issue:43874|issue:46617", - "issue:44387|issue:44556", "issue:43576|issue:46132", + "issue:44387|issue:44556", "issue:30333|issue:33453", "issue:45125|issue:45478", "issue:44355|issue:44749", @@ -18027,20 +17967,20 @@ "issue:44315|issue:46209", "issue:44222|issue:44230" ], - "prepared_review_unit_hash": "235c9998a184dfb79b8568175ba6e4cc72a605b72a44fdd8d767b1bbfab02077", + "prepared_review_unit_hash": "a2eab54e66023c374e8d4fc855f1a34af28e9c94aa7eb375401273f076b9890e", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 39, - "serialized_chars": 38291, + "serialized_chars": 38292, "estimated_input_tokens": 9573, "estimated_eval_tokens": 19402 }, @@ -18049,228 +17989,246 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is largely heterogeneous: a mix of unrelated loading/import regressions, docs issues, model-family bugs, and environment/runtime problems. None of the soft pairs look like safe duplicates, so they should remain separate.", - "confidence": 0.93, - "canonical_issue_reason": "No single issue clearly subsumes the cluster; the items span different subsystems, models, and failure modes, so a canonical issue would be misleading.", + "summary": "This cluster is heterogeneous: most items are unrelated Transformers regression, import, loading, docs, or model-specific bugs. None of the soft pairs look like the same underlying issue closely enough to merge, so I would reject all candidate duplicate edges.", + "confidence": 0.78, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "If one issue must represent the cluster, issue:43576 is the broadest and most generally framed regression report, with clear user impact and some downstream references; however, it still does not duplicate the others.", + "best_issue_reason": "If forced to pick a representative, issue 43576 is the broadest v5 regression report in the set, but the cluster is too mixed for it to serve as a true canonical duplicate target.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44222", "right": "issue:45125", "accept": false, - "reason": "Both mention FP8/MoE, but one is about save_pretrained and the other tensor-parallel model loading; different bugs and code paths." + "reason": "FP8 save_pretrained/serialization vs tensor-parallel plan/model-loading; different failure modes and code paths." }, { "left": "issue:43824", "right": "issue:46617", "accept": false, - "reason": "Both are loading-related, but one is a missing exported model class and the other is a cache FileNotFoundError for trust_remote_code." + "reason": "Missing top-level class export vs cache/trust_remote_code FileNotFoundError; unrelated causes." }, { "left": "issue:43502", "right": "issue:44485", "accept": false, - "reason": "One is about local_files_only still triggering network calls; the other is a RoPE implementation discussion for GLM-5." + "reason": "Local-files-only network leakage vs GLM-5 RoPE implementation discussion; not the same bug." }, { "left": "issue:43010", "right": "issue:43065", "accept": false, - "reason": "Cache/layer no_grad decoration and a dummy Conv2d in Sam3PixelDecoder are unrelated fixes." + "reason": "Cache update no_grad behavior vs SAM3 dummy Conv2d; completely different subsystems." }, { "left": "issue:43874", "right": "issue:46617", "accept": false, - "reason": "Missing image-patch counting in a processor and a cache FileNotFoundError are different failures." + "reason": "Missing image-patch helper vs cache load failure; different underlying problems." }, { - "left": "issue:44387", - "right": "issue:44556", + "left": "issue:43576", + "right": "issue:46132", "accept": false, - "reason": "Both concern version regressions, but one is int4 memory/OOM and the other is checkpoint reload compatibility." + "reason": "Broken env CLI command vs AttentionInterface.register behavior change; not the same issue." }, { - "left": "issue:43576", - "right": "issue:46132", + "left": "issue:44387", + "right": "issue:44556", "accept": false, - "reason": "A broken env CLI command and AttentionInterface.register behavior changes are unrelated changes." + "reason": "Int4 OOM/memory regression vs checkpoint reload breakage after upgrade; distinct bugs." }, { "left": "issue:30333", "right": "issue:33453", "accept": false, - "reason": "MLFlow reporting on failed jobs and tokenizer loading regression are different subsystems and symptoms." + "reason": "MLflow reporting on failed jobs vs tokenizer loading regression; unrelated." }, { "left": "issue:45125", "right": "issue:45478", "accept": false, - "reason": "Both involve Qwen3.5 MoE, but one is missing _tp_plan for tensor parallelism while the other is a broader from_pretrained error." + "reason": "Both involve Qwen3.5 MoE, but one is missing _tp_plan and the other is a from_pretrained error; too little evidence of one concrete bug." }, { "left": "issue:44355", "right": "issue:44749", "accept": false, - "reason": "Compiled Python file errors and slower filtering after upgrade are unrelated regressions." + "reason": "Compiled Python file errors vs slowdown during data filtering after upgrade; different symptoms and paths." }, { "left": "issue:42548", "right": "issue:44351", "accept": false, - "reason": "Different import failures for different symbols; no evidence they share the same underlying export bug." + "reason": "Both are import errors, but for different symbols and likely different export failures." }, { "left": "issue:30990", "right": "issue:31795", "accept": false, - "reason": "Sentence Transformers loading hang and forward() documentation confusion are unrelated." + "reason": "Sentence Transformers loading hang vs docs confusion about model.forward arguments; unrelated." }, { "left": "issue:45125", "right": "issue:45375", "accept": false, - "reason": "Missing tensor-parallel plan and missing config field handling are separate Qwen3.5 MoE issues." + "reason": "Tensor-parallel plan issue vs missing config field silently dropped by strict validation; related model family, not same bug." }, { "left": "issue:43502", "right": "issue:44222", "accept": false, - "reason": "Local-files-only network leakage and FP8 save_pretrained issues are not the same bug." + "reason": "Local_files_only network requests vs FP8 save_pretrained; different code paths." }, { "left": "issue:43502", "right": "issue:44961", "accept": false, - "reason": "A network-policy bug and an unrelated 'racoon' issue are clearly distinct." + "reason": "Network leakage bug vs unrelated placeholder issue title; no shared underlying defect." }, { "left": "issue:43010", "right": "issue:43232", "accept": false, - "reason": "Both touch generation internals, but no_grad decoration and sync_gpus/update_model_kwargs behavior are different code-path problems." + "reason": "No-grad decoration for cache updates vs generation kwargs handling after sync_gpus; unrelated generation internals." }, { "left": "issue:46664", "right": "issue:46726", "accept": false, - "reason": "A video num_frames semantic mismatch and an AutoProcessor backend-missing import error are unrelated." + "reason": "Video num_frames semantics vs AutoProcessor backend-missing import message; different processor bugs." }, { "left": "issue:43827", "right": "issue:44961", "accept": false, - "reason": "Documentation references to pipeline() after v5 removals do not match the unrelated 'racoon' issue." + "reason": "Docs still referencing pipeline() vs unrelated placeholder issue; not a duplicate." }, { "left": "issue:30333", "right": "issue:36296", "accept": false, - "reason": "MLFlow failure reporting and tensor-parallel training bugs are different." + "reason": "MLflow reporting vs tensor-parallel training bug; unrelated." }, { "left": "issue:29870", "right": "issue:31795", "accept": false, - "reason": "An UnboundLocalError in generation and a docs confusion issue are not duplicates." + "reason": "Bart generate UnboundLocalError vs documentation confusion; different artifact types and problems." }, { "left": "issue:43299", "right": "issue:44556", "accept": false, - "reason": "Both mention version regressions, but they affect different models and have different loading symptoms." + "reason": "Qwen3VL/Qwen3.5 MoE load regression vs checkpoint reload after version upgrade; both version-related, but not the same concrete failure." }, { "left": "issue:33453", "right": "issue:34689", "accept": false, - "reason": "Tokenizer loading regression and Llama 3.2 vision model loading breakage are different concrete failures." + "reason": "Tokenizer loading regression vs Llama 3.2 Vision model loading breakage; similar theme, different model-specific bugs." }, { "left": "issue:44355", "right": "issue:45071", "accept": false, - "reason": "Compiled-file errors and PretrainedConfig type-checking regression are unrelated." + "reason": "Compiled-file loading errors vs PretrainedConfig type checking regression; unrelated." }, { "left": "issue:43576", "right": "issue:45071", "accept": false, - "reason": "A broken CLI env command and a PretrainedConfig typing regression do not share the same bug." + "reason": "Broken env command vs typing regression; no shared code path." }, { "left": "issue:42757", "right": "issue:43576", "accept": false, - "reason": "Missing is_offline_mode export from huggingface_hub and the env command regression are separate issues." + "reason": "huggingface_hub import issue vs transformers env command bug; different packages and failures." }, { "left": "issue:44315", "right": "issue:44360", "accept": false, - "reason": "Liger Kernel application with model_init and a DSA indexer ReLU discussion are unrelated." + "reason": "Liger Kernel/model_init application vs DSA indexer ReLU discussion; unrelated." }, { "left": "issue:31795", "right": "issue:38617", "accept": false, - "reason": "Docs about forward arguments and an import error for layer_type_validation are different problems." + "reason": "Documentation confusion vs missing layer_type_validation import; different problem classes." }, { "left": "issue:43576", "right": "issue:43824", "accept": false, - "reason": "A broken env command and a missing Qwen2.5-VL export are not the same underlying bug." + "reason": "CLI env regression vs missing model class export; not the same bug." }, { "left": "issue:44961", "right": "issue:45468", "accept": false, - "reason": "The 'racoon' issue is unrelated to Gemma-4 audio positional encoding." + "reason": "Placeholder title vs Gemma4 audio positional encoding bug; no relation." }, { "left": "issue:44485", "right": "issue:44961", "accept": false, - "reason": "GLM-5 RoPE implementation discussion and an unrelated placeholder issue are not duplicates." + "reason": "GLM-5 RoPE implementation discussion vs placeholder title; unrelated." + }, + { + "left": "issue:44351", + "right": "issue:46617", + "accept": false, + "reason": "Missing HybridCache export vs cache FileNotFoundError; different root causes." + }, + { + "left": "issue:36296", + "right": "issue:43901", + "accept": false, + "reason": "Tensor-parallel training bug vs text-classification docs mismatch; unrelated." }, { "left": "issue:43352", "right": "issue:44961", "accept": false, - "reason": "Flash Attention support for Nemotron and the unrelated 'racoon' issue do not match." + "reason": "Flash Attention 2.0 support error vs placeholder title; no overlap." + }, + { + "left": "issue:43576", + "right": "issue:44938", + "accept": false, + "reason": "Env command broken in v5 vs Python 3.14 load failure; different issues." }, { "left": "issue:43475", "right": "issue:46617", "accept": false, - "reason": "SAM 3 Video missing fpn_position_embeddings and a cache FileNotFoundError are different failures." + "reason": "SAM3VisionEncoderOutput attribute error vs trust_remote_code cache loading failure; unrelated." }, { "left": "issue:46143", "right": "issue:46144", "accept": false, - "reason": "Both concern RoFormer, but one is kwargs propagation and the other is attention-interface usage; related area, not the same bug." + "reason": "kwargs passthrough bug vs attention-interface usage in RoFormer; related area, but not the same concrete defect." }, { "left": "issue:41628", "right": "issue:42548", "accept": false, - "reason": "Missing AutoImageProcessor and missing PreTrainedModel are distinct import/export regressions." + "reason": "Different missing top-level imports; same broad symptom, but not enough to treat as one bug." }, { "left": "issue:44315", "right": "issue:46209", "accept": false, - "reason": "Tensor-parallel application of Liger Kernel and FP8 inference failure on B200 are unrelated." + "reason": "Liger Kernel/model_init issue vs FP8 inference failure on B200; different code paths." }, { "left": "issue:44222", "right": "issue:44230", "accept": false, - "reason": "Both mention FP8 MoE, but save_pretrained support and inference support are different concrete problems." + "reason": "FP8 save_pretrained bug vs FP8 inference support request; same buzzwords, different problems." } ] }, @@ -18352,8 +18310,8 @@ "issue:44273|issue:45003", "issue:44360|issue:46664", "issue:44038|issue:45125", - "issue:44393|issue:45071", "issue:41669|issue:43673", + "issue:44393|issue:45071", "issue:44393|issue:45753", "issue:43576|issue:44556", "issue:43525|issue:45850", @@ -18385,20 +18343,20 @@ "issue:44315|issue:44829", "issue:43519|issue:46655" ], - "prepared_review_unit_hash": "2908815b9b6847a70c8bc285c25a5d75d85980a6145d739c56d8c3a1cc9521f8", + "prepared_review_unit_hash": "4cf768a63b492611751e7c2fdd10ee3c2061375f5f19c364c96fe103f68149d6", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 40, - "serialized_chars": 39085, + "serialized_chars": 39088, "estimated_input_tokens": 9772, "estimated_eval_tokens": 19800 }, @@ -18407,252 +18365,252 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a heterogeneous set of issues: broad v5 regressions, model-specific bugs, import/performance problems, and unrelated API breakages. None of the soft pairs looks like the same underlying bug/change closely enough to merge as duplicates.", - "confidence": 0.88, - "canonical_issue_reason": null, - "canonical_pr_reason": null, - "best_issue_reason": "issue:45850 is the closest umbrella candidate because it broadly describes v5 breakage, but it is far too vague to be a precise duplicate target.", - "best_pr_reason": null, + "summary": "These are mostly unrelated closed issues around Transformers v5 regressions, model-specific bugs, and import/loading problems. No soft pair is close enough to be treated as the same underlying bug or change.", + "confidence": 0.34, + "canonical_issue_reason": "issue:45850 is the broadest umbrella for the v5 regression theme, but it is still too generic to represent a true duplicate cluster.", + "canonical_pr_reason": "No pull requests are present in the provided items.", + "best_issue_reason": "issue:45850 is the best overall representative because it captures the widest version-regression complaint, even though it is not a precise duplicate target.", + "best_pr_reason": "No pull requests are present, so there is no best PR candidate.", "soft_edge_verdicts": [ { "left": "issue:45362", "right": "issue:45850", "accept": false, - "reason": "Different scope: a specific Qwen3.5 chat crash vs a vague general v5 breakage complaint." + "reason": "Same general v5 frustration, but one is a specific Qwen3.5 chat crash and the other is a broad umbrella complaint." }, { "left": "issue:43673", "right": "issue:44492", "accept": false, - "reason": "Different bugs: missing GenerationMixin cache in chunked prefill vs a typo in cache strategy docs/code." + "reason": "Different cache-related topics: chunked_prefill cache handling vs a typo in cache strategy wording." }, { "left": "issue:44485", "right": "issue:46664", "accept": false, - "reason": "Unrelated subsystems: GLM-5 RoPE implementation vs SmolVLM video num_frames semantics." + "reason": "Unrelated model areas: GLM-5 RoPE implementation vs SmolVLM video-frame semantics." }, { "left": "issue:44273", "right": "issue:45542", "accept": false, - "reason": "Different failure modes: lazy loading problems vs undefined TensorFlow backend from tensorboard-only install." + "reason": "Both involve loading/import paths, but they are different bugs in different subsystems." }, { "left": "issue:44623", "right": "issue:45230", "accept": false, - "reason": "One is a concrete processor save_pretrained bug; the other is an empty generic report." + "reason": "One is a specific missing-files bug in processor.save_pretrained; the other is a generic bug report with no clear overlap." }, { "left": "issue:44273", "right": "issue:45003", "accept": false, - "reason": "Different code paths: lazy loading behavior vs unsafe sys.modules access in modeling_utils." + "reason": "Both touch import/lazy-loading behavior, but the concrete failure modes are different." }, { "left": "issue:44360", "right": "issue:46664", "accept": false, - "reason": "Different models and bugs: DSA indexer ReLU discussion vs SmolVLM video processor frame handling." + "reason": "Completely different areas: DSA indexer behavior vs video processor frame handling." }, { "left": "issue:44038", "right": "issue:45125", "accept": false, - "reason": "Qwen3-VL-Moe bug vs missing tensor-parallel plan for Qwen3_5Moe; related family, different concrete issues." + "reason": "Both mention Qwen3/MoE, but one is a model-support bug and the other is tensor-parallel metadata." }, { - "left": "issue:44393", - "right": "issue:45071", + "left": "issue:41669", + "right": "issue:43673", "accept": false, - "reason": "Different layers: Qwen3-VL bounding-box output problem vs PretrainedConfig type-checking regression." + "reason": "Import-time performance regression and cache-missing runtime bug are not the same code-path problem." }, { - "left": "issue:41669", - "right": "issue:43673", + "left": "issue:44393", + "right": "issue:45071", "accept": false, - "reason": "Import-time performance regression from import * is not the same as a chunked-prefill cache bug." + "reason": "Different layers entirely: Qwen3-VL output behavior vs PretrainedConfig type checking." }, { "left": "issue:44393", "right": "issue:45753", "accept": false, - "reason": "Both are Qwen3-VL-family issues, but one is bbox hallucination/error and the other is an infinite loop on a specific image." + "reason": "Both are Qwen3-VL-family issues, but one is bbox output error and the other is an infinite loop on a specific image." }, { "left": "issue:43576", "right": "issue:44556", "accept": false, - "reason": "Different breakages: transformers env command vs checkpoint reload compatibility after upgrade." + "reason": "Both are v5 breakages, but one is the env command and the other is checkpoint reload compatibility." }, { "left": "issue:43525", "right": "issue:45850", "accept": false, - "reason": "A specific Llama4 pad_token_id AttributeError is not the same as a broad statement that v5 breaks many models." + "reason": "A specific Llama4Config attribute error is not the same as a broad 'many models break' complaint." }, { "left": "issue:43572", "right": "issue:45850", "accept": false, - "reason": "Specific StableLm pad_token_idx regression vs a generic v5 compatibility complaint." + "reason": "StableLmConfig pad token field bug is model-specific, not the same as the generic v5 regression umbrella." }, { "left": "issue:44556", "right": "issue:45092", "accept": false, - "reason": "Both concern v5 compatibility, but one is checkpoint reload and the other is remote-code meta-initialization incompatibility." + "reason": "Related upgrade/reload symptoms, but the checkpoint compatibility context is different and not clearly the same bug." }, { "left": "issue:44492", "right": "issue:45850", "accept": false, - "reason": "A typo in cache strategies is not the same underlying bug as a general 'v5 breaks many models' report." + "reason": "A typo in cache strategies is unrelated to the broad v5 regression report." }, { "left": "issue:45278", "right": "issue:45850", "accept": false, - "reason": "Broad import errors after upgrade overlaps in theme, but not the same concrete bug as the generic breakage report." + "reason": "General upgrade breakage overlaps in theme only; it is not the same concrete issue." }, { "left": "issue:42371", "right": "issue:44492", "accept": false, - "reason": "TF32 configuration guidance is unrelated to cache strategy typos." + "reason": "TF32 API guidance and cache strategy typo are unrelated." }, { "left": "issue:44230", "right": "issue:46095", "accept": false, - "reason": "Different requests entirely: fp8 inference support for Qwen3-VL/Qwen3.5 Moe vs a DeepSeek weight conversion question." + "reason": "Different request types: FP8 support for Qwen3-VL/Qwen3.5 MoE vs a DeepSeekV4 weight conversion question." }, { "left": "issue:41669", "right": "issue:46218", "accept": false, - "reason": "Import-speed regression from import * is not the same as an actual module import failure for Qwen3_5ForCausalLM." + "reason": "Import-related, but one is import-time slowdown and the other is a missing module import error." }, { "left": "issue:43408", "right": "issue:44315", "accept": false, - "reason": "Different subsystems and bugs: sam3 tracker/model-type warning vs Liger Kernel not applied with model_init." + "reason": "Different models and different failures: model-type mismatch warning vs Liger Kernel not applied." }, { "left": "issue:44246", "right": "issue:44273", "accept": false, - "reason": "Both mention import/loading, but one is intermittent import slowness and the other is lazy loading correctness." + "reason": "Both concern imports/loading, but one is intermittent slow import and the other is broken lazy loading." }, { "left": "issue:42890", "right": "issue:44908", "accept": false, - "reason": "Test seeding instability is unrelated to inverse_sqrt scheduler kwargs being ignored." + "reason": "Test flakiness from missing set_seed calls is unrelated to scheduler kwargs handling." }, { "left": "issue:43576", "right": "issue:44908", "accept": false, - "reason": "Different APIs and symptoms: env CLI breakage vs scheduler parameter plumbing." + "reason": "Different bugs in different features: env CLI vs inverse_sqrt scheduler kwargs." }, { "left": "issue:43352", "right": "issue:44393", "accept": false, - "reason": "Flash Attention 2.0 unsupported is a capability check issue, not a Qwen3-VL bbox hallucination bug." + "reason": "Flash-Attention-2 support error is unrelated to Qwen3-VL bbox output hallucination." }, { "left": "issue:43976", "right": "issue:44393", "accept": false, - "reason": "Python-version compatibility regression is unrelated to Qwen3-VL 2D bounding box output errors." + "reason": "Python-version compatibility regression is unrelated to the Qwen3-VL output bug." }, { "left": "issue:44908", "right": "issue:45071", "accept": false, - "reason": "Scheduler kwargs handling and PretrainedConfig typing are different code paths with no shared bug." + "reason": "Scheduler kwargs handling is unrelated to PretrainedConfig type checking." }, { "left": "issue:44295", "right": "issue:45230", "accept": false, - "reason": "Concrete position_ids buffer access error vs an empty generic bug report." + "reason": "Specific position_ids buffer error does not match a generic bug report." }, { "left": "issue:44297", "right": "issue:45230", "accept": false, - "reason": "Tokenizer save_pretrained metadata bug vs generic bug report; no common underlying defect." + "reason": "Tokenizer save_pretrained class mismatch is a distinct serializer bug, not the same as a generic report." }, { "left": "issue:44492", "right": "issue:46328", "accept": false, - "reason": "Cache strategy typo is unrelated to SequenceFeatureExtractor.pad performance behavior." + "reason": "Cache strategy typo and feature-extractor padding conversion are unrelated." }, { "left": "issue:44704", "right": "issue:44908", "accept": false, - "reason": "AutoProcessor cached_file kwargs plumbing is unrelated to inverse_sqrt scheduler kwargs." + "reason": "Both mention kwargs, but they affect different APIs and different code paths." }, { "left": "issue:41669", "right": "issue:44351", "accept": false, - "reason": "Import * performance regression is not the same as a HybridCache import failure." + "reason": "Import performance regression is not the same as a missing HybridCache symbol." }, { "left": "issue:41669", "right": "issue:45341", "accept": false, - "reason": "Different areas entirely: model import performance vs a testing_utils bug." + "reason": "No substantive overlap between import-star performance and a testing_utils bug." }, { "left": "issue:43976", "right": "issue:45850", "accept": false, - "reason": "Python 3.9+ compatibility issue is broader but not the same bug as the generic v5 breakage complaint." + "reason": "A Python-version support issue is broader regression noise, not the same underlying bug." }, { "left": "issue:43976", "right": "issue:46218", "accept": false, - "reason": "Python-version incompatibility and a specific missing import are different failure modes." + "reason": "Python support and missing module import errors are different failure classes." }, { "left": "issue:44077", "right": "issue:44861", "accept": false, - "reason": "PatchTSMixer post_init policy and tied-weight-key list handling are unrelated implementation bugs." + "reason": "Optional post_init validation and tied-weights list handling are unrelated." }, { "left": "issue:31795", "right": "issue:33453", "accept": false, - "reason": "Documentation confusion about forward args does not match a tokenizer loading regression." + "reason": "Documentation confusion and tokenizer loading regression are different issues." }, { "left": "issue:43576", "right": "issue:44273", "accept": false, - "reason": "CLI env command breakage and lazy loading correctness are different issues." + "reason": "CLI env command breakage and lazy loading failure only overlap at a very high level." }, { "left": "issue:44315", "right": "issue:44829", "accept": false, - "reason": "Liger Kernel application with model_init is unrelated to flash_attention_3 causing degenerate training." + "reason": "Liger Kernel application in model_init is unrelated to flash_attention_3 training degeneration." }, { "left": "issue:43519", "right": "issue:46655", "accept": false, - "reason": "Both are processor-related, but timestamp calculation and video padding calculation are different concrete bugs." + "reason": "Both are processor-related, but timestamp calculation and video padding are distinct bugs." } ] }, @@ -18770,14 +18728,14 @@ "issue:43298|issue:44230", "issue:43717|issue:46144" ], - "prepared_review_unit_hash": "253b2b0dac0d01691ebb84e596776e43eb84364856bbed982896e2256583f642", + "prepared_review_unit_hash": "cb5a2f096d7f4f61484a67a85256ca265bb0b959728ccd57dfc03fcba51c539c", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 48, @@ -18792,264 +18750,270 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is highly heterogeneous with no strong duplicate group overall. The only clear recurring theme is Flash Attention 2.0 support requests, but the rest are mostly unrelated model-loading, tokenizer, scheduling, import, and training issues. I would anchor the cluster on the broad Flash Attention 2.0 issue, but keep nearly all soft links rejected.", - "confidence": 0.82, - "canonical_issue_reason": "issue:45250 is the broadest, most central Flash Attention 2.0 request in the cluster and best captures the small recurring theme shared by the other FA2-related issues.", + "summary": "This cluster is mostly heterogeneous: the soft matches share broad themes, but none look like the same concrete bug or change. I would not merge any of the candidate pairs; 42913 is the best representative only because it is the clearest broad tokenizer-v5 regression report.", + "confidence": 0.79, + "canonical_issue_reason": "42913 is the broadest and clearest issue in the set, and the only plausible near-match is 43066; it is the best anchor if this cluster must be represented by one issue.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45250 is the best representative issue because it is generic enough to subsume model-specific FA2 support requests, unlike the more narrow GPT2/Nemotron variants.", + "best_issue_reason": "42913 is the most descriptive user-facing report and the best general representative among otherwise unrelated issues.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44360", "right": "issue:45468", "accept": false, - "reason": "Different topics: DSA indexer/ReLU vs Gemma-4 audio positional encoding." + "reason": "Different model/component areas: DSA indexer vs Gemma-4 audio positional encoding." }, { "left": "issue:44273", "right": "issue:44908", "accept": false, - "reason": "Lazy loading failure and inverse_sqrt scheduler kwargs are unrelated bugs." + "reason": "Lazy loading versus scheduler kwargs are unrelated failure modes." }, { "left": "issue:45230", "right": "issue:45310", "accept": false, - "reason": "Generic bug report vs a specific Qwen3.5 MoE from_pretrained regression." + "reason": "Generic bug report does not match a specific Qwen3.5 MoE from_pretrained regression." }, { "left": "issue:44655", "right": "issue:44861", "accept": false, - "reason": "Saving pipeline objects and tied-weight key handling are different code paths." + "reason": "Saving Pipeline objects and tied-weight key handling are different code paths." }, { "left": "issue:45468", "right": "issue:46894", "accept": false, - "reason": "Gemma-4 audio positional encoding and CohereASR training-loss shifting are unrelated." + "reason": "Different model families and symptoms: Gemma-4 audio encoding vs CohereASR training loss shift." }, { "left": "issue:43519", "right": "issue:44485", "accept": false, - "reason": "Timestamp calculation in Qwen3VL is not the same as GLM-5 RoPE implementation." + "reason": "Qwen3VL timestamp math and GLM-5 RoPE implementation are distinct bugs." }, { "left": "issue:43576", "right": "issue:46218", "accept": false, - "reason": "Broken env command and missing Qwen3_5 import are different failures." + "reason": "Transformers env CLI breakage is unrelated to a Qwen3_5 import failure." }, { "left": "issue:43519", "right": "issue:44908", "accept": false, - "reason": "Processor timestamp math and scheduler kwargs handling do not match." + "reason": "Processor timestamp calculation and inverse_sqrt scheduler kwargs do not share a code path." }, { "left": "issue:41669", "right": "issue:43704", "accept": false, - "reason": "Import-star cleanup and VRAM leak in Qwen3ForCausalLM are unrelated." + "reason": "Import-star cleanup/perf issue is unrelated to a VRAM leak in Qwen3ForCausalLM." }, { "left": "issue:31795", "right": "issue:34689", "accept": false, - "reason": "Documentation confusion around inputs/past_key_values is not a model-loading breakage." + "reason": "Documentation confusion about inputs differs from a Llama 3.2 model-loading regression." }, { "left": "issue:44485", "right": "issue:45468", "accept": false, - "reason": "Both are model-related, but they target different implementations and bugs." + "reason": "GLM-5 RoPE math and Gemma-4 audio positional encoding are separate implementation problems." }, { "left": "issue:43704", "right": "issue:44485", "accept": false, - "reason": "VRAM leak and RoPE implementation issue are different concrete problems." + "reason": "VRAM leak in Qwen3ForCausalLM is not the same as GLM-5 RoPE behavior." }, { "left": "issue:46018", "right": "issue:46172", "accept": false, - "reason": "DeepSeek-V4 gating logic and a comb transpose/dtype issue are not the same bug." + "reason": "DeepSeek shared-expert gating and a comb transpose/dtype bug are different concrete issues." }, { "left": "issue:43502", "right": "issue:43519", "accept": false, - "reason": "Unexpected network requests with local_files_only=True is unrelated to timestamp calculation." + "reason": "Local-files-only network access and Qwen3VL timestamp calculation are unrelated." }, { "left": "issue:42371", "right": "issue:43704", "accept": false, - "reason": "TF32 API guidance and Qwen3 memory leak are unrelated." + "reason": "TF32 settings and a dataloader-thread VRAM leak are not the same bug." }, { "left": "issue:45850", "right": "issue:46218", "accept": false, - "reason": "Broad 5.0 breakage complaint and a specific import error are not the same issue." + "reason": "A broad v5 breakage complaint does not specifically match a Qwen3_5 import error." }, { "left": "issue:43519", "right": "issue:44623", "accept": false, - "reason": "Processor timestamp math and missing save_pretrained files are different bugs." + "reason": "Timestamp computation and missing files in processor.save_pretrained are different failures." }, { "left": "issue:44263", "right": "issue:45468", "accept": false, - "reason": "torch.split return values in GlmMoeDsaIndexer is unrelated to Gemma-4 audio encoding." + "reason": "torch.split return handling in GlmMoeDsaIndexer is unrelated to Gemma-4 audio encoding." }, { "left": "issue:43606", "right": "issue:45468", "accept": false, - "reason": "CPU offload device mismatch and audio positional encoding are different code paths." + "reason": "CPU offload device mismatch in bark-small is unrelated to Gemma-4 audio positional encoding." }, { "left": "issue:43298", "right": "issue:46095", "accept": false, - "reason": "Continuous batching support for audio models and DeepSeek-V4 weight conversion are unrelated." + "reason": "Continuous batching support for audio models and DeepSeek-V4 weight conversion are not the same change." }, { "left": "issue:43519", "right": "issue:43704", "accept": false, - "reason": "Timestamp calculation bug and VRAM leak are different failures." + "reason": "Processor timestamp math and VRAM leakage are unrelated symptoms." }, { "left": "issue:44351", "right": "issue:46218", "accept": false, - "reason": "Generic HybridCache import failure and Qwen3_5 module import failure are not the same concrete problem." + "reason": "Both are import errors, but they are missing different symbols from different modules." }, { "left": "issue:43352", "right": "issue:46218", "accept": false, - "reason": "Missing Flash Attention support for Nemotron and Qwen3_5 import error are unrelated." + "reason": "Flash Attention 2.0 support for Nemotron is unrelated to importing Qwen3_5ForCausalLM." }, { "left": "issue:42913", "right": "issue:43066", "accept": false, - "reason": "Both involve tokenizer v5 behavior, but the reports are too broad vs a specific decoder-type bug to treat as duplicates." + "reason": "Both concern tokenizer v5 behavior, but one is a broad regression report and the other is a specific decoder-type bug; not clearly the same concrete issue." }, { "left": "issue:28348", "right": "issue:45250", "accept": false, - "reason": "Both mention Flash Attention 2.0, but one is GPT2LMHeadModel-specific while the other is a generic request." + "reason": "Specific GPT2LMHeadModel Flash Attention support request is not the same as a generic Flash Attention 2.0 topic." }, { "left": "issue:44908", "right": "issue:44998", "accept": false, - "reason": "Scheduler kwargs handling and the unrelated 'Unemployment' issue do not match." + "reason": "Scheduler kwargs and an unrelated low-signal issue title share no concrete bug." }, { "left": "issue:44492", "right": "issue:44998", "accept": false, - "reason": "A cache-strategy typo is unrelated to the nonsensical 'Unemployment' report." + "reason": "A cache-strategy typo and a generic low-signal issue are not duplicates." }, { "left": "issue:43618", "right": "issue:44998", "accept": false, - "reason": "CLIPOutput attentions assignment and the unrelated issue are not duplicates." + "reason": "CLIPOutput attentions assignment and a generic low-signal issue are unrelated." + }, + { + "left": "issue:44998", + "right": "issue:45850", + "accept": false, + "reason": "A generic low-signal issue is not the same as broad v5 breakage." }, { "left": "issue:44938", "right": "issue:44998", "accept": false, - "reason": "Python 3.14 loading failure is unrelated to the other issue." + "reason": "Python 3.14 loading failure is unrelated to a generic low-signal issue." }, { "left": "issue:44861", "right": "issue:45290", "accept": false, - "reason": "Tied weight key crashes and chat template crashes are different bugs." + "reason": "Tied-weight key handling and chat-template tool-call crashes are different code paths." }, { "left": "issue:31515", "right": "issue:38617", "accept": false, - "reason": "Slow from_pretrained loading and an import error are different problem types." + "reason": "Checkpoint load slowness and missing layer_type_validation import are unrelated." }, { "left": "issue:43572", "right": "issue:44861", "accept": false, - "reason": "Missing pad_token_idx in a config and tied-weight key handling are unrelated." + "reason": "Missing pad_token_idx and tied-weight key crashes are different configuration/model-save problems." }, { "left": "issue:43525", "right": "issue:44861", "accept": false, - "reason": "Llama4Config pad_token_id absence is not the same as tied-weight key crashes." + "reason": "Llama4Config missing pad_token_id is unrelated to tied-weight key list handling." }, { "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 files are different problems." }, { "left": "issue:28348", "right": "issue:43352", "accept": false, - "reason": "Both are Flash Attention 2.0 related, but they target different models and feature gaps." + "reason": "Both mention Flash Attention 2.0, but they target different models and likely different fixes." }, { "left": "issue:29870", "right": "issue:31515", "accept": false, - "reason": "BART generate decoder_attention_mask failure and slow checkpoint loading are unrelated." + "reason": "BART generate() decoder_attention_mask bug is unrelated to from_pretrained checkpoint slowness." }, { "left": "issue:30333", "right": "issue:31515", "accept": false, - "reason": "MLFlow reporting on failed jobs is unrelated to checkpoint load speed." + "reason": "MLFlow reporting on job failure and checkpoint loading performance are different issues." }, { "left": "issue:31515", "right": "issue:36296", "accept": false, - "reason": "Slow from_pretrained loading and tensor parallel training bug are different issues." + "reason": "from_pretrained slowness and tensor-parallel training bugs are not the same bug." }, { "left": "issue:31515", "right": "issue:31795", "accept": false, - "reason": "Load performance and forward documentation confusion are not the same underlying bug." + "reason": "Checkpoint loading performance and forward-argument documentation are unrelated." }, { "left": "issue:44230", "right": "issue:45310", "accept": false, - "reason": "FP8 support request and Qwen3.5 MoE from_pretrained error are different concerns." + "reason": "FP8 inference support and Qwen3.5 MoE from_pretrained failure are different feature/breakage reports." }, { "left": "issue:43298", "right": "issue:44230", "accept": false, - "reason": "Continuous batching for audio models and FP8 support are unrelated feature requests." + "reason": "Audio continuous batching support and FP8 inference support are separate feature requests." }, { "left": "issue:43717", "right": "issue:46144", "accept": false, - "reason": "Mamba init_weights discussion and RoFormer attention-interface issue are unrelated." + "reason": "Mamba init_weights usage and RoFormer attention-interface usage are unrelated implementation concerns." } ] }, @@ -19095,20 +19059,20 @@ "issue:37934|issue:43668", "issue:42936|issue:43668" ], - "prepared_review_unit_hash": "a4d4fce4710b6f1a2df604e30614f7e6db1f24e3ff84d0bb9205714972b1be6d", + "prepared_review_unit_hash": "770c8c78cc8e52be1473494a44abd6df8294f402b453d710733f3c6349d9f14d", "component_budget": { "node_count": 506, "item_count": 506, - "soft_pair_count": 1860, - "serialized_chars": 965072, - "estimated_input_tokens": 241268, - "estimated_eval_tokens": 482792 + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 }, "budget": { "node_count": 8, "item_count": 8, "soft_pair_count": 11, - "serialized_chars": 7773, + "serialized_chars": 7775, "estimated_input_tokens": 1944, "estimated_eval_tokens": 4144 }, @@ -19117,78 +19081,78 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These issues are not duplicates of one another; they span different models, configs, tokenizer warnings, and attention/loss code paths. No soft edge looks like the same underlying bug or change.", + "summary": "This cluster is not a duplicate set; it contains several unrelated bugs and one type-hint/typo issue across different models and subsystems. No single canonical issue or PR stands out.", "confidence": 0.97, - "canonical_issue_reason": "No single issue subsumes the others; each report targets a different model/component and a different defect type.", + "canonical_issue_reason": "No canonical issue: the items span unrelated topics (norm behavior, ignore_value handling, kwargs propagation, type hints, typos, tokenizer warning, DeepSeek gating, RoFormer attention), so they are not the same underlying bug.", "canonical_pr_reason": null, - "best_issue_reason": "No best issue for deduplication purposes, since the set is heterogeneous rather than one bug cluster.", + "best_issue_reason": "None is a strong global representative because the cluster is heterogeneous; if forced, issue 42936 is a concrete self-contained bug report, but it does not unify the set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:42936", "right": "issue:46018", "accept": false, - "reason": "Different models and failures: Mask2Former ignore_value handling vs DeepSeek-V4 shared expert gating." + "reason": "Different models and different bugs: Mask2Former ignore_value handling vs DeepSeek-V4 shared expert gating." }, { "left": "issue:43668", "right": "issue:44031", "accept": false, - "reason": "Unrelated bug classes: ModernBERTConfig type hint vs tokenizer regex warning regression." + "reason": "Unrelated problems: a ModernBERT config type hint vs tokenizer regex warning behavior." }, { "left": "issue:42936", "right": "issue:43240", "accept": false, - "reason": "Different code paths and symptoms: Mask2Former ignore_value usage vs kwargs propagation in loss calculation." + "reason": "Mask2Former ignore_value usage is unrelated to kwargs not being forwarded to a loss function." }, { "left": "issue:43240", "right": "issue:46144", "accept": false, - "reason": "Different subsystems: loss kwargs forwarding vs RoFormer attention interface usage." + "reason": "Different code paths: loss-function kwargs propagation vs RoFormer attention interface usage." }, { "left": "issue:42936", "right": "issue:44031", "accept": false, - "reason": "Mask2Former loss-argument behavior is unrelated to tokenizer regex warnings." + "reason": "Different subsystems and symptoms: segmentation loss config vs tokenizer warning regression." }, { "left": "issue:42936", "right": "issue:46144", "accept": false, - "reason": "Different models and defects: Mask2Former ignore_value vs RoFormer attention implementation." + "reason": "No shared underlying bug; these concern different models and different implementation layers." }, { "left": "issue:42936", "right": "issue:43813", "accept": false, - "reason": "A semantic bug in Mask2Former is not the same as a typo in a quantization attribute name." + "reason": "Mask2Former loss handling and a typo in a quantization attribute are unrelated." }, { "left": "issue:43668", "right": "issue:43813", "accept": false, - "reason": "Different areas entirely: config type annotation vs quantization typo." + "reason": "A type-hint bug in ModernBERTConfig is not the same as a typo in quantization code." }, { "left": "issue:43240", "right": "issue:43668", "accept": false, - "reason": "Loss-function kwargs forwarding is unrelated to a config type-hint correction." + "reason": "Loss kwargs forwarding and an incorrect config type hint are different kinds of issues." }, { "left": "issue:37934", "right": "issue:43668", "accept": false, - "reason": "Llama4TextL2Norm RMS-norm question is a model implementation concern, not a ModernBERT config typing issue." + "reason": "Llama4TextL2Norm/RMS norm question is unrelated to ModernBERTConfig norm_eps typing." }, { "left": "issue:42936", "right": "issue:43668", "accept": false, - "reason": "Mask2Former ignore_value behavior and ModernBERTConfig norm_eps typing are unrelated defects." + "reason": "Different models and bug classes; no shared concrete code-path problem." } ] }, @@ -19221,12 +19185,12 @@ "issue:10519|issue:11005", "issue:11005|issue:11307" ], - "prepared_review_unit_hash": "df2a8f85bb99f1beae3927f9b8d79049f68a89bfef711ece07fbea7e0eb70175", + "prepared_review_unit_hash": "c2fe418a88ac6a3a9dcbacc242f797d6255c0904363002830a5e08449378fc23", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 2876, + "serialized_chars": 2875, "estimated_input_tokens": 719, "estimated_eval_tokens": 1694 }, @@ -19234,7 +19198,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 2876, + "serialized_chars": 2875, "estimated_input_tokens": 719, "estimated_eval_tokens": 1694 }, @@ -19243,30 +19207,30 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These three issues are not duplicates; they cover unrelated topics in tokenization truncation, learning-rate scheduling, and Wav2Vec2 word offsets. The similarity scores appear to be superficial rather than describing the same underlying bug or feature request.", + "summary": "These three closed issues are unrelated: tokenization truncation behavior, a ReduceLROnPlateau-style feature request, and Wav2Vec2 word boundary timestamps. They should not be merged as duplicates.", "confidence": 0.98, - "canonical_issue_reason": null, + "canonical_issue_reason": "No canonical issue: the items describe different features/requests in different parts of the library.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 10519 is the most concrete and self-contained request, but it is not a canonical duplicate target for the others.", + "best_issue_reason": "If a representative had to be chosen, issue 10519 is the most technically specific and discussion-heavy, but it still does not represent the others.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:10519", "right": "issue:11307", "accept": false, - "reason": "Truncation-direction option and Wav2Vec2 word boundary offsets are unrelated features in different components." + "reason": "Different problem domains: tokenizer truncation direction vs. Wav2Vec2 time offsets." }, { "left": "issue:10519", "right": "issue:11005", "accept": false, - "reason": "Tokenizer truncation behavior and ReduceLROnPlateau-like functionality address completely different APIs and problems." + "reason": "Unrelated feature requests: truncation option vs. ReduceLROnPlateau-like scheduling behavior." }, { "left": "issue:11005", "right": "issue:11307", "accept": false, - "reason": "Learning-rate scheduler functionality and speech-model word time offsets do not describe the same bug or change." + "reason": "No shared underlying bug or change; one is about learning-rate scheduling, the other about word timestamp extraction." } ] }, @@ -19298,12 +19262,12 @@ "issue:13244|issue:16618", "issue:15980|issue:16618" ], - "prepared_review_unit_hash": "478f0913acffed45e9d572187ad0b693c27538f28e0220ad0cf846d83e433809", + "prepared_review_unit_hash": "95d5c4a94b8695c0ce6a23d989c0acc670b337bb928a3807fe58c2040b3278ee", "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 }, @@ -19311,7 +19275,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 2478, + "serialized_chars": 2477, "estimated_input_tokens": 620, "estimated_eval_tokens": 1496 }, @@ -19320,24 +19284,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These three issues do not appear to describe the same underlying bug. #13244 is about TAPAS tokenization differing from TensorFlow behavior, #15980 is about a misleading auth/download error for private models, and #16618 is a model-loading failure for bert-base-uncased. The soft similarities are too broad to treat as duplicates.", - "confidence": 0.94, - "canonical_issue_reason": "No strong canonical issue emerges because the reports cover different failure modes and subsystems; if one must be chosen, #16618 is the broadest/generalized user-facing loading failure.", + "summary": "The cluster is loosely about model loading/tokenization, but the issues describe different problems: one tokenizer discrepancy, one private-model auth/error-message issue, and one generic model-loading failure. I do not see a true duplicate set here.", + "confidence": 0.82, + "canonical_issue_reason": "Issue #16618 is the broadest model-loading report, but the cluster is not a single duplicate bug because #15980 is specifically about private-model authentication and #13244 is about Tapas tokenization behavior.", "canonical_pr_reason": null, - "best_issue_reason": "#16618 is the most general and user-facing of the three, but it still does not subsume the others; it is only the best representative by breadth, not a duplicate target.", + "best_issue_reason": "#16618 is the best representative of the cluster\u2019s shared theme of model loading, but it is still not a clean canonical duplicate for the other issues.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:13244", "right": "issue:16618", "accept": false, - "reason": "Different problems: tokenizer output mismatch for TAPAS versus a model-loading failure for bert-base-uncased." + "reason": "Tokenizer mismatch for Tapas is a different underlying problem from a failure to load bert-base-uncased." }, { "left": "issue:15980", "right": "issue:16618", "accept": false, - "reason": "Different problems: misleading private-model auth/download error versus failure to load a public model." + "reason": "Private-model download/auth error is not the same bug as a generic inability to load a public model." } ] }, @@ -19367,12 +19331,12 @@ "soft_pairs": [ "issue:26350|issue:33260" ], - "prepared_review_unit_hash": "0678c993e311c24e8c4fef1daa2bca7744ae93920a82fedc309996c75f153728", + "prepared_review_unit_hash": "9a3de37a3e401568baec5273f2b8c33e879ec84080137e0b9ce86cec111db10f", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1590, + "serialized_chars": 1589, "estimated_input_tokens": 398, "estimated_eval_tokens": 1052 }, @@ -19380,7 +19344,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1590, + "serialized_chars": 1589, "estimated_input_tokens": 398, "estimated_eval_tokens": 1052 }, @@ -19389,18 +19353,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two unrelated feature/community-contribution issues: one for Flash Attention 2 support on more architectures, the other for GGUF support on more architectures. They share a vague theme of broader architecture support but not the same underlying work item.", - "confidence": 0.98, + "summary": "The two issues are both community contribution requests, but they target different feature areas: Flash Attention 2 support vs. GGUF support. They are not duplicates and should remain separate.", + "confidence": 0.96, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:26350 is the more established thread by discussion activity and age, but it is not a duplicate of issue:33260.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:26350", "right": "issue:33260", "accept": false, - "reason": "Different support efforts with different targets (Flash Attention 2 vs GGUF); the shared 'more architectures' wording is too broad to indicate the same underlying issue." + "reason": "Different underlying feature requests with different target architectures/format support; they do not describe the same bug or change." } ] }, @@ -19465,217 +19429,217 @@ "issue:46222|issue:46931", "issue:46226|issue:46931" ], - "prepared_review_unit_hash": "df1f345e288b0c8f9b8c1dd62c2877e8337ffdc95b71b983ad329cc3a4e8a019", + "prepared_review_unit_hash": "248e905d114ad4cb2b28f89b24b0c4f5a08e4c1c14ad3ab0d9a12b4d0cb6f72e", "component_budget": { "node_count": 9, "item_count": 9, "soft_pair_count": 29, - "serialized_chars": 15981, - "estimated_input_tokens": 3996, - "estimated_eval_tokens": 8248 + "serialized_chars": 15977, + "estimated_input_tokens": 3995, + "estimated_eval_tokens": 8246 }, "budget": { "node_count": 9, "item_count": 9, "soft_pair_count": 29, - "serialized_chars": 15981, - "estimated_input_tokens": 3996, - "estimated_eval_tokens": 8248 + "serialized_chars": 15977, + "estimated_input_tokens": 3995, + "estimated_eval_tokens": 8246 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster has one clear exact duplicate pair (issue:46222 / issue:46226) and several additional i18n translation issues with very high template similarity, including repeated placeholder-title entries, but the rest do not have enough evidence to safely merge as duplicates.", - "confidence": 0.81, - "canonical_issue_reason": "issue:46222 is the best canonical issue because it forms an exact duplicate pair with issue:46226, is still open, and already has inbound references; the other near-identical i18n issues are only template-similar, not proven duplicates.", + "summary": "Most items are i18n docs-translation requests, but they split into separate language-specific issues rather than one duplicate. The only clear duplicate sets are the three generic placeholder-title issues (43834/44365/44370) and the identical Thai pair (46222/46226).", + "confidence": 0.68, + "canonical_issue_reason": "Issue 43834 is the best representative of the largest exact-duplicate subgroup: it matches the other placeholder-title translation issues and is the most generic/central of that set.", "canonical_pr_reason": null, - "best_issue_reason": "issue:46222 is the best representative issue for the exact Thai duplicate pair and the most suitable anchor for triage because it is open and already referenced.", + "best_issue_reason": "Among the available issues, 43834 is the cleanest representative of a repeated placeholder translation request; it sits in the densest duplicate pocket and is the most canonical-looking issue here.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:27088", "right": "issue:44370", "accept": false, - "reason": "Different language-specific translation requests; the similarity is just the shared docs-translation template." + "reason": "Different target language/request; not the same concrete translation issue." }, { "left": "issue:27088", "right": "issue:44365", "accept": false, - "reason": "Different language-specific translation requests; not the same underlying issue." + "reason": "Different target language/request; not a duplicate." }, { "left": "issue:27088", "right": "issue:43834", "accept": false, - "reason": "Turkish request vs a generic placeholder translation issue; template similarity is not enough." + "reason": "Turkish vs generic placeholder translation request; not the same issue." }, { "left": "issue:27088", "right": "issue:38435", "accept": false, - "reason": "Turkish vs Romanian translation request; distinct issues." + "reason": "Different language translation requests; separate issues." }, { "left": "issue:27088", "right": "issue:28936", "accept": false, - "reason": "Turkish vs Spanish translation request; separate language-specific work." + "reason": "Turkish vs Spanish translation request; not a duplicate." }, { "left": "issue:28936", "right": "issue:43834", "accept": false, - "reason": "Spanish request vs generic placeholder issue; high similarity is template-driven only." + "reason": "Spanish request is distinct from the generic placeholder translation issue." }, { "left": "issue:28936", "right": "issue:44365", "accept": false, - "reason": "Spanish request vs generic placeholder issue; not enough evidence of the same concrete request." + "reason": "Spanish vs generic placeholder request; not the same underlying issue." }, { "left": "issue:28936", "right": "issue:38435", "accept": false, - "reason": "Spanish vs Romanian translation request; different underlying issues." + "reason": "Different languages, so these are separate translation requests." }, { "left": "issue:28936", "right": "issue:44370", "accept": false, - "reason": "Spanish request vs generic placeholder issue; only superficial similarity." + "reason": "Spanish vs generic placeholder request; not a duplicate." }, { "left": "issue:38435", "right": "issue:44370", "accept": false, - "reason": "Romanian request vs generic placeholder translation issue; body similarity appears template-level." + "reason": "Romanian request is distinct from the generic placeholder issue." }, { "left": "issue:38435", "right": "issue:44365", "accept": false, - "reason": "Romanian request vs generic placeholder issue; distinct translation targets." + "reason": "Romanian vs generic placeholder translation request; not the same issue." }, { "left": "issue:38435", "right": "issue:43834", "accept": false, - "reason": "Romanian request vs generic placeholder issue; not the same underlying request." + "reason": "Romanian request is separate from the placeholder-title subgroup." }, { "left": "issue:38435", "right": "issue:46222", "accept": false, - "reason": "Romanian translation issue vs Thai-titled issue; different requests." + "reason": "Different language/request and different title; not a duplicate." }, { "left": "issue:38435", "right": "issue:46226", "accept": false, - "reason": "Romanian translation issue vs Thai-titled issue; different requests." + "reason": "Different language/request and different title; not a duplicate." }, { "left": "issue:43834", "right": "issue:44370", - "accept": false, - "reason": "Both are template-like i18n issues, but there is no concrete evidence they target the same language or request." + "accept": true, + "reason": "Same generic placeholder translation request; looks like an exact duplicate." }, { "left": "issue:43834", "right": "issue:44365", - "accept": false, - "reason": "Generic placeholder translation issues with similar bodies, but not enough to call them duplicates." + "accept": true, + "reason": "Same placeholder-title issue with no meaningful difference." }, { "left": "issue:43834", "right": "issue:46222", "accept": false, - "reason": "Generic placeholder translation issue vs a different Thai-titled issue; no same underlying bug/change evidence." + "reason": "Different title/content and not clearly the same translation request." }, { "left": "issue:43834", "right": "issue:46226", "accept": false, - "reason": "Generic placeholder translation issue vs a different Thai-titled issue; not clearly the same request." + "reason": "Different title/content and not clearly the same translation request." }, { "left": "issue:43834", "right": "issue:46931", "accept": false, - "reason": "Different Thai content/title; similarity is too broad to treat as a duplicate." + "reason": "Different title and likely different underlying request; too weak to merge." }, { "left": "issue:44365", "right": "issue:44370", - "accept": false, - "reason": "Near-identical template text, but the actual language target is not identifiable as the same issue." + "accept": true, + "reason": "Same placeholder-title translation issue; effectively duplicate." }, { "left": "issue:44365", "right": "issue:46222", "accept": false, - "reason": "Generic i18n template issue vs Thai-titled issue; not enough to say same underlying request." + "reason": "Different title/content; not enough to treat as the same issue." }, { "left": "issue:44365", "right": "issue:46226", "accept": false, - "reason": "Generic i18n template issue vs Thai-titled issue; only superficial similarity." + "reason": "Different title/content; not the same concrete issue." }, { "left": "issue:44365", "right": "issue:46931", "accept": false, - "reason": "Different Thai content/title; not the same issue." + "reason": "Different title and likely different request; reject." }, { "left": "issue:44370", "right": "issue:46222", "accept": false, - "reason": "Generic template issue vs Thai-titled issue; no concrete duplicate evidence." + "reason": "Different title/content; not a safe duplicate match." }, { "left": "issue:44370", "right": "issue:46226", "accept": false, - "reason": "Generic template issue vs Thai-titled issue; similarity is not enough." + "reason": "Different title/content; not the same issue." }, { "left": "issue:44370", "right": "issue:46931", "accept": false, - "reason": "Different Thai content/title; does not look like the same underlying request." + "reason": "Different title and underlying request; reject." }, { "left": "issue:46222", "right": "issue:46226", "accept": true, - "reason": "Same exact Thai title and matching body size, created one day apart; this is the clear duplicate pair." + "reason": "Identical Thai-titled issue; clear duplicate." }, { "left": "issue:46222", "right": "issue:46931", "accept": false, - "reason": "Different Thai content/title; it does not appear to be the same request as the exact duplicate pair." + "reason": "Similar language/theme but different title and not clearly the same concrete issue." }, { "left": "issue:46226", "right": "issue:46931", "accept": false, - "reason": "Different Thai content/title; not the same underlying issue as the exact duplicate pair." + "reason": "Different title and not enough evidence they are the same underlying request." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The Thai exact-duplicate call for issue:46222 / issue:46226 is well grounded, but the summary is too strong in saying this is the *only* clear duplicate pair. The packet also shows other issues with identical placeholder titles ([i18n-] Translating docs to ) and very high soft similarity, so dismissing all of them as merely template-level similarity overstates certainty. The soft-edge false verdicts are conservative, but the overall summary is not fully grounded." + "accept": true, + "feedback": "The summary is grounded in the packet: the only clearly exact-duplicate clusters are the three identical placeholder-title issues (43834/44365/44370) and the identical Thai pair (46222/46226). The remaining pair rejections are conservative given that the packet only shows soft similarity and no shared issue links or concrete same-bug evidence. The canonical issue choice for the larger exact-duplicate subgroup is reasonable." }, "evaluator_used": true, - "retried": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -19765,12 +19729,12 @@ "issue:42503|issue:46653", "issue:45027|issue:46653" ], - "prepared_review_unit_hash": "9ee21c6cb2cb50423011415959e9ebf74d72efd89b7f05e36d03bb1f077541cb", + "prepared_review_unit_hash": "e917d1021d3bca9f7af3891e01eee639061d5c948cdb4d7b1fe74eb939f54b0b", "component_budget": { "node_count": 21, "item_count": 21, "soft_pair_count": 48, - "serialized_chars": 28617, + "serialized_chars": 28618, "estimated_input_tokens": 7155, "estimated_eval_tokens": 14566 }, @@ -19778,7 +19742,7 @@ "node_count": 21, "item_count": 21, "soft_pair_count": 48, - "serialized_chars": 28617, + "serialized_chars": 28618, "estimated_input_tokens": 7155, "estimated_eval_tokens": 14566 }, @@ -19787,300 +19751,282 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are mostly unrelated model/feature request issues that only share broad wording like \"Add\" or model-family terms. I found no true duplicate pair among the soft candidates; the set is better treated as separate requests in the same general area.", - "confidence": 0.88, - "canonical_issue_reason": "issue:46133 is the strongest anchor only in the sense that it is a detailed, active open model-request thread; however, it is not a duplicate of the others and the cluster does not contain a true canonical duplicate.", + "summary": "This set is only loosely related by being model/architecture request issues; the titles span unrelated models and features, so there are no true duplicates here. All soft edges should be rejected.", + "confidence": 0.89, + "canonical_issue_reason": "Issue 36879 is the clearest representative of the group: a concrete, concise model-add request, and it sits near the center of the soft-similarity web. It is still only a loose representative, not a true duplicate target.", "canonical_pr_reason": null, - "best_issue_reason": "issue:46133 has the most complete request framing and discussion activity, so it is the best representative issue for this mixed cluster.", + "best_issue_reason": "36879 is the best issue anchor because it is specific, closed, and broadly representative of the 'add model support' pattern that links these items. However, the underlying requests are still distinct.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:36718", "right": "issue:47004", "accept": false, - "reason": "Similar keypoint-task wording, but RoMa and LoMa are different models and likely different implementations." + "reason": "RoMa keypoint matcher vs LoMa keypoint detector/matcher: related naming, but different models and different requested changes." }, { "left": "issue:36718", "right": "issue:41115", "accept": false, - "reason": "Different domains: keypoint matching vs MiniCPM3 model support." + "reason": "Keypoint matcher request vs MiniCPM3 architecture request; unrelated model families and code paths." }, { "left": "issue:36817", "right": "issue:41211", "accept": false, - "reason": "Different model requests with no shared concrete change." + "reason": "EuroBert config addition and DEIMv2 are different model requests, not the same change." }, { "left": "issue:36817", "right": "issue:43671", "accept": false, - "reason": "EuroBert config support and Qwen3-TTS support are unrelated." + "reason": "EuroBert config vs Qwen3-TTS support; different model types and implementation areas." }, { "left": "issue:36817", "right": "issue:36879", "accept": false, - "reason": "Different models and different request targets." + "reason": "EuroBert config and RF-DETR model support are unrelated model additions." }, { "left": "issue:36817", "right": "issue:42954", "accept": false, - "reason": "Mirko Privitera is not the same underlying model/config request as EuroBert." + "reason": "EuroBert model config vs a person-name issue; no shared underlying bug or feature." }, { "left": "issue:36879", "right": "issue:41211", "accept": false, - "reason": "RF-DETR and DEIMv2 are distinct model additions." + "reason": "RF-DETR and DEIMv2 are different model additions with different integration work." }, { "left": "issue:36879", "right": "issue:43909", "accept": false, - "reason": "Object detection model request vs audio model request." + "reason": "RF-DETR vision model vs LFM2.5 audio model; distinct artifacts and code paths." }, { "left": "issue:36879", "right": "issue:44121", "accept": false, - "reason": "Different request types: detector support vs weight-sparse transformer support." + "reason": "RF-DETR model support is unrelated to OpenAI weight-sparse transformer support." }, { "left": "issue:36879", "right": "issue:43837", "accept": false, - "reason": "RF-DETR and Qwen3-ASR are unrelated models and code paths." + "reason": "RF-DETR and Qwen3-ASR are different models and different tasks." }, { "left": "issue:36879", "right": "issue:43671", "accept": false, - "reason": "RF-DETR and Qwen3-TTS are unrelated." + "reason": "RF-DETR vision model vs Qwen3-TTS audio support; not the same change." }, { "left": "issue:37015", "right": "issue:43837", "accept": false, - "reason": "NeoBERT and Qwen3-ASR are different model families and features." + "reason": "NeoBERT addition and Qwen3-ASR support are different model requests." }, { "left": "issue:37015", "right": "issue:44121", "accept": false, - "reason": "NeoBERT request vs sparse-transformer request; no same bug/change." + "reason": "NeoBERT vs weight-sparse transformer support; unrelated architectures." }, { "left": "issue:37015", "right": "issue:43909", "accept": false, - "reason": "BERT model addition and audio model addition are unrelated." + "reason": "NeoBERT and LFM2.5 Audio address different model families." }, { "left": "issue:36879", "right": "issue:37015", "accept": false, - "reason": "RF-DETR and NeoBERT are different model-support issues." + "reason": "RF-DETR model support and NeoBERT support are separate model integrations." }, { "left": "issue:37015", "right": "issue:42503", "accept": false, - "reason": "NeoBERT and ModernVBERT are different model requests." + "reason": "NeoBERT and ModernVBERT are different model additions, not duplicate requests." }, { "left": "issue:41115", "right": "issue:43837", "accept": false, - "reason": "MiniCPM3 model architecture and Qwen3-ASR support are unrelated." + "reason": "MiniCPM3 architecture request vs Qwen3-ASR support; different models and tasks." }, { "left": "issue:41115", "right": "issue:44121", "accept": false, - "reason": "MiniCPM3 support and sparse transformer support are not the same change." + "reason": "MiniCPM3 architecture work is not the same as adding weight-sparse transformer support." }, { "left": "issue:41115", "right": "issue:42503", "accept": false, - "reason": "MiniCPM3 and ModernVBERT are different model requests." + "reason": "MiniCPM3 architecture and ModernVBERT models are separate model requests." }, { "left": "issue:41115", "right": "issue:45027", "accept": false, - "reason": "MiniCPM3 and Voxtral TTS are different models/modalities." + "reason": "MiniCPM3 vs Voxtral TTS support; different model families and code paths." }, { "left": "issue:41211", "right": "issue:43671", "accept": false, - "reason": "DEIMv2 and Qwen3-TTS are unrelated requests." + "reason": "DEIMv2 and Qwen3-TTS are unrelated model additions." }, { "left": "issue:42503", "right": "issue:43837", "accept": false, - "reason": "ModernVBERT and Qwen3-ASR are different model additions." + "reason": "ModernVBERT models and Qwen3-ASR support are different requested features." }, { "left": "issue:42503", "right": "issue:43671", "accept": false, - "reason": "ModernVBERT and Qwen3-TTS are unrelated." + "reason": "ModernVBERT and Qwen3-TTS are unrelated model integrations." }, { "left": "issue:42503", "right": "issue:42954", "accept": false, - "reason": "ModernVBERT models vs Mirko Privitera are not the same underlying change." + "reason": "ModernVBERT models and the Mirko Privitera issue are clearly unrelated." }, { "left": "issue:42738", "right": "issue:45027", "accept": false, - "reason": "BERT-like RoPE support is not the same as Voxtral TTS support." + "reason": "BERT-like models with RoPE and Voxtral TTS support are different model requests." }, { "left": "issue:42503", "right": "issue:42738", "accept": false, - "reason": "Related only at a broad BERT-family level; one asks for specific model support, the other for a RoPE architecture option." + "reason": "ModernVBERT models vs a general BERT-like RoPE request; related theme, but not the same change." }, { "left": "issue:42738", "right": "issue:42954", "accept": false, - "reason": "No shared concrete model or code path." + "reason": "BERT-like RoPE discussion and a person-name issue have no common code-path." }, { "left": "issue:42738", "right": "issue:46653", "accept": false, - "reason": "BERT-like RoPE support and Mirko Privitera are unrelated." + "reason": "BERT-like RoPE support is unrelated to the Mirko Privitera issue." }, { "left": "issue:42954", "right": "issue:44205", "accept": false, - "reason": "Mirko Privitera and SAM3-LiteText are different requests." + "reason": "Person-name issue vs SAM3-LiteText; no shared underlying request." }, { "left": "issue:42954", "right": "issue:43671", "accept": false, - "reason": "No same underlying model or feature." + "reason": "Mirko Privitera issue and Qwen3-TTS support are unrelated." }, { "left": "issue:42954", "right": "issue:45177", "accept": false, - "reason": "Mirko Privitera and DeepSeek-OCR2 are unrelated." + "reason": "Mirko Privitera issue vs DeepSeek-OCR2 request; not the same change." }, { "left": "issue:42954", "right": "issue:46653", "accept": false, - "reason": "No evidence these are the same request." + "reason": "Mirko Privitera issue and another person-name issue are not a code change duplicate." }, { "left": "issue:43671", "right": "issue:43837", "accept": false, - "reason": "Both are Qwen3 support requests, but TTS and ASR are different concrete model changes." + "reason": "Qwen3-TTS support and Qwen3-ASR support are different modalities and likely different implementations." }, { "left": "issue:43671", "right": "issue:45027", "accept": false, - "reason": "Different vendors/models and different modality support." + "reason": "Qwen3-TTS support and Voxtral-4B-TTS-2603 support are different model-specific requests." }, { "left": "issue:43671", "right": "issue:45177", "accept": false, - "reason": "Qwen3-TTS and DeepSeek-OCR2 are unrelated." + "reason": "Qwen3-TTS and DeepSeek-OCR2 are unrelated model additions." }, { "left": "issue:43671", "right": "issue:44121", "accept": false, - "reason": "TTS model support and sparse-transformer request are not the same change." + "reason": "Qwen3-TTS support is not the same as weight-sparse transformer support." }, { "left": "issue:43837", "right": "issue:43909", "accept": false, - "reason": "Qwen3-ASR and LFM2.5 Audio are different audio-model requests." + "reason": "Qwen3-ASR support and LFM2.5 Audio are different audio models with different integration work." }, { "left": "issue:43837", "right": "issue:45322", "accept": false, - "reason": "Qwen3-ASR support and EUPE support are unrelated." + "reason": "Qwen3-ASR support and EUPE are unrelated model requests." }, { "left": "issue:43837", "right": "issue:44121", "accept": false, - "reason": "Different target features: ASR model support vs sparse-transformer support." - }, - { - "left": "issue:43909", - "right": "issue:44205", - "accept": false, - "reason": "Audio model support and SAM3-LiteText are different." - }, - { - "left": "issue:43909", - "right": "issue:46645", - "accept": false, - "reason": "Audio model addition vs Granite SWA+sink option are unrelated." - }, - { - "left": "issue:44121", - "right": "issue:45322", - "accept": false, - "reason": "Sparse-transformer request and EUPE model request are different." + "reason": "ASR support and weight-sparse transformer support are different code paths." }, { "left": "issue:43837", "right": "issue:45027", "accept": false, - "reason": "ASR model support and Voxtral TTS support are distinct modalities/models." + "reason": "Qwen3-ASR and Voxtral-4B-TTS-2603 are different speech models and tasks." }, { "left": "issue:37015", "right": "issue:45027", "accept": false, - "reason": "NeoBERT and Voxtral TTS are unrelated." + "reason": "NeoBERT and Voxtral TTS are unrelated model requests." }, { "left": "issue:43837", "right": "issue:45177", "accept": false, - "reason": "ASR support and DeepSeek-OCR2 are different model additions." + "reason": "Qwen3-ASR support and DeepSeek-OCR2 are different model additions." }, { "left": "issue:45322", "right": "issue:46133", "accept": false, - "reason": "EUPE and TIPSv2 are different model requests with different architectures and goals." + "reason": "EUPE and TIPSv2 are separate research models with different integration surfaces." }, { "left": "issue:42503", "right": "issue:46653", "accept": false, - "reason": "ModernVBERT and Mirko Privitera are unrelated." + "reason": "ModernVBERT models and the Mirko Privitera issue are unrelated." }, { "left": "issue:45027", "right": "issue:46653", "accept": false, - "reason": "Voxtral TTS support and Mirko Privitera are not the same request." + "reason": "Voxtral TTS support and the Mirko Privitera issue are not the same underlying change." } ] }, @@ -20110,12 +20056,12 @@ "soft_pairs": [ "issue:38740|issue:46443" ], - "prepared_review_unit_hash": "b4074ca30683346e9e31eeab8ec756faea23bf3d98051315af3d1df5dfd29442", + "prepared_review_unit_hash": "75e35e3aee90439a253f2b98fd9c68800d8285105b6528cad84e0a2e15391ca1", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1506, + "serialized_chars": 1505, "estimated_input_tokens": 377, "estimated_eval_tokens": 1010 }, @@ -20123,7 +20069,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1506, + "serialized_chars": 1505, "estimated_input_tokens": 377, "estimated_eval_tokens": 1010 }, @@ -20132,18 +20078,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are not duplicates: one is a docs request for adding Pruna, the other is a Vision RoPE refactor. The soft similarity appears to be a false positive.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "These two issues are not duplicates: one is a docs request about adding Pruna as an optimization framework, while the other is a Vision RoPE refactor in model code. They concern different subsystems and different kinds of work.", + "confidence": 0.99, + "canonical_issue_reason": "Issue #46443 is the better canonical issue only because it is the more concrete technical change request; the other is a documentation-oriented enhancement and does not appear related.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 46443 is the more substantial technical item and the better anchor if one must be chosen, but there is no true canonical issue for this cluster because the topics are unrelated.", + "best_issue_reason": "#46443 is the best issue representative for the cluster because it describes an actual code refactor, whereas #38740 is a standalone docs item with no overlap.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:38740", "right": "issue:46443", "accept": false, - "reason": "Different domains and intents: documentation/framework listing vs a vision model RoPE code refactor. No evidence they describe the same underlying bug or change." + "reason": "The titles and context are unrelated: documentation for adding Pruna as an optimization framework versus a Vision RoPE refactor. No shared code path, filenames, or explicit target suggests the same underlying bug/change." } ] }, @@ -20173,12 +20119,12 @@ "soft_pairs": [ "issue:41084|issue:43037" ], - "prepared_review_unit_hash": "e0e92708d1d5a715c58c33dcf7df54688f3389bcb8e0e97480a157d8b34e8aea", + "prepared_review_unit_hash": "e5835d87d37ad645790103d308718457ce31f2af65018185dfda214a61cba6d3", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1480, + "serialized_chars": 1479, "estimated_input_tokens": 370, "estimated_eval_tokens": 996 }, @@ -20186,7 +20132,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1480, + "serialized_chars": 1479, "estimated_input_tokens": 370, "estimated_eval_tokens": 996 }, @@ -20195,18 +20141,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are not duplicates: one is a vague technical request for block decoding, the other is a feature/support request for DeepSeek v3.2. They do not appear to describe the same underlying problem.", - "confidence": 0.95, - "canonical_issue_reason": "No canonical issue: the titles and limited body signals point to different requests rather than duplicate reports.", + "summary": "These two issues are not duplicates: one is a generic request about block decoding, while the other is a feature request for DeepSeek v3.2 support. They reference different underlying problems and likely different implementation work.", + "confidence": 0.97, + "canonical_issue_reason": "Issue #41084 is the broader, more generic software-behavior request and is the better candidate to represent the cluster if forced to pick an issue.", "canonical_pr_reason": null, - "best_issue_reason": "Issue #43037 is the better representative of the cluster only in the sense that it is more specific, but it still does not match #41084 as a duplicate.", + "best_issue_reason": "#41084 is the more general issue and has clearer potential to subsume related block-decoding discussion; #43037 is model-specific and narrower.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:41084", "right": "issue:43037", "accept": false, - "reason": "The issues are about different topics: block decoding vs. DeepSeek v3.2 support. There is no evidence they refer to the same bug or change." + "reason": "Different topics and code paths: block decoding versus adding support for a specific model version. The titles and context do not indicate the same underlying bug or change." } ] }, @@ -20236,12 +20182,12 @@ "soft_pairs": [ "issue:41306|issue:41826" ], - "prepared_review_unit_hash": "11673b895be932dc3418972ed621398e6041910f0c6bf471a5fd1ed9f0ab8f47", + "prepared_review_unit_hash": "a944f1b5968727fb48b28f01f8504178cec4a2d207393d2caa6d2c841d9d8298", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1584, + "serialized_chars": 1583, "estimated_input_tokens": 396, "estimated_eval_tokens": 1048 }, @@ -20249,7 +20195,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1584, + "serialized_chars": 1583, "estimated_input_tokens": 396, "estimated_eval_tokens": 1048 }, @@ -20258,18 +20204,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These issues are about different Liger/Tiled model features: one requests a `from_pretrained()` flag for TiledFusedLogitsLoss/FLCE, the other requests TiledMLP integration for reduced memory footprint. They do not look like the same underlying bug or change.", - "confidence": 0.91, - "canonical_issue_reason": null, + "summary": "The two issues are both about memory-efficient tiled model components, but they describe different features: one is about a `from_pretrained()` flag for FLCE/LigerFusedLinearCrossEntropy, the other about integrating TiledMLP. They do not look like duplicates or the same underlying change.", + "confidence": 0.17, + "canonical_issue_reason": "Issue #41306 is the more concrete, feature-specific request and is easier to anchor as the representative item, but it is not actually the same problem as the other issue.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "#41306 is the more narrowly scoped and actionable issue, while #41826 is a separate enhancement request for a different tiled component.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:41306", "right": "issue:41826", "accept": false, - "reason": "Different requested changes: one is a loader flag for TiledFusedLogitsLoss/FLCE, the other is a separate TiledMLP memory-footprint feature request; no shared concrete code-path issue." + "reason": "Different components and different requested changes: FLCE/from_pretrained flag vs TiledMLP integration. They share a broad memory-efficiency theme only." } ] }, @@ -20299,7 +20245,7 @@ "soft_pairs": [ "issue:43598|issue:43600" ], - "prepared_review_unit_hash": "608682e9cd6797934d161e19111434db3a5bc8eb75c8cd822a445098f3164228", + "prepared_review_unit_hash": "6578ed3b63c377822c0b3b37d037b3f24dc3b3e0e072c7f0b75ce5815141eee5", "component_budget": { "node_count": 2, "item_count": 2, @@ -20321,18 +20267,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two closed issues both touch Trainer customization, but they describe different problems: one is about `remove_unused_column` behavior, the other about FLOPs estimation for custom trainers. They are related in subsystem, not the same underlying bug.", - "confidence": 0.78, - "canonical_issue_reason": "Issue #43598 is the clearer canonical item because it describes a concrete Trainer customization behavior that could subsume a broader design discussion.", + "summary": "These two issues are thematically related to Trainer custom trainers, but they describe different problems: one is about `remove_unused_column` customizability, the other about FLOPS estimation accounting. They should not be treated as duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue #43598 is the best representative for the cluster since it is more specific and actionable than the FLOPs-estimation request.", + "best_issue_reason": "Issue 43598 is the cleaner representative of the cluster theme because it is a concrete Trainer-behavior request, but it is not a duplicate of 43600.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43598", "right": "issue:43600", "accept": false, - "reason": "They are about different Trainer behaviors: column removal/customizability vs FLOPs counting for custom trainers. Related area, but not the same fix or bug." + "reason": "Related to Trainer extensibility, but they target different code paths and outcomes: column pruning customization vs FLOPS estimation for custom trainers." } ] }, @@ -20362,12 +20308,12 @@ "soft_pairs": [ "issue:43684|issue:44405" ], - "prepared_review_unit_hash": "75cd5a02b791308c754afdc853844360443940c37e4fe49e728bda902b431f8b", + "prepared_review_unit_hash": "601b9829fb9ab71f3910c5afaf3a465f815642cc928a8aded997a9141adb270c", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1612, + "serialized_chars": 1611, "estimated_input_tokens": 403, "estimated_eval_tokens": 1062 }, @@ -20375,7 +20321,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1612, + "serialized_chars": 1611, "estimated_input_tokens": 403, "estimated_eval_tokens": 1062 }, @@ -20384,18 +20330,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are about different Qwen model support gaps: one requests model registration for Qwen3-Omni in AutoModel/AutoModelForConditionalGeneration, while the other requests AutoModelForSequenceClassification support for Qwen3.5. They are related by model family, but not the same underlying bug or change.", - "confidence": 0.94, - "canonical_issue_reason": "Issue 43684 is the broader model-registration request and is the better representative of the cluster\u2019s general theme, though it is not a duplicate of 44405.", + "summary": "These are not duplicates: #43684 asks for Qwen3-Omni registration in AutoModel/AutoModelForConditionalGeneration, while #44405 requests AutoModelForSequenceClassification support for Qwen3.5 (Qwen3_5Config). They involve different model/config families and different auto-head behavior.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue for this pair; they target different model families and different feature surfaces.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43684 is the most general and central item here, but the cluster is not truly a duplicate set; 44405 is a separate sequence-classification support request.", + "best_issue_reason": "#43684 is the broader auto-registration request, but it still does not subsume #44405\u2019s sequence-classification support for Qwen3.5.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43684", "right": "issue:44405", "accept": false, - "reason": "Different requested capabilities and code paths: Qwen3-Omni registration in AutoModel/AutoModelForConditionalGeneration versus Qwen3.5 AutoModelForSequenceClassification support. Related subsystem, but not the same change." + "reason": "Different underlying requests: Qwen3-Omni model registration vs Qwen3.5 sequence-classification support. Same broad ecosystem, but not the same bug/change and would not be merged into one issue." } ] }, @@ -20432,7 +20378,7 @@ "issue:43845|issue:43859", "issue:43846|issue:43859" ], - "prepared_review_unit_hash": "bfa59791a5937f390bbc22c9eaae943bdd8e53c2e1daf2b2bd6f11c6f7b0ee40", + "prepared_review_unit_hash": "79c3ee06538e7a2dd1fe5bcc84f0d114606746d0210fbdb4328f69f5206cb1d1", "component_budget": { "node_count": 4, "item_count": 4, @@ -20454,55 +20400,58 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The three \"huggingface\" issues look like likely duplicates by title, but there isn\u2019t enough concrete problem detail to confidently merge them; the \"Fett\" issue appears separate.", - "confidence": 0.62, - "canonical_issue_reason": "Issue 43845 is the best canonical candidate among the set because it is the earliest of the near-identical \"huggingface\" reports and has the same title as 43846 and 43859. Issue 43835 has a different title and is likely unrelated.", + "summary": "Three issues share the exact same bare title \"huggingface \" and, with no other distinguishing evidence, look like duplicate reports; \"Fett\" appears unrelated. No PRs are present.", + "confidence": 0.9, + "canonical_issue_reason": "Issue 43845 is a reasonable canonical representative because it is one of the repeated identical-title reports and is the earliest of that cluster, with no evidence distinguishing it from the other two.", "canonical_pr_reason": null, - "best_issue_reason": "43845 is the strongest issue anchor overall: earliest timestamp, same repeated title as the other two \"huggingface\" issues, and no evidence that 43835 belongs to that cluster.", + "best_issue_reason": "Issue 43845 is the best overall issue candidate since it anchors the duplicate cluster of identical \"huggingface \" reports and is the earliest among them.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43835", "right": "issue:43845", "accept": false, - "reason": "Different titles and no body/detail evidence tying them to the same underlying problem." + "reason": "Different titles and no shared body, file, or target evidence; \"Fett\" does not look like the same issue as \"huggingface \"." }, { "left": "issue:43835", "right": "issue:43846", "accept": false, - "reason": "Different titles and no concrete shared bug context beyond generic similarity." + "reason": "Different titles and no shared body, file, or target evidence; \"Fett\" does not look like the same issue as \"huggingface \"." }, { "left": "issue:43835", "right": "issue:43859", "accept": false, - "reason": "Different titles and no concrete shared bug context beyond generic similarity." + "reason": "Different titles and no shared body, file, or target evidence; \"Fett\" does not look like the same issue as \"huggingface \"." }, { "left": "issue:43845", "right": "issue:43846", - "accept": false, - "reason": "Titles are identical, but there is no substantive issue description to confirm they report the same bug rather than two separate reports about the same topic." + "accept": true, + "reason": "Same exact title text and no distinguishing metadata or content evidence; these likely report the same issue." }, { "left": "issue:43845", "right": "issue:43859", - "accept": false, - "reason": "Too little concrete detail to safely treat these as the same underlying issue." + "accept": true, + "reason": "Same exact title text and no distinguishing metadata or content evidence; this is a plausible duplicate match." }, { "left": "issue:43846", "right": "issue:43859", - "accept": false, - "reason": "Titles match, but the available metadata is insufficient to verify a true duplicate relationship." + "accept": true, + "reason": "Same exact title text and no distinguishing metadata or content evidence; this is a plausible duplicate match." } ] }, - "evaluator_result": null, - "evaluator_used": false, - "retried": false, - "accepted_nontrivial_soft_edge": false, + "evaluator_result": { + "accept": false, + "feedback": "The summary is mostly grounded, but two soft-edge verdict reasons overstate the evidence. For issue:43845 vs issue:43859, the packet only supports identical titles and lack of other distinguishing metadata; it does not support \u201cnear-identical metadata.\u201d For issue:43846 vs issue:43859, they are not in very close time proximity, so that wording is also too strong. The duplicate judgments may be reasonable as soft matches, but the rationale must stay more conservative." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -20536,88 +20485,88 @@ "issue:44572|issue:44937", "issue:44573|issue:44937" ], - "prepared_review_unit_hash": "de7354f9d45bf63475dddb18a917d8e8de7daa598d4e8573c84421ed403e6e51", + "prepared_review_unit_hash": "287c6a7dafed58d00f051b211284112621300b6d48a75043a7f93f3b37f33fbd", "component_budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 9, - "serialized_chars": 5791, - "estimated_input_tokens": 1448, - "estimated_eval_tokens": 3152 + "serialized_chars": 5785, + "estimated_input_tokens": 1447, + "estimated_eval_tokens": 3150 }, "budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 9, - "serialized_chars": 5791, - "estimated_input_tokens": 1448, - "estimated_eval_tokens": 3152 + "serialized_chars": 5785, + "estimated_input_tokens": 1447, + "estimated_eval_tokens": 3150 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All items appear to be unrelated low-information/spam issues rather than duplicates; there is no concrete shared bug or change to cluster on. The soft-similarity signals are not enough to justify merging any pair.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "All items look like tiny closed spam/noise issues with no substantive body, links, or code context. There is no strong evidence that any pair refers to the same underlying bug, so no soft merges are justified.", + "confidence": 0.94, + "canonical_issue_reason": "No issue stands out as a true canonical duplicate target; the set is dominated by spammy or content-free issues, so there is no reliable representative bug report.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44418 is the least overtly spam-like title, but it is still too generic and unsupported to serve as a strong canonical representative.", + "best_issue_reason": "issue:44418 is the least spam-like and has the only vaguely descriptive title, but it is still too empty and generic to serve as a strong canonical issue.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44371", "right": "issue:44572", "accept": false, - "reason": "Both are generic spam placeholders with no shared concrete subject, so they are not the same underlying issue." + "reason": "Same placeholder title, but both are content-free closed issues with no evidence they describe the same underlying problem; could just be separate spam posts." }, { "left": "issue:44371", "right": "issue:44573", "accept": false, - "reason": "Generic spam-like titles only; no evidence they describe the same bug or request." + "reason": "Identical spam-style titles are not enough to merge; there is no body, linkage, or code context tying them to one bug." }, { "left": "issue:44371", "right": "issue:44418", "accept": false, - "reason": "One is explicit spam and the other is a vague documentation suggestion; no concrete overlap." + "reason": "One is a spam placeholder and the other is a generic documentation suggestion; no concrete shared defect or change target." }, { "left": "issue:44371", "right": "issue:44937", "accept": false, - "reason": "Completely different titles and apparent intent; no shared issue content." + "reason": "Completely different titles and no supporting details; no sign of a shared underlying issue." }, { "left": "issue:44418", "right": "issue:44572", "accept": false, - "reason": "A generic documentation request and a spam placeholder are not the same underlying issue." + "reason": "Generic documentation suggestion versus placeholder spam; insufficient evidence of the same report." }, { "left": "issue:44418", "right": "issue:44573", "accept": false, - "reason": "No substantive overlap beyond low-quality similarity; not a duplicate pair." + "reason": "Generic documentation suggestion versus placeholder spam; no common bug description or target." }, { "left": "issue:44572", "right": "issue:44573", "accept": false, - "reason": "Both are spam placeholders, but there is no evidence they refer to the same specific report or bug." + "reason": "Both are spam placeholders, but that does not establish the same underlying issue; they look like separate low-effort submissions." }, { "left": "issue:44572", "right": "issue:44937", "accept": false, - "reason": "Different apparent intent and no shared concrete topic." + "reason": "Placeholder spam versus a specific-sounding title; no evidence they refer to the same defect." }, { "left": "issue:44573", "right": "issue:44937", "accept": false, - "reason": "Titles do not indicate the same issue, and there is no supporting detail to connect them." + "reason": "Placeholder spam versus a specific-sounding title; too little overlap to treat as duplicates." } ] }, @@ -20647,12 +20596,12 @@ "soft_pairs": [ "issue:45145|issue:45151" ], - "prepared_review_unit_hash": "eb7b8db1628ccc7087a6264145e7ef1d036966d3db1f06cba44677658e846a9a", + "prepared_review_unit_hash": "6c32a6a5efac5cbb34a8758cb5d55201428cb54bcf5b4db9758eb26b3f6d5de9", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1646, + "serialized_chars": 1645, "estimated_input_tokens": 412, "estimated_eval_tokens": 1080 }, @@ -20660,7 +20609,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1646, + "serialized_chars": 1645, "estimated_input_tokens": 412, "estimated_eval_tokens": 1080 }, @@ -20669,28 +20618,28 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The issues have identical titles and are strongly similar, but the available evidence is still only soft similarity, so they should be treated as likely duplicates rather than definitively identical proposals.", - "confidence": 0.89, - "canonical_issue_reason": "Issue #45145 is the better canonical target because it is earlier and already has an inbound reference.", + "summary": "These two issues look like a probable duplicate pair: they have the same title, very similar body size/content, and no evidence of distinct scope. However, the evidence is limited to soft similarity only, so this should be treated as likely duplicate rather than definitive.", + "confidence": 0.8, + "canonical_issue_reason": "Issue #45145 is slightly earlier and already has an inbound reference, so it is the safest canonical representative for this likely duplicate issue pair.", "canonical_pr_reason": null, - "best_issue_reason": "Issue #45145 is the strongest representative issue because it is earlier and has the slightly stronger linkage signal.", + "best_issue_reason": "Issue #45145 is the best overall issue anchor because it appeared first and already has a reference pointing to it, making it the most stable deduplication target.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45145", "right": "issue:45151", - "accept": false, - "reason": "The matching title and soft similarity make this a plausible duplicate, but the packet does not provide direct body-level evidence or a deterministic match, so this is not strong enough to confidently merge as the same issue." + "accept": true, + "reason": "Same title and highly similar long-form content, with no signs of different scope; this is a likely duplicate, though the evidence is only soft_similarity and not definitive." } ] }, "evaluator_result": { "accept": false, - "feedback": "The canonical issue choice is grounded, but the duplicate verdict is a bit too strong for the evidence. The packet supports that the two issues have identical titles and soft similarity, but it does not directly show that the bodies are \"highly similar\" or that they are definitively the same proposal. Since the pair only has soft_similarity evidence and deterministic_accept is false, the soft-edge verdict should be phrased more conservatively." + "feedback": "The packet supports a likely duplicate pair, but the analyst overstates certainty: evidence is only soft_similarity with no diff/file/link corroboration, so calling them \"near-exact duplicates\" and assigning 0.99 confidence is too strong. Keep the conclusion conservative (probable duplicate, not definitive)." }, "evaluator_used": true, "retried": true, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -20767,7 +20716,7 @@ "soft_pairs": [ "issue:45762|issue:45763" ], - "prepared_review_unit_hash": "c6082a2b16e9c006a20f287b16a6f716709daa75d9c0499c6488e0112804d9d8", + "prepared_review_unit_hash": "6b0913f086f89aa32485370b2e2d5e90219266124d3720f31b9eaabafed85853", "component_budget": { "node_count": 2, "item_count": 2, @@ -20789,8 +20738,8 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two closed issues appear unrelated personal-name placeholders with different names and dates; there is no sign of the same underlying bug or request, so the soft duplicate link should be rejected.", - "confidence": 0.98, + "summary": "These two issues do not appear to be duplicates; they are separate, unrelated issue reports identified only by different personal-name titles and dates.", + "confidence": 0.97, "canonical_issue_reason": null, "canonical_pr_reason": null, "best_issue_reason": null, @@ -20800,7 +20749,7 @@ "left": "issue:45762", "right": "issue:45763", "accept": false, - "reason": "Different subject names and dates, no shared description or issue target, and no evidence they refer to the same underlying problem." + "reason": "Different names and different dates; no evidence they describe the same underlying bug or request." } ] }, @@ -20830,7 +20779,7 @@ "soft_pairs": [ "issue:45764|issue:45765" ], - "prepared_review_unit_hash": "26f5b4796d481f29c702ad7195b9b80d0d0cbc80d7ebd851211a1e6ebf6f6592", + "prepared_review_unit_hash": "77c92d7cf664e91d272e496099f9fc33a4120d08c81b510177a889503d4593fe", "component_budget": { "node_count": 2, "item_count": 2, @@ -20852,18 +20801,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both items look like unrelated low-information issues with personal-name/date titles, not the same underlying bug or request.", + "summary": "The two issues appear to be unrelated closed posts with only personal-name/date titles and no shared technical content, so this is not a duplicate cluster.", "confidence": 0.98, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "If a representative issue is needed, issue:45764 is marginally earlier, but neither issue is a good canonical choice because the titles do not indicate a shared problem.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45764", "right": "issue:45765", "accept": false, - "reason": "Different names and dates, with no body, targets, filenames, or discussion tying them to the same bug or change. They do not look like duplicates." + "reason": "Different personal-name/date titles with no body, links, or technical details; no evidence they describe the same underlying bug or request." } ] }, @@ -20893,40 +20842,40 @@ "soft_pairs": [ "issue:45810|issue:46295" ], - "prepared_review_unit_hash": "edb31b02b2117c18f090d2e60fd7b4fc86ac23b53d18b6b9dd309b605b53382d", + "prepared_review_unit_hash": "13309a802f8475cd2acd343252158eb45ae9dd9984edcfd7327851d82f0e59a4", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1539, - "estimated_input_tokens": 385, - "estimated_eval_tokens": 1026 + "serialized_chars": 1541, + "estimated_input_tokens": 386, + "estimated_eval_tokens": 1028 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1539, - "estimated_input_tokens": 385, - "estimated_eval_tokens": 1026 + "serialized_chars": 1541, + "estimated_input_tokens": 386, + "estimated_eval_tokens": 1028 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are about different features: one requests a Qwen token-classification/value-model class, the other asks for FP8 training support. They do not look like duplicates.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "The two issues are unrelated: one requests adding Qwen3_5ForTokenClassification as a value-model option, while the other requests FP8 training support. They should not be merged as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": "Issue #46295 is the broader, more general feature request and is the better representative of the cluster, though it is not a duplicate of #45810.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "#46295 is the stronger standalone issue because it describes a broader training capability request; #45810 is a model-specific API addition request.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45810", "right": "issue:46295", "accept": false, - "reason": "Different underlying requests and code paths: model class support for a specific Qwen architecture vs training-time FP8 support. They are not the same bug or change." + "reason": "Different underlying changes: one is about adding a specific token-classification/value-model class, the other is about FP8 training support. They do not describe the same bug or feature." } ] }, @@ -20966,88 +20915,88 @@ "issue:45871|issue:45873", "issue:45872|issue:45873" ], - "prepared_review_unit_hash": "09b4ae85709f93ee2c575866400f75f4370bef8acb0a1d6043aa5dee6da4ceec", + "prepared_review_unit_hash": "58f7f2037af75f7ba89a00ab84c167a49456bcbe0578785778f8956a0e9a1bd6", "component_budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 8, - "serialized_chars": 5722, - "estimated_input_tokens": 1431, - "estimated_eval_tokens": 3118 + "serialized_chars": 5728, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 }, "budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 8, - "serialized_chars": 5722, - "estimated_input_tokens": 1431, - "estimated_eval_tokens": 3118 + "serialized_chars": 5728, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The cluster contains two clear duplicate issue pairs: 45870\u219445871 and 45872\u219445873. Issue 45869 is a separate dataset-sample request, and the two title variants are different feature requests, not duplicates of each other.", - "confidence": 0.96, - "canonical_issue_reason": "Issue 45870 is the best canonical for the first duplicate pair because it is the earliest of the two identical 'Example: Financial time series prediction with DataForge' reports.", + "summary": "Two separate exact-duplicate issue pairs are present: 45870/45871 and 45872/45873. Issue 45869 is a different dataset offer and is not a duplicate of the example/service requests. There is no single canonical issue for the whole cluster.", + "confidence": 0.94, + "canonical_issue_reason": "No single canonical issue spans the whole cluster; the items split into two distinct duplicate pairs with different underlying requests.", "canonical_pr_reason": null, - "best_issue_reason": "45870 is the strongest overall anchor in this cluster: it is an exact duplicate target with the earliest timestamp, while the other items split into a different duplicate pair and one unrelated request.", + "best_issue_reason": "Issue 45870 is the best representative of its duplicate pair because it is the earlier of the two identical \"Example: Financial time series prediction with DataForge\" reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45869", "right": "issue:45870", "accept": false, - "reason": "Different requests: a financial data bundle/sample vs an example prediction workflow. Same general domain, but not the same underlying issue." + "reason": "Different topics: a financial dataset bundle offer vs an example request; similar finance wording is not enough to make them the same issue." }, { "left": "issue:45869", "right": "issue:45871", "accept": false, - "reason": "Different requests: dataset bundle/sample vs example prediction workflow. Similar theme only, not a duplicate." + "reason": "Different topics: a financial dataset bundle offer vs an example request; not the same underlying issue." }, { "left": "issue:45870", "right": "issue:45871", "accept": true, - "reason": "Same title and same apparent content/body size; this looks like a direct duplicate of the same example request." + "reason": "Same title and effectively the same content; these are direct duplicates of the same example issue." }, { "left": "issue:45870", "right": "issue:45872", "accept": false, - "reason": "Both are finance-related DataForge ideas, but one is a prediction example and the other is a scraping-service feature request. Different underlying change." + "reason": "Both mention financial time series/DataForge, but one is an example request and the other adds a scraping-service component; not the same concrete issue." }, { "left": "issue:45870", "right": "issue:45873", "accept": false, - "reason": "Same as 45870\u219445872: related domain, but different feature request and not mergeable as one issue." + "reason": "Example request vs scraping-service request; related theme, but different underlying issue and likely separate work." }, { "left": "issue:45871", "right": "issue:45872", "accept": false, - "reason": "Different feature requests: example prediction workflow versus scraping-service integration." + "reason": "Same reasoning as 45870/45872: related finance/DataForge theme, but different requested change." }, { "left": "issue:45871", "right": "issue:45873", "accept": false, - "reason": "Different feature requests: example prediction workflow versus scraping-service integration." + "reason": "Same reasoning as 45870/45873: similar broad area, but not the same underlying issue." }, { "left": "issue:45872", "right": "issue:45873", "accept": true, - "reason": "Same title and same apparent content/body size; this looks like a direct duplicate of the same scraping-service request." + "reason": "Same title and effectively the same content; these are direct duplicates of the same DataForge scraping-service issue." } ] }, "evaluator_result": { "accept": true, - "feedback": "The cluster summary is grounded in the packet and stays conservative: 45870/45871 and 45872/45873 are supported as duplicate-looking pairs by identical titles, identical body lengths, and strong similarity, while 45869 is reasonably treated as separate. The soft-edge rejections are also appropriately cautious." + "feedback": "The summary is well grounded in the packet. The two exact-duplicate pairs are supported by identical titles/body sizes and close timestamps: 45870/45871 and 45872/45873. The decision to treat 45869 as separate is conservative and consistent with its distinct title/content. The soft-edge rejections between the example-request pair and the scraping-service pair are also appropriately cautious, and the conclusion that there is no single canonical issue for the whole cluster is supported." }, "evaluator_used": true, "retried": false, @@ -21074,7 +21023,7 @@ "soft_pairs": [ "issue:46439|issue:46441" ], - "prepared_review_unit_hash": "70fa6b161cf89c1a14246f3fa03fb50ff837ad40c574b8daad4d872db1e73d1d", + "prepared_review_unit_hash": "44d301303d7add82cb4dbe1531fa7bb1a675ed989c5d7079dd1d39f9ce24b6fa", "component_budget": { "node_count": 2, "item_count": 2, @@ -21096,18 +21045,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both issues concern `cache_implementation=\"static\"` and `linear_attention`, but they describe different failure modes: one is cache `early_initialization` leading to shape corruption and `RuntimeError`, the other is a `KeyError: 'linear_attention'` in Qwen3.5. They look related, but not clearly the same underlying bug.", - "confidence": 0.76, - "canonical_issue_reason": "Issue 46439 is the better canonical issue because it is more mechanism-focused and broadly describes the static-cache/linear-attention corruption path, rather than a model-specific symptom.", + "summary": "Both issues point at static-cache initialization around linear attention, but they describe different observed failures (shape corruption vs. `KeyError: 'linear_attention'`). I would not merge them without clearer evidence they share the exact same root cause.", + "confidence": 0.78, + "canonical_issue_reason": "Issue 46439 is the more specific root-cause report: it names `early_initialization`, identifies linear-attention shape corruption, and includes the concrete `RuntimeError` behavior.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 46439 is the strongest representative for the cluster: it states the likely root cause and the concrete impact on linear-attention layers, making it more useful for deduping and tracking.", + "best_issue_reason": "Issue 46439 is the best single issue to keep because it is more diagnostic and likely broader than the model-specific crash report.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46439", "right": "issue:46441", "accept": false, - "reason": "Related static-cache linear-attention crashes, but one is shape corruption during early initialization and the other is a `KeyError` for `linear_attention`; not enough evidence they are the same bug." + "reason": "They both involve static cache and linear attention, but the failure modes differ materially (`wrong shapes`/`RuntimeError` vs `KeyError: 'linear_attention'`). That is not enough to treat them as the same underlying bug." } ] }, @@ -21137,7 +21086,7 @@ "soft_pairs": [ "issue:46551|issue:46552" ], - "prepared_review_unit_hash": "6fff14897a234860398fed1ed74af4840fa45b62d98b92ca22f2c975fc935180", + "prepared_review_unit_hash": "7329e8f22fa8cbb2f5326ff7eb65d820a3e6dee1b5e43d0c25101eb071a0efd4", "component_budget": { "node_count": 2, "item_count": 2, @@ -21159,24 +21108,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are duplicates with identical titles and matching problem descriptions about Wav2Vec2CTCTokenizer v5 leaking the word delimiter into special-token handling. Issue #46552 is the better canonical tracker because it is still open and already has inbound references.", + "summary": "The two issues are near-identical duplicates about the same Wav2Vec2CTCTokenizer v5 special-token bug; one is likely the cleaner canonical issue because it is open and already referenced elsewhere.", "confidence": 0.99, - "canonical_issue_reason": "Pick issue #46552: it is open, has existing inbound references, and matches the same bug report exactly.", + "canonical_issue_reason": "issue:46552 is the better canonical issue because it is open, has inbound references, and has the same title/body as issue:46551, making it the active tracking thread.", "canonical_pr_reason": null, - "best_issue_reason": "Issue #46552 is the best issue to keep as the primary tracker because it is current and already being referenced.", + "best_issue_reason": "issue:46552 is the best issue to keep as the primary record for this bug since it is still open and appears to be the version others are already pointing to.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46551", "right": "issue:46552", "accept": true, - "reason": "Same title and same underlying tokenizer bug; they are the same issue reported twice." + "reason": "Same title and same reported tokenizer bug; both describe the identical v5 word-delimiter special-token leakage into get_special_tokens_mask and convert_ids_to_tokens." } ] }, "evaluator_result": { "accept": true, - "feedback": "The assessment is grounded in the packet: the two issues have identical titles, identical body length, and the pair is flagged with soft similarity/Jaccard 1.0. Choosing #46552 as the canonical issue is also supported because it is open and has inbound references. No overstatement beyond the evidence is apparent." + "feedback": "The conclusion is grounded in the packet: both issues have identical titles, the same body length, and soft-similarity evidence; 46552 is open and has inbound references while 46551 is closed, so choosing 46552 as the canonical issue is conservative. The soft-edge accept is also supported by the provided evidence. No overstatement rises to the level of requiring rejection." }, "evaluator_used": true, "retried": false, @@ -21203,7 +21152,7 @@ "soft_pairs": [ "issue:46657|issue:46692" ], - "prepared_review_unit_hash": "5b0f34b9b411e9f6d91e793ce0c0e19a202b840b4232f7dffb3bbe9a0a041d78", + "prepared_review_unit_hash": "5106752f9772cece1b8e2381c301aa47c8828c1cd4ea66db8aaa57e071fe38c6", "component_budget": { "node_count": 2, "item_count": 2, @@ -21225,18 +21174,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues appear to describe different training failures: one is a tensor size mismatch in object detection training, the other is an out-of-memory problem after several epochs in RT-DETR v2. They should not be deduplicated.", - "confidence": 0.91, - "canonical_issue_reason": "Issue 46657 is the broader, more generic training error report and is the better cluster anchor; issue 46692 describes a separate RT-DETR v2 OOM symptom.", + "summary": "These two issues are only loosely related by object-detection training, but they describe different failure modes: a tensor shape mismatch versus an out-of-memory condition after several epochs. They should not be merged as duplicates.", + "confidence": 0.89, + "canonical_issue_reason": "Issue #46657 is the better representative of the cluster because it is a concrete, broadly framed training failure report; #46692 appears to be a separate RT-DETR v2 memory-growth problem.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 46657 is the best issue representative because it captures a concrete code-path failure (tensor shape mismatch) rather than a capacity-related runtime symptom.", + "best_issue_reason": "#46657 is the most canonical issue here because it states a specific runtime error directly tied to training and is less specialized than the RT-DETR v2 OOM report.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46657", "right": "issue:46692", "accept": false, - "reason": "Different underlying bugs: tensor-shape mismatch during object-detection training versus OOM after some epochs in RT-DETR v2. No evidence they share the same fix path." + "reason": "The reports do not look like the same underlying bug: one is a tensor size mismatch during object detection training, while the other is an out-of-memory failure after some epochs in RT-DETR v2. Different symptoms and likely different root causes." } ] }, @@ -21266,7 +21215,7 @@ "soft_pairs": [ "issue:46822|issue:46823" ], - "prepared_review_unit_hash": "64299c4d273f118e3f5717f091f3b0d82250c92e0261823ca78be07a8131928b", + "prepared_review_unit_hash": "9c1658f34440ee95cffdc66a0c22dd074f0dfbfc6c33f5eda70a0b85337cf88d", "component_budget": { "node_count": 2, "item_count": 2, @@ -21288,18 +21237,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues both involve model loading failures, but they describe different bugs in different code paths: ZeRO-3 zero.Init partitioning versus 4-bit device_map/max_memory placement logic. They should not be triaged as duplicates.", - "confidence": 0.93, - "canonical_issue_reason": "Issue 46822 is the more specific and self-contained bug report, with a clear ZeRO-3 / zero.Init failure mode and a concrete offending submodule.", + "summary": "The two issues are not duplicates: one is about ZeRO-3 zero.Init failing to partition a composite model submodule and causing OOM during multi-GPU load, while the other is about 4-bit loading with device_map='auto' and max_memory incorrectly dispatching modules to CPU and raising ValueError.", + "confidence": 0.98, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue 46822 best represents the cluster because it states a precise loading/OOM defect tied to a specific subsystem and model structure.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46822", "right": "issue:46823", "accept": false, - "reason": "Different underlying bugs: one is ZeRO-3 zero.Init failing to partition a composite submodule, the other is 4-bit auto device mapping incorrectly sending modules to CPU despite available memory. The symptoms overlap only at a high level ('load failure'), not at the code-path level." + "reason": "Different failure modes and code paths: ZeRO-3 partitioning/OOM during initialization vs 4-bit auto device mapping with max_memory causing CPU dispatch on a fitting model. They share only broad loading/dispatch terminology, not the same underlying bug." } ] }, @@ -21321,21 +21270,22 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-46906-6", + "cluster_id": "cluster-46906-7", "nodes": [ "issue:46906", "issue:46940", "issue:46948", "issue:46969", "issue:46985", - "issue:46998" + "issue:46998", + "issue:47037" ], "soft_pairs": [ "issue:46906|issue:46940", "issue:46906|issue:46948", "issue:46906|issue:46969", "issue:46906|issue:46985", - "issue:46906|issue:46998", + "issue:46906|issue:47037", "issue:46940|issue:46948", "issue:46940|issue:46969", "issue:46940|issue:46985", @@ -21344,127 +21294,168 @@ "issue:46948|issue:46985", "issue:46948|issue:46998", "issue:46969|issue:46985", + "issue:46969|issue:47037", "issue:46969|issue:46998", - "issue:46985|issue:46998" + "issue:46985|issue:47037", + "issue:46985|issue:46998", + "issue:46998|issue:47037", + "issue:46948|issue:47037", + "issue:46940|issue:47037" ], - "prepared_review_unit_hash": "a1f6634efb05baec69fa8e6eb830474b34d4bcd9a2cc8e346f6bfcfad73f4077", + "prepared_review_unit_hash": "ebc05b2bcfdbed493824eaf87a7cd44b1976f17753e4a2ced5f52aa7ff53894a", "component_budget": { - "node_count": 6, - "item_count": 6, - "soft_pair_count": 15, - "serialized_chars": 8796, - "estimated_input_tokens": 2199, - "estimated_eval_tokens": 4654 + "node_count": 7, + "item_count": 7, + "soft_pair_count": 20, + "serialized_chars": 11154, + "estimated_input_tokens": 2789, + "estimated_eval_tokens": 5834 }, "budget": { - "node_count": 6, - "item_count": 6, - "soft_pair_count": 15, - "serialized_chars": 8796, - "estimated_input_tokens": 2199, - "estimated_eval_tokens": 4654 + "node_count": 7, + "item_count": 7, + "soft_pair_count": 20, + "serialized_chars": 11154, + "estimated_input_tokens": 2789, + "estimated_eval_tokens": 5834 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are near-identical dated triage issues for the same area, but they read as separate daily integration-failure reports rather than duplicates of one underlying bug.", - "confidence": 0.93, - "canonical_issue_reason": "Issue 46906 is the earliest in the series and has the strongest inbound reference signal, so it is the best anchor if one issue must represent this cluster.", + "summary": "These are daily \"integration failure triage\" issues for different dates, so they look like recurring triage reports rather than duplicates of one underlying bug. Use the earliest open issue as the representative thread; reject all soft-duplicate links.", + "confidence": 0.94, + "canonical_issue_reason": "issue:46906 is the earliest issue in the series and has the strongest discussion/inbound reference activity, making it the best representative for this recurring triage topic.", "canonical_pr_reason": null, - "best_issue_reason": "46906 is the most central representative of the repeated triage thread: earliest created and most referenced among the dated reports.", + "best_issue_reason": "As a cluster representative, issue:46906 is the clearest anchor because it starts the date-stamped sequence and appears to have the most engagement.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46906", "right": "issue:46940", "accept": false, - "reason": "Same template and subsystem, but these are date-stamped triage reports for different days, not the same underlying issue." + "reason": "Same template title, but different date-stamped triage entries; no evidence they report the same underlying failure." }, { "left": "issue:46906", "right": "issue:46948", "accept": false, - "reason": "Both are recurring '[serge] integration failure triage' entries, but the differing dates indicate separate reports." + "reason": "Recurring triage post on a different day; similarity is structural, not a shared bug report." }, { "left": "issue:46906", "right": "issue:46969", "accept": false, - "reason": "These appear to be daily triage snapshots, so they should not be merged as one bug." + "reason": "Separate daily triage issue, not the same incident or code-path failure." }, { "left": "issue:46906", "right": "issue:46985", "accept": false, - "reason": "Different daily triage issue, not evidence of the same concrete failure instance." + "reason": "Another date-specific triage entry; no concrete duplication signal beyond wording." }, { "left": "issue:46906", - "right": "issue:46998", + "right": "issue:47037", "accept": false, - "reason": "Same broad title, but this is a later dated triage report rather than a duplicate of the earlier one." + "reason": "Same series pattern, but a later day's triage note rather than the same issue." }, { "left": "issue:46940", "right": "issue:46948", "accept": false, - "reason": "The shared wording is templated; the date progression suggests distinct daily reports." + "reason": "Consecutive daily triage reports; likely distinct snapshots of failures, not one bug thread." }, { "left": "issue:46940", "right": "issue:46969", "accept": false, - "reason": "These are separate integration triage issues for different days, not one bug report." + "reason": "Different day/date indicates separate triage instances, not a duplicate issue." }, { "left": "issue:46940", "right": "issue:46985", "accept": false, - "reason": "No sign of the same concrete failure; just another dated triage entry." + "reason": "No concrete shared failure described; only the recurring triage wrapper matches." }, { "left": "issue:46940", "right": "issue:46998", "accept": false, - "reason": "Recurring triage thread, but the later date means it is not the same issue instance." + "reason": "Distinct daily report, not a single underlying defect." + }, + { + "left": "issue:46940", + "right": "issue:47037", + "accept": false, + "reason": "Later triage entry with the same format, but not evidence of the same root cause." }, { "left": "issue:46948", "right": "issue:46969", "accept": false, - "reason": "High textual similarity comes from the template; the reports appear date-specific." + "reason": "Separate date-stamped triage issues; similarity alone is insufficient to merge." }, { "left": "issue:46948", "right": "issue:46985", "accept": false, - "reason": "These look like separate daily failure triage tickets, not duplicates." + "reason": "Same broad triage series, but no shared concrete bug or fix target." }, { "left": "issue:46948", "right": "issue:46998", "accept": false, - "reason": "Same series, but different day and likely different failure snapshot." + "reason": "Different daily triage report; no indication of a duplicate incident." }, { "left": "issue:46969", "right": "issue:46985", "accept": false, - "reason": "The titles match a repeated triage pattern, not one underlying bug." + "reason": "Consecutive daily triage posts are recurring operational notes, not the same issue." + }, + { + "left": "issue:46969", + "right": "issue:47037", + "accept": false, + "reason": "Different day and no shared underlying failure details; reject as duplicate." }, { "left": "issue:46969", "right": "issue:46998", "accept": false, - "reason": "Later dated triage entry; no evidence they describe the exact same failure." + "reason": "Distinct triage snapshot, not one concrete bug report." + }, + { + "left": "issue:46985", + "right": "issue:47037", + "accept": false, + "reason": "Same recurring triage pattern, but no evidence they describe the same failure." }, { "left": "issue:46985", "right": "issue:46998", "accept": false, - "reason": "Both are integration triage reports, but they are distinct date-stamped issues." + "reason": "Separate date-specific triage issue, not a duplicate." + }, + { + "left": "issue:46998", + "right": "issue:47037", + "accept": false, + "reason": "Later daily triage entry; title overlap is expected and not enough to merge." + }, + { + "left": "issue:46948", + "right": "issue:47037", + "accept": false, + "reason": "Recurring triage threads for different dates; no shared concrete bug evidence." + }, + { + "left": "issue:46940", + "right": "issue:47037", + "accept": false, + "reason": "Same template title, but these are distinct daily triage reports rather than duplicates." } ] }, @@ -21494,7 +21485,7 @@ "soft_pairs": [ "issue:46983|issue:46984" ], - "prepared_review_unit_hash": "7cb5abd819ebaef960d78e3f6ff8937c135e108e544855a4c123a95f0b26e72a", + "prepared_review_unit_hash": "fdbfa86d003963b31aac085b7b008818d0f35f26a39a899c71eac94ab8ba749e", "component_budget": { "node_count": 2, "item_count": 2, @@ -21516,28 +21507,28 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both items are effectively the same feature request: adding RWKV7 model support. The later open issue is the better canonical tracker, while the older closed issue looks like a duplicate or superseded report.", - "confidence": 0.99, - "canonical_issue_reason": "Issue #46984 is the better canonical issue because it is open and likely the active tracker for the same RWKV7 support request; #46983 is older and already closed.", + "summary": "Both issues ask for RWKV7 model support, but the packet only provides identical titles and soft similarity. That is enough to group them as closely related, but not enough to confidently prove duplication or that #46983 is an earlier duplicate. #46984 is still the better live tracker because it is open.", + "confidence": 0.72, + "canonical_issue_reason": "#46984 is open and therefore the better active tracker for the RWKV7 support request. The evidence does not conclusively show duplicate status, so this is based on tracker suitability rather than a proven earlier/later duplicate relationship.", "canonical_pr_reason": null, - "best_issue_reason": "Issue #46984 is the best issue to keep because it remains open, has the same title, and is the most plausible current discussion point for implementing RWKV7 support.", + "best_issue_reason": "#46984 is the best issue to keep as the canonical live discussion because it is open and matches the same feature request. The available evidence is too limited to justify stronger duplicate claims.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46983", "right": "issue:46984", - "accept": true, - "reason": "Same title and same underlying feature request (RWKV7 model support); no evidence of distinct scope or different bug/change." + "accept": false, + "reason": "The issues have the same title and look related, but the packet lacks body-level or code-level evidence to establish that they are the same underlying report. Soft similarity alone is not enough to accept a duplicate edge." } ] }, "evaluator_result": { - "accept": true, - "feedback": "Grounded: both issues have the same title, were opened minutes apart, and the packet only provides soft-similarity evidence, so treating the open one as the better canonical tracker is reasonable. The duplicate/superseded language is appropriately cautious, and the soft-edge verdict is conservative." + "accept": false, + "feedback": "The canonical choice of the open issue is grounded, but the summary and soft-edge verdict overstate the evidence. The packet only shows same title, close timestamps, and soft similarity; it does not establish that #46983 is a duplicate or an earlier duplicate. Confidence 0.98 is also too strong for this evidence level." }, "evaluator_used": true, - "retried": false, - "accepted_nontrivial_soft_edge": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -21560,7 +21551,7 @@ "soft_pairs": [ "issue:47007|issue:47015" ], - "prepared_review_unit_hash": "7d18f8e8d5f8616340d7fe68dc048542b5e4cca807734bc3f3bab0ae57f8f0bc", + "prepared_review_unit_hash": "bb1ea77f541ecd2c07cfafdbe808298e465cad3a247a1b4aa4f411975b6d76bc", "component_budget": { "node_count": 2, "item_count": 2, @@ -21582,18 +21573,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are thematically similar AI/cybersecurity writeups, but they do not describe the same concrete bug, task, or change. Treat them as related topical content rather than duplicates.", - "confidence": 0.89, - "canonical_issue_reason": "Issue #47015 has the cleaner, broader title and is the better representative of this small thematic cluster, but the pair is not a true duplicate set.", + "summary": "The two issues are only loosely similar: both discuss AI-driven vulnerability discovery, but one is about lessons from using advanced LLMs while the other is about a transformer-based cybersecurity discovery system. They read like related but distinct writeups, not duplicates.", + "confidence": 0.84, + "canonical_issue_reason": "Issue 47015 is the broader, more generic title and is the better representative of the shared AI-vulnerability-discovery theme.", "canonical_pr_reason": null, - "best_issue_reason": "#47015 is the most general and concise of the two issue titles, so it is the best single issue to stand in for the shared topic if one must be chosen.", + "best_issue_reason": "If a single representative is needed, 47015 is the closest fit because its wording is more general; 47007 is more specifically framed around lessons from advanced LLM usage.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:47007", "right": "issue:47015", "accept": false, - "reason": "The titles overlap around AI-assisted vulnerability discovery, but the bodies/titles suggest separate article-like topics rather than the same underlying bug or change. There are no shared files or explicit targets to indicate a duplicate." + "reason": "The titles overlap on AI cybersecurity, but they describe different angles and likely different submissions: LLM lessons vs transformer-based system. Similar topic, not the same underlying issue." } ] }, @@ -21732,412 +21723,400 @@ "pull_request:43451|pull_request:46933", "pull_request:46658|pull_request:46961" ], - "prepared_review_unit_hash": "d6a048e31a7600184c4012c44e9d177409bb782e97666153d1e2cc687ef06ae8", + "prepared_review_unit_hash": "9aa406b5b3e68b9acae4dfdc16d8edd0c081e44fc3fbf36c591c58d54a2a583a", "component_budget": { "node_count": 42, "item_count": 42, "soft_pair_count": 131, - "serialized_chars": 149678, - "estimated_input_tokens": 37420, - "estimated_eval_tokens": 75096 + "serialized_chars": 149662, + "estimated_input_tokens": 37416, + "estimated_eval_tokens": 75088 }, "budget": { "node_count": 40, "item_count": 40, "soft_pair_count": 72, - "serialized_chars": 113804, - "estimated_input_tokens": 28451, - "estimated_eval_tokens": 57158 + "serialized_chars": 113799, + "estimated_input_tokens": 28450, + "estimated_eval_tokens": 57156 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is mostly a grab-bag of unrelated PRs (new model additions, test fixes, docs, and a few model-specific bugfixes). The only clear duplicate/change-equivalent pair is the CohereASR training-loss double-shift fix: 46895 and 46958. I\u2019d treat 46958 as the best representative of that duplicate pair.", - "confidence": 0.95, + "summary": "Almost all candidate pairs are unrelated fixes across different models or docs. The only clear duplicate-style pair is the two CohereASR double-shift PRs, which address the same loss computation bug in the same code path.", + "confidence": 0.94, "canonical_issue_reason": null, - "canonical_pr_reason": "46958 is the clearest direct fix for the CohereASR double-shift loss bug; 46895 is an alternate implementation of the same underlying change.", + "canonical_pr_reason": "PR 46895 is the cleanest canonical choice: it directly targets issue 46894 and fixes the CohereASR training-loss double-shift in-place with a narrowly scoped code change.", "best_issue_reason": null, - "best_pr_reason": "There is no single global best PR for this set because most items are unrelated. If forced to pick one duplicate representative, 46958 is the cleanest.", + "best_pr_reason": "PR 46895 is the best representative of the underlying bug because it is explicitly issue-linked, narrowly scoped, and fixes the exact CohereASR loss path without broad unrelated edits.", "soft_edge_verdicts": [ { "left": "pull_request:46895", "right": "pull_request:46958", "accept": true, - "reason": "Same CohereASR training-loss double-shift bug in the same forward/loss path; the two patches are alternate fixes for the same concrete issue." + "reason": "Same CohereASR training-loss double-shift bug in the same forward/loss path; one is just an alternative implementation." }, { "left": "pull_request:46903", "right": "pull_request:46958", "accept": false, - "reason": "Both touch CohereASR loss handling, but 46903 is a broader multi-model fix (also PPFormulaNet) and is not the same concrete change as 46958." + "reason": "46903 also changes PPFormulaNet, so it is broader than the CohereASR fix and not the same concrete change." }, { "left": "pull_request:46950", "right": "pull_request:46951", "accept": false, - "reason": "Different models and different failure modes: Mamba2 OOM vs PEFT integration loading behavior." + "reason": "Different problems: Mamba2 OOM/runtime behavior vs PEFT loading integration." }, { "left": "pull_request:46927", "right": "pull_request:46950", "accept": false, - "reason": "Unrelated model fixes: GLM-46V input casting vs Mamba2 OOM/test behavior." + "reason": "Unrelated model bugs: GLM46V input dtype handling vs Mamba2 OOM." }, { "left": "pull_request:46950", "right": "pull_request:47006", "accept": false, - "reason": "Different subsystems and bugs: Mamba2 OOM vs Gemma3 generation cache behavior." + "reason": "Different code paths and models; Mamba2 OOM fix is not the same as Gemma3 generation cache logic." }, { "left": "pull_request:46927", "right": "pull_request:46951", "accept": false, - "reason": "Different models and fixes; only superficial similarity from being small integration-test-related PRs." + "reason": "GLM46V dtype/input handling is unrelated to PEFT integration loading." }, { "left": "pull_request:46951", "right": "pull_request:46971", "accept": false, - "reason": "PEFT loading change vs Mamba2 test dtype argument change; not the same underlying bug." + "reason": "PEFT integration load fix and Mamba2 test argument change are different issues." }, { "left": "pull_request:46949", "right": "pull_request:46971", "accept": false, - "reason": "Whisper cache/state fix vs Mamba2 test dtype/load tweak; unrelated." + "reason": "Whisper cache update bug and Mamba2 test loading change are unrelated." }, { "left": "pull_request:46927", "right": "pull_request:47006", "accept": false, - "reason": "GLM-46V input handling vs Gemma3 generation cache logic; different code paths." + "reason": "Different models and failures; no shared concrete bug." }, { "left": "pull_request:46951", "right": "pull_request:47006", "accept": false, - "reason": "PEFT integration loading vs Gemma3 generation cache behavior; unrelated." + "reason": "PEFT loading change does not match Gemma3 cache-generation fix." }, { "left": "pull_request:46971", "right": "pull_request:47006", "accept": false, - "reason": "Mamba2 test-only change vs Gemma3 generation fix; not mergeable as one PR." + "reason": "Test-side dtype/load change is unrelated to Gemma3 generation cache behavior." }, { "left": "pull_request:46941", "right": "pull_request:46970", "accept": false, - "reason": "Both are serge-generated test-fix PRs, but they target different models and different concrete failures." + "reason": "Both are large test-fix style PRs, but they address different models and failure modes." }, { "left": "pull_request:46908", "right": "pull_request:47006", "accept": false, - "reason": "DeepSeekV32 test load-error workaround vs Gemma3 generation cache fix; unrelated." + "reason": "DeepSeekV32 load/offload fix is unrelated to Gemma3 generation caching." }, { "left": "pull_request:46908", "right": "pull_request:46928", "accept": false, - "reason": "Different models and changes: DeepSeekV32 test offload workaround vs BLIP-2 tokenizer registry update." + "reason": "DeepSeekV32 loading issue and BLIP-2 tokenizer list update are unrelated." }, { "left": "pull_request:46907", "right": "pull_request:46926", "accept": false, - "reason": "EdgeTam runtime kwarg cleanup vs MiniMax-M3-VL test load_error workaround; unrelated." + "reason": "EdgeTam import/config kwarg fix and MiniMax M3 VL load-offload test fix are unrelated." }, { "left": "pull_request:46970", "right": "pull_request:46971", "accept": false, - "reason": "Musicgen-Melody output mismatch fix vs Mamba2 test load/dtype workaround; not the same bug." + "reason": "Musicgen Melody output mismatch and Mamba2 OOM are different bugs." }, { "left": "pull_request:46926", "right": "pull_request:46928", "accept": false, - "reason": "MiniMax-M3-VL test load_error workaround vs BLIP-2 tokenizer registry update; unrelated." + "reason": "Different targets: MiniMax M3 VL load error vs BLIP-2 tokenizer auto mapping." }, { "left": "pull_request:46927", "right": "pull_request:46986", "accept": false, - "reason": "GLM-46V input casting vs GLM-OCR output mismatch; different models and failure modes." + "reason": "Different model families and failures: GLM46V dtype issue vs GLM OCR output mismatch." }, { "left": "pull_request:46941", "right": "pull_request:46986", "accept": false, - "reason": "DAC output mismatch fix vs GLM-OCR output mismatch fix; different concrete code paths." + "reason": "Broader dtype/quantization fixes versus GLM OCR output mismatch are not the same change." }, { "left": "pull_request:46926", "right": "pull_request:46941", "accept": false, - "reason": "MiniMax-M3-VL load_error workaround vs DAC output mismatch fix; unrelated." + "reason": "MiniMax M3 VL load_error and DAC-related output mismatch are unrelated." }, { - "left": "pull_request:46933", - "right": "pull_request:46942", + "left": "pull_request:46928", + "right": "pull_request:46941", "accept": false, - "reason": "Gemma4 quantized vision/audio dtype casting vs TimesFM 2.5 quantized block casting; similar theme, but different models and code paths." + "reason": "Tokenizer auto-list update and DAC output mismatch are unrelated." }, { - "left": "pull_request:46954", - "right": "pull_request:46961", + "left": "pull_request:46907", + "right": "pull_request:46908", "accept": false, - "reason": "Pytest-v9-related GLMASR strict fp32 flag change vs unrelated workflow cleanup branch; not a duplicate bug fix." + "reason": "EdgeTam import_or_config failure and DeepSeekV32 load_error are different issues." }, { - "left": "pull_request:46954", - "right": "pull_request:46968", + "left": "pull_request:46908", + "right": "pull_request:46951", "accept": false, - "reason": "GLMASR test/runtime compatibility fix vs docs autolink cleanup; unrelated." + "reason": "DeepSeekV32 loading fix is unrelated to PEFT integration loading." }, { - "left": "pull_request:46954", - "right": "pull_request:46958", + "left": "pull_request:46907", + "right": "pull_request:46941", "accept": false, - "reason": "GLMASR pytest compatibility fix vs CohereASR loss fix; different models and bugs." + "reason": "EdgeTam kwarg cleanup and DAC output mismatch are unrelated." }, { - "left": "pull_request:46958", - "right": "pull_request:46961", + "left": "pull_request:46926", + "right": "pull_request:46927", "accept": false, - "reason": "CohereASR loss fix vs workflow cleanup; no shared underlying bug." + "reason": "Different model-specific test/load issues with no shared code path." }, { - "left": "pull_request:46961", - "right": "pull_request:46974", + "left": "pull_request:46954", + "right": "pull_request:46961", "accept": false, - "reason": "Workflow housekeeping vs Musicgen generate typo fix; unrelated." + "reason": "Pytest-v9 compatibility fix in GlmASR is unrelated to the Serge normalization workflow." }, { - "left": "pull_request:46958", - "right": "pull_request:46974", + "left": "pull_request:46954", + "right": "pull_request:46968", "accept": false, - "reason": "CohereASR loss bug vs Musicgen generate typo bug; different code paths." + "reason": "GlmASR fp32-module tweak and docs autolink cleanup are unrelated." }, { "left": "pull_request:46954", - "right": "pull_request:46971", + "right": "pull_request:46958", "accept": false, - "reason": "GLMASR pytest compatibility fix vs Mamba2 test load workaround; unrelated." + "reason": "Different models and different bugs; no shared concrete fix." }, { - "left": "pull_request:46920", - "right": "pull_request:46968", + "left": "pull_request:46958", + "right": "pull_request:46961", "accept": false, - "reason": "Voxtral Realtime audio API adaptation vs docs autolink cleanup; different areas." + "reason": "CohereASR loss fix and workflow automation are unrelated." }, { - "left": "pull_request:46971", + "left": "pull_request:46961", "right": "pull_request:46974", "accept": false, - "reason": "Mamba2 test workaround vs Musicgen generate logic fix; unrelated." + "reason": "Workflow cleanup and Musicgen generate typo are unrelated." }, { "left": "pull_request:46958", - "right": "pull_request:46968", - "accept": false, - "reason": "CohereASR loss fix vs docs autolink cleanup; unrelated." - }, - { - "left": "pull_request:46968", "right": "pull_request:46974", "accept": false, - "reason": "Docs autolink cleanup vs Musicgen generate bug fix; no shared bug." + "reason": "CohereASR training loss and Musicgen cache-length typo are different bugs." }, { "left": "pull_request:46920", - "right": "pull_request:46961", + "right": "pull_request:46968", "accept": false, - "reason": "Voxtral Realtime API adaptation vs workflow cleanup; unrelated." + "reason": "Voxtral realtime API adaptation and docs autolink cleanup are unrelated." }, { "left": "pull_request:46961", "right": "pull_request:46968", "accept": false, - "reason": "Workflow cleanup vs docs autolink cleanup; different repositories concerns and no same code path." + "reason": "Workflow automation and docs link cleanup are not the same change." }, { "left": "pull_request:46961", "right": "pull_request:46971", "accept": false, - "reason": "Workflow cleanup vs Mamba2 test workaround; unrelated." + "reason": "Workflow cleanup is unrelated to Mamba2 test loading changes." }, { "left": "pull_request:46949", "right": "pull_request:46968", "accept": false, - "reason": "Whisper cache/update fix vs docs autolink cleanup; unrelated." + "reason": "Whisper cache bug and docs autolink cleanup are unrelated." }, { "left": "pull_request:46266", "right": "pull_request:46957", "accept": false, - "reason": "DINOv2 checkpoint-conversion modernization vs finegrained FP8 generation/runtime changes; different goals and code paths." + "reason": "DINO modernization and finegrained FP8 generation changes are unrelated." }, { "left": "pull_request:47034", "right": "pull_request:47035", "accept": false, - "reason": "Neuron SFT experiment vs torch-compile backend default; related to Neuron, but not the same bug/change and not mergeable as one PR." + "reason": "47034 is a broader Neuron SFT experiment; 47035 is a separate training_args backend default tweak, not the same concrete change." }, { "left": "pull_request:47019", "right": "pull_request:47025", "accept": false, - "reason": "Masking utility correctness guard vs DiffusionGemma test adjustments; unrelated." + "reason": "Masking-utils contract regression and Diffusion Gemma test skips are unrelated." }, { "left": "pull_request:46957", "right": "pull_request:46965", "accept": false, - "reason": "Finegrained FP8 generation/kernel tuning vs DeepSeek V4 use_cache=False support; different feature areas." + "reason": "Generation optimization / FP8 work and DeepSeek V4 use_cache support are different issues." }, { "left": "pull_request:44517", "right": "pull_request:46417", "accept": false, - "reason": "Different new model integrations (Qwen3-TTS vs HunYuan-VL); unrelated." + "reason": "Different model additions (Qwen3-TTS vs HunYuan VL); not the same artifact." + }, + { + "left": "pull_request:46933", + "right": "pull_request:47035", + "accept": false, + "reason": "Gemma4 quantized vision/audio casting fix is unrelated to Neuron backend defaults." + }, + { + "left": "pull_request:46266", + "right": "pull_request:47003", + "accept": false, + "reason": "DINO modernization and OpenVINO exporter work are unrelated." + }, + { + "left": "pull_request:46933", + "right": "pull_request:46942", + "accept": false, + "reason": "Both touch dtype casting, but on different models and code paths; too broad to treat as the same change." + }, + { + "left": "pull_request:46912", + "right": "pull_request:47025", + "accept": false, + "reason": "RoPE test coverage and Diffusion Gemma test fixes are different test-only changes." }, { "left": "pull_request:46196", "right": "pull_request:46995", "accept": false, - "reason": "Wav2Vec2 SDPA dispatch support vs MLU availability detection for flash linear attention; unrelated." + "reason": "Wav2Vec2 SDPA dispatch and flash-linear-attention MLU support are unrelated backend changes." }, { "left": "pull_request:46707", "right": "pull_request:46961", "accept": false, - "reason": "Native FSDP2 module work vs workflow cleanup; different subsystems." + "reason": "FSDP2 migration and workflow cleanup do not address the same problem." }, { "left": "pull_request:46968", "right": "pull_request:47019", "accept": false, - "reason": "Docs autolink cleanup vs masking utility regression fix; unrelated." + "reason": "Docs autolink cleanup and masking_utils regression fix are unrelated." }, { "left": "pull_request:46707", "right": "pull_request:46968", "accept": false, - "reason": "FSDP2 migration work vs docs autolink cleanup; no shared underlying bug/change." + "reason": "FSDP2 work and docs autolink cleanup are unrelated." }, { "left": "pull_request:46707", "right": "pull_request:46957", "accept": false, - "reason": "FSDP2 migration vs finegrained FP8 generation/kernel work; unrelated." + "reason": "FSDP2 migration and finegrained FP8 generation optimization are unrelated." }, { "left": "pull_request:46912", - "right": "pull_request:47025", + "right": "pull_request:46942", "accept": false, - "reason": "RoPE test improvements vs DiffusionGemma test tweaks; both test-related but not the same bug or fix." + "reason": "RoPE test coverage and TimesFM 2.5 quantized cast fix are unrelated." }, { "left": "pull_request:46196", "right": "pull_request:46965", "accept": false, - "reason": "Wav2Vec2 SDPA dispatch vs DeepSeek V4 cache handling; unrelated." - }, - { - "left": "pull_request:46961", - "right": "pull_request:47019", - "accept": false, - "reason": "Workflow cleanup vs masking utility fix; unrelated." - }, - { - "left": "pull_request:46707", - "right": "pull_request:47019", - "accept": false, - "reason": "FSDP2 migration vs masking utility regression fix; different code paths." - }, - { - "left": "pull_request:46266", - "right": "pull_request:47034", - "accept": false, - "reason": "DINO modernization vs Neuron SFT experiment; unrelated." + "reason": "Wav2Vec2 SDPA dispatch and DeepSeek V4 use_cache support are unrelated." }, { "left": "pull_request:46836", "right": "pull_request:46961", "accept": false, - "reason": "Unlimited OCR model addition vs workflow cleanup; unrelated." + "reason": "Unlimited OCR model addition and workflow cleanup are unrelated." }, { "left": "pull_request:43451", "right": "pull_request:47019", "accept": false, - "reason": "Molmo2 model addition vs masking utility regression fix; unrelated." + "reason": "Molmo2 model addition and masking-utils regression fix are unrelated." }, { "left": "pull_request:46836", "right": "pull_request:46965", "accept": false, - "reason": "Unlimited OCR model addition vs DeepSeek V4 cache support; unrelated." + "reason": "Unlimited OCR model addition and DeepSeek V4 cache behavior are unrelated." }, { "left": "pull_request:46957", "right": "pull_request:46961", "accept": false, - "reason": "Finegrained FP8 runtime work vs workflow cleanup; unrelated." + "reason": "Finegrained FP8 work and workflow normalization are unrelated." }, { "left": "pull_request:43451", "right": "pull_request:46266", "accept": false, - "reason": "Different model additions (Molmo2 vs DINOv2 modernization); not a duplicate bug/change." + "reason": "Molmo2 model addition and DINO modernization are unrelated." }, { "left": "pull_request:43451", "right": "pull_request:46995", "accept": false, - "reason": "Molmo2 model addition vs flash-linear-attention availability tweak; unrelated." + "reason": "Molmo2 model addition and flash-linear-attention availability are unrelated." }, { "left": "pull_request:46266", "right": "pull_request:47003", "accept": false, - "reason": "DINO modernization vs OpenVINO HF Exporter; different subsystems and goals." + "reason": "DINO modernization and OpenVINO exporter work are unrelated." }, { "left": "pull_request:46961", - "right": "pull_request:47035", - "accept": false, - "reason": "Workflow cleanup vs Neuron torch-compile backend default; unrelated." - }, - { - "left": "pull_request:46836", "right": "pull_request:46995", "accept": false, - "reason": "Unlimited OCR model addition vs MLU support for flash-linear-attention availability; unrelated." - }, - { - "left": "pull_request:43451", - "right": "pull_request:46933", - "accept": false, - "reason": "Molmo2 model addition vs Gemma4 dtype-casting bug fix; unrelated." + "reason": "Workflow automation and flash-linear-attention availability are unrelated." }, { - "left": "pull_request:46658", - "right": "pull_request:46961", + "left": "pull_request:46968", + "right": "pull_request:47025", "accept": false, - "reason": "Step 3.7 Flash implementation work vs workflow cleanup; unrelated." + "reason": "Docs autolink cleanup and Diffusion Gemma test changes are unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative. The only accepted duplicate edge, 46895 \u2194 46958, is well supported by both PR titles and diffs as the same CohereASR double-shift loss bug. The rejected edges are mostly unrelated or only loosely similar, and I don\u2019t see any overclaiming that would justify rejection." + "feedback": "The summary is grounded in the packet: 46895 is explicitly issue-linked and the CohereASR pair is the only clear same-bug duplicate-style match. The other pairwise rejects are conservative and supported by the titles/diffs, with no obvious overstatement." }, "evaluator_used": true, "retried": false, @@ -22201,8 +22180,8 @@ "pull_request:44517|pull_request:47003", "pull_request:45694|pull_request:46707", "pull_request:43451|pull_request:46957", - "pull_request:43451|pull_request:46417", "pull_request:43451|pull_request:47025", + "pull_request:43451|pull_request:46417", "pull_request:46196|pull_request:46961", "pull_request:43451|pull_request:46961", "pull_request:46961|pull_request:47003", @@ -22250,352 +22229,382 @@ "pull_request:46912|pull_request:47003", "pull_request:46836|pull_request:46912" ], - "prepared_review_unit_hash": "65ffe0904b98e275f5d5b1ccecf942abe5b674a81bcbd2433ae75c20c55d6e4a", + "prepared_review_unit_hash": "a83a01dadfa39eae035985f1ff4394bac675ca6d24c87588d946357eed79d86c", "component_budget": { "node_count": 42, "item_count": 42, "soft_pair_count": 131, - "serialized_chars": 149678, - "estimated_input_tokens": 37420, - "estimated_eval_tokens": 75096 + "serialized_chars": 149662, + "estimated_input_tokens": 37416, + "estimated_eval_tokens": 75088 }, "budget": { "node_count": 30, "item_count": 30, "soft_pair_count": 59, - "serialized_chars": 89550, - "estimated_input_tokens": 22388, - "estimated_eval_tokens": 45032 + "serialized_chars": 89539, + "estimated_input_tokens": 22385, + "estimated_eval_tokens": 45026 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a loose lexical cluster, not a true duplicate cluster: it mixes unrelated model additions, docs/auto-registration changes, trainer/infrastructure tweaks, and one-off integration-test fixes. The soft edges are mostly shared patterns or broad subsystem overlap, so I reject them all as duplicates.", + "summary": "This cluster is mostly a mix of unrelated PRs: new model integrations, isolated bug fixes, docs-only edits, and automation/workflow changes. The soft similarities are driven by shared repo scaffolding and recurring filenames, not the same underlying bug or change.", "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "PR 43451 is the most central concrete change in the set: a full model addition with substantial code, docs, auto-mapping, and tests. It has the strongest structural overlap with other model-integration PRs, so it is the best anchor for this cluster.", "best_issue_reason": null, - "best_pr_reason": null, + "best_pr_reason": "43451 is the best overall PR anchor because it is substantial, self-contained, and representative of the shared model-registration scaffolding seen across the cluster. The other PRs are narrower fixes, drafts, or unrelated subsystem changes.", "soft_edge_verdicts": [ { "left": "pull_request:46961", "right": "pull_request:46965", "accept": false, - "reason": "Different fixes in different areas: branch-normalization automation vs DeepSeek V4 cache semantics." + "reason": "Automation/workflow cleanup vs a DeepSeek V4 cache/use_cache fix; unrelated change targets." }, { "left": "pull_request:46658", "right": "pull_request:46965", "accept": false, - "reason": "Unrelated Step-3.7 Flash implementation work vs DeepSeek V4 cache semantics." + "reason": "Step 3.7 Flash model work vs DeepSeek V4 behavior fix; different models and code paths." }, { "left": "pull_request:46196", "right": "pull_request:47025", "accept": false, - "reason": "Different models and failures; wav2vec2 SDPA dispatch vs Diffusion Gemma test cleanup." + "reason": "wav2vec2 sdpa dispatch support vs diffusion gemma test adjustments; unrelated subsystems." }, { "left": "pull_request:43451", "right": "pull_request:44517", "accept": false, - "reason": "Different model additions; only shared auto-mapping/docs boilerplate." + "reason": "Both add models, but Molmo2 and Qwen3-TTS are different integrations with different code paths." }, { "left": "pull_request:46707", "right": "pull_request:47025", "accept": false, - "reason": "Different subsystems: native FSDP2 migration vs Diffusion Gemma tests." + "reason": "Native FSDP2 migration vs diffusion gemma test cleanup; no shared bug or change." }, { "left": "pull_request:45694", "right": "pull_request:47035", "accept": false, - "reason": "Same file, but this is a TrainingArguments formatting/doc tweak and the other sets a Neuron compile backend; not the same bug." + "reason": "Batch-size split_batches fix vs torch-compile-neuron backend default; different training arguments." }, { "left": "pull_request:46995", "right": "pull_request:47003", "accept": false, - "reason": "Different subsystems: import_utils capability check vs OpenVINO exporter." + "reason": "MLU flash-linear-attention availability vs OpenVINO exporter work; different features entirely." }, { "left": "pull_request:44517", "right": "pull_request:46836", "accept": false, - "reason": "Different model additions; shared docs/tree registration only." + "reason": "Qwen3-TTS and Unlimited OCR are separate model additions." }, { "left": "pull_request:44517", "right": "pull_request:47003", "accept": false, - "reason": "Different model addition vs exporter infrastructure; not a duplicate." + "reason": "Qwen3-TTS model integration vs OpenVINO exporter platform work; unrelated." }, { "left": "pull_request:45694", "right": "pull_request:46707", "accept": false, - "reason": "Trainer arg formatting/doc tweak vs FSDP2 migration; unrelated." + "reason": "TrainingArguments batch-size semantics vs FSDP2 module migration; unrelated code paths." }, { "left": "pull_request:43451", "right": "pull_request:46957", "accept": false, - "reason": "Different model addition vs FP8 generation optimization." + "reason": "Molmo2 model addition vs finegrained FP8/generation optimization work; different concerns." }, { "left": "pull_request:43451", - "right": "pull_request:46417", + "right": "pull_request:47025", "accept": false, - "reason": "Different model additions; shared registration boilerplate only." + "reason": "Molmo2 model addition vs diffusion gemma test fixes; unrelated models." }, { "left": "pull_request:43451", - "right": "pull_request:47025", + "right": "pull_request:46417", "accept": false, - "reason": "Model addition vs Diffusion Gemma test cleanup; unrelated." + "reason": "Molmo2 and HunYuan VL are distinct model integrations, not duplicates." }, { "left": "pull_request:46196", "right": "pull_request:46961", "accept": false, - "reason": "Different models and fixes; wav2vec2 dispatch vs branch normalization." + "reason": "wav2vec2 sdpa support vs branch-normalization workflow; unrelated." }, { "left": "pull_request:43451", "right": "pull_request:46961", "accept": false, - "reason": "Model addition vs branch normalization; unrelated." + "reason": "Model addition vs CI workflow automation; no shared underlying issue." }, { "left": "pull_request:46961", "right": "pull_request:47003", "accept": false, - "reason": "Same branch-normalization theme as 46961, but OpenVINO exporter is unrelated." + "reason": "Workflow normalization PR vs OpenVINO exporter PR; completely different scopes." }, { "left": "pull_request:46961", "right": "pull_request:47034", "accept": false, - "reason": "Branch-normalization automation vs Neuron SFT experiment; unrelated." + "reason": "Workflow automation vs Neuron SFT experiment files; unrelated." }, { "left": "pull_request:46912", "right": "pull_request:46961", "accept": false, - "reason": "RoPE test coverage vs branch normalization; no shared code-path." + "reason": "RoPE test improvements vs workflow automation; unrelated." }, { "left": "pull_request:46417", "right": "pull_request:46836", "accept": false, - "reason": "Different model additions; only shared docs/auto-registration boilerplate." + "reason": "HunYuan VL and Unlimited OCR are separate model additions." }, { "left": "pull_request:43451", "right": "pull_request:45694", "accept": false, - "reason": "Model addition vs TrainingArguments tweak; unrelated." + "reason": "Molmo2 model integration vs TrainingArguments batch-size fix; unrelated." }, { "left": "pull_request:46417", "right": "pull_request:47003", "accept": false, - "reason": "Different model addition vs OpenVINO exporter." + "reason": "HunYuan VL model integration vs OpenVINO exporter work; different domains." }, { "left": "pull_request:45694", "right": "pull_request:47003", "accept": false, - "reason": "TrainingArguments tweak vs OpenVINO exporter; unrelated." + "reason": "TrainingArguments split_batches fix vs OpenVINO exporter changes; unrelated." }, { "left": "pull_request:46417", "right": "pull_request:46965", "accept": false, - "reason": "Different model addition vs DeepSeek V4 cache fix." + "reason": "HunYuan VL model addition vs DeepSeek V4 cache/use_cache fix; unrelated." }, { "left": "pull_request:46196", "right": "pull_request:46836", "accept": false, - "reason": "Different models; wav2vec2 dispatch vs Unlimited OCR." + "reason": "wav2vec2 sdpa dispatch vs Unlimited OCR model addition; different targets." }, { "left": "pull_request:45694", "right": "pull_request:46836", "accept": false, - "reason": "TrainingArguments tweak vs Unlimited OCR model addition." + "reason": "TrainingArguments fix vs Unlimited OCR model addition; no duplicate underlying change." }, { "left": "pull_request:46912", "right": "pull_request:47034", "accept": false, - "reason": "RoPE tests vs Neuron SFT experiment; unrelated." + "reason": "RoPE test coverage vs Neuron SFT experiment; unrelated." }, { "left": "pull_request:43451", "right": "pull_request:46968", "accept": false, - "reason": "Model addition vs docs autolink cleanup; unrelated." + "reason": "Molmo2 model addition vs docs autolink cleanup; unrelated." }, { "left": "pull_request:46968", "right": "pull_request:47003", "accept": false, - "reason": "Docs autolink cleanup vs OpenVINO exporter; unrelated." + "reason": "Documentation autolink fix vs exporter implementation; unrelated." }, { "left": "pull_request:46950", "right": "pull_request:46971", "accept": false, - "reason": "Same model/failure family, but one changes the model kernel path and the other only adjusts test loading args; not the same concrete fix." + "reason": "Both mention Mamba2 OOM, but one changes model code and the other only adjusts tests/load dtype; not the same concrete fix." }, { "left": "pull_request:46949", "right": "pull_request:46950", "accept": false, - "reason": "Different models and failure modes: Whisper tests vs Mamba2 OOM." + "reason": "Whisper/audioflamingo cache-state fix vs Mamba2 OOM work; unrelated models." }, { "left": "pull_request:46949", "right": "pull_request:46951", "accept": false, - "reason": "Different models and subsystems: Whisper tests vs PEFT integration loading." + "reason": "Cache initialization fix vs PEFT loading mismatch fix; unrelated." }, { "left": "pull_request:46927", "right": "pull_request:46949", "accept": false, - "reason": "Different models; GLM46V vs Whisper." + "reason": "GLM46V input handling vs Whisper/audioflamingo cache fix; unrelated models." }, { "left": "pull_request:46907", "right": "pull_request:46927", "accept": false, - "reason": "Different models; EdgeTam vs GLM46V." + "reason": "EdgeTam processor-kwarg fix vs GLM46V input casting fix; different models and bugs." }, { "left": "pull_request:46927", "right": "pull_request:46971", "accept": false, - "reason": "Different models; GLM46V vs Mamba2." + "reason": "GLM46V input casting vs Mamba2 OOM test fix; unrelated." }, { "left": "pull_request:46908", "right": "pull_request:46926", "accept": false, - "reason": "Different models; DeepSeek-V3.2 test load_error vs MiniMax-M3-VL test load_error." + "reason": "DeepSeekV3.2 load-error test fix vs MiniMax-M3-VL load-error test fix; different models." }, { "left": "pull_request:46986", "right": "pull_request:47006", "accept": false, - "reason": "Different models; GLM OCR vs Gemma3 generation." + "reason": "GLM OCR output mismatch fix vs Gemma3 generation-config change; unrelated." + }, + { + "left": "pull_request:46970", + "right": "pull_request:47006", + "accept": false, + "reason": "MusicGen Melody test failures vs Gemma3 generation behavior; different code paths." }, { "left": "pull_request:46970", + "right": "pull_request:46986", + "accept": false, + "reason": "MusicGen Melody and GLM OCR are separate model test fixes." + }, + { + "left": "pull_request:46907", + "right": "pull_request:46949", + "accept": false, + "reason": "EdgeTam processor kwarg fix vs Whisper/audioflamingo cache fix; unrelated." + }, + { + "left": "pull_request:46907", "right": "pull_request:47006", "accept": false, - "reason": "Different models; MusicGen Melody vs Gemma3." + "reason": "EdgeTam bug fix vs Gemma3 generation-utils fix; unrelated." }, { "left": "pull_request:46907", "right": "pull_request:46970", "accept": false, - "reason": "Different models; EdgeTam vs MusicGen Melody." + "reason": "EdgeTam and MusicGen Melody target different model families and bugs." + }, + { + "left": "pull_request:46907", + "right": "pull_request:46986", + "accept": false, + "reason": "EdgeTam processor fix vs GLM OCR mismatch fix; unrelated." }, { "left": "pull_request:46707", "right": "pull_request:47034", "accept": false, - "reason": "FSDP2 migration vs Neuron SFT experiment; unrelated." + "reason": "Native FSDP2 migration vs Neuron SFT experiment; unrelated." }, { "left": "pull_request:46658", "right": "pull_request:46836", "accept": false, - "reason": "Different build/implementation areas; Step-3.7 Flash script vs Unlimited OCR model addition." + "reason": "Step 3.7 Flash model implementation vs Unlimited OCR model addition; different models." }, { "left": "pull_request:43451", "right": "pull_request:46836", "accept": false, - "reason": "Different model additions; only shared docs/auto-registration boilerplate." - }, - { - "left": "pull_request:46836", - "right": "pull_request:47003", - "accept": false, - "reason": "Model addition vs exporter bundle; unrelated." + "reason": "Molmo2 and Unlimited OCR are separate model integrations." }, { "left": "pull_request:40546", "right": "pull_request:43451", "accept": false, - "reason": "Different model additions; VibeVoice vs Molmo2." + "reason": "VibeVoice and Molmo2 are unrelated model additions." }, { "left": "pull_request:43451", "right": "pull_request:46658", "accept": false, - "reason": "Model addition vs Step-3.7 Flash build script; unrelated." + "reason": "Molmo2 and Step 3.7 Flash are distinct model additions." + }, + { + "left": "pull_request:46658", + "right": "pull_request:47003", + "accept": false, + "reason": "Step 3.7 Flash model work vs OpenVINO exporter infrastructure; unrelated." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46196", + "accept": false, + "reason": "Molmo2 model addition vs wav2vec2 sdpa dispatch; unrelated." }, { "left": "pull_request:43451", "right": "pull_request:47003", "accept": false, - "reason": "Model addition vs exporter bundle; unrelated." + "reason": "Molmo2 model addition vs OpenVINO exporter changes; unrelated." }, { "left": "pull_request:43451", "right": "pull_request:47034", "accept": false, - "reason": "Model addition vs Neuron SFT experiment; unrelated." + "reason": "Molmo2 model addition vs Neuron training experiment; unrelated." }, { "left": "pull_request:43451", "right": "pull_request:47035", "accept": false, - "reason": "Model addition vs Neuron compile backend tweak; unrelated." + "reason": "Molmo2 model addition vs torch-compile backend default; unrelated." }, { "left": "pull_request:43451", "right": "pull_request:46912", "accept": false, - "reason": "Model addition vs RoPE test coverage; unrelated." + "reason": "Molmo2 model addition vs RoPE test improvements; unrelated." }, { "left": "pull_request:46196", "right": "pull_request:47003", "accept": false, - "reason": "Different models; wav2vec2 attention dispatch vs OpenVINO exporter." + "reason": "wav2vec2 sdpa support vs OpenVINO exporter work; unrelated." }, { "left": "pull_request:47003", "right": "pull_request:47034", "accept": false, - "reason": "Exporter bundle vs Neuron compile backend tweak; unrelated." + "reason": "OpenVINO exporter project vs Neuron SFT experiment; unrelated." }, { "left": "pull_request:47003", "right": "pull_request:47035", "accept": false, - "reason": "Exporter bundle vs Neuron compile backend tweak; unrelated." + "reason": "OpenVINO exporter work vs torch-compile neuron default; unrelated." }, { "left": "pull_request:46912", "right": "pull_request:47003", "accept": false, - "reason": "RoPE tests vs exporter bundle; unrelated." + "reason": "RoPE tests vs OpenVINO exporter infrastructure; unrelated." }, { "left": "pull_request:46836", "right": "pull_request:46912", "accept": false, - "reason": "Different model addition vs RoPE test coverage; unrelated." + "reason": "Unlimited OCR model addition vs RoPE test improvements; unrelated." } ] }, @@ -22625,7 +22634,7 @@ "soft_pairs": [ "pull_request:40755|pull_request:41224" ], - "prepared_review_unit_hash": "1aa217df47c4cb6cd94254879c1cf7e3b0e6611b328965c777c77335895529e6", + "prepared_review_unit_hash": "69ba97c99602f06d1ea230cec72e2e712beb101c436ad0aa9eba0ec5633789c3", "component_budget": { "node_count": 2, "item_count": 2, @@ -22647,18 +22656,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two open PRs in the same library, but they address different features: one adds TimesFM covariate forecasting support, the other adds DINOv3 image-classification support. No duplicate relationship is evident.", + "summary": "The two PRs are unrelated feature additions in different model families: TimesFM covariate forecasting vs. DINOv3 image classification support. They should not be clustered as duplicates.", "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "Neither PR is a duplicate of the other. If one must be treated as the cluster representative, #41224 is the narrower, more self-contained feature PR; #40755 is a broader TimesFM feature addition with docs, xreg utilities, and tests.", + "canonical_pr_reason": "PR #41224 is the better representative PR in this cluster because it is a more focused, concrete model feature addition with clear code and test changes for a single capability.", "best_issue_reason": null, - "best_pr_reason": "#41224 is the better standalone representative because it targets a single concrete capability addition (DINOv3 image classification) with a focused code path and API surface. #40755 is a larger, unrelated TimesFM enhancement.", + "best_pr_reason": "PR #41224 is the strongest standalone PR here: it introduces a specific new classification path for DINOv3 with supporting implementation and tests, making it the clearest representative of a distinct change.", "soft_edge_verdicts": [ { "left": "pull_request:40755", "right": "pull_request:41224", "accept": false, - "reason": "Different models and different code paths: TimesFM covariate forecasting versus DINOv3 image classification. They may both be feature PRs, but they are not the same underlying change and would not plausibly be merged into one PR." + "reason": "Different models, different tasks, and different code paths; this is not the same underlying bug or change." } ] }, @@ -22779,148 +22788,148 @@ "pull_request:43823|pull_request:45213", "pull_request:43251|pull_request:43254" ], - "prepared_review_unit_hash": "f6ebd0975d9b0614a919e41563ed83d52ccbeed922b593aaaa29f9df98c0ed3f", + "prepared_review_unit_hash": "97afea65aff25649f61c87f1679e40de5d4f39b3f460558d6f418f9441b741f0", "component_budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 18, - "serialized_chars": 34788, - "estimated_input_tokens": 8697, - "estimated_eval_tokens": 17650 + "serialized_chars": 34782, + "estimated_input_tokens": 8696, + "estimated_eval_tokens": 17648 }, "budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 18, - "serialized_chars": 34788, - "estimated_input_tokens": 8697, - "estimated_eval_tokens": 17650 + "serialized_chars": 34782, + "estimated_input_tokens": 8696, + "estimated_eval_tokens": 17648 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Only one soft edge looks like a true duplicate: the two fixed_cross_entropy PRs. All other pairs are merely similar by subsystem or generic model-creation/refactor themes, not the same concrete change.", - "confidence": 0.96, + "summary": "This cluster is mostly heterogeneous. The only clear duplicate pair is the two `fixed_cross_entropy` PRs (#43251 and #43254), which both add extra kwargs to the same loss call. The other pairs are just broad topical similarity (models, auto-mapping, tests, docs) and do not look like the same concrete fix.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #43251 is the better canonical duplicate representative: it directly targets the reported issue, is more complete about supported cross_entropy kwargs, and includes the same concrete code-path fix as #43254.", + "canonical_pr_reason": "PR #43251 is the better canonical representative of the only true duplicate pair: it is the more explicit implementation of the cross-entropy kwargs fix, has an issue target, and more discussion/review context. #43254 is a narrower near-copy of the same change.", "best_issue_reason": null, - "best_pr_reason": "PR #43251 is the best overall representative because it is the most explicit and complete fix for the shared cross_entropy kwargs bug; #43254 is a narrower variant of the same patch.", + "best_pr_reason": "PR #43251 is the best global representative because it addresses a concrete bug report, has the clearer issue linkage, and is the more fully reviewed version of the shared loss-function fix.", "soft_edge_verdicts": [ { "left": "pull_request:41855", "right": "pull_request:43823", "accept": false, - "reason": "Different features in different areas: Mistral tokenizer methods vs adding a MobileLLM model." + "reason": "Unrelated PRs: Mistral tokenizer method additions vs adding a new MobileLLM model. Same general repo area, different changes." }, { "left": "pull_request:41855", "right": "pull_request:43424", "accept": false, - "reason": "Unrelated changes: tokenizer API additions vs an ExecuTorch dynamic-shape export test." + "reason": "Different code paths: tokenizer API additions vs ExecuTorch dynamic-shape export test." }, { "left": "pull_request:42668", "right": "pull_request:43823", "accept": false, - "reason": "Both touch preprocessing/model-loading plumbing, but they address different models and different problems." + "reason": "Both are model/processor related, but one is auto/processor robustness and the other is a new MobileLLM model. Not the same fix." }, { "left": "pull_request:43251", "right": "pull_request:43823", "accept": false, - "reason": "Cross-entropy kwargs fix is unrelated to adding a new MobileLLM model." + "reason": "Cross-entropy kwargs fix vs adding a new model; no shared underlying bug or change." }, { "left": "pull_request:43251", "right": "pull_request:44827", "accept": false, - "reason": "Different concrete fixes: loss utility kwargs vs Mistral4 test/model updates." + "reason": "Loss-function kwargs change and Mistral4 test/model fixes are distinct changes with no concrete overlap." }, { "left": "pull_request:43424", "right": "pull_request:44827", "accept": false, - "reason": "A dynamic export test is not the same change as Mistral4 model/test fixes." + "reason": "ExecuTorch export test and Mistral4 fixes are separate test concerns, not one mergeable change." }, { "left": "pull_request:43424", "right": "pull_request:43823", "accept": false, - "reason": "ExecuTorch export testing and MobileLLM model addition are unrelated." + "reason": "Dynamic-shape export test vs new MobileLLM model; different features and failure modes." }, { "left": "pull_request:43823", "right": "pull_request:45655", "accept": false, - "reason": "New MobileLLM model support vs a base-class config resolution ordering fix are unrelated." + "reason": "Adding MobileLLM and fixing `cls.config` resolution are unrelated model-creation vs core metaclass behavior changes." }, { "left": "pull_request:43823", "right": "pull_request:43995", "accept": false, - "reason": "Different models and different code paths: MobileLLM addition vs Falcon output-interface refactor." + "reason": "New model addition vs Falcon output-collection refactor; same broad area but not the same concrete change." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45213", + "accept": false, + "reason": "New model addition vs an agent skill/docs/model-creation scaffolding PR; not the same underlying bug or code path." }, { "left": "pull_request:43995", "right": "pull_request:45213", "accept": false, - "reason": "Falcon model refactor is unrelated to the new-model skill/documentation PR." + "reason": "Falcon forward/refactor changes are unrelated to the model-creation skill PR." }, { "left": "pull_request:43995", "right": "pull_request:44827", "accept": false, - "reason": "Both are model code changes, but they fix different model families and different issues." + "reason": "Falcon output interface refactor and Mistral4 fixes are separate model-specific changes." }, { "left": "pull_request:44827", "right": "pull_request:45170", "accept": false, - "reason": "Mistral4 test/model changes are not the same as the CLIP-like layrnorm-to-layernorm rename fix." + "reason": "Mistral4 test fixes vs CLIP-family `layrnorm` rename/model conversion fix; only broad model-family similarity." }, { "left": "pull_request:44827", "right": "pull_request:46267", "accept": false, - "reason": "Unrelated subsystems: Mistral4 tests vs serve chat parsing updates." + "reason": "Mistral4 fixes and chat response parsing updates are unrelated subsystems." }, { "left": "pull_request:44827", "right": "pull_request:46048", "accept": false, - "reason": "Mistral4 test fixes and adding a TDT loss kernel are different concrete changes." + "reason": "Mistral4 fixes vs adding a TDT loss kernel; different features and code paths." }, { "left": "pull_request:44827", "right": "pull_request:45655", "accept": false, - "reason": "Mistral4-specific fixes are unrelated to the config-resolution ordering change." + "reason": "Mistral4 model fixes and `cls.config` resolution are unrelated core changes." }, { "left": "pull_request:44827", "right": "pull_request:45213", "accept": false, - "reason": "A model bugfix is not the same as the new-model skill PR." - }, - { - "left": "pull_request:43823", - "right": "pull_request:45213", - "accept": false, - "reason": "Adding a specific model implementation is not the same as adding a generic model-creation skill and docs." + "reason": "Mistral4 tests/fixes vs model-creation skill PR; no shared underlying bug." }, { "left": "pull_request:43251", "right": "pull_request:43254", "accept": true, - "reason": "Same underlying bug and same concrete fix path: passing additional kwargs through fixed_cross_entropy to nn.functional.cross_entropy." + "reason": "Same underlying bug: both modify `fixed_cross_entropy` to pass supported kwargs through to `nn.functional.cross_entropy`, and both target the same issue." } ] }, "evaluator_result": { "accept": true, - "feedback": "The duplicate call on 43251/43254 is well grounded by the shared file, shared issue target, and same fixed_cross_entropy bugfix path. The other pairs are conservatively rejected as only loosely similar by subsystem/theme, not the same mergeable bugfix." + "feedback": "The summary is grounded in the packet and stays conservative. The only accepted duplicate pair is correctly supported by shared file, shared issue target, and nearly identical loss-function diffs. The other pair rejections are appropriately cautious and do not overstate evidence." }, "evaluator_used": true, "retried": false, @@ -22947,7 +22956,7 @@ "soft_pairs": [ "pull_request:42975|pull_request:42976" ], - "prepared_review_unit_hash": "c5c83f596b2f303b052383c06bc721cb55e3a8b04a80ec381795e9a160f18b3d", + "prepared_review_unit_hash": "64d11cd6698e2b37fced03084e2bdc76cd33c1783974a8d1d8fd27bcda195ee8", "component_budget": { "node_count": 2, "item_count": 2, @@ -22969,18 +22978,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two workflow-maintenance PRs, but they upgrade different GitHub Actions families and should not be treated as duplicates. PR 42975 is a broad Node 24 compatibility bump for core actions; PR 42976 is a separate docker/build-push-action version upgrade.", + "summary": "Both items are broad GitHub Actions upgrade PRs, but they target different actions and failure modes: one updates checkout/cache/upload-artifact for Node 24 compatibility, while the other updates docker/build-push-action and related workflow actions. They are related maintenance work, not the same underlying change.", "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 42975 is the broader, more representative workflow-actions compatibility update, spanning multiple core actions and many workflows.", + "canonical_pr_reason": "PR 42975 is the better canonical representative because it is the broader workflow-actions compatibility upgrade and its title more clearly states the maintenance goal.", "best_issue_reason": null, - "best_pr_reason": "PR 42975 is the best representative of the cluster because it addresses the larger cross-workflow Actions runtime compatibility shift rather than a single action bump.", + "best_pr_reason": "PR 42975 is the best representative of the cluster overall since it covers many workflow action version bumps and is the more general GitHub Actions upgrade effort.", "soft_edge_verdicts": [ { "left": "pull_request:42975", "right": "pull_request:42976", "accept": false, - "reason": "Both are workflow dependency bumps, but they change different actions (core GitHub Actions runtime compatibility vs docker/build-push-action) and do not look like the same underlying fix." + "reason": "Different concrete upgrades: 42975 focuses on checkout/cache/upload-artifact for Node 24 compatibility, while 42976 upgrades docker/build-push-action and other workflow actions. They touch overlapping workflows but not the same code-path problem." } ] }, @@ -23015,12 +23024,12 @@ "pull_request:43448|pull_request:45490", "pull_request:45490|pull_request:46562" ], - "prepared_review_unit_hash": "ab3a2c83ad40380d1f86e140523c6a0e29417e8b6391c37861880495957fa009", + "prepared_review_unit_hash": "d60010f34e0e92464645af4870a0870d68865a6c4ae7b29a40fd572e49abe4be", "component_budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 4, - "serialized_chars": 13047, + "serialized_chars": 13046, "estimated_input_tokens": 3262, "estimated_eval_tokens": 6780 }, @@ -23028,7 +23037,7 @@ "node_count": 4, "item_count": 4, "soft_pair_count": 4, - "serialized_chars": 13047, + "serialized_chars": 13046, "estimated_input_tokens": 3262, "estimated_eval_tokens": 6780 }, @@ -23037,36 +23046,36 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All four pull requests are distinct model-addition efforts (ViT NEPA, Molmo, CTSM, Step3p7). They share some generated boilerplate and registry/docs updates, but they do not fix the same code-path problem or represent a mergeable duplicate PR set.", + "summary": "All four pull requests are distinct model additions, not duplicates. The overlap is limited to shared repository boilerplate such as docs, auto registries, and test scaffolding; the actual model targets and code paths differ.", "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: these are separate model integrations for different architectures and code paths, not duplicates of one another.", + "canonical_pr_reason": "No canonical PR: these PRs add different models (ViT NEPA, Molmo, CTSM, Step3p7) and only overlap on shared infra files, not the same underlying change.", "best_issue_reason": null, - "best_pr_reason": "No single best PR for deduplication purposes; each PR targets a different model and would not be merged into one change.", + "best_pr_reason": "No single PR is a good cluster representative for duplicate triage; 46562 is the most developed by activity, but it still addresses a different model than the others.", "soft_edge_verdicts": [ { "left": "pull_request:42978", "right": "pull_request:43448", "accept": false, - "reason": "ViT NEPA vs Molmo: different model families, files, and implementation paths; shared auto/docs updates are just boilerplate." + "reason": "ViT NEPA vs Molmo are unrelated model additions; shared auto/docs files are boilerplate, not the same fix." }, { "left": "pull_request:42978", "right": "pull_request:45490", "accept": false, - "reason": "ViT NEPA vs CTSM are unrelated model additions; overlap is limited to common registry/documentation changes." + "reason": "ViT NEPA and CTSM add different model implementations with different code paths and tests." }, { "left": "pull_request:43448", "right": "pull_request:45490", "accept": false, - "reason": "Molmo and CTSM add different models with different configs/modeling/processing code, so they are not the same underlying change." + "reason": "Molmo and CTSM are separate model ports; the common files reflect registry/docs updates, not duplicate functionality." }, { "left": "pull_request:45490", "right": "pull_request:46562", "accept": false, - "reason": "CTSM and Step3p7 are separate model integrations; shared auto-mapping/docs edits do not indicate a duplicate fix or mergeable single PR." + "reason": "CTSM and Step3p7 are different model additions; despite some overlapping auto-mapping/docs changes, they do not fix the same concrete problem." } ] }, @@ -23115,130 +23124,130 @@ "pull_request:46539|pull_request:46922", "pull_request:46886|pull_request:46922" ], - "prepared_review_unit_hash": "93f1737a19a295cd07ed334612090963ca9b1e62a8257a56e9359650ca385383", + "prepared_review_unit_hash": "7a3a96288ec154605a843968f2dbc393e9582290a6b09f0d7d73fbed5f836ddf", "component_budget": { "node_count": 7, "item_count": 7, "soft_pair_count": 15, - "serialized_chars": 19595, - "estimated_input_tokens": 4899, - "estimated_eval_tokens": 10054 + "serialized_chars": 19601, + "estimated_input_tokens": 4901, + "estimated_eval_tokens": 10058 }, "budget": { "node_count": 7, "item_count": 7, "soft_pair_count": 15, - "serialized_chars": 19595, - "estimated_input_tokens": 4899, - "estimated_eval_tokens": 10054 + "serialized_chars": 19601, + "estimated_input_tokens": 4901, + "estimated_eval_tokens": 10058 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Only one pair looks like a true duplicate: the two YOLOS fixes (46886 and 46922) both target the same device-mismatch issue under `device_map=\"auto\"`. The rest are unrelated PRs that merely share superficial similarity or the word \u201cfix\u201d.", - "confidence": 0.96, + "summary": "Most items are unrelated PRs; the only strong duplicate-looking pair is the YOLOS device-mismatch fixes, which target the same issue and file. The rest differ by model, subsystem, or bug shape and should stay separate.", + "confidence": 0.89, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46922 is the best representative because it addresses the YOLOS bug directly in the forward path and includes the split-module adjustment; 46886 looks like an earlier, narrower precursor to the same fix.", + "canonical_pr_reason": "pull_request:46922 is the best representative of the duplicate YOLOS fix: it addresses the concrete runtime device mismatch in the forward path and also includes the supporting no-split-module change. pull_request:46886 is a narrower variant of the same fix.", "best_issue_reason": null, - "best_pr_reason": "46922 is the clearest and most complete concrete code fix in the set, whereas the other PRs are unrelated or test/CI-only changes.", + "best_pr_reason": "pull_request:46922 is the strongest overall PR in the cluster because it fixes the specific YOLOS failure mode directly and subsumes the related auto device-map adjustment, unlike the other PRs which are clearly unrelated.", "soft_edge_verdicts": [ { "left": "pull_request:43291", "right": "pull_request:43665", "accept": false, - "reason": "Different models and different failure modes; one is Whisper tokenizer/test work, the other is unrelated CLIP/ViT test edits." + "reason": "Both are test-only PRs, but they target different models and different failures (Whisper decoding/tests vs CLIP/ViT config tests)." }, { "left": "pull_request:43291", "right": "pull_request:44815", "accept": false, - "reason": "Whisper decoding/tests vs FP8 dequantization/loading changes; no shared code path or bug." + "reason": "Whisper tokenizer/test changes are unrelated to FP8 dequantization and model loading logic." }, { "left": "pull_request:43291", "right": "pull_request:46380", "accept": false, - "reason": "Whisper test/tokenization fix is unrelated to generation watermarking validation and other model test updates." + "reason": "Different subsystems entirely: Whisper decoding tests vs generation config validation and watermarking handling." }, { "left": "pull_request:43291", "right": "pull_request:46922", "accept": false, - "reason": "Whisper decode/test changes and YOLOS device-placement changes are unrelated." + "reason": "Whisper test fixes and YOLOS device-mismatch fixes are unrelated model-specific problems." }, { "left": "pull_request:43665", "right": "pull_request:46539", "accept": false, - "reason": "Dummy test assertions in CLIP/ViT are not the same issue as Phimoe integration test/CI memory handling." + "reason": "Both touch tests, but one is CLIP/ViT test stubbing and the other is Phimoe CI/integration setup; not the same bug." }, { "left": "pull_request:43665", "right": "pull_request:44815", "accept": false, - "reason": "Test-only CLIP/ViT edits do not match FP8 dequantization/model-loading fixes." + "reason": "CLIP/ViT test edits do not match the FP8 dequant/model-loading code path." }, { "left": "pull_request:43665", "right": "pull_request:46380", "accept": false, - "reason": "Unrelated test files and unrelated generation/config/model-loading changes." + "reason": "No shared concrete failure: model config tests vs generation watermarking validation." }, { "left": "pull_request:43665", "right": "pull_request:46922", "accept": false, - "reason": "No shared concrete bug; test placeholders vs YOLOS runtime device mismatch." + "reason": "Different model families and fixes; the similarity is only superficial test/model code touch." }, { "left": "pull_request:44815", "right": "pull_request:46539", "accept": false, - "reason": "FP8 quantization/dequant code is unrelated to CI runner memory limits and Phimoe integration testing." + "reason": "FP8 dequantization/loading changes are unrelated to Phimoe CI runner memory/device setup." }, { "left": "pull_request:44815", "right": "pull_request:46380", "accept": false, - "reason": "Different subsystems and different bugs: FP8 dequantization versus generation config validation." + "reason": "Different code paths: quantization/dequant loading vs generation configuration validation." }, { "left": "pull_request:44815", "right": "pull_request:46922", "accept": false, - "reason": "No common code-path problem; FP8 dequant fix and YOLOS device mismatch are separate." + "reason": "FP8 quantization fixes and YOLOS device placement are separate bugs." }, { "left": "pull_request:46380", "right": "pull_request:46539", "accept": false, - "reason": "Generation config/watermarking changes are unrelated to Phimoe integration test infrastructure." + "reason": "Generation watermarking validation and Phimoe integration test infrastructure are unrelated." }, { "left": "pull_request:46380", "right": "pull_request:46922", "accept": false, - "reason": "Generation validation fix and YOLOS device mismatch fix do not share the same underlying bug." + "reason": "Different models and different runtime issues; no shared concrete fix." }, { "left": "pull_request:46539", "right": "pull_request:46922", "accept": false, - "reason": "CI/test-environment changes are unrelated to YOLOS model device placement." + "reason": "Phimoe CI/device-map test infrastructure is not the same underlying bug as YOLOS mid-position embedding placement." }, { "left": "pull_request:46886", "right": "pull_request:46922", "accept": true, - "reason": "Both PRs address the same YOLOS device-mismatch issue for `device_map=\"auto\"` in the same file and could plausibly be merged into one fix." + "reason": "Same YOLOS device-mismatch issue on the same file and same explicit target (46869); both changes address the same concrete failure path and could plausibly be merged into one PR." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet, and the soft-edge judgments are conservative overall. The only accepted pair, 46886/46922, is plausibly mergeable as a single YOLOS bugfix since they touch the same file, share the same issue target, and 46922 extends the earlier no-split-module fix with the forward-path device move. The remaining pairs are reasonably rejected as unrelated." + "feedback": "The summary is grounded in the packet and stays conservative on the soft edges. The only clearly duplicate-looking pair is the YOLOS fixes (46886/46922), which share the same file and target the same issue; the other pairwise rejections are well-supported by different models, subsystems, or bug shapes. No overstatement stands out." }, "evaluator_used": true, "retried": false, @@ -23268,52 +23277,52 @@ "pull_request:43488|pull_request:43532", "pull_request:43532|pull_request:46747" ], - "prepared_review_unit_hash": "c494be56a5730115f91b9d64b01651fad05cd37f594688c547ca6eaef0a08871", + "prepared_review_unit_hash": "917daa183669b57e36a7f5ac3c2b228b2da406b1ed7fa002ef6292511d27191d", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 6349, - "estimated_input_tokens": 1588, - "estimated_eval_tokens": 3432 + "serialized_chars": 6348, + "estimated_input_tokens": 1587, + "estimated_eval_tokens": 3430 }, "budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 6349, - "estimated_input_tokens": 1588, - "estimated_eval_tokens": 3432 + "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": "These three pull requests are unrelated: one tweaks packaging/test deps, one rewires several CI/workflow/test files, and one only changes a security-gate workflow trigger. They share broad testing/CI context but not the same underlying bug or change, so none should be merged as duplicates.", + "summary": "These are three unrelated '[don't merge]' test/demo PRs. They all have superficial similarity, but each changes a different area of the repo, so none should be merged as duplicates.", "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "pull_request:43532 is the broadest and most representative of the cluster\u2019s test/CI demo changes.", "best_issue_reason": null, - "best_pr_reason": null, + "best_pr_reason": "It is only a weak placeholder canonical: the three PRs do not fix the same concrete problem or implement the same change.", "soft_edge_verdicts": [ { "left": "pull_request:43488", "right": "pull_request:46747", "accept": false, - "reason": "Different changesets: 43488 is packaging/dependency/test setup work, while 46747 only changes the workflow trigger for the PR security gate. Same general repo area, but not the same fix." + "reason": "Different changes: #43488 is a setup/dependency/version bump with a Vit test tweak, while #46747 only changes the PR security-gate workflow. No shared code-path or mergeable common fix." }, { "left": "pull_request:43488", "right": "pull_request:43532", "accept": false, - "reason": "43488 adds pytest-random-order and bumps version metadata; 43532 is a broader CI/workflow and testing-utils diff. Shared test filename is incidental, not a shared code-path bug." + "reason": "#43488 is about packaging/version/dependency table changes; #43532 is about workflows, testing utilities, and a Vit test. The shared test filename is incidental, not the same underlying bug/change." }, { "left": "pull_request:43532", "right": "pull_request:46747", "accept": false, - "reason": "43532 modifies multiple CI workflows plus testing utilities, whereas 46747 makes a single workflow trigger change. They are not addressing the same concrete problem and would not be merged into one PR." + "reason": "#43532 spans workflow/test utility updates and a test file, but #46747 only edits the PR security-gate workflow. They are different CI changes with different intent and code paths." } ] }, @@ -23343,12 +23352,12 @@ "soft_pairs": [ "pull_request:43612|pull_request:43613" ], - "prepared_review_unit_hash": "30f4a7970318ae8c69dfbac40e1c9fc104df562295b4eb9bd6bc4c5652b70bee", + "prepared_review_unit_hash": "eece3935d47f91dd761df227e50132568bf0a121ea89472c872a554fd5f40f9e", "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 }, @@ -23356,7 +23365,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6082, + "serialized_chars": 6083, "estimated_input_tokens": 1521, "estimated_eval_tokens": 3298 }, @@ -23365,18 +23374,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two distinct feature PRs: one adds a Promptable Concept Segmentation pipeline, the other adds a Promptable Visual Segmentation pipeline. They share infrastructure and docs updates, but they target different pipelines, model integrations, and task docs, so they are not duplicates.", - "confidence": 0.93, + "summary": "These are not duplicates: both add new pipeline-related docs/code, but one is for Promptable Concept Segmentation and the other for Promptable Visual Segmentation, with different pipeline classes and different model/test integrations.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "No clear canonical duplicate here; the PRs add different pipelines and touch different concrete model paths. If a representative is needed, #43613 is the broader visual-segmentation addition.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "#43613 is the stronger representative because it has broader model coverage and a larger implementation surface, but it is still a separate feature from #43612 rather than a duplicate.", + "best_pr_reason": "pull_request:43613 is the more substantial standalone PR, with broader model/test coverage and a clearly defined visual-segmentation code path; pull_request:43612 is a separate concept-segmentation addition, not a duplicate.", "soft_edge_verdicts": [ { "left": "pull_request:43612", "right": "pull_request:43613", "accept": false, - "reason": "They are related but not the same change: one implements promptable concept segmentation, the other promptable visual segmentation. Different pipeline modules, task docs, and model integrations mean they should not be merged as duplicates." + "reason": "Different underlying changes: concept segmentation vs visual segmentation. The overlap is mostly boilerplate/docs/registration; they do not target the same concrete bug or code path, so they should not be merged as duplicates." } ] }, @@ -23406,40 +23415,40 @@ "soft_pairs": [ "pull_request:43651|pull_request:43663" ], - "prepared_review_unit_hash": "b2fda3262f35020a11e00360fcca519c1be346b725f8e65a32493a53f6487261", + "prepared_review_unit_hash": "26f377ca772e361c2ca5d3704d14ee2e8b39ae43441d2c49de38a71efeb7cc7b", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4292, - "estimated_input_tokens": 1073, - "estimated_eval_tokens": 2402 + "serialized_chars": 4293, + "estimated_input_tokens": 1074, + "estimated_eval_tokens": 2404 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4292, - "estimated_input_tokens": 1073, - "estimated_eval_tokens": 2402 + "serialized_chars": 4293, + "estimated_input_tokens": 1074, + "estimated_eval_tokens": 2404 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two PRs both touch `trainer.py` but address different trainer extension points: one changes gradient-accumulation loss scaling, the other refactors input column filtering/signature inspection. They are not duplicates.", - "confidence": 0.95, + "summary": "These two PRs both touch `trainer.py` but fix unrelated Trainer extension points: one changes gradient-accumulation loss scaling, the other extracts signature-column filtering for PEFT/custom trainers. Shared file alone is not enough to treat them as duplicates.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "No single canonical PR: PR #43651 and PR #43663 fix unrelated behaviors in different parts of `Trainer` and are not mergeable into one duplicate change.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "Neither PR is a better duplicate representative of the other; they target separate code paths and should be triaged independently.", + "best_pr_reason": "No single PR is a duplicate/canonical representative of the other; they target different code paths and could be merged independently.", "soft_edge_verdicts": [ { "left": "pull_request:43651", "right": "pull_request:43663", "accept": false, - "reason": "Same file, but different underlying problems: #43651 changes loss normalization during training steps, while #43663 adds an overridable signature-column helper for column filtering. They do not fix the same bug or change." + "reason": "Different behaviors: 43651 adjusts loss normalization during training, while 43663 refactors input-signature column filtering. Same file, but not the same bug/change." } ] }, @@ -23469,7 +23478,7 @@ "soft_pairs": [ "pull_request:43656|pull_request:43842" ], - "prepared_review_unit_hash": "bf8649c53afc9e8824551e9c70c92387cfea822d3bdaa2a5466483a52f61a16a", + "prepared_review_unit_hash": "f9a0502e8d58276da58810ee2965909a1e9881a6f1a037b1bba80a6da604bd6a", "component_budget": { "node_count": 2, "item_count": 2, @@ -23491,25 +23500,28 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two open PRs touch the same CLI module but they fix different NameError/typing issues: one broad annotation/runtime typing cleanup, and one specific TypeAdapter import problem when pydantic is absent. They are related but not the same underlying change.", - "confidence": 0.87, + "summary": "Both pull requests target the same runtime TypeAdapter annotation failure in src/transformers/cli/serve.py; 43842 is the more focused fix, while 43656 is a broader typing cleanup around the same code path.", + "confidence": 0.84, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #43842 is the clearest, most targeted fix for the explicit TypeAdapter NameError path: it confines the import to TYPE_CHECKING and quotes the annotation, matching the reported bug closely.", + "canonical_pr_reason": "Prefer PR 43842: it directly addresses the TypeAdapter NameError when pydantic is unavailable by moving the import under TYPE_CHECKING and quoting the annotation, making it the tighter fix for the reported runtime bug.", "best_issue_reason": null, - "best_pr_reason": "PR #43842 is the best representative because it fixes a concrete, narrow code path with minimal surface area. PR #43656 appears to address a different annotation/name-resolution problem in the same file, so it is less suitable as the canonical change.", + "best_pr_reason": "PR 43842 is the best representative because it is the most explicit, minimal correction for the concrete failure mode; PR 43656 is related but broader and less directly scoped.", "soft_edge_verdicts": [ { "left": "pull_request:43656", "right": "pull_request:43842", - "accept": false, - "reason": "Same file and similar theme, but the fixes target different NameError causes: #43656 is a broader typing/annotation cleanup, while #43842 specifically addresses TypeAdapter import handling when pydantic is not installed. They are related but not the same bug." + "accept": true, + "reason": "Same underlying serve.py TypeAdapter NameError in the CLI, and the two fixes are compatible enough that they could plausibly be combined into one PR." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "Grounded overall: both PRs are in src/transformers/cli/serve.py and the packet supports treating them as alternative fixes for the same TypeAdapter NameError. The soft-pair verdict is conservative enough here because the changes are compatible and could be combined in one PR, even if 43656 is broader." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -23694,286 +23706,382 @@ "pull_request:44044|pull_request:44129", "pull_request:44044|pull_request:44154" ], - "prepared_review_unit_hash": "b93b3e2b1e06a914808e85c56d9662e23f16dff67de313680c08dbce842e2d44", + "prepared_review_unit_hash": "20769dffad22993f340d151e961f32da8368a1395aa22cf77e56a4f1056f0385", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 78, - "serialized_chars": 84897, - "estimated_input_tokens": 21225, - "estimated_eval_tokens": 42706 + "serialized_chars": 84894, + "estimated_input_tokens": 21224, + "estimated_eval_tokens": 42704 }, "budget": { "node_count": 27, "item_count": 27, "soft_pair_count": 72, - "serialized_chars": 75092, - "estimated_input_tokens": 18773, - "estimated_eval_tokens": 37802 + "serialized_chars": 75093, + "estimated_input_tokens": 18774, + "estimated_eval_tokens": 37804 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of parallel output-tracing refactors across different model families. Most soft links should be rejected because they share only the general theme, not the same concrete code path. The clear duplicates are the GPT-Neo pair and the GPT-J/CodeGen line of PRs.", - "confidence": 0.89, - "canonical_issue_reason": null, - "canonical_pr_reason": "PR 44722 is the strongest canonical PR candidate: it targets the same standardized output-tracing refactor, touches the core GPT-J/CodeGen path, has the strongest overlap/evidence, and appears to be the most complete later variant of that change set.", - "best_issue_reason": null, - "best_pr_reason": "PR 44722 is the best representative PR for the cluster because it most fully captures the underlying decorator-based output-tracing change in the shared GPT-J/CodeGen code path and has the strongest duplicate evidence among the listed PRs.", + "summary": "This cluster is mostly a set of parallel model-by-model refactors for standardized output tracing around issue 43979, not one single bug fix. The only clear duplicate-like overlaps are within the same model code paths, especially GPT-J/GPT-Neo variants.", + "confidence": 0.8, + "canonical_issue_reason": "43979 is the umbrella tracking issue for the standardized output-tracing refactor family, and many of the PRs explicitly target it.", + "canonical_pr_reason": "44722 is the strongest representative PR in the cluster: it is the latest GPT-J/CodeGen refactor, hits the same model code path as nearby GPT-J variants, and appears to be the most complete version of that subcluster.", + "best_issue_reason": "No issue artifact is present in the items, so 43979 is best treated as the organizing issue inferred from the PR targets.", + "best_pr_reason": "44722 is the best overall PR exemplar because it is later, issue-linked, and captures the same GPT-J output-tracing refactor as the overlapping GPT-J PRs while also covering CodeGen.", "soft_edge_verdicts": [ { "left": "pull_request:43996", "right": "pull_request:44085", "accept": false, - "reason": "Different model families and concrete code paths; same output-tracing theme is not enough to treat these as duplicates." + "reason": "Different models and code paths; sharing the same refactor theme and issue target is not enough." }, { "left": "pull_request:43996", "right": "pull_request:44044", "accept": false, - "reason": "Different model families and different implementation paths; only the refactor theme overlaps." + "reason": "Different model families (FNet/CVT vs DeBERTaV2); only the umbrella issue is shared." }, { "left": "pull_request:44066", "right": "pull_request:44085", "accept": true, - "reason": "Same GPT-J output-tracing refactor in the same underlying file/code path; this looks like a duplicate/superset variant of the same change." + "reason": "Same GPT-J output-tracing code path in modeling_gptj.py; one PR is just a broader variant of the other." }, { "left": "pull_request:44007", "right": "pull_request:44072", "accept": false, - "reason": "RegNet vs EfficientNet are different model code paths; same issue target alone does not make them duplicates." + "reason": "Different models; both are output-tracing refactors but not the same concrete change." }, { "left": "pull_request:44072", "right": "pull_request:44722", "accept": false, - "reason": "EfficientNet and GPT-J/CodeGen are unrelated concrete code paths; these are only thematically similar." + "reason": "EfficientNet and GPT-J/CodeGen are unrelated code paths." }, { "left": "pull_request:44066", "right": "pull_request:44072", "accept": false, - "reason": "GPT-J/CodeGen versus EfficientNet are different model implementations; not the same underlying bug/change." + "reason": "Different model code paths; same refactor pattern only." }, { "left": "pull_request:44013", "right": "pull_request:44044", "accept": false, - "reason": "MobileNetV2 and DeBERTa V2 are unrelated code paths; shared tracking context is not enough." + "reason": "MobileNetV2 and DeBERTaV2 are unrelated implementations." }, { "left": "pull_request:44066", "right": "pull_request:44086", "accept": false, - "reason": "GPT-J/CodeGen output tracing and MGP-STR output tracing are different model-specific changes." + "reason": "Different models; no shared code path." }, { "left": "pull_request:44066", "right": "pull_request:44071", "accept": false, - "reason": "GPT-J/CodeGen and MPT are different concrete implementations; only the refactor pattern overlaps." + "reason": "GPT-J/CodeGen and MPT are unrelated model implementations." }, { "left": "pull_request:44018", "right": "pull_request:44068", "accept": true, - "reason": "Same GPT-Neo output-tracing refactor in the same file/test area; these are the same concrete change path." + "reason": "Both are GPT-Neo output-tracing refactors on the same model file and are effectively overlapping revisions." }, { "left": "pull_request:44066", "right": "pull_request:44068", "accept": false, - "reason": "GPT-J/CodeGen and GPT-Neo are different model code paths, so they are not duplicates." + "reason": "GPT-J and GPT-Neo are different model code paths." }, { "left": "pull_request:44018", "right": "pull_request:44066", "accept": false, - "reason": "GPT-Neo and GPT-J/CodeGen are different model families; same standardized-output theme is too broad." + "reason": "GPT-Neo vs GPT-J; same family of refactor, but not the same underlying change." }, { "left": "pull_request:44019", "right": "pull_request:44722", "accept": false, - "reason": "ResNet and GPT-J/CodeGen are unrelated concrete code paths." + "reason": "ResNet and GPT-J/CodeGen are different code paths." }, { "left": "pull_request:44086", "right": "pull_request:44722", "accept": false, - "reason": "MGP-STR and GPT-J/CodeGen are different model implementations; not the same underlying change." + "reason": "MGP-STR and GPT-J/CodeGen are unrelated models." }, { "left": "pull_request:44019", "right": "pull_request:44085", "accept": false, - "reason": "ResNet and GPT-J are different model families; only the refactor style is similar." + "reason": "ResNet and GPT-J are different concrete implementations." }, { "left": "pull_request:44018", "right": "pull_request:44019", "accept": false, - "reason": "GPT-Neo and ResNet are different code paths; these do not look like the same bug/change." + "reason": "GPT-Neo and ResNet are unrelated model code paths." }, { "left": "pull_request:44019", "right": "pull_request:44071", "accept": false, - "reason": "ResNet and MPT are unrelated model implementations." + "reason": "ResNet and MPT are unrelated; only the high-level refactor theme matches." }, { "left": "pull_request:44068", "right": "pull_request:44722", "accept": false, - "reason": "GPT-Neo and GPT-J/CodeGen are different concrete code paths, even though both use the same tracing pattern." + "reason": "GPT-Neo and GPT-J/CodeGen are different model families." }, { "left": "pull_request:44018", "right": "pull_request:44116", "accept": false, - "reason": "GPT-Neo and Flaubert are different model-specific implementations; not a duplicate." + "reason": "GPT-Neo and Flaubert are different code paths." }, { "left": "pull_request:44068", "right": "pull_request:44116", "accept": false, - "reason": "GPT-Neo and Flaubert are unrelated code paths." - }, - { - "left": "pull_request:44071", - "right": "pull_request:44722", - "accept": false, - "reason": "MPT and GPT-J/CodeGen are different model families and code paths." - }, - { - "left": "pull_request:44018", - "right": "pull_request:44085", - "accept": false, - "reason": "GPT-Neo and GPT-J are different model-specific changes, so they should not be merged as duplicates." + "reason": "GPT-Neo and Flaubert are unrelated models." }, { "left": "pull_request:44085", "right": "pull_request:44086", "accept": false, - "reason": "GPT-J and MGP-STR are different concrete code paths." + "reason": "GPT-J and MGP-STR are unrelated code paths." }, { "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 different models." }, { "left": "pull_request:44098", "right": "pull_request:44101", "accept": false, - "reason": "ViLT and Flaubert/XLM are different model code paths; only the broad refactor theme overlaps." + "reason": "ViLT and Flaubert/XLM are unrelated implementations." }, { "left": "pull_request:44059", "right": "pull_request:44722", "accept": false, - "reason": "GPT-2 and GPT-J/CodeGen are different model families; not the same underlying change." + "reason": "GPT-2 and GPT-J/CodeGen are different concrete code paths." }, { "left": "pull_request:44017", "right": "pull_request:44722", "accept": false, - "reason": "SegFormer and GPT-J/CodeGen are different concrete implementations." + "reason": "Segformer and GPT-J/CodeGen are unrelated models." }, { "left": "pull_request:44017", "right": "pull_request:44161", "accept": false, - "reason": "SegFormer and LongT5 are unrelated model code paths." + "reason": "Segformer and LongT5 are different code paths." }, { "left": "pull_request:44017", "right": "pull_request:44073", "accept": false, - "reason": "SegFormer and VisualBert are different model-specific refactors." + "reason": "Segformer and VisualBert are unrelated models." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-Neo and MPT are unrelated code paths." }, { "left": "pull_request:44071", "right": "pull_request:44076", "accept": false, - "reason": "MPT and ImageGPT are different concrete code paths." + "reason": "MPT and ImageGPT are different model implementations." }, { "left": "pull_request:44074", "right": "pull_request:44101", "accept": false, - "reason": "TextNet and Flaubert/XLM are unrelated model implementations." + "reason": "TextNet and Flaubert/XLM are unrelated models." }, { "left": "pull_request:43997", - "right": "pull_request:44018", + "right": "pull_request:44017", "accept": false, - "reason": "RegNet and GPT-Neo are different model code paths." + "reason": "RegNet and Segformer are different code paths." }, { - "left": "pull_request:44056", - "right": "pull_request:44101", + "left": "pull_request:43997", + "right": "pull_request:44056", "accept": false, - "reason": "MPNet and Flaubert/XLM are different concrete changes." + "reason": "RegNet and MPNet are unrelated implementations." }, { - "left": "pull_request:43997", + "left": "pull_request:44071", + "right": "pull_request:44129", + "accept": false, + "reason": "MPT and SpeechT5 are different models." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44161", + "accept": false, + "reason": "FNet/CVT and LongT5 are unrelated code paths." + }, + { + "left": "pull_request:43996", "right": "pull_request:44017", "accept": false, - "reason": "RegNet and SegFormer are different model families and code paths." + "reason": "FNet/CVT and Segformer are different model families." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44071", + "accept": false, + "reason": "SqueezeBert and MPT are unrelated implementations." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44073", + "accept": false, + "reason": "SqueezeBert and VisualBert are different code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44018", + "accept": false, + "reason": "RegNet and GPT-Neo are unrelated models." }, { "left": "pull_request:43997", "right": "pull_request:44161", "accept": false, - "reason": "RegNet and LongT5 are unrelated model implementations." + "reason": "RegNet and LongT5 are different code paths." }, { "left": "pull_request:44071", "right": "pull_request:44076", "accept": false, - "reason": "MPT and ImageGPT are different concrete code paths." + "reason": "MPT and ImageGPT are unrelated model implementations." }, { "left": "pull_request:44074", "right": "pull_request:44101", "accept": false, - "reason": "TextNet and Flaubert/XLM are unrelated model-specific refactors." + "reason": "TextNet and Flaubert/XLM are different concrete changes." }, { - "left": "pull_request:43997", + "left": "pull_request:43996", + "right": "pull_request:44161", + "accept": false, + "reason": "FNet/CVT and LongT5 are unrelated code paths." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44071", + "accept": false, + "reason": "ResNet and MPT are different model families." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44101", + "accept": false, + "reason": "MPNet and Flaubert/XLM are unrelated code paths." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-2 and MPT are different models." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44101", + "accept": false, + "reason": "MPNet and Flaubert/XLM are unrelated implementations." + }, + { + "left": "pull_request:43996", "right": "pull_request:44017", "accept": false, - "reason": "RegNet and SegFormer are different model families; same issue target alone is insufficient." + "reason": "FNet/CVT and Segformer are different model code paths." }, { - "left": "pull_request:43997", - "right": "pull_request:44161", + "left": "pull_request:44010", + "right": "pull_request:44044", "accept": false, - "reason": "RegNet and LongT5 are different concrete code paths." + "reason": "SqueezeBert and DeBERTaV2 are unrelated models." }, { - "left": "pull_request:44071", - "right": "pull_request:44076", + "left": "pull_request:44044", + "right": "pull_request:44129", + "accept": false, + "reason": "DeBERTaV2 and SpeechT5 are different code paths." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44154", + "accept": false, + "reason": "DeBERTaV2 and Vits are unrelated model implementations." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44722", + "accept": true, + "reason": "Same GPT-J/CodeGen output-tracing refactor family on the same underlying code path; 44722 is the broader later variant." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44722", + "accept": true, + "reason": "Both touch GPT-J output tracing in modeling_gptj.py; 44722 is the broader variant that subsumes the same change." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-J and SpeechT5 are unrelated models." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44007", + "accept": false, + "reason": "FNet/CVT and RegNet/ResNet/RT-DETR are different code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44129", + "accept": false, + "reason": "FNet/CVT and SpeechT5 are unrelated models." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44129", "accept": false, - "reason": "MPT and ImageGPT are different model-specific changes." + "reason": "DeBERTaV2 and SpeechT5 are different code paths." }, { "left": "pull_request:44044", "right": "pull_request:44154", "accept": false, - "reason": "DeBERTa V2 and ViT are different model code paths; only the output-handling theme overlaps." + "reason": "DeBERTaV2 and Vits are unrelated model implementations." } ] }, "evaluator_result": { "accept": true, - "feedback": "Grounded and conservative overall. The summary correctly distinguishes broad output-tracing similarity from actual duplicate/mergeable PRs, and the accepted soft-edge pairs are within the same concrete GPT-Neo and GPT-J/CodeGen paths. The canonical PR choice of 44722 is supported by the packet\u2019s shared gptj/codegen evidence and later timestamp." + "feedback": "The summary stays grounded in the packet: 43979 is clearly the shared umbrella target across many PRs, and 44722 is a reasonable representative because it is later, issue-linked, and spans both GPT-J and CodeGen. The soft-edge decisions are mostly conservative and the accepted pairs are plausibly mergeable as overlapping revisions of the same output-tracing bugfix within the same model code paths." }, "evaluator_used": true, "retried": false, @@ -24011,70 +24119,70 @@ "pull_request:43997|pull_request:44010", "pull_request:44010|pull_request:44129" ], - "prepared_review_unit_hash": "006e08c6d1a861eea82677912780118d385557d9a1a29f44ba068b6b3a0bd0f9", + "prepared_review_unit_hash": "f3bddcf8f81f7b7e4553a86d6b906b62cc5546e91429cb580876d45675123a57", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 78, - "serialized_chars": 84897, - "estimated_input_tokens": 21225, - "estimated_eval_tokens": 42706 + "serialized_chars": 84894, + "estimated_input_tokens": 21224, + "estimated_eval_tokens": 42704 }, "budget": { "node_count": 8, "item_count": 8, "soft_pair_count": 6, - "serialized_chars": 17126, - "estimated_input_tokens": 4282, - "estimated_eval_tokens": 8820 + "serialized_chars": 17122, + "estimated_input_tokens": 4281, + "estimated_eval_tokens": 8818 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These PRs are all in the same broad initiative to standardize output capture, but they target different model implementations and different code paths. They are not duplicates of one another and should not be merged together as a single PR.", - "confidence": 0.95, + "summary": "A cluster of model-specific PRs refactoring output tracing to standardized capture_outputs/can_return_tuple patterns. They are thematically similar, but each touches a different model and code path, so they are not duplicates of the same concrete change.", + "confidence": 0.84, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #44129 is the strongest canonical representative because it has an explicit issue target, is clearly scoped to one model, and shows a concrete refactor of SpeechT5 output tracing rather than a generic template change.", + "canonical_pr_reason": "PR 44129 is the strongest canonical representative: it is a focused output-tracing refactor, has an explicit issue target, and includes a substantial implementation change rather than only import/decorator plumbing.", "best_issue_reason": null, - "best_pr_reason": "PR #44129 is the best standalone PR candidate because it is the most explicitly anchored to a tracked problem and appears self-contained and concrete for one code path.", + "best_pr_reason": "PR 44129 is the best overall fit because it is concrete, self-contained, and directly tied to a tracking issue; the other PRs are similar refactors but for different model implementations.", "soft_edge_verdicts": [ { "left": "pull_request:44024", "right": "pull_request:44076", "accept": false, - "reason": "Both are standardized output/capture refactors, but they touch different models (FocalNet vs ImageGPT) and different implementation paths. Same theme, not the same underlying bug/change." + "reason": "Both are standardized output-tracing refactors, but for different model families (FocalNet vs ImageGPT) and different classes/forward paths; not the same underlying bug or mergeable into one PR." }, { "left": "pull_request:44073", "right": "pull_request:44074", "accept": false, - "reason": "VisualBert and TextNet are separate model-specific output tracing changes. They may share the same refactor pattern, but they are not the same concrete fix." + "reason": "VisualBert and TextNet are separate model implementations with different output contracts; this is a shared refactor pattern, not one concrete code-path fix." }, { "left": "pull_request:44056", "right": "pull_request:44073", "accept": false, - "reason": "MPNet and VisualBert are unrelated model code paths. Shared decorator/capture-output migration theme is too broad to treat as duplicate." + "reason": "MPNet and VisualBert touch different encoder stacks and tracing hooks; they are similar cleanup PRs, not duplicates of the same change." }, { "left": "pull_request:44056", "right": "pull_request:44074", "accept": false, - "reason": "MPNet and TextNet are different model implementations with separate output plumbing. This is a common refactor pattern, not one duplicate PR." + "reason": "MPNet vs TextNet is only a broad output-tracing modernization theme; the code paths and model-specific hook mappings differ." }, { "left": "pull_request:43997", "right": "pull_request:44010", "accept": false, - "reason": "RegNet and SqueezeBert are distinct models and the diffs target different forward/output handling. The overlap is only in the standardized tracing migration." + "reason": "RegNet and SqueezeBert are unrelated architectures and forward implementations; no shared concrete bug or patch to merge together." }, { "left": "pull_request:44010", "right": "pull_request:44129", "accept": false, - "reason": "SqueezeBert and SpeechT5 are separate model-specific changes. The shared output-capture machinery does not make them the same change or bug fix." + "reason": "SqueezeBert and SpeechT5 both adjust output collection, but they are separate model-specific refactors with different APIs and internals, so they are not duplicates." } ] }, @@ -24154,274 +24262,274 @@ "pull_request:44028|pull_request:44030", "pull_request:44029|pull_request:44030" ], - "prepared_review_unit_hash": "28fedce9024713b12095a380c29284b6024e4c1fcab5cf52fb285b37750f09e9", + "prepared_review_unit_hash": "f9b48bb67c0197c3ef5bb26685902bf8a6be6aebdbf38975bcebdf6f51c5e395", "component_budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 40, - "serialized_chars": 40170, - "estimated_input_tokens": 10043, - "estimated_eval_tokens": 20342 + "serialized_chars": 40173, + "estimated_input_tokens": 10044, + "estimated_eval_tokens": 20344 }, "budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 40, - "serialized_chars": 40170, - "estimated_input_tokens": 10043, - "estimated_eval_tokens": 20342 + "serialized_chars": 40173, + "estimated_input_tokens": 10044, + "estimated_eval_tokens": 20344 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All candidates are similar output-tracing refactors, but they target different model implementations and code paths. I do not see any true duplicate PRs here; this looks like a theme cluster rather than the same underlying change.", - "confidence": 0.93, + "summary": "These are separate output-tracing refactors across unrelated model families. They share a maintenance theme, but not the same concrete bug or code-path, so I do not treat any pair as duplicates.", + "confidence": 0.94, "canonical_issue_reason": null, - "canonical_pr_reason": "No PR pair is a true duplicate. If one PR must represent the cluster, 44003 is the strongest representative because it is the broadest and most concrete tracing change, spanning Mamba/FalconMamba plumbing.", + "canonical_pr_reason": "PR 44003 is the strongest representative of the set: it makes the broadest output-tracing plumbing change and spans two closely related Mamba-family implementations.", "best_issue_reason": null, - "best_pr_reason": "44003 is the best single representative: it has the largest, most concrete change set and touches multiple closely related files, making it the most substantive example of this refactor theme.", + "best_pr_reason": "PR 44003 is the best single representative of this cluster because it is the most substantive and central example of the refactor pattern here.", "soft_edge_verdicts": [ { "left": "pull_request:43998", "right": "pull_request:44002", "accept": false, - "reason": "Same refactor theme, but different models and code paths (timm_backbone vs upernet)." + "reason": "Different model families and files (timm_backbone vs upernet); same refactor pattern only, not the same concrete change." }, { "left": "pull_request:43998", "right": "pull_request:44000", "accept": false, - "reason": "Different model implementations; similar tracing cleanup, not the same underlying change." + "reason": "Different model families and files (timm_backbone vs vision_text_dual_encoder); only a shared output-tracing theme." }, { "left": "pull_request:43998", "right": "pull_request:44001", "accept": false, - "reason": "Different architectures and files; only the refactor pattern is shared." + "reason": "Different code paths in unrelated models (timm_backbone vs univnet); not the same underlying bug/change." }, { "left": "pull_request:43998", "right": "pull_request:44003", "accept": false, - "reason": "Different model families and tracing plumbing; not the same code-path fix." + "reason": "Different subsystems and model families (timm_backbone vs mamba/falcon_mamba); same maintenance pattern only." }, { "left": "pull_request:43998", "right": "pull_request:44029", "accept": false, - "reason": "Different models entirely; same output-tracing motif is too broad to be duplicate." + "reason": "Different model families and implementations (timm_backbone vs rwkv); not a duplicate fix." }, { "left": "pull_request:43999", "right": "pull_request:44001", "accept": false, - "reason": "Different models (MobileNetV1 vs UnivNet); not one concrete shared change." + "reason": "MobileNetV1 and UnivNet are unrelated code paths; both refactor output handling but do not fix the same issue." }, { "left": "pull_request:43999", "right": "pull_request:44002", "accept": false, - "reason": "Separate model code paths; shared wording is only a refactor pattern." + "reason": "Different models and files (mobilenet_v1 vs upernet); shared refactor theme is too broad to merge as one duplicate." }, { "left": "pull_request:43999", "right": "pull_request:44025", "accept": false, - "reason": "Different model stacks and different files; not the same bug/change." + "reason": "MobileNetV1 vs Depth Anything are unrelated model families; not the same underlying bug or change." }, { "left": "pull_request:43999", "right": "pull_request:44026", "accept": false, - "reason": "MobileNetV1 vs VisionEncoderDecoder are unrelated implementations." + "reason": "Different model wrappers and code paths (mobilenet_v1 vs vision_encoder_decoder); only superficial similarity." }, { "left": "pull_request:43999", "right": "pull_request:44027", "accept": false, - "reason": "Different model code paths; both are refactors, not duplicates." + "reason": "MobileNetV1 and Speech Encoder-Decoder are unrelated; not the same fix." }, { "left": "pull_request:44000", "right": "pull_request:44001", "accept": false, - "reason": "Different models (vision-text dual encoder vs UnivNet); not a shared fix." + "reason": "Vision-Text Dual Encoder and UnivNet are unrelated; both use can_return_tuple-style refactors, but not the same bug." }, { "left": "pull_request:44000", "right": "pull_request:44002", "accept": false, - "reason": "Different architectures and output logic; only the refactor style matches." + "reason": "Different architectures and files (vision_text_dual_encoder vs upernet); not a duplicate change." }, { "left": "pull_request:44000", "right": "pull_request:44003", "accept": false, - "reason": "Distinct model families; not the same concrete tracing problem." + "reason": "Vision-Text Dual Encoder and Mamba-family code paths are unrelated; same refactor motif only." }, { "left": "pull_request:44000", "right": "pull_request:44025", "accept": false, - "reason": "Vision-text dual encoder and Depth Anything are unrelated code paths." + "reason": "Different model families (vision_text_dual_encoder vs depth_anything); not the same underlying change." }, { "left": "pull_request:44001", "right": "pull_request:44002", "accept": false, - "reason": "UnivNet and UperNet are unrelated; no shared underlying bug." + "reason": "UnivNet and UperNet are unrelated model code paths; not a duplicate bug fix." }, { "left": "pull_request:44001", "right": "pull_request:44003", "accept": false, - "reason": "Different model stacks and different output-capture implementations." + "reason": "UnivNet and Mamba-family changes address different implementations; only broad output-tracing similarity." }, { "left": "pull_request:44001", "right": "pull_request:44025", "accept": false, - "reason": "Different models and different forward signatures; not mergeable as one PR." + "reason": "UnivNet vs Depth Anything are unrelated; same pattern, different code paths." }, { "left": "pull_request:44002", "right": "pull_request:44029", "accept": false, - "reason": "UperNet vs RWKV are unrelated code paths; similarity is only stylistic." + "reason": "UperNet and RWKV are unrelated models; not the same concrete issue." }, { "left": "pull_request:44002", "right": "pull_request:44003", "accept": false, - "reason": "Separate model families and unrelated tracing changes." + "reason": "UperNet and Mamba-family code are different subsystems; not mergeable as one duplicate fix." }, { "left": "pull_request:44003", "right": "pull_request:44028", "accept": false, - "reason": "Mamba/FalconMamba and SuperPoint are unrelated implementations." + "reason": "Mamba-family and SuperPoint are unrelated; same refactor theme only." }, { "left": "pull_request:44003", "right": "pull_request:44004", "accept": false, - "reason": "Different models and different forward contracts; not the same concrete fix." + "reason": "Mamba-family output tracing changes and CodeGen output tracing changes affect different model code paths; not the same change." }, { "left": "pull_request:43998", "right": "pull_request:44004", "accept": false, - "reason": "timm_backbone and CodeGen target different subsystems and code paths." + "reason": "TimmBackbone and CodeGen are unrelated; they do not fix the same underlying issue." }, { "left": "pull_request:44001", "right": "pull_request:44004", "accept": false, - "reason": "Different model types; similar refactor theme is not enough for duplication." + "reason": "UnivNet and CodeGen are different model families and implementations; no shared concrete bug." }, { "left": "pull_request:44002", "right": "pull_request:44004", "accept": false, - "reason": "UperNet and CodeGen are unrelated changes." + "reason": "UperNet and CodeGen are unrelated code paths; only the refactor style matches." }, { "left": "pull_request:44000", "right": "pull_request:44004", "accept": false, - "reason": "Vision-text dual encoder vs CodeGen are different code paths and behaviors." + "reason": "Vision-Text Dual Encoder and CodeGen are unrelated models; not a duplicate pair." }, { "left": "pull_request:44025", "right": "pull_request:44026", "accept": false, - "reason": "Depth Anything and VisionEncoderDecoder are unrelated model implementations." + "reason": "Depth Anything and Vision-Encoder-Decoder are different model families; same output-tracing cleanup theme only." }, { "left": "pull_request:44025", "right": "pull_request:44027", "accept": false, - "reason": "Different model families; not the same underlying change." + "reason": "Depth Anything and Speech Encoder-Decoder are unrelated code paths; not the same bug/change." }, { "left": "pull_request:44025", "right": "pull_request:44028", "accept": false, - "reason": "Depth Anything and SuperPoint are separate model code paths." + "reason": "Depth Anything and SuperPoint are different model implementations; only broad similarity in output handling." }, { "left": "pull_request:44025", "right": "pull_request:44029", "accept": false, - "reason": "Different architectures; only the tracing refactor pattern is shared." + "reason": "Depth Anything and RWKV are unrelated models; not a duplicate fix." }, { "left": "pull_request:44025", "right": "pull_request:44030", "accept": false, - "reason": "Different models and different forward behavior; not duplicates." + "reason": "Depth Anything and DPR are different model code paths; same refactor theme only." }, { "left": "pull_request:44026", - "right": "pull_request:44027", + "right": "pull_request:44029", "accept": false, - "reason": "VisionEncoderDecoder and SpeechEncoderDecoder are separate wrappers, not one change." + "reason": "Vision-Encoder-Decoder and RWKV are unrelated; no shared concrete bug." }, { "left": "pull_request:44026", - "right": "pull_request:44028", + "right": "pull_request:44027", "accept": false, - "reason": "Different model stacks and output contracts." + "reason": "Vision-Encoder-Decoder and Speech Encoder-Decoder are similar wrappers but still different concrete code paths and files." }, { "left": "pull_request:44026", - "right": "pull_request:44029", + "right": "pull_request:44028", "accept": false, - "reason": "Unrelated models; same refactor motif is too broad." + "reason": "Vision-Encoder-Decoder and SuperPoint are unrelated model families; not the same change." }, { "left": "pull_request:44026", "right": "pull_request:44030", "accept": false, - "reason": "Different code paths and model behavior." + "reason": "Vision-Encoder-Decoder and DPR are unrelated implementations; not a duplicate pair." }, { "left": "pull_request:44027", "right": "pull_request:44028", "accept": false, - "reason": "SpeechEncoderDecoder and SuperPoint are unrelated implementations." + "reason": "Speech Encoder-Decoder and SuperPoint are unrelated models; same refactor motif only." }, { "left": "pull_request:44027", "right": "pull_request:44029", "accept": false, - "reason": "Different model families; not the same concrete bug/change." + "reason": "Speech Encoder-Decoder and RWKV are different code paths; not the same underlying bug/change." }, { "left": "pull_request:44027", "right": "pull_request:44030", "accept": false, - "reason": "Different model wrappers and unrelated forward logic." + "reason": "Speech Encoder-Decoder and DPR are unrelated model families; not a duplicate fix." }, { "left": "pull_request:44028", "right": "pull_request:44029", "accept": false, - "reason": "SuperPoint and RWKV are unrelated models." + "reason": "SuperPoint and RWKV are unrelated implementations; no shared concrete issue." }, { "left": "pull_request:44028", "right": "pull_request:44030", "accept": false, - "reason": "Different architectures and tracing behavior." + "reason": "SuperPoint and DPR are different model code paths; not mergeable as one fix." }, { "left": "pull_request:44029", "right": "pull_request:44030", "accept": false, - "reason": "RWKV and DPR are unrelated code paths; not a duplicate PR pair." + "reason": "RWKV and DPR are unrelated model families; only the output-tracing refactor style matches." } ] }, @@ -24451,7 +24559,7 @@ "soft_pairs": [ "pull_request:44662|pull_request:45401" ], - "prepared_review_unit_hash": "c5bd9a9e4ee84f4d8411fbecf358783cc4539634fde3f73f32177b8701df6892", + "prepared_review_unit_hash": "61401d655252433fd7eeaf00573644d740052924175f480b4c4112c0b39aff66", "component_budget": { "node_count": 2, "item_count": 2, @@ -24473,18 +24581,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two unrelated model-addition PRs. They share boilerplate transformer integration files, but they target different models and different code paths, so they are not duplicates.", + "summary": "These are two unrelated model-addition PRs that only overlap on standard Transformers boilerplate (docs, auto registries, model package wiring). They add different model families and do not fix the same concrete code path.", "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45401 is the cleaner canonical PR because it has an explicit issue target and a tightly scoped Voxtral TTS implementation with associated tests and conversion code.", + "canonical_pr_reason": "PR #45401 is the better representative of the cluster because it is more focused, explicitly tied to an issue, and has a narrower, self-contained Voxtral TTS implementation.", "best_issue_reason": null, - "best_pr_reason": "PR 45401 is the best representative of the cluster because it is more clearly anchored to a specific upstream issue and has the more self-contained model-support change.", + "best_pr_reason": "PR #45401 is the strongest standalone PR: it has an explicit target issue, a coherent scope, and a clearly bounded model-support change. PR #44662 is larger and unrelated to it.", "soft_edge_verdicts": [ { "left": "pull_request:44662", "right": "pull_request:45401", "accept": false, - "reason": "Both are model integration PRs, but one adds PenguinVL and the other adds Voxtral TTS; they touch shared framework registration files only as part of separate model implementations, not the same underlying bug or change." + "reason": "Do not merge: they add different models (PenguinVL vs Voxtral TTS). Shared files are only generic registry/docs plumbing, not the same underlying bug or change." } ] }, @@ -24514,7 +24622,7 @@ "soft_pairs": [ "pull_request:44965|pull_request:46845" ], - "prepared_review_unit_hash": "7a92a5c1b38e9859aa64b9b95623935012e1ebde67dc6fe699378e3f782821c5", + "prepared_review_unit_hash": "2780d2e768d80fbedcf78032f46e16b1e40bcb092746dc330e344fe8f67b4b6a", "component_budget": { "node_count": 2, "item_count": 2, @@ -24536,18 +24644,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two pull requests are not duplicates: one changes the PR documentation workflow into a draft-conversion workflow, while the other switches CI to a smaller runner and includes unrelated code/test edits. They do not appear to fix the same concrete problem.", - "confidence": 0.95, + "summary": "These two open PRs are not duplicates: one rewires the PR documentation workflow to convert PRs to draft, while the other tweaks CI branch routing and includes an unrelated placeholder code change in `modeling_utils.py`. They do not appear to fix the same bug or implement the same change.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "No issue is present, so there is no canonical issue target. As a PR representative, `#46845` is the better anchor because it has broader context and a more substantial CI-related change set, but it is still unrelated to `#44965`.", "best_issue_reason": null, - "best_pr_reason": "PR 46845 is the better representative of the cluster only because it is the more substantive and active change set, but it is still not a duplicate of 44965.", + "best_pr_reason": "`#46845` is the most representative PR in the cluster because it has more discussion and touches the main CI path, whereas `#44965` is a small workflow-only change. Neither should be treated as a duplicate of the other.", "soft_edge_verdicts": [ { "left": "pull_request:44965", "right": "pull_request:46845", "accept": false, - "reason": "Different targets and intent: 44965 edits the PR documentation workflow to convert PRs to draft, while 46845 changes CI runner selection and touches model/utils code. They do not share the same underlying bug or mergeable fix." + "reason": "Different underlying changes: `#44965` changes the PR documentation workflow to convert PRs to draft, while `#46845` changes CI branch selection and adds an unrelated code edit. They do not appear mergeable as one fix." } ] }, @@ -24577,12 +24685,12 @@ "soft_pairs": [ "pull_request:44979|pull_request:45363" ], - "prepared_review_unit_hash": "a1a3330903c49866139f9d160e165eb5a7dc6008478bc31a047f94f83afde308", + "prepared_review_unit_hash": "0130d286cf3ab76063dc525ffafa85d74a0eebb4f4a6189e3a2df06bff309283", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4407, + "serialized_chars": 4408, "estimated_input_tokens": 1102, "estimated_eval_tokens": 2460 }, @@ -24590,7 +24698,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4407, + "serialized_chars": 4408, "estimated_input_tokens": 1102, "estimated_eval_tokens": 2460 }, @@ -24599,18 +24707,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs are about fusion, but they implement different layers of the stack: #44979 adds a generic Module Fusion API and tests, while #45363 adds n-to-1 kernel fusion plumbing through KernelConfig and model-loading/integration code. They are related in theme, not the same underlying change.", - "confidence": 0.91, + "summary": "These PRs are related to fusion, but they are not the same change: #44979 introduces a generic module-fusion API, while #45363 adds KernelConfig-based n-to-1 kernel fusion integration and an example. They share theme, not a concrete code path.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "#44979 is the broader foundational API proposal; #45363 is a separate downstream feature built around KernelConfig and integration hooks, not a duplicate of that API work.", + "canonical_pr_reason": "No single canonical PR: #44979 is a foundational API addition, while #45363 is a narrower integration built around that idea and does not appear to be a duplicate of it.", "best_issue_reason": null, - "best_pr_reason": "#44979 is the best cluster anchor because it defines the core fusion abstraction and is the more general change; #45363 is narrower and depends on a different implementation path.", + "best_pr_reason": "#44979 is the more central and reusable change; it defines the module fusion abstraction that a later integration PR would build on.", "soft_edge_verdicts": [ { "left": "pull_request:44979", "right": "pull_request:45363", "accept": false, - "reason": "Reject: these PRs target different concrete code paths and feature layers. #44979 introduces a new module-fusion API, while #45363 wires n-to-1 kernel fusion through KernelConfig/model loading. Shared fusion theme is too broad to treat as the same change." + "reason": "Different levels of the stack: generic module-fusion API vs a specific KernelConfig-based kernel-fusion integration/example. Related, but not mergeable as one duplicate PR." } ] }, @@ -24640,40 +24748,40 @@ "soft_pairs": [ "pull_request:45086|pull_request:45317" ], - "prepared_review_unit_hash": "939fb09c6cb5a9f18e7421a98f5f2cf2299206ccab3017ea4e490549932f0879", + "prepared_review_unit_hash": "eb4b8714e20c7044c6f0b93a8b944240da709720f69527a83e3448255ccb3383", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4412, - "estimated_input_tokens": 1103, - "estimated_eval_tokens": 2462 + "serialized_chars": 4414, + "estimated_input_tokens": 1104, + "estimated_eval_tokens": 2464 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4412, - "estimated_input_tokens": 1103, - "estimated_eval_tokens": 2462 + "serialized_chars": 4414, + "estimated_input_tokens": 1104, + "estimated_eval_tokens": 2464 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests address the same Mistral regex AttributeError in the same code path (`src/transformers/tokenization_utils_tokenizers.py`) and target the same upstream issue #45081; one is just a slimmer version of the other, while the second adds tests.", + "summary": "Both PRs target the same Mistral regex AttributeError fix in `src/transformers/tokenization_utils_tokenizers.py` and share the same issue target; `45317` is the more complete version because it adds test coverage.", "confidence": 0.97, - "canonical_issue_reason": "No issue node is present in the cluster, but both PRs explicitly target issue #45081, so that is the implied canonical issue.", - "canonical_pr_reason": "PR #45317 is the better canonical PR because it contains the same fix as #45086 plus accompanying tests, making it the more complete and mergeable patch.", - "best_issue_reason": "No standalone issue artifact is included here; issue #45081 is only referenced as the shared target.", - "best_pr_reason": "PR #45317 is the strongest representative: same bug fix, same file, and added test coverage.", + "canonical_issue_reason": null, + "canonical_pr_reason": "`45317` is the stronger representative PR: it implements the same code fix as `45086` and also adds a test file, making it the more complete change set.", + "best_issue_reason": null, + "best_pr_reason": "`45317` is the best PR for the cluster because it covers the same underlying bug fix with additional tests, so it is the most suitable single reference for the change.", "soft_edge_verdicts": [ { "left": "pull_request:45086", "right": "pull_request:45317", "accept": true, - "reason": "Same underlying AttributeError fix in `_patch_mistral_regex`, same file and identical core code change; #45317 is a superset with tests." + "reason": "Same concrete bug fix in the same file, same issue target, and the second PR is just the more complete variant with tests; these are duplicates of the same change." } ] }, @@ -24703,12 +24811,12 @@ "soft_pairs": [ "pull_request:45115|pull_request:45569" ], - "prepared_review_unit_hash": "6fe149a99d48e9fb0b543f92c1426c2d258282edaed69871ab2492be24beb893", + "prepared_review_unit_hash": "1db25c72be192eb0e642ff4b759f0467ff12fff34d64e7efb5f07458af63311a", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6066, + "serialized_chars": 6065, "estimated_input_tokens": 1517, "estimated_eval_tokens": 3290 }, @@ -24716,7 +24824,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6066, + "serialized_chars": 6065, "estimated_input_tokens": 1517, "estimated_eval_tokens": 3290 }, @@ -24725,18 +24833,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two open PRs both touch Nemotron-H, but they are not the same change: one is a refactor toward GraniteMoeHybrid-compatible conversion/model wiring, while the other splits Nemotron-H into separate dense/sparse implementations with docs and auto-mapping updates.", - "confidence": 0.87, + "summary": "Two open PRs in the NemotronH area overlap on shared model files, but they target different changes: one is a refactor to inherit GraniteMoeHybrid and add conversion mappings, while the other introduces separate NemotronH dense/sparse model variants and docs. They are related but not duplicates.", + "confidence": 0.9, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45569 is the broader, more complete architecture change: it adds the dense/sparse model packages, updates docs and auto mappings, and makes the existing nemotron_h entry a dispatcher. It is the better cluster anchor than the narrower refactor PR.", + "canonical_pr_reason": "No issue is present. If forced to pick one PR as the primary anchor, 45569 is broader and more complete, covering the concrete productization of NemotronH into dense/sparse variants plus docs and auto-mapping updates.", "best_issue_reason": null, - "best_pr_reason": "PR #45569 is the best representative of the cluster because it subsumes the model-structure work and includes the visible user-facing/model registry changes. PR #45115 is narrower and focused on conversion/refactor plumbing.", + "best_pr_reason": "45569 is the better global candidate because it is the larger, more user-facing implementation with explicit new model families, documentation, and tests. 45115 is narrower and mainly a refactor/conversion-mapping change.", "soft_edge_verdicts": [ { "left": "pull_request:45115", "right": "pull_request:45569", "accept": false, - "reason": "Shared Nemotron-H files are not enough: #45115 is a conversion/refactor to GraniteMoeHybrid-style naming, while #45569 splits Nemotron-H into dense/sparse model variants and updates docs/auto-mapping. They are related but not the same concrete code-path change." + "reason": "Shared NemotronH files alone are not enough: 45115 focuses on inheritance/conversion-key renames, while 45569 adds distinct dense/sparse model packages and docs. They do not look like the same underlying code change and would not plausibly merge as one PR." } ] }, @@ -24766,7 +24874,7 @@ "soft_pairs": [ "pull_request:45332|pull_request:45333" ], - "prepared_review_unit_hash": "3c12deebec90967848c43d75a9cc839afca670aa90442af321159e639531bad7", + "prepared_review_unit_hash": "01317ef4f18c5039b2b461224d9d5d8a32b96997c67eb1fb12ba1c59e801c3dd", "component_budget": { "node_count": 2, "item_count": 2, @@ -24788,18 +24896,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs are in the heterogeneous-modeling feature area and share the configuration/docs scaffold, but #45332 is a broader end-to-end implementation while #45333 is a narrower per-layer configuration subset. They look related, not clearly duplicates.", - "confidence": 0.81, + "summary": "Two open PRs are closely related heterogeneous-configuration work, but one is a broader model-support implementation and the other is a narrower config-only slice. They overlap on shared configuration files and docs, yet they are not the same concrete change.", + "confidence": 0.84, "canonical_issue_reason": null, - "canonical_pr_reason": "#45332 is the better canonical PR because it covers the full heterogeneous model support stack (config, modeling, masking, cache, utilities, tests), making it the most complete representative of the feature.", + "canonical_pr_reason": "PR #45332 is the broader, more complete change: it includes heterogeneous config support plus modeling, masking, cache, skip, and model integration updates, so it subsumes the narrower config-only work.", "best_issue_reason": null, - "best_pr_reason": "#45332 is the strongest overall PR candidate since it includes the shared configuration work plus the larger modeling/runtime changes that define the feature.", + "best_pr_reason": "PR #45332 is the best representative because it covers the full heterogeneous-model feature set and incorporates the shared configuration/doc pieces that PR #45333 focuses on.", "soft_edge_verdicts": [ { "left": "pull_request:45332", "right": "pull_request:45333", "accept": false, - "reason": "They overlap on heterogeneous configuration files, but #45333 is only the config subset whereas #45332 adds broader modeling/runtime support. Different scope and not clearly the same concrete change, so do not merge as duplicates." + "reason": "They are related, but not the same underlying PR. #45333 is a config-only subset while #45332 adds broader modeling and runtime integration changes, so they should not be merged as duplicates." } ] }, @@ -24829,40 +24937,40 @@ "soft_pairs": [ "pull_request:45415|pull_request:45425" ], - "prepared_review_unit_hash": "b97cdceced0ab140b3bd366010730d96210678dd4192229bb3b453d9553dc6f8", + "prepared_review_unit_hash": "e9e0c09a0f5f0fd41782355d2c1d8713b2b75826b58432db5edbda976461f31c", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5512, - "estimated_input_tokens": 1378, - "estimated_eval_tokens": 3012 + "serialized_chars": 5514, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5512, - "estimated_input_tokens": 1378, - "estimated_eval_tokens": 3012 + "serialized_chars": 5514, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related to typing work in transformers, but they are not the same change. PR 45415 is a broad repo-wide type-checking sweep with many file edits and pyproject ignore adjustments, while PR 45425 adds concrete typing helpers/protocols and updates a smaller, different set of typing-sensitive modules. The shared files are not enough to make them duplicates.", - "confidence": 0.93, + "summary": "Both PRs are typing-focused and overlap in a few files, but they address different scopes: 45415 is a broad type-checking sweep across many transformers modules, while 45425 adds specific typing protocols/aliases and updates a smaller set of code paths around modeling_utils and related integrations. They are related, but not the same change set.", + "confidence": 0.9, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: 45415 is a broad type-checking/enforcement sweep across many modules, whereas 45425 is a narrower typing-API/helper change centered on _typing.py, modeling_utils, and PEFT-related protocol support.", + "canonical_pr_reason": "No canonical PR: 45415 is a broad repository-wide typing cleanup, whereas 45425 is a narrower typing refactor for specific interfaces and modeling_utils-related annotations.", "best_issue_reason": null, - "best_pr_reason": null, + "best_pr_reason": "45425 is the better representative because it is more self-contained and specific; 45415 is a large umbrella typing pass that touches many unrelated files.", "soft_edge_verdicts": [ { "left": "pull_request:45415", "right": "pull_request:45425", "accept": false, - "reason": "Reject: the PRs address different concrete work. 45415 is a large typing cleanup pass with pyproject ignore changes and many unrelated module edits; 45425 introduces specific typing aliases/protocols and a smaller set of targeted fixes. Shared filenames like modeling_utils/check_types do not make them the same underlying fix." + "reason": "Shared typing-related files are not enough to make these duplicates. 45415 is a broad mypy/type-checking enablement PR, while 45425 introduces concrete typing aliases/protocols and targeted fixes for specific modules. They would not plausibly merge as one single change." } ] }, @@ -24892,7 +25000,7 @@ "soft_pairs": [ "pull_request:45548|pull_request:45662" ], - "prepared_review_unit_hash": "bb3a30d9e06380e27b0a460d170cd5cea1e9ce7eec8da1ef5ddebd04220d43ac", + "prepared_review_unit_hash": "c4ca9af17cc81199a582c4ff69b38fa66061f2244cf56eb89e24e2aa753803c1", "component_budget": { "node_count": 2, "item_count": 2, @@ -24914,18 +25022,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two PRs are related by expert-parallelism code, but they fix different concrete bugs: one is about preserving/loading distributed_config for EP + ZeRO-3 under accelerate launch, while the other prevents expert parameters from being overwritten during FSDP2/rank-0 broadcast via post-shard wrapping. They should not be deduplicated.", - "confidence": 0.92, + "summary": "Both PRs are about expert-parallel loading, but they fix different failure modes: one targets DeepSpeed ZeRO-3 / accelerate launch state handling, the other targets FSDP2 expert parameters being overwritten during rank-0 broadcast. Related area, not the same change.", + "confidence": 0.89, "canonical_issue_reason": null, - "canonical_pr_reason": "No issue artifacts are present. If a canonical PR must be chosen for the cluster, PR #45662 is the better representative because it targets a more specific, well-defined EP failure mode and contains the more explicit code-path fix.", + "canonical_pr_reason": "PR #45662 is the better representative of the cluster because it addresses the more concrete runtime corruption bug in EP loading and spans the relevant loading paths, whereas #45548 is a narrower DeepSpeed/serialization fix.", "best_issue_reason": null, - "best_pr_reason": "PR #45662 is the stronger standalone PR for this cluster: it addresses a concrete overwrite bug in the load/shard path and includes the relevant distributed-tensor handling changes, whereas PR #45548 is a different loading/config-serialization fix.", + "best_pr_reason": "PR #45662 is the strongest standalone PR: it fixes a specific code-path problem with expert parameters being silently overwritten and includes the loading-path changes needed for that behavior.", "soft_edge_verdicts": [ { "left": "pull_request:45548", "right": "pull_request:45662", "accept": false, - "reason": "Shared EP terminology and one shared file are not enough: #45548 fixes distributed_config serialization and has_ep for ZeRO-3/accelerate loading, while #45662 fixes expert parameters being overwritten during FSDP2/rank-0 broadcast. Different bug, different code path." + "reason": "Different concrete bugs and code paths: #45548 is about preserving distributed_config / EP state for DeepSpeed ZeRO-3 loading, while #45662 is about preventing expert params from being overwritten during FSDP2-style loading. They are related but not mergeable as one fix." } ] }, @@ -24955,7 +25063,7 @@ "soft_pairs": [ "pull_request:45639|pull_request:45645" ], - "prepared_review_unit_hash": "7b98dae0ee1180c081566a5d500031afa679da0904cd81462cf0f2162c9dbf2d", + "prepared_review_unit_hash": "07241d8686e2aeef0e7d8f7036fc2eb3649fcad4314f1ebdafa7a64ef5fb3b68", "component_budget": { "node_count": 2, "item_count": 2, @@ -24977,18 +25085,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs address the same xdist-related captured_info logging problem around issue #45561; PR 45645 is the more complete change because it includes the code fix plus the CI workflow update needed to consume multiple artifact files.", - "confidence": 0.91, + "summary": "Both PRs target the same xdist-related captured_info artifact collision in testing/debug logging. PR 45645 appears to be the broader follow-up that includes the core fix plus CI log collection, so it is the better canonical representative.", + "confidence": 0.87, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45645 is the better canonical representative because it covers the full xdist-safe captured_info flow: generating worker-specific logs and updating the workflow to read all captured_info files.", + "canonical_pr_reason": "PR 45645 is the more complete change set: it includes the xdist-safe output handling in testing_utils and also updates the workflow to collect multiple captured_info artifacts, whereas 45639 only contains the core library-side fix.", "best_issue_reason": null, - "best_pr_reason": "PR 45645 is the strongest overall match for the cluster since it addresses the underlying bug and the downstream CI log-collection path in one change.", + "best_pr_reason": "PR 45645 best represents the underlying problem because it covers the same captured_info xdist collision and the CI artifact handling needed to preserve/debug those logs.", "soft_edge_verdicts": [ { "left": "pull_request:45639", "right": "pull_request:45645", "accept": true, - "reason": "Same underlying xdist collision fix for captured_info artifacts; 45645 is essentially the same change plus the required workflow consumer update, so they could plausibly be merged into one PR." + "reason": "Both PRs fix the same concrete captured_info xdist collision path; 45645 is a superset with the same core code change plus workflow handling, so they could plausibly be merged into one PR." } ] }, @@ -25018,7 +25126,7 @@ "soft_pairs": [ "pull_request:45720|pull_request:46259" ], - "prepared_review_unit_hash": "a07bed4281fd334d8b0c5f7a12da2605ac0f6d67a7795f655b6363ac7b2fc173", + "prepared_review_unit_hash": "fe53914da7af8a58e84175d2033625056c92bb35bb939c8d8dcfc4e179345379", "component_budget": { "node_count": 2, "item_count": 2, @@ -25040,18 +25148,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two unrelated linter/modeling-rule PRs overlap on some model files, but they target different code paths: one adjusts forward return-dict/tuple handling, the other changes model weight initialization rules.", + "summary": "These are two unrelated linter/modeling-rule PRs. PR #45720 changes model forward signatures/return behavior for TRF003, while PR #46259 changes weight initialization logic for TRF018. They share some broad model files, but not the same underlying fix.", "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46259 is the broader, more established change set with much higher review activity and a larger implementation footprint, so it is the best representative PR in this cluster.", + "canonical_pr_reason": "PR #45720 is the more focused, concrete code-path change: it adds `can_return_tuple` and adjusts forward-return behavior across specific models. PR #46259 is a separate, broader initialization-rule change with different semantics.", "best_issue_reason": null, - "best_pr_reason": "PR 46259 is the strongest standalone candidate because it has more discussion, touches many files, and is the more substantial rule implementation.", + "best_pr_reason": "PR #45720 is the best representative of the cluster because it is the more narrowly scoped and directly verifiable change, whereas #46259 is a different linter rule affecting a different part of model behavior.", "soft_edge_verdicts": [ { "left": "pull_request:45720", "right": "pull_request:46259", "accept": false, - "reason": "Shared model filenames and a linter-driven theme are not enough: 45720 changes forward() return-tuple/return_dict behavior for TRF003, while 46259 refactors _init_weights behavior for TRF018. They do not fix the same bug and would not plausibly merge into one PR." + "reason": "Do not merge: these fix different problems. #45720 targets tuple/return-dict handling in model forwards; #46259 targets weight initialization behavior. Shared model files are incidental and not evidence of the same underlying change." } ] }, @@ -25147,7 +25255,7 @@ "soft_pairs": [ "pull_request:45932|pull_request:46100" ], - "prepared_review_unit_hash": "44067c82497bb20dcfc2c5b9b0ae13af917320dc10d959bd927c0876d7f2572c", + "prepared_review_unit_hash": "dfc08ca0411ffdbfbd2f25982497b29bf0827b4e1245ecc0d3344e3b88a6a2fc", "component_budget": { "node_count": 2, "item_count": 2, @@ -25169,18 +25277,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two PRs are both non-user-facing maintenance changes, but they are not the same change: one is a Grafana/CI + model-config patch, the other is a test-migration PR adding DDP/test files. The soft duplicate link should be rejected.", - "confidence": 0.93, + "summary": "These two pull requests are not duplicates: PR #45932 is a broad CI/telemetry + model config sweep, while PR #46100 is a test-migration PR adding DDP config and model tests. They do not target the same concrete bug or change and would not plausibly merge into one PR.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46100 is the cleaner anchor for this cluster because it is a self-contained test-migration change; PR 45932 mixes CI secret plumbing with unrelated model/config edits, so it is less representative.", + "canonical_pr_reason": "No clear canonical PR: the PRs address different concerns and code paths. If one must be picked, #45932 is the more distinctive infra/CI-oriented change, but it is not a duplicate of #46100.", "best_issue_reason": null, - "best_pr_reason": "PR 46100 is the best overall representative because its scope is narrower and coherent, while PR 45932 spans several unrelated edits across CI and multiple models.", + "best_pr_reason": "#45932 is the better representative of its own change set because it is a single infra/model-config sweep with a clear experimental intent; #46100 is a separate test migration and should not be collapsed into it.", "soft_edge_verdicts": [ { "left": "pull_request:45932", "right": "pull_request:46100", "accept": false, - "reason": "Different underlying changes and code paths: 45932 updates CI/Grafana secrets and several model/config files, while 46100 adds DDP test infrastructure and migrates model tests. They are not plausibly the same PR." + "reason": "Different underlying work: #45932 touches CI secrets and multiple model config files, while #46100 adds DDP/test scaffolding and model tests. Shared similarity is only broad test/model maintenance, not the same concrete fix." } ] }, @@ -25210,12 +25318,12 @@ "soft_pairs": [ "pull_request:46126|pull_request:46269" ], - "prepared_review_unit_hash": "b515b036ef212003dcc11d4b43380a53d64cf4b8b64d7ea5b0d35d73dc43cb8e", + "prepared_review_unit_hash": "50f5cbdb62a5d95edf59db30f155eae109b15044c556c456157801b2de71bbf3", "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 }, @@ -25223,7 +25331,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5526, + "serialized_chars": 5525, "estimated_input_tokens": 1382, "estimated_eval_tokens": 3020 }, @@ -25232,18 +25340,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related by branch ancestry and shared distributed filenames, but they are not the same underlying change. PR #46126 is a focused EP/Trainer integration feature, while PR #46269 is a broad distributed branch base touching many unrelated areas.", - "confidence": 0.93, + "summary": "The two PRs overlap in distributed code, but they are not the same change. PR #46126 is a focused EP/trainer integration feature, while PR #46269 is a broad distributed branch base that touches many unrelated areas. The shared filenames are not enough to treat them as duplicates.", + "confidence": 0.94, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #46126 is the more specific, self-contained change centered on EP + Trainer integration; PR #46269 is a broad base branch with many extra edits and is not a clean duplicate target.", + "canonical_pr_reason": "PR #46126 is the better representative of a single concrete change: it targets EP routing and trainer integration with a tight scope and coherent purpose.", "best_issue_reason": null, - "best_pr_reason": "PR #46126 is the best representative because it has a clear, concrete code-path goal and a narrow scope. PR #46269 is too broad and mixed-purpose to serve as the canonical cluster representative.", + "best_pr_reason": "PR #46126 is the strongest standalone candidate because it is narrowly scoped and clearly addresses one feature path, unlike #46269 which is an umbrella branch-base PR.", "soft_edge_verdicts": [ { "left": "pull_request:46126", "right": "pull_request:46269", "accept": false, - "reason": "Shared files alone are not enough here: the titles and diffs indicate different scopes and purposes, with #46126 implementing EP/Trainer integration and #46269 acting as a broad distributed branch base. They do not look mergeable as one PR fixing the same concrete problem." + "reason": "They share some distributed files, but #46126 is a specific EP/trainer feature PR and #46269 is a broad branch-base PR spanning many unrelated distributed subsystems. Not the same underlying change or bug fix." } ] }, @@ -25311,12 +25419,12 @@ "pull_request:46504|pull_request:46716", "pull_request:46432|pull_request:46716" ], - "prepared_review_unit_hash": "9c948adcddad2492df2cc501b50870f5c27c927861c8fd44b06c9403e4b96233", + "prepared_review_unit_hash": "32c2ce950f62904d13b31c8b404aae8c8d31602002356c66e5f678025c8ec1d2", "component_budget": { "node_count": 10, "item_count": 10, "soft_pair_count": 31, - "serialized_chars": 26678, + "serialized_chars": 26677, "estimated_input_tokens": 6670, "estimated_eval_tokens": 13596 }, @@ -25324,7 +25432,7 @@ "node_count": 10, "item_count": 10, "soft_pair_count": 31, - "serialized_chars": 26678, + "serialized_chars": 26677, "estimated_input_tokens": 6670, "estimated_eval_tokens": 13596 }, @@ -25333,198 +25441,198 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is mostly a set of unrelated workflow-maintenance PRs. The only near-duplicate relationship is within `.github/workflows/pr-ci-security-gate.yml`, where two PRs harden the same security-gate workflow and could plausibly be combined.", - "confidence": 0.9, + "summary": "This cluster is a heterogeneous set of workflow maintenance PRs. The only obvious mini-cluster is the pr-ci-security-gate.yml sequence, but the individual PRs make different changes (logic, permissions, comments, secrets), so I reject all soft edges as true duplicates.", + "confidence": 0.86, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #46554 is the best canonical PR anchor because it makes the most substantive change to the shared `pr-ci-security-gate.yml` workflow. The other same-file PRs are narrower follow-ups or adjacent hardening tweaks rather than clearer representatives.", + "canonical_pr_reason": "No single PR cleanly represents the whole cluster; 46554 is the best anchor because it makes the most substantive pr-ci-security-gate behavior change and sits in the only near-duplicate subgroup.", "best_issue_reason": null, - "best_pr_reason": "PR #46554 is the strongest global representative: it is the most behavior-changing workflow-security update in the set and closest to the one accepted same-file duplicate candidate.", + "best_pr_reason": "46554 is the strongest representative PR: it changes actual gate logic rather than only adjacent permissions/comments/secrets, and is the most substantive item in the only related subgroup.", "soft_edge_verdicts": [ { "left": "pull_request:46138", "right": "pull_request:46554", "accept": false, - "reason": "Different workflows and different fixes; same broad CI area, but not the same underlying change." + "reason": "Different workflows and different fixes: stdin handling for self-comment-ci vs security-gate logic changes." }, { "left": "pull_request:46138", "right": "pull_request:46558", "accept": false, - "reason": "Unrelated workflow file and change scope; no shared concrete bug." + "reason": "Unrelated workflow changes; one fixes PR comment piping, the other adjusts security-gate permissions/approval flow." }, { "left": "pull_request:46138", "right": "pull_request:46563", "accept": false, - "reason": "Different workflow and behavior change; not a duplicate." + "reason": "Different workflow areas and different concerns; not the same concrete change." }, { "left": "pull_request:46138", "right": "pull_request:46716", "accept": false, - "reason": "Different workflow and unrelated update to the self-comment CI script invocation." + "reason": "Self-comment CI input handling is unrelated to pr-ci-security-gate secrets handling." }, { "left": "pull_request:46138", "right": "pull_request:46504", "accept": false, - "reason": "Separate workflow logic change in a different file; no same underlying bug." + "reason": "Both touch shell quoting, but in different workflows and different code paths; not a duplicate fix." }, { "left": "pull_request:46432", "right": "pull_request:46503", "accept": false, - "reason": "Different workflow files and distinct behavior changes; only superficial CI similarity." + "reason": "SSH runner hardening is unrelated to benchmark trigger conditions." }, { "left": "pull_request:46432", "right": "pull_request:46501", "accept": false, - "reason": "Different workflows and different security/permission changes." + "reason": "Different workflows and different changes; no shared bug or patch." }, { "left": "pull_request:46432", "right": "pull_request:46502", "accept": false, - "reason": "Different workflow automation task; not the same concrete problem." + "reason": "SSH runner changes do not overlap with assign-reviewers checkout ref handling." }, { "left": "pull_request:46432", "right": "pull_request:46554", "accept": false, - "reason": "Both are workflow hardening, but they address different jobs and different code paths." + "reason": "Different workflows and different security concerns; not the same fix." }, { "left": "pull_request:46432", "right": "pull_request:46558", "accept": false, - "reason": "Distinct workflow changes in different files; not mergeable as one fix." + "reason": "SSH runner secret cleanup is unrelated to pr-ci-security-gate permissions changes." }, { "left": "pull_request:46501", "right": "pull_request:46503", "accept": false, - "reason": "Different workflow files and unrelated updates." + "reason": "Both are workflow chores, but they change different jobs and different triggers." }, { "left": "pull_request:46501", "right": "pull_request:46502", "accept": false, - "reason": "Separate workflow targets; same repo area but different fixes." + "reason": "Different workflows and different behavior; no shared underlying bug." }, { "left": "pull_request:46501", "right": "pull_request:46504", "accept": false, - "reason": "Different workflow files and distinct logic changes." + "reason": "Anti-slop permission adjustment is unrelated to build-ci-docker-images shell quoting." }, { "left": "pull_request:46501", "right": "pull_request:46554", "accept": false, - "reason": "No shared concrete bug; one is anti-slop permissions, the other is PR CI security-gate logic." + "reason": "Different files, different workflows, different code paths." }, { "left": "pull_request:46502", "right": "pull_request:46503", "accept": false, - "reason": "Different workflow responsibilities; not the same underlying change." + "reason": "Assign-reviewers checkout ref change is unrelated to benchmark trigger logic." }, { "left": "pull_request:46502", "right": "pull_request:46504", "accept": false, - "reason": "Different workflow files and separate changes." + "reason": "Different workflow scripts and unrelated fixes." }, { "left": "pull_request:46502", "right": "pull_request:46554", "accept": false, - "reason": "Different workflows; no single code-path problem." + "reason": "No shared concrete bug; these are distinct workflow updates." }, { "left": "pull_request:46503", "right": "pull_request:46504", "accept": false, - "reason": "Separate workflow behavior changes in different files." + "reason": "Benchmark dispatch condition and docker-image commit-message quoting are different changes." }, { "left": "pull_request:46503", "right": "pull_request:46554", "accept": false, - "reason": "Different workflow files and unrelated fixes." + "reason": "Different workflows and different logic; not a duplicate." }, { "left": "pull_request:46504", "right": "pull_request:46554", "accept": false, - "reason": "Different workflow files; same CI umbrella but not the same concrete bug." + "reason": "Both are workflow tweaks, but one changes commit-message parsing and the other changes security-gate behavior." }, { "left": "pull_request:46504", "right": "pull_request:46558", "accept": false, - "reason": "No shared underlying change; different automation paths." + "reason": "Different workflows and different fixes; not the same underlying change." }, { "left": "pull_request:46554", "right": "pull_request:46558", - "accept": true, - "reason": "Same `pr-ci-security-gate.yml` workflow and same security-hardening theme; these changes could plausibly be merged into one PR." + "accept": false, + "reason": "Same file, but different changes: gate logic vs permissions/approval flow." }, { "left": "pull_request:46554", "right": "pull_request:46563", "accept": false, - "reason": "Same file, but the diffs are different kinds of edits; comment-only caller note is not the same fix as workflow security checks." + "reason": "One changes security-gate behavior; the other only adds explanatory comments." }, { "left": "pull_request:46554", "right": "pull_request:46716", "accept": false, - "reason": "Related security context, but the actual changes are different: workflow checks versus caller secret handling." + "reason": "Different security-gate modifications: workflow logic vs secrets inheritance." }, { "left": "pull_request:46558", "right": "pull_request:46563", "accept": false, - "reason": "Different edits to the same file family, but not the same underlying bug/change." + "reason": "Permissions/approval changes are distinct from comment-only updates." }, { "left": "pull_request:46558", "right": "pull_request:46716", "accept": false, - "reason": "Both touch PR CI security, but they fix different aspects and are not one concrete change." + "reason": "Different adjustments to the same workflow, but not the same concrete fix." }, { "left": "pull_request:46563", "right": "pull_request:46716", "accept": false, - "reason": "Different modifications to the reusable workflow caller; not the same duplicate change." + "reason": "Comments-only review note is unrelated to switching secrets to empty." }, { "left": "pull_request:46504", "right": "pull_request:46563", "accept": false, - "reason": "Different workflow files and unrelated fixes." + "reason": "Different workflows and different purposes; not a duplicate." }, { "left": "pull_request:46432", "right": "pull_request:46563", "accept": false, - "reason": "Distinct workflows with different purposes; no shared code-path problem." + "reason": "SSH runner hardening and security-gate comments are unrelated." }, { "left": "pull_request:46504", "right": "pull_request:46716", "accept": false, - "reason": "Different workflow files and separate changes." + "reason": "Different workflow files and different fixes." }, { "left": "pull_request:46432", "right": "pull_request:46716", "accept": false, - "reason": "Separate workflow changes; not a duplicate." + "reason": "No shared code path or bug; different workflow changes." } ] }, @@ -25617,7 +25725,7 @@ "soft_pairs": [ "pull_request:46281|pull_request:46890" ], - "prepared_review_unit_hash": "eda23a4ab5172db0ab7bf75f8faea1fd4bcfe6dcd9ff1e7fafd1c2e47d1e33e1", + "prepared_review_unit_hash": "e8b6b3eb82a7c106d9c267c060e1724316881c9cf360ee15daf0d0cbb8808498", "component_budget": { "node_count": 2, "item_count": 2, @@ -25639,18 +25747,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are thematically similar security fixes, but they address different path-traversal bugs in different code paths: one in sharded checkpoint loading from `weight_map`, the other in explicit `transformers_weights` resolution.", - "confidence": 0.95, + "summary": "Both PRs harden Transformers against path traversal, but they patch different code paths and different untrusted inputs: one validates shard filenames from a checkpoint index, the other validates an explicitly configured weights filename. They are related security fixes, not duplicates.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "No single canonical PR: `pull_request:46281` hardens `src/transformers/utils/hub.py` for shard filenames from an index, while `pull_request:46890` hardens `src/transformers/modeling_utils.py` for `transformers_weights`; they are separate fixes.", + "canonical_pr_reason": "PR 46281 is the cleaner representative of the cluster theme: it fixes traversal in the sharded-checkpoint loader via `weight_map` filenames. PR 46890 targets a separate explicit-filename path in `modeling_utils.py`, so it is not a duplicate.", "best_issue_reason": null, - "best_pr_reason": "Neither PR is a global best duplicate target for the other; they are distinct patches for distinct loader inputs and should remain separate.", + "best_pr_reason": "PR 46281 is the best representative artifact because it addresses a concrete, self-contained loader vulnerability with a focused patch and tests; PR 46890 is a distinct hardening change in another code path.", "soft_edge_verdicts": [ { "left": "pull_request:46281", "right": "pull_request:46890", "accept": false, - "reason": "Different underlying bugs and code paths. PR 46281 blocks traversal via sharded checkpoint `weight_map` filenames in `hub.py`; PR 46890 blocks traversal via the explicit `transformers_weights` field in `modeling_utils.py`. They are related hardening work, but not the same concrete fix and not mergeable as one PR." + "reason": "Same broad security theme, but different inputs and different loading code paths (`get_checkpoint_shard_files` vs `_get_resolved_checkpoint_files`); they would not plausibly be merged as one PR." } ] }, @@ -25680,7 +25788,7 @@ "soft_pairs": [ "pull_request:46574|pull_request:46580" ], - "prepared_review_unit_hash": "27d5c30e88afea8913c96bfb5de556d3ded3a925e12e47d73d0401ceeb5332d0", + "prepared_review_unit_hash": "8e692a8b0894ef7e10956ab9ae6475ae4f07452dab786110ed79a07e0b38528a", "component_budget": { "node_count": 2, "item_count": 2, @@ -25702,18 +25810,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two PRs are not duplicates: one fixes CodeLlama decode-time stripping of real leading spaces, while the other fixes DeepSeek-Coder/DeepSeek-LLM loading as the wrong tokenizer class via auto-tokenization heuristics. They touch different code paths, different files, and different underlying bugs.", + "summary": "These PRs are unrelated: one fixes CodeLlama decode stripping of real leading spaces, while the other fixes DeepSeek-Coder/DeepSeek-LLM loading the wrong tokenizer class. They should not be clustered as duplicates.", "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46580 is the better cluster representative only because it is a clearer standalone fix with an explicit targeted loading-path change and a direct issue link (#46489); PR 46574 is a separate tokenizer-decoding bug in CodeLlama.", + "canonical_pr_reason": "No canonical PR: the patches address different bugs in different code paths and would not plausibly be merged into one PR.", "best_issue_reason": null, - "best_pr_reason": "PR 46580 is the strongest standalone PR in this set: it has a precise problem statement, a narrowly scoped fix in tokenization_auto.py, and a concrete affected-model list. PR 46574 is also valid, but it addresses a different bug and should not be grouped with 46580.", + "best_pr_reason": "No single best PR for duplication purposes; each PR is a standalone fix for a different model family, so neither is a better representative of the other.", "soft_edge_verdicts": [ { "left": "pull_request:46574", "right": "pull_request:46580", "accept": false, - "reason": "Different bugs and code paths: 46574 changes CodeLlama decoding to preserve leading spaces, while 46580 changes auto-tokenizer selection for DeepSeek checkpoints. They are not the same underlying change and would not be merged into one PR." + "reason": "Different problems and different fixes: CodeLlama decoder stripping vs DeepSeek checkpoint tokenizer-class override. No shared concrete code-path bug." } ] }, @@ -25745,12 +25853,12 @@ "pull_request:46781|pull_request:46993", "pull_request:46993|pull_request:46994" ], - "prepared_review_unit_hash": "1f0471f50fadc1917f5befffd5252762a8aacfbc4414ea128ac6a99e17635151", + "prepared_review_unit_hash": "e4b259d8016efdf659fd702780f0adaf0eb64ce8445be6822b81e848f6c89f16", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 6603, + "serialized_chars": 6604, "estimated_input_tokens": 1651, "estimated_eval_tokens": 3558 }, @@ -25758,7 +25866,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 6603, + "serialized_chars": 6604, "estimated_input_tokens": 1651, "estimated_eval_tokens": 3558 }, @@ -25767,24 +25875,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These three PRs are not duplicates: they fix three different bugs in different code paths (quantized cache layer validation, prompt lookup decoding length capping, and generation argument validation for empty token lists). The soft similarities are superficial and mostly due to shared generation-related surfaces.", + "summary": "All three PRs are generation-related but they fix different code paths: cache validation, prompt lookup decoding length capping, and logits processor input validation. They are not duplicates, so both soft edges should be rejected.", "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46994 is the best cluster representative only in the weak sense that it is the clearest, self-contained validation fix and is closest to the other generation-related change; however, it does not subsume the other PRs and this is not a true duplicate cluster.", + "canonical_pr_reason": "PR 46994 is the clearest standalone bugfix: it addresses a specific validation gap in logits_process and includes a narrowly targeted change plus tests.", "best_issue_reason": null, - "best_pr_reason": "PR 46994 is the most suitable single representative because it has the cleanest scope and most explicit validation hardening, but it is still unrelated to PR 46781 and only loosely adjacent to PR 46993.", + "best_pr_reason": "PR 46994 is the best representative of the cluster because it has the most self-contained, precisely scoped fix and the title matches the concrete failure mode.", "soft_edge_verdicts": [ { "left": "pull_request:46781", "right": "pull_request:46993", "accept": false, - "reason": "Different bugs and code paths: cache layer-type validation in QuantizedCache vs prompt lookup candidate length calculation in generation." + "reason": "Different underlying problems and code paths: unsupported QuantizedCache layer-type validation vs prompt lookup decoding overshooting max_length. Similar subsystem only, not the same fix." }, { "left": "pull_request:46993", "right": "pull_request:46994", "accept": false, - "reason": "Both touch generation, but one fixes max_length handling in candidate generation while the other adds validation for empty bad_words_ids/sequence_bias inputs; they are not the same underlying change." + "reason": "Both are generation fixes, but one changes candidate generation length math while the other hardens logits processor validation for empty lists. They do not fix the same bug and would not plausibly merge into one PR." } ] }, @@ -25814,7 +25922,7 @@ "soft_pairs": [ "pull_request:46847|pull_request:47009" ], - "prepared_review_unit_hash": "0e85cc683751c0bde3f69166de254da9ef6fad386f903868958ab2cdbcd7fb31", + "prepared_review_unit_hash": "5945467acf106e30b18efa60862683b7613f485dba66fc72e3272a53bfbe8bf0", "component_budget": { "node_count": 2, "item_count": 2, @@ -25836,18 +25944,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two Qwen3.5 PRs are effectively duplicates: both add the same linear_attn tensor-parallel entries in the same files and target the same issue.", - "confidence": 0.99, - "canonical_issue_reason": "Both PRs point to the same issue (#46846), but no issue node is present in this cluster to choose from.", - "canonical_pr_reason": "PR #46847 is the better canonical representative: it is slightly earlier and has more review/inbound activity, while making the same code change as #47009.", + "summary": "Two near-identical Qwen3.5 PRs add the same linear_attn tensor-parallel entries to the same files for the same issue target; they look like duplicate attempts at the same change.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #46847 is the better canonical PR: it is the earlier submission, has more discussion/review activity and inbound references, and already explicitly targets the tracking issue for this change.", "best_issue_reason": null, - "best_pr_reason": "PR #46847 is the best representative because it is the earlier, more discussed version of the same fix and covers the exact same files and TP-plan entries.", + "best_pr_reason": "PR #46847 is the best representative of the change because it appears to be the original, more established PR for the same fix, while #47009 is a later duplicate with no added substantive scope.", "soft_edge_verdicts": [ { "left": "pull_request:46847", "right": "pull_request:47009", "accept": true, - "reason": "Same underlying fix: identical Qwen3.5 linear_attn TP-plan additions in the same files, with only minor rebase/context differences." + "reason": "Both PRs touch the same Qwen3.5 files and add the same linear_attn base_model_tp_plan entries for the same issue target, so they are the same underlying code change." } ] }, @@ -25877,12 +25985,12 @@ "soft_pairs": [ "pull_request:47032|pull_request:47033" ], - "prepared_review_unit_hash": "e6fb0025cab0286672450cb5975ec7ec426d31bd7b0f73d497468f9a6d3b903a", + "prepared_review_unit_hash": "e029b231e5ada9892fe9af845ff0a7788ab7c06233e299b7287492d50bd1614f", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4301, + "serialized_chars": 4302, "estimated_input_tokens": 1076, "estimated_eval_tokens": 2408 }, @@ -25890,7 +25998,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4301, + "serialized_chars": 4302, "estimated_input_tokens": 1076, "estimated_eval_tokens": 2408 }, @@ -25899,18 +26007,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two unrelated PRs: one fixes image-text-to-text stop_sequence handling, the other fixes serving tool-call parsing for legacy schemas. They do not look like the same underlying bug or a PR pair that should be merged together.", + "summary": "The two pull requests address unrelated fixes in different areas of Transformers: image-text-to-text stop sequence handling vs. serve tool parsing for legacy response schemas. They are not duplicates and should not be clustered as the same change.", "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: PR 47032 is a pipeline stop_sequence fix, while PR 47033 changes CLI serving tool parsing; they address different code paths and different bugs.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "No best PR for duplication purposes: neither PR is a duplicate of the other, so there is no single representative change for the cluster.", + "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "pull_request:47032", "right": "pull_request:47033", "accept": false, - "reason": "Different subsystems and different fixes: image-text-to-text stop_sequence conversion vs. legacy serve tool-call parsing." + "reason": "Different subsystems, different bug shapes, and different touched files; they do not appear to fix the same underlying code-path problem and would not plausibly merge into one PR." } ] }, diff --git a/analysis/current/manifest.json b/analysis/current/manifest.json index dd83efe49a1c117e7c016c07edf5b912e91934fb..b0ae37c6f804e553e2a4503b56f50d69ababbce7 100644 --- a/analysis/current/manifest.json +++ b/analysis/current/manifest.json @@ -1,8 +1,8 @@ { - "analysis_id": "hybrid-model-20260703t000025z", + "analysis_id": "hybrid-model-20260703t060023z", "archived_artifacts": { - "hybrid": "snapshots/20260703T000025Z/analysis-runs/hybrid-model-20260703t000025z/analysis-report-hybrid.json", - "hybrid_reviews": "snapshots/20260703T000025Z/analysis-runs/hybrid-model-20260703t000025z/analysis-report-hybrid.llm-reviews.json" + "hybrid": "snapshots/20260703T060023Z/analysis-runs/hybrid-model-20260703t060023z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260703T060023Z/analysis-runs/hybrid-model-20260703t060023z/analysis-report-hybrid.llm-reviews.json" }, "artifacts": { "hybrid": "analysis/current/analysis-report-hybrid.json", @@ -10,9 +10,9 @@ }, "channel": "canonical", "model": null, - "published_at": "2026-07-03T00:11:04Z", + "published_at": "2026-07-03T06:11:41Z", "repo": "huggingface/transformers", "schema_version": 1, - "snapshot_id": "20260703T000025Z", + "snapshot_id": "20260703T060023Z", "variant": "hybrid" } diff --git a/snapshots/20260703T060023Z/analysis-runs/hybrid-model-20260703t060023z/analysis-report-hybrid.json b/snapshots/20260703T060023Z/analysis-runs/hybrid-model-20260703t060023z/analysis-report-hybrid.json new file mode 100644 index 0000000000000000000000000000000000000000..897c9b9b60a8cc24ee44da58330c5fd67ecfc7af --- /dev/null +++ b/snapshots/20260703T060023Z/analysis-runs/hybrid-model-20260703t060023z/analysis-report-hybrid.json @@ -0,0 +1,2657 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260703T060023Z", + "generated_at": "2026-07-03T06:11:39Z", + "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-46894-6", + "summary": "Cluster of 3 issues and 3 PRs centered on issue #46894.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 46894, + "canonical_pr_number": 46903, + "issue_numbers": [ + 46894, + 46897, + 46901 + ], + "pr_numbers": [ + 46895, + 46903, + 46958 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 46895, + "right_pr_number": 46903, + "code_similarity": 0.362, + "size_similarity": 0.443, + "file_overlap": 0.333, + "area_overlap": 0.419, + "patch_similarity": 0.178, + "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, + 651 + ] + ] + }, + { + "filename": "src/transformers/models/cohere_asr/modular_cohere_asr.py", + "left_ranges": [ + [ + 478, + 490 + ], + [ + 502, + 517 + ] + ], + "right_ranges": [ + [ + 500, + 514 + ] + ] + } + ] + }, + { + "left_pr_number": 46895, + "right_pr_number": 46958, + "code_similarity": 0.715, + "size_similarity": 0.833, + "file_overlap": 1.0, + "area_overlap": 0.366, + "patch_similarity": 0.802, + "shared_filenames": [ + "src/transformers/models/cohere_asr/modeling_cohere_asr.py", + "src/transformers/models/cohere_asr/modular_cohere_asr.py", + "tests/models/cohere_asr/test_modeling_cohere_asr.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/cohere_asr/modeling_cohere_asr.py", + "left_ranges": [ + [ + 615, + 627 + ], + [ + 639, + 654 + ] + ], + "right_ranges": [ + [ + 22, + 28 + ], + [ + 638, + 645 + ] + ] + }, + { + "filename": "src/transformers/models/cohere_asr/modular_cohere_asr.py", + "left_ranges": [ + [ + 478, + 490 + ], + [ + 502, + 517 + ] + ], + "right_ranges": [ + [ + 16, + 22 + ], + [ + 501, + 508 + ] + ] + }, + { + "filename": "tests/models/cohere_asr/test_modeling_cohere_asr.py", + "left_ranges": [ + [ + 35, + 41 + ], + [ + 176, + 214 + ] + ], + "right_ranges": [ + [ + 15, + 21 + ], + [ + 180, + 221 + ] + ] + } + ] + }, + { + "left_pr_number": 46903, + "right_pr_number": 46958, + "code_similarity": 0.325, + "size_similarity": 0.369, + "file_overlap": 0.333, + "area_overlap": 0.364, + "patch_similarity": 0.156, + "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": [ + [ + 637, + 651 + ] + ], + "right_ranges": [ + [ + 22, + 28 + ], + [ + 638, + 645 + ] + ] + }, + { + "filename": "src/transformers/models/cohere_asr/modular_cohere_asr.py", + "left_ranges": [ + [ + 500, + 514 + ] + ], + "right_ranges": [ + [ + 16, + 22 + ], + [ + 501, + 508 + ] + ] + } + ] + } + ] + }, + { + "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-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": 0.711, + "size_similarity": 0.857, + "file_overlap": 1.0, + "area_overlap": 0.5, + "patch_similarity": 0.429, + "shared_filenames": [ + "src/transformers/models/yolos/modeling_yolos.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/yolos/modeling_yolos.py", + "left_ranges": [ + [ + 446, + 452 + ] + ], + "right_ranges": [ + [ + 434, + 440 + ], + [ + 446, + 452 + ] + ] + } + ] + } + ] + }, + { + "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-46846-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #46846.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 46846, + "canonical_pr_number": 46847, + "issue_numbers": [ + 46846 + ], + "pr_numbers": [ + 46847, + 47009 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 46847, + "right_pr_number": 47009, + "code_similarity": 1.0, + "size_similarity": 1.0, + "file_overlap": 1.0, + "area_overlap": 1.0, + "patch_similarity": 1.0, + "shared_filenames": [ + "src/transformers/models/qwen3_5/configuration_qwen3_5.py", + "src/transformers/models/qwen3_5/modular_qwen3_5.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/qwen3_5/configuration_qwen3_5.py", + "left_ranges": [ + [ + 66, + 76 + ] + ], + "right_ranges": [ + [ + 66, + 76 + ] + ] + }, + { + "filename": "src/transformers/models/qwen3_5/modular_qwen3_5.py", + "left_ranges": [ + [ + 103, + 113 + ] + ], + "right_ranges": [ + [ + 103, + 113 + ] + ] + } + ] + } + ] + }, + { + "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-45081-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45081.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 45081, + "canonical_pr_number": 45317, + "issue_numbers": [ + 45081 + ], + "pr_numbers": [ + 45086, + 45317 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 45086, + "right_pr_number": 45317, + "code_similarity": 0.589, + "size_similarity": 0.34, + "file_overlap": 0.5, + "area_overlap": 1.0, + "patch_similarity": 0.136, + "shared_filenames": [ + "src/transformers/tokenization_utils_tokenizers.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/tokenization_utils_tokenizers.py", + "left_ranges": [ + [ + 1360, + 1370 + ], + [ + 1374, + 1380 + ] + ], + "right_ranges": [ + [ + 1360, + 1370 + ], + [ + 1374, + 1380 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-43698-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #43698.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 43698, + "canonical_pr_number": 43779, + "issue_numbers": [ + 43698 + ], + "pr_numbers": [ + 43779, + 43816 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 43779, + "right_pr_number": 43816, + "code_similarity": 0.418, + "size_similarity": 0.538, + "file_overlap": 1.0, + "area_overlap": 0.02, + "patch_similarity": 0.02, + "shared_filenames": [ + "src/transformers/integrations/integration_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/integrations/integration_utils.py", + "left_ranges": [ + [ + 2227, + 2241 + ], + [ + 2303, + 2309 + ] + ], + "right_ranges": [ + [ + 2278, + 2291 + ], + [ + 2309, + 2322 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-46836-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 46836, + "issue_numbers": [], + "pr_numbers": [ + 46836, + 47003 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 46836, + "right_pr_number": 47003, + "code_similarity": 0.034, + "size_similarity": 0.034, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.182, + "shared_filenames": [], + "shared_file_areas": [] + } + ] + }, + { + "cluster_id": "cluster-43656-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 43656, + "issue_numbers": [], + "pr_numbers": [ + 43656, + 43842 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 43656, + "right_pr_number": 43842, + "code_similarity": 0.405, + "size_similarity": 0.5, + "file_overlap": 1.0, + "area_overlap": 0.0, + "patch_similarity": 0.036, + "shared_filenames": [ + "src/transformers/cli/serve.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/cli/serve.py", + "left_ranges": [ + [ + 11, + 18 + ], + [ + 30, + 36 + ], + [ + 315, + 323 + ], + [ + 665, + 671 + ], + [ + 931, + 937 + ], + [ + 1843, + 1849 + ], + [ + 1868, + 1874 + ] + ], + "right_ranges": [ + [ + 54, + 61 + ], + [ + 587, + 593 + ] + ] + } + ] + } + ] + }, + { + "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-46894-6", + "canonical_issue_number": 46894, + "duplicate_issue_numbers": [ + 46897, + 46901 + ], + "reason": "Issues in cluster-46894-6 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-43656-2", + "canonical_pr_number": 43656, + "duplicate_pr_numbers": [ + 43842 + ], + "target_issue_number": null, + "reason": "PRs in cluster-43656-2 are treated as duplicates because they share soft_similarity evidence." + }, + { + "cluster_id": "cluster-43698-3", + "canonical_pr_number": 43779, + "duplicate_pr_numbers": [ + 43816 + ], + "target_issue_number": 43698, + "reason": "PRs in cluster-43698-3 are treated as duplicates because they converge on issue #43698 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-43979-11", + "canonical_pr_number": 44007, + "duplicate_pr_numbers": [ + 43996, + 44013, + 44044, + 44066, + 44072, + 44085, + 44129, + 44154, + 44722 + ], + "target_issue_number": 43979, + "reason": "PRs in cluster-43979-11 are treated as duplicates because they converge on issue #43979 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-44018-2", + "canonical_pr_number": 44068, + "duplicate_pr_numbers": [ + 44018 + ], + "target_issue_number": null, + "reason": "PRs in cluster-44018-2 are treated as duplicates because they share soft_similarity evidence." + }, + { + "cluster_id": "cluster-45081-3", + "canonical_pr_number": 45317, + "duplicate_pr_numbers": [ + 45086 + ], + "target_issue_number": 45081, + "reason": "PRs in cluster-45081-3 are treated as duplicates because they converge on issue #45081 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-45561-3", + "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-46836-2", + "canonical_pr_number": 46836, + "duplicate_pr_numbers": [ + 47003 + ], + "target_issue_number": null, + "reason": "PRs in cluster-46836-2 are treated as duplicates because they share soft_similarity evidence." + }, + { + "cluster_id": "cluster-46846-3", + "canonical_pr_number": 46847, + "duplicate_pr_numbers": [ + 47009 + ], + "target_issue_number": 46846, + "reason": "PRs in cluster-46846-3 are treated as duplicates because they converge on issue #46846 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-46894-6", + "canonical_pr_number": 46903, + "duplicate_pr_numbers": [ + 46895, + 46958 + ], + "target_issue_number": 46894, + "reason": "PRs in cluster-46894-6 are treated as duplicates because they converge on issue #46894 with closing_reference, shared_issue_target, soft_similarity evidence." + } + ], + "best_issue": { + "cluster_id": "cluster-46894-6", + "issue_number": 46894, + "reason": "Issue #46894 is the strongest global issue candidate because it is open, belongs to a cluster with 6 artifacts, and carries 0 discussion comments plus 6 inbound references.", + "score": 166.0 + }, + "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/20260703T060023Z/analysis-runs/hybrid-model-20260703t060023z/analysis-report-hybrid.llm-reviews.json b/snapshots/20260703T060023Z/analysis-runs/hybrid-model-20260703t060023z/analysis-report-hybrid.llm-reviews.json new file mode 100644 index 0000000000000000000000000000000000000000..f73e9329e6981145c3632796df01e5b4f38c53a3 --- /dev/null +++ b/snapshots/20260703T060023Z/analysis-runs/hybrid-model-20260703t060023z/analysis-report-hybrid.llm-reviews.json @@ -0,0 +1,26033 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260703T060023Z", + "generated_at": "2026-07-03T06:11:39Z", + "model": "gpt-5.4-mini?service_tier=flex", + "reviews": [ + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "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:35545|issue:36010", + "issue:44509|issue:44512", + "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:45685|issue:46869", + "issue:45859|issue:45902", + "issue:43638|issue:45161", + "issue:34928|issue:36331", + "issue:43596|issue:45161", + "issue:43716|issue:45507", + "issue:45874|issue:46294", + "issue:44292|issue:45237", + "issue:39290|issue:45081", + "issue:45823|issue:45874", + "issue:45800|issue:46515", + "issue:34928|issue:41720", + "issue:43425|issue:44292", + "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:43716|issue:45685", + "issue:44918|issue:45161", + "issue:43039|issue:44918", + "issue:41762|issue:43638", + "issue:43827|issue:44512", + "issue:43828|issue:45685", + "issue:43039|issue:43828", + "issue:30277|issue:36331", + "issue:34928|issue:36010", + "issue:45237|issue:46015", + "issue:43039|issue:44805" + ], + "prepared_review_unit_hash": "b55cc8dde3b86133a4f9417694cb8ff71a12da67f9d06e947c7d74d33a0eff18", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 63, + "serialized_chars": 46572, + "estimated_input_tokens": 11643, + "estimated_eval_tokens": 23542 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly a mixed cluster of unrelated runtime, model-specific, and docs issues. Only a couple of true duplicates stand out: the DeepSpeed ZeRO-3/BERT index-0 failure pair and the torch 2.4.1 compatibility wording pair.", + "confidence": 0.73, + "canonical_issue_reason": "issue:43596 is the clearest duplicate-worthy bug in the set: it matches issue:43638 on the same DeepSpeed ZeRO-3 / BERT \"index 0 is out of bounds\" failure path and is phrased more generally.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43596 is the best issue anchor because it represents one of the few exact duplicate pairs here and is more canonical than the typo-ridden follow-up report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44322", + "right": "issue:45464", + "accept": false, + "reason": "Different Qwen3.5 bugs: missing config field vs streaming chat/completions failure." + }, + { + "left": "issue:45237", + "right": "issue:45685", + "accept": false, + "reason": "Different models and failure modes: AMD GPU support vs MPS histogram op missing." + }, + { + "left": "issue:43716", + "right": "issue:45237", + "accept": false, + "reason": "Different model/runtime problems: image preprocessor dtype mismatch vs GPT-OSS on AMD GPUs." + }, + { + "left": "issue:43828", + "right": "issue:45237", + "accept": false, + "reason": "Both are hardware/runtime issues, but the concrete bug and code path differ." + }, + { + "left": "issue:45237", + "right": "issue:46468", + "accept": false, + "reason": "Different models/examples and no shared concrete failure path." + }, + { + "left": "issue:45237", + "right": "issue:45507", + "accept": false, + "reason": "Different MoE/device issues; not the same underlying bug." + }, + { + "left": "issue:30277", + "right": "issue:35545", + "accept": false, + "reason": "DeepSpeed collective mismatch vs ONNX export error are unrelated." + }, + { + "left": "issue:35545", + "right": "issue:42915", + "accept": false, + "reason": "Different failures: ONNX export vs FineGrainedFP8Config runtime issue." + }, + { + "left": "issue:35545", + "right": "issue:41720", + "accept": false, + "reason": "Different code paths: export error vs auto device-mapping CUDA assert." + }, + { + "left": "issue:45901", + "right": "issue:46015", + "accept": false, + "reason": "Task crash vs outdated RAG docs; not the same issue." + }, + { + "left": "issue:35545", + "right": "issue:36010", + "accept": false, + "reason": "Export failure and GenerationMixin import failure are unrelated." + }, + { + "left": "issue:44509", + "right": "issue:44512", + "accept": false, + "reason": "Both are v5 docs cleanup items, but they target different removed docs content." + }, + { + "left": "issue:44918", + "right": "issue:46165", + "accept": false, + "reason": "Different training issues: TRL SFT embedding unpacking vs resume/trust-remote-code behavior." + }, + { + "left": "issue:46165", + "right": "issue:46869", + "accept": false, + "reason": "Different scenarios and failures: resume-training remote code vs YOLO multi-GPU device map." + }, + { + "left": "issue:44805", + "right": "issue:45823", + "accept": false, + "reason": "Mask-shape indexing error vs device mismatch in Gemma4 PLE." + }, + { + "left": "issue:43638", + "right": "issue:44805", + "accept": false, + "reason": "Both involve shape/indexing errors, but the specific bug and stack path differ." + }, + { + "left": "issue:43638", + "right": "issue:46165", + "accept": false, + "reason": "Completely different training issues." + }, + { + "left": "issue:43596", + "right": "issue:44805", + "accept": false, + "reason": "Different failures: ZeRO-3/BERT init index error vs generic mask/tensor mismatch." + }, + { + "left": "issue:44805", + "right": "issue:46294", + "accept": false, + "reason": "Different multimodal and tensor-shape bugs." + }, + { + "left": "issue:44805", + "right": "issue:45161", + "accept": false, + "reason": "Different failures: mask indexing vs GPT-OSS TP/MoE issue." + }, + { + "left": "issue:46165", + "right": "issue:46468", + "accept": false, + "reason": "Different problems and code paths." + }, + { + "left": "issue:44560", + "right": "issue:46165", + "accept": false, + "reason": "Video chat-template failure vs training resume issue." + }, + { + "left": "issue:34634", + "right": "issue:35545", + "accept": false, + "reason": "Different subsystems: BarkProcessor voice preset vs ONNX export." + }, + { + "left": "issue:30277", + "right": "issue:43638", + "accept": false, + "reason": "DeepSpeed collective mismatch and BERT zero3 index error are not the same bug." + }, + { + "left": "issue:43596", + "right": "issue:43638", + "accept": true, + "reason": "Same DeepSpeed ZeRO-3/BERT index-0 out-of-bounds failure; these look like duplicate reports." + }, + { + "left": "issue:43039", + "right": "issue:45161", + "accept": false, + "reason": "Different bugs: Liger Kernel cross-entropy being called vs GPT-OSS TP failure." + }, + { + "left": "issue:43975", + "right": "issue:44457", + "accept": false, + "reason": "Different issues: detokenization regression vs LoRA merge/save mismatch." + }, + { + "left": "issue:44292", + "right": "issue:44457", + "accept": false, + "reason": "Different model/runtime bugs." + }, + { + "left": "issue:44841", + "right": "issue:46661", + "accept": false, + "reason": "Different processor failure vs remote-code model_class/config_class error." + }, + { + "left": "issue:43716", + "right": "issue:46869", + "accept": false, + "reason": "Different model and failure mode; only broad device-map similarity." + }, + { + "left": "issue:24643", + "right": "issue:30277", + "accept": false, + "reason": "Both mention DeepSpeed, but the concrete runtime errors are different." + }, + { + "left": "issue:34928", + "right": "issue:44805", + "accept": false, + "reason": "Both are tensor-size/shape issues, but not the same underlying bug." + }, + { + "left": "issue:45685", + "right": "issue:46869", + "accept": false, + "reason": "Different hardware/runtime errors with no shared code-path bug." + }, + { + "left": "issue:45859", + "right": "issue:45902", + "accept": false, + "reason": "Same class name, but one is CPU-offload compatibility and the other is uninitialized-memory materialization." + }, + { + "left": "issue:43638", + "right": "issue:45161", + "accept": false, + "reason": "Different bugs: ZeRO-3/BERT index error vs GPT-OSS TP issue." + }, + { + "left": "issue:34928", + "right": "issue:36331", + "accept": false, + "reason": "Activation-checkpointing/FSDP shape mismatch vs Trainer API signature mismatch." + }, + { + "left": "issue:43596", + "right": "issue:45161", + "accept": false, + "reason": "Different bugs and models." + }, + { + "left": "issue:43716", + "right": "issue:45507", + "accept": false, + "reason": "Different model-specific runtime failures." + }, + { + "left": "issue:45874", + "right": "issue:46294", + "accept": false, + "reason": "Different Gemma3/Gemma4 feature bugs; not the same failure." + }, + { + "left": "issue:44292", + "right": "issue:45237", + "accept": false, + "reason": "Different model/runtime errors." + }, + { + "left": "issue:39290", + "right": "issue:45081", + "accept": false, + "reason": "Gemma3 vLLM config issue vs Mistral tokenizer regex patch crash." + }, + { + "left": "issue:45823", + "right": "issue:45874", + "accept": false, + "reason": "Different Gemma4 issues: device mismatch vs expensive reverse embedding lookup." + }, + { + "left": "issue:45800", + "right": "issue:46515", + "accept": true, + "reason": "Same torch 2.4.1 compatibility regression; effectively duplicate wording." + }, + { + "left": "issue:34928", + "right": "issue:41720", + "accept": false, + "reason": "Different bugs: FSDP checkpointing mismatch vs Qwen3 auto device-map CUDA assert." + }, + { + "left": "issue:43425", + "right": "issue:44292", + "accept": false, + "reason": "Generic torch compatibility vs Qwen3 NVFP4 runtime error are not the same issue." + }, + { + "left": "issue:43828", + "right": "issue:45685", + "accept": false, + "reason": "Different hardware-specific dtype vs MPS op-availability issues." + }, + { + "left": "issue:43039", + "right": "issue:44918", + "accept": false, + "reason": "Liger Kernel cross-entropy call vs TRL embedding unpacking are unrelated." + }, + { + "left": "issue:41762", + "right": "issue:43638", + "accept": false, + "reason": "Both involve ZeRO-3, but one is Gemma3 loading and the other is BERT init; different bugs." + }, + { + "left": "issue:43827", + "right": "issue:44512", + "accept": false, + "reason": "Both are docs regressions from v5 removals, but they cover different removed content." + }, + { + "left": "issue:43828", + "right": "issue:45685", + "accept": false, + "reason": "Different runtime problems; same broad area but not the same bug." + }, + { + "left": "issue:43039", + "right": "issue:44805", + "accept": false, + "reason": "Different failure modes and code paths." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The two accepted soft-edge pairs are well supported by the issue titles and are plausible duplicates, and the rejected pairs are phrased cautiously without overstating evidence. No mergeability concerns apply since these are issue pairs, not PR pairs." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 2, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "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:45800|issue:46304", + "issue:43866|issue:46402", + "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:44292|issue:46015", + "issue:43866|issue:44863", + "issue:43872|issue:46304", + "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:45362|issue:45464", + "issue:43792|issue:45070", + "issue:43596|issue:45800", + "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": "e6090c211c2a079bc0810899150ad46a26ebcbeea8e335c1d0adf051387c4562", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "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": "This cluster is heterogeneous rather than one true duplicate set. The only strong duplicate-like subclusters are the DeepSpeed ZeRO-3 loading IndexErrors and the AutoTokenizer wrong-token-ID regressions; most other pairs are just same-subsystem or same-error-shape but different bugs.", + "confidence": 0.73, + "canonical_issue_reason": "Issue 45920 is the best canonical anchor for the tokenizer regression subcluster: it states the broadest, most user-facing form of the bug and covers multiple affected model families, making it a better umbrella than the narrower Granite-only report.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45920 is the strongest overall issue candidate because it describes a broad v5 regression with clear symptom wording and multi-model impact, which makes it the most representative item among the actual duplicate-like reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45800", + "right": "issue:46304", + "accept": false, + "reason": "Both are version-compatibility complaints, but they concern different incompatibilities and different failure modes; no shared concrete code-path bug." + }, + { + "left": "issue:43866", + "right": "issue:46402", + "accept": false, + "reason": "Checkpoint corruption vs missing text-encoder weights are unrelated model-loading problems." + }, + { + "left": "issue:43296", + "right": "issue:46304", + "accept": false, + "reason": "Different stacks and symptoms: vLLM/PaddleOCR-VL load failure versus PyTorch distributed support regression." + }, + { + "left": "issue:41762", + "right": "issue:43596", + "accept": true, + "reason": "Both report the same ZeRO-3 loading failure pattern: index 0 out of bounds during model init/load, plausibly the same underlying bug in the shared load path." + }, + { + "left": "issue:39290", + "right": "issue:41720", + "accept": false, + "reason": "Missing config attribute in Gemma3/vLLM is unrelated to Qwen3 auto-device-map cuda assert." + }, + { + "left": "issue:43901", + "right": "issue:44512", + "accept": false, + "reason": "Both touch docs, but they are different documentation issues about different APIs and different removed behaviors." + }, + { + "left": "issue:44589", + "right": "issue:46304", + "accept": false, + "reason": "Float8 storage lookup failure is unrelated to the PyTorch USE_DISTRIBUTED=0 regression." + }, + { + "left": "issue:45237", + "right": "issue:45901", + "accept": false, + "reason": "Different features and different failures; no shared code-path bug." + }, + { + "left": "issue:43975", + "right": "issue:46015", + "accept": false, + "reason": "Model detokenization regression 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." + }, + { + "left": "issue:45823", + "right": "issue:46294", + "accept": false, + "reason": "Both involve device/input handling, but one is a device mismatch in forward and the other is a video feature/token mismatch during chat templating." + }, + { + "left": "issue:44841", + "right": "issue:45823", + "accept": false, + "reason": "Processor failure for Voxtral and Gemma4 device mismatch are different model-path bugs." + }, + { + "left": "issue:43866", + "right": "issue:46772", + "accept": false, + "reason": "Checkpoint corruption is not the same as NVFP4/modelopt compatibility." + }, + { + "left": "issue:44863", + "right": "issue:46015", + "accept": false, + "reason": "NemotronH checkpoint loading and RAG docs are unrelated." + }, + { + "left": "issue:45020", + "right": "issue:46304", + "accept": false, + "reason": "Remote-code breakage and distributed-build breakage are different regressions." + }, + { + "left": "issue:45812", + "right": "issue:45920", + "accept": true, + "reason": "Both describe the same AutoTokenizer v5 regression producing wrong token IDs, just reported against different model families; they plausibly share one fix." + }, + { + "left": "issue:44292", + "right": "issue:46015", + "accept": false, + "reason": "NVFP4 model execution failure and RAG documentation are unrelated." + }, + { + "left": "issue:43866", + "right": "issue:44863", + "accept": false, + "reason": "Corrupted checkpoint and checkpoint loading implementation issue are not the same bug." + }, + { + "left": "issue:43872", + "right": "issue:46304", + "accept": false, + "reason": "bitsandbytes constructor incompatibility is unrelated to the USE_DISTRIBUTED regression." + }, + { + "left": "issue:39290", + "right": "issue:43531", + "accept": false, + "reason": "Both mention sliding_window, but one is a missing config attribute in vLLM and the other is a model-specific sliding_window issue." + }, + { + "left": "issue:41093", + "right": "issue:43638", + "accept": false, + "reason": "Both are IndexErrors, but the shapes and triggering paths differ; this is too generic to be a duplicate." + }, + { + "left": "issue:43792", + "right": "issue:45901", + "accept": false, + "reason": "Whisper load/runtime failure and table-question-answering crash are unrelated." + }, + { + "left": "issue:43854", + "right": "issue:43866", + "accept": false, + "reason": "Model loading test failure and checkpoint corruption are different problems." + }, + { + "left": "issue:43425", + "right": "issue:45070", + "accept": false, + "reason": "Torch version incompatibility and pydantic/config regression are unrelated." + }, + { + "left": "issue:43975", + "right": "issue:45901", + "accept": false, + "reason": "Detokenization regression and table QA crash do not share the same bug." + }, + { + "left": "issue:45362", + "right": "issue:45464", + "accept": false, + "reason": "Different Qwen3.5 issues with different symptoms and code paths." + }, + { + "left": "issue:43792", + "right": "issue:45070", + "accept": false, + "reason": "Whisper failure and pydantic field breakage are unrelated." + }, + { + "left": "issue:43596", + "right": "issue:45800", + "accept": false, + "reason": "Both mention loading/runtime problems, but one is ZeRO-3 index error and the other is a torch/transformers compatibility complaint." + }, + { + "left": "issue:43541", + "right": "issue:46869", + "accept": false, + "reason": "Different model families and different errors; no shared fix is plausible." + }, + { + "left": "issue:44292", + "right": "issue:45901", + "accept": false, + "reason": "Different task and failure mode." + }, + { + "left": "issue:39290", + "right": "issue:45823", + "accept": false, + "reason": "Gemma3 sliding-window config error is unrelated to Gemma4 device-mapping mismatch." + }, + { + "left": "issue:45464", + "right": "issue:45685", + "accept": false, + "reason": "Chat/completions streaming failure and MPS histogram error are unrelated." + }, + { + "left": "issue:36010", + "right": "issue:39290", + "accept": false, + "reason": "Import path breakage and Gemma3/vLLM config mismatch are unrelated." + }, + { + "left": "issue:43541", + "right": "issue:45685", + "accept": false, + "reason": "Different MoE/runtime problems with no shared concrete code path." + }, + { + "left": "issue:43541", + "right": "issue:43828", + "accept": false, + "reason": "Grouped_mm tracing error and autocast dtype mismatch are different problems despite both involving MoE models." + }, + { + "left": "issue:44841", + "right": "issue:45084", + "accept": false, + "reason": "Processor failure and template-node compile error are unrelated." + }, + { + "left": "issue:43404", + "right": "issue:43866", + "accept": false, + "reason": "Weight-tying bug in Mistral3 is unrelated to a corrupted checkpoint report." + }, + { + "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 IndexErrors, but the failure shapes and model-loading contexts differ enough that this is not the same bug." + }, + { + "left": "issue:45399", + "right": "issue:46661", + "accept": false, + "reason": "Flash-attn fallback gating and remote-code AutoModel config_class errors are different bugs." + }, + { + "left": "issue:43572", + "right": "issue:45070", + "accept": false, + "reason": "Missing pad_token_idx and pydantic field breakage are unrelated." + }, + { + "left": "issue:45800", + "right": "issue:45823", + "accept": false, + "reason": "Torch/transformers incompatibility and Gemma4 device mismatch are different regression classes." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: it uses only the issue titles to identify two plausible duplicate-like subclusters (DeepSpeed ZeRO-3 index errors and AutoTokenizer v5 wrong-token regressions) and keeps the rest conservative. The accepted soft edges are supported by strong title-level similarity, and the rejected pairs do not overclaim a shared bug based on the available evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 3, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "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:45507", + "issue:45632", + "issue:45698", + "issue:45759", + "issue:45800", + "issue:45902", + "issue:46165", + "issue:46294", + "issue:46463", + "issue:46468", + "issue:46640", + "issue:46661", + "issue:46869" + ], + "soft_pairs": [ + "issue:43296|issue:43366", + "issue:44805|issue:44918", + "issue:43541|issue:45161", + "issue:43828|issue:45464", + "issue:44387|issue:45005", + "issue:43366|issue:45464", + "issue:44291|issue:45070", + "issue:43531|issue:45070", + "issue:41093|issue:43596", + "issue:43716|issue:46294", + "issue:45020|issue:45800", + "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:44360|issue:44512", + "issue:44530|issue:45005", + "issue:44492|issue:44512", + "issue:44863|issue:45005", + "issue:45183|issue:46463", + "issue:44918|issue:45507" + ], + "prepared_review_unit_hash": "da132711c161a507e1b51a4df9b205a91effcad86edc673a0b928ba8b7f91a6f", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36241, + "estimated_input_tokens": 9061, + "estimated_eval_tokens": 18378 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous model/regression reports with a loose overlap around Transformers v5 loading/custom-code behavior. None of the soft pairs look like true duplicates; they share themes but not the same concrete bug.", + "confidence": 0.23, + "canonical_issue_reason": "Issue 45020 is the broadest representative of the cluster\u2019s strongest recurring theme: recent Transformers versions breaking remote-code/custom-model loading.", + "canonical_pr_reason": null, + "best_issue_reason": "45020 is the best single issue to anchor the cluster because it is broad, recent, and central to the loading-regression pattern seen in several nearby reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43296", + "right": "issue:43366", + "accept": false, + "reason": "Different failures: PaddleOCR-VL load regression vs GGUF gpt-oss support request." + }, + { + "left": "issue:44805", + "right": "issue:44918", + "accept": false, + "reason": "Both are shape/embedding related, but the concrete bugs are different." + }, + { + "left": "issue:43541", + "right": "issue:45161", + "accept": false, + "reason": "Both involve MoE/parallelism, but one is grouped_mm tracing and the other is TP support." + }, + { + "left": "issue:43828", + "right": "issue:45464", + "accept": false, + "reason": "Different code paths: autocast dtype mismatch vs streaming chat/completions failure." + }, + { + "left": "issue:44387", + "right": "issue:45005", + "accept": false, + "reason": "OOM under int4 quantization is unrelated to tied-weights regressions." + }, + { + "left": "issue:43366", + "right": "issue:45464", + "accept": false, + "reason": "gpt-oss GGUF support and Qwen3.5 streaming failures are unrelated." + }, + { + "left": "issue:44291", + "right": "issue:45070", + "accept": false, + "reason": "init_empty_weights argument handling is distinct from the pydantic PretrainedConfig regression." + }, + { + "left": "issue:43531", + "right": "issue:45070", + "accept": false, + "reason": "Qwen3-MoE sliding_window bug is unrelated to config-field breakage." + }, + { + "left": "issue:41093", + "right": "issue:43596", + "accept": false, + "reason": "Both are IndexErrors, but the tensor/mask mismatch and zero3 init failure are different bugs." + }, + { + "left": "issue:43716", + "right": "issue:46294", + "accept": false, + "reason": "Image-preprocessor dtype mismatch is not the same as video token/feature mismatch." + }, + { + "left": "issue:45020", + "right": "issue:45800", + "accept": false, + "reason": "Both are version regressions, but one is remote_code breakage and the other is torch 2.4.1 incompatibility." + }, + { + "left": "issue:43054", + "right": "issue:43994", + "accept": false, + "reason": "Both mention SigLIP2, but one is embedding quality and the other is bad AutoModel/pipeline outputs." + }, + { + "left": "issue:44479", + "right": "issue:46294", + "accept": false, + "reason": "General v5.3.0 video regression is different from a specific Qwen3.5-9B template mismatch." + }, + { + "left": "issue:46468", + "right": "issue:46869", + "accept": false, + "reason": "Broken example docs and multi-GPU device-map failure are not the same bug." + }, + { + "left": "issue:45759", + "right": "issue:45902", + "accept": false, + "reason": "One is text_config unwrapping, the other is uninitialized rotary embedding memory." + }, + { + "left": "issue:43596", + "right": "issue:46165", + "accept": false, + "reason": "Zero3/BertModel indexing and remote-code resume issues do not share the same root cause." + }, + { + "left": "issue:46640", + "right": "issue:46661", + "accept": false, + "reason": "Flash Attention fallback and remote-code model_class/config_class errors are unrelated." + }, + { + "left": "issue:43716", + "right": "issue:44918", + "accept": false, + "reason": "Different multimodal/training failures: image dtype mismatch vs Qwen3.5 embedding unpacking." + }, + { + "left": "issue:39186", + "right": "issue:44918", + "accept": false, + "reason": "FSDP weight dimensionality and Qwen3.5 embedding unpacking are different problems." + }, + { + "left": "issue:40990", + "right": "issue:44960", + "accept": false, + "reason": "Perplexity regression on gpt-oss and GLM5 model support are unrelated." + }, + { + "left": "issue:43645", + "right": "issue:45399", + "accept": false, + "reason": "Jupyter custom-model initialization and flash-attn fallback blocking are different issues." + }, + { + "left": "issue:45632", + "right": "issue:45698", + "accept": false, + "reason": "Both involve remote-code loading, but cache-path collision and wrong module loading after save are not the same bug." + }, + { + "left": "issue:43404", + "right": "issue:45005", + "accept": false, + "reason": "Tied lm_head in Mistral3 and translation-model tied weights regressions are separate concrete cases." + }, + { + "left": "issue:42915", + "right": "issue:45399", + "accept": false, + "reason": "Qwen3-MoE FP8 failure is unrelated to flash-attn fallback logic." + }, + { + "left": "issue:43873", + "right": "issue:45005", + "accept": false, + "reason": "Quantization offloading behavior and tied-weights regression are different code paths." + }, + { + "left": "issue:44360", + "right": "issue:44512", + "accept": false, + "reason": "DSA indexer activation bug and docs wording change are unrelated." + }, + { + "left": "issue:44530", + "right": "issue:45005", + "accept": false, + "reason": "PagedAttentionCache group-type crash is unrelated to tied-weights issues." + }, + { + "left": "issue:44492", + "right": "issue:44512", + "accept": false, + "reason": "A cache-strategy typo and a removed-doc-command note are not duplicates." + }, + { + "left": "issue:44863", + "right": "issue:45005", + "accept": false, + "reason": "NemotronH checkpoint loading and translation-model tied weights are unrelated." + }, + { + "left": "issue:45183", + "right": "issue:46463", + "accept": false, + "reason": "Both are validation-message bugs, but one is XOR logic across models and the other is T5 decoder-input validation." + }, + { + "left": "issue:44918", + "right": "issue:45507", + "accept": false, + "reason": "Qwen3.5 embedding unpacking and GraniteMoEHybrid invalid method 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": 26, + "review_unit_index": 4, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "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:44928", + "issue:44960", + "issue:44991", + "issue:45020", + "issue:45399", + "issue:45520", + "issue:45663", + "issue:45701", + "issue:45859", + "issue:45902", + "issue:45920", + "issue:45987", + "issue:46489", + "issue:46491", + "issue:46858", + "issue:46899", + "issue:46932" + ], + "soft_pairs": [ + "issue:43975|issue:44292", + "issue:46489|issue:46491", + "issue:39692|issue:43054", + "issue:43950|issue:44534", + "issue:36331|issue:41762", + "issue:43644|issue:43950", + "issue:43872|issue:44589", + "issue:46899|issue:46932", + "issue:43828|issue:44928", + "issue:45920|issue:46489", + "issue:43054|issue:45701", + "issue:45987|issue:46858", + "issue:38175|issue:43054", + "issue:43950|issue:45902", + "issue:44910|issue:45859", + "issue:43054|issue:43493", + "issue:42947|issue:44387", + "issue:41553|issue:43883", + "issue:45663|issue:45902", + "issue:44403|issue:44451", + "issue:43975|issue:45520", + "issue:24643|issue:28530", + "issue:42947|issue:43856", + "issue:44403|issue:44488", + "issue:43643|issue:45020", + "issue:41762|issue:42915", + "issue:41553|issue:43881", + "issue:44991|issue:45399", + "issue:43824|issue:43872", + "issue:43782|issue:44960" + ], + "prepared_review_unit_hash": "cf0634f2e240521ccb9c9033aea1b9081b0b8e2f85d35144ac5e1ce08ba85312", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36283, + "estimated_input_tokens": 9071, + "estimated_eval_tokens": 18398 + }, + "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 bug reports that only share broad themes like loading, tokenization, cache handling, and quantization. A few pairs are true duplicates, especially the non-persistent-buffer regression, the StaticCache get_seq_length contract breakage, and the Gemma4 quantized embedder dtype-cast bug.", + "confidence": 0.82, + "canonical_issue_reason": "43950 is the strongest canonical anchor: it describes the broadest and clearest regression class (non-persistent buffers being corrupted/junk-filled in Transformers 5.x), and it best captures the shared underlying bug seen in the closest duplicate pair.", + "canonical_pr_reason": null, + "best_issue_reason": "43950 is the best representative issue because it is broadly worded, regression-focused, and central to one of the clearest duplicate groups in the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43975", + "right": "issue:44292", + "accept": false, + "reason": "Different failures on different models: DeepSeek-Coder detokenization vs Qwen NVFP4 load/runtime error. Same broad area, not the same bug." + }, + { + "left": "issue:46489", + "right": "issue:46491", + "accept": false, + "reason": "Both are tokenizer regressions, but they describe different concrete model-specific symptoms and no shared code path is evident from the titles alone." + }, + { + "left": "issue:39692", + "right": "issue:43054", + "accept": false, + "reason": "Both mention SigLIP2, but one is a docs/example mismatch and quantization failure, while the other is a model quality/embedding discrepancy." + }, + { + "left": "issue:43950", + "right": "issue:44534", + "accept": true, + "reason": "Same underlying non-persistent-buffer regression: both describe v5 filling/corrupting register_buffer(persistent=False) state with junk." + }, + { + "left": "issue:36331", + "right": "issue:41762", + "accept": false, + "reason": "Unrelated errors: custom trainer compute_loss kwarg mismatch vs Gemma3 loading failure under DeepSpeed ZeRO-3." + }, + { + "left": "issue:43643", + "right": "issue:43950", + "accept": false, + "reason": "Remote-code config field loss and non-persistent-buffer corruption are different bugs affecting different parts of the loading stack." + }, + { + "left": "issue:43872", + "right": "issue:44589", + "accept": false, + "reason": "Different quantization/storage failures: bitsandbytes Int8Params constructor incompatibility vs missing Float8 storage object." + }, + { + "left": "issue:46899", + "right": "issue:46932", + "accept": true, + "reason": "These are the same Gemma4 quantized vision/audio embedder dtype-casting bug; the second explicitly calls it the same issue." + }, + { + "left": "issue:43828", + "right": "issue:44928", + "accept": false, + "reason": "Different model and failure mode: autocast dtype mismatch vs RLHF NaN/SDPA fallback from 3D position_ids." + }, + { + "left": "issue:45920", + "right": "issue:46489", + "accept": false, + "reason": "Both are tokenizer regressions, but the first is a broad AutoTokenizer token-id issue across many models while the second is a specific DeepSeek-Coder output bug tied to a prior fix gap." + }, + { + "left": "issue:43054", + "right": "issue:45701", + "accept": false, + "reason": "SigLIP2 embedding quality complaint vs general tokenizer-change report; no concrete same bug is shown." + }, + { + "left": "issue:45987", + "right": "issue:46858", + "accept": true, + "reason": "Same StaticCache.get_seq_length contract breakage: one report says it returns a shape-(1,) tensor, the other says it returns a device tensor and breaks slice arithmetic." + }, + { + "left": "issue:38175", + "right": "issue:43054", + "accept": false, + "reason": "Both involve SigLIP2 behavior, but one reports zero probabilities and the other lower text-embedding quality; too little evidence of the same bug." + }, + { + "left": "issue:43950", + "right": "issue:45902", + "accept": false, + "reason": "Related around buffer initialization, but one is a broad from_pretrained non-persistent-buffer corruption regression and the other is a specific model/materialization uninitialized-memory read." + }, + { + "left": "issue:44910", + "right": "issue:45859", + "accept": false, + "reason": "Different problems: Flash Attention crash from packed-sequence misinterpretation vs CPU-offload incompatibility in a Qwen3_5MoE rotary embedding." + }, + { + "left": "issue:43054", + "right": "issue:43493", + "accept": false, + "reason": "Both discuss SigLIP2, but one is about degraded embeddings and the other is a discrepancy versus JAX implementation; not clearly the same concrete bug." + }, + { + "left": "issue:42947", + "right": "issue:44387", + "accept": false, + "reason": "Different training pathologies: LoRA gradient checkpointing ineffective vs int4 quantization causing higher CUDA reserved memory/OOM." + }, + { + "left": "issue:41553", + "right": "issue:43883", + "accept": false, + "reason": "Unrelated model-loading issues: bad AutoTokenizer error message for Voxtral vs missing all_tied_weights_keys on MolmoForCausalLM." + }, + { + "left": "issue:45663", + "right": "issue:45902", + "accept": false, + "reason": "Different code paths and symptoms: Gemma4 FSDP2 shared_kv_states KeyError vs non-persistent buffer materialization reading junk." + }, + { + "left": "issue:44403", + "right": "issue:44451", + "accept": false, + "reason": "One is about noisy loading logs, the other is a model-loading failure for ScandiBERT." + }, + { + "left": "issue:43975", + "right": "issue:45520", + "accept": false, + "reason": "Different issues: tokenizer detokenization bug vs Python 3.13 import_utils flash_attn KeyError." + }, + { + "left": "issue:24643", + "right": "issue:28530", + "accept": false, + "reason": "Training with DeepSpeed weight shape error vs early-stopping metric configuration; unrelated failures." + }, + { + "left": "issue:42947", + "right": "issue:43856", + "accept": false, + "reason": "Both concern memory/training efficiency, but they are different model/workflow problems and not the same concrete bug." + }, + { + "left": "issue:44403", + "right": "issue:44488", + "accept": false, + "reason": "Different symptom severity and scope: noisy transformer loading vs actual loading failure for a specific model." + }, + { + "left": "issue:43643", + "right": "issue:45020", + "accept": false, + "reason": "Remote-code handling is the common theme, but one is missing fields in AutoConfig output while the other is a broad report of model breakage; not the same concrete defect." + }, + { + "left": "issue:41762", + "right": "issue:42915", + "accept": false, + "reason": "Different load-time failures on different models and configurations: Gemma3 ZeRO-3 index error vs Qwen3Moe FineGrainedFP8Config failure." + }, + { + "left": "issue:41553", + "right": "issue:43881", + "accept": false, + "reason": "Unrelated model-loading complaints: Voxtral tokenizer error message vs glm-4v-9b load failure." + }, + { + "left": "issue:44991", + "right": "issue:45399", + "accept": false, + "reason": "Tokenizer loading regression vs flash-attn fallback gating; different subsystems and failure modes." + }, + { + "left": "issue:43824", + "right": "issue:43872", + "accept": false, + "reason": "Import error for a missing model class is unrelated to the bitsandbytes Int8Params constructor incompatibility." + }, + { + "left": "issue:43782", + "right": "issue:44960", + "accept": false, + "reason": "Different models and failure modes: weight_only=True loading error vs a generic GLM5 issue." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary stays grounded in the packet and is appropriately conservative about the cluster being mostly loose/unrelated with only a few likely duplicate pairs. The accepted soft-edge pairs are supported by closely matching titles, and the rejected pairs avoid overclaiming beyond the evidence available in the packet." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 5, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "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:43881", + "issue:43883", + "issue:43940", + "issue:43950", + "issue:44220", + "issue:44291", + "issue:44368", + "issue:44479", + "issue:44509", + "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:45685", + "issue:45715", + "issue:45759", + "issue:45823", + "issue:45923", + "issue:46315", + "issue:46489", + "issue:46661", + "issue:46869", + "issue:46894", + "issue:46897" + ], + "soft_pairs": [ + "issue:45685|issue:45823", + "issue:45081|issue:45399", + "issue:45020|issue:45823", + "issue:44661|issue:45020", + "issue:43881|issue:43883", + "issue:43828|issue:45823", + "issue:42491|issue:44368", + "issue:45020|issue:45356", + "issue:45715|issue:46315", + "issue:43872|issue:46869", + "issue:45759|issue:46661", + "issue:45488|issue:46489", + "issue:43827|issue:44509", + "issue:43404|issue:43950", + "issue:46894|issue:46897", + "issue:43872|issue:44291", + "issue:45020|issue:46661", + "issue:43792|issue:44220", + "issue:43257|issue:45305", + "issue:42947|issue:45923", + "issue:36331|issue:41093", + "issue:43940|issue:45313", + "issue:41553|issue:44991", + "issue:43854|issue:44863", + "issue:44928|issue:45588", + "issue:41093|issue:44805", + "issue:44479|issue:44560", + "issue:24643|issue:30064" + ], + "prepared_review_unit_hash": "cc1d6fbcfefd55e3ff8420db373cd92525024dd532cda1e0057a7c0c7f76b7e8", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36753, + "estimated_input_tokens": 9189, + "estimated_eval_tokens": 18634 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly a heterogeneous cluster of unrelated bug reports, with a few true duplicate pockets: docs about removed pipelines, init_empty_weights/_is_hf_initialized loading failures, identical mask-shape IndexErrors, and the Florence/Cohere double-shift pattern. The rest are similarity-only false positives across different models or subsystems.", + "confidence": 0.82, + "canonical_issue_reason": "No single cluster-wide canonical issue exists because the items span many unrelated bugs; if one issue must anchor the only clearly duplicated concrete bug here, 46894 is the best choice as the earlier, more explicit report paired with 46897.", + "canonical_pr_reason": null, + "best_issue_reason": "46894 is the strongest issue to keep as the representative anchor: it is the earlier member of the clearest true duplicate pair and states the underlying bug cleanly.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45685", + "right": "issue:45823", + "accept": false, + "reason": "Different symptoms and code paths: MPS histogram int error vs Gemma4 device-mapping mismatch." + }, + { + "left": "issue:45081", + "right": "issue:45399", + "accept": false, + "reason": "Both touch attention/tokenizer areas, but one is a Mistral regex/tokenizer crash and the other is a flash-attn fallback blocker." + }, + { + "left": "issue:45020", + "right": "issue:45823", + "accept": false, + "reason": "Broad \"recent versions break remote_code\" report is not the same bug as a Gemma4 forward device mismatch." + }, + { + "left": "issue:44661", + "right": "issue:45020", + "accept": false, + "reason": "Tokenizer mapping edge case vs remote-code model loading regressions; too different to treat as one bug." + }, + { + "left": "issue:43881", + "right": "issue:43883", + "accept": false, + "reason": "Both are model-load failures, but for different models and different missing attributes." + }, + { + "left": "issue:43828", + "right": "issue:45823", + "accept": false, + "reason": "Dtype mismatch under autocast is a different failure mode from a device placement bug." + }, + { + "left": "issue:42491", + "right": "issue:44368", + "accept": false, + "reason": "Both mention Qwen/LoRA, but one is a cross-version compatibility failure and the other is just a warning about tie_word_embeddings." + }, + { + "left": "issue:45020", + "right": "issue:45356", + "accept": false, + "reason": "General remote_code regressions and a Kimi tokenizer codec regression are not the same underlying bug." + }, + { + "left": "issue:45715", + "right": "issue:46315", + "accept": false, + "reason": "Same tokenizer performance theme, but different functions and likely different hot paths." + }, + { + "left": "issue:43872", + "right": "issue:46869", + "accept": false, + "reason": "Bitsandbytes init-empty-weights incompatibility is unrelated to YOLO multi-GPU device-map failure." + }, + { + "left": "issue:45759", + "right": "issue:46661", + "accept": false, + "reason": "Both involve model construction, but one is config unwrapping for composite multimodal configs and the other is remote-code config_class resolution." + }, + { + "left": "issue:45488", + "right": "issue:46489", + "accept": false, + "reason": "Both are tokenizer regressions around DeepSeek-family models, but they describe different failures and different root causes." + }, + { + "left": "issue:43827", + "right": "issue:44509", + "accept": true, + "reason": "Same docs cleanup: both say removed pipeline tasks are still referenced after v5 removals." + }, + { + "left": "issue:43404", + "right": "issue:43950", + "accept": false, + "reason": "Tied lm_head weights vs non-persistent buffer corruption are distinct model-state bugs." + }, + { + "left": "issue:46894", + "right": "issue:46897", + "accept": true, + "reason": "Near-identical duplicate reports of the same training-loss double-shift bug." + }, + { + "left": "issue:43872", + "right": "issue:44291", + "accept": true, + "reason": "Same unexpected _is_hf_initialized / init_empty_weights loading failure." + }, + { + "left": "issue:45020", + "right": "issue:46661", + "accept": false, + "reason": "Generic remote-code breakage and an AutoModel config_class AttributeError are related only at a high level." + }, + { + "left": "issue:43792", + "right": "issue:44220", + "accept": false, + "reason": "Whisper not running and a specific fbank extraction issue are plausibly related, but not clearly the same concrete bug." + }, + { + "left": "issue:43257", + "right": "issue:45305", + "accept": false, + "reason": "Both involve DeepSpeed/Qwen training, but one is weight conversion at load time and the other is gradient averaging during training." + }, + { + "left": "issue:42947", + "right": "issue:45923", + "accept": false, + "reason": "Gradient checkpointing problems, but one is PEFT LoRA ineffectiveness and the other is a missing support flag on a remote-code variant." + }, + { + "left": "issue:36331", + "right": "issue:41093", + "accept": false, + "reason": "Unexpected compute_loss kwarg and mask-shape IndexError are unrelated training failures." + }, + { + "left": "issue:43940", + "right": "issue:45313", + "accept": false, + "reason": "Both are ZeRO-3 weight-loading failures, but on different Qwen variants with different reported failure shapes; not enough to merge." + }, + { + "left": "issue:41553", + "right": "issue:44991", + "accept": false, + "reason": "Bad tokenizer error message for Voxtral vs tokenizer loading failure for EMBEDDIA/est-roberta are different issues." + }, + { + "left": "issue:43854", + "right": "issue:44863", + "accept": false, + "reason": "Different model families and different load/test failures." + }, + { + "left": "issue:44928", + "right": "issue:45588", + "accept": false, + "reason": "RLHF gradient explosion from 3D position_ids is unrelated to a flash_attention s_aux=None crash." + }, + { + "left": "issue:41093", + "right": "issue:44805", + "accept": true, + "reason": "Same generic mask-shape mismatch IndexError; likely the same label/mask alignment bug reported twice." + }, + { + "left": "issue:44479", + "right": "issue:44560", + "accept": false, + "reason": "Both are Qwen3 VL/video regressions, but they describe different broken paths and symptoms." + }, + { + "left": "issue:24643", + "right": "issue:30064", + "accept": false, + "reason": "DeepSpeed training weight shape error and void segmentation map processing are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The cluster read is grounded in the packet and stays mostly conservative. The three accepted soft pairs are well supported by highly similar titles and error patterns, and the representative choice of 46894 over 46897 is reasonable as the earlier duplicate. The summary appropriately characterizes the rest as heterogeneous without over-claiming a cluster-wide canonical issue." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 6, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:36010", + "issue:38175", + "issue:41553", + "issue:42491", + "issue:42617", + "issue:42915", + "issue:43366", + "issue:43531", + "issue:43550", + "issue:43596", + "issue:43632", + "issue:43638", + "issue:43645", + "issue:43646", + "issue:43716", + "issue:43720", + "issue:43824", + "issue:43828", + "issue:43950", + "issue:43957", + "issue:43994", + "issue:44155", + "issue:44488", + "issue:44530", + "issue:44560", + "issue:44589", + "issue:44661", + "issue:44683", + "issue:44805", + "issue:44863", + "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:46402", + "issue:46489", + "issue:46515", + "issue:46858", + "issue:46899" + ], + "soft_pairs": [ + "issue:44530|issue:44863", + "issue:44863|issue:46402", + "issue:41553|issue:44488", + "issue:43366|issue:43828", + "issue:45325|issue:45902", + "issue:43366|issue:43531", + "issue:43596|issue:46515", + "issue:44155|issue:45584", + "issue:45663|issue:45823", + "issue:43366|issue:44589", + "issue:43638|issue:46515", + "issue:43950|issue:45834", + "issue:43824|issue:43957", + "issue:42915|issue:43957", + "issue:43716|issue:44560", + "issue:44910|issue:46858", + "issue:43645|issue:43646", + "issue:42617|issue:43366", + "issue:44560|issue:45685", + "issue:43550|issue:44910", + "issue:36010|issue:42915", + "issue:44683|issue:46515", + "issue:44560|issue:44918", + "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" + ], + "prepared_review_unit_hash": "dc15327b9947e058094318053db3510dd3026b90e571361f5004d185320969d8", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 38059, + "estimated_input_tokens": 9515, + "estimated_eval_tokens": 19286 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly heterogeneous, with only two clear duplicate subclusters: Transformers v5 custom model initialization, and SigLIP2 bad-output reports. Most other pairs share a subsystem or model family but not the same concrete bug.", + "confidence": 0.63, + "canonical_issue_reason": "issue:43646 is the broadest clean duplicate target for the true custom-model-initialization regression; issue:43645 is just a notebook-specific repro of the same problem.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43646 is the best representative of the accepted duplicate set because it names the underlying regression cleanly and broadly.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44530", + "right": "issue:44863", + "accept": false, + "reason": "Both are model-specific loading failures, but they concern different architectures and different failure modes." + }, + { + "left": "issue:44863", + "right": "issue:46402", + "accept": false, + "reason": "Both mention weights not loading, but they are different models and likely different code paths." + }, + { + "left": "issue:41553", + "right": "issue:44488", + "accept": false, + "reason": "Both are loading-related, but one is an AutoTokenizer error message and the other is a separate model-loading regression." + }, + { + "left": "issue:43366", + "right": "issue:43828", + "accept": false, + "reason": "Different models and different runtime failures; only a loose MoE-family similarity." + }, + { + "left": "issue:45325", + "right": "issue:45902", + "accept": false, + "reason": "Both involve positional/rotary internals, but they are on different models with different bugs." + }, + { + "left": "issue:43366", + "right": "issue:43531", + "accept": false, + "reason": "One is GGUF support for gpt-oss; the other is a Qwen3-MoE sliding_window bug." + }, + { + "left": "issue:43596", + "right": "issue:46515", + "accept": false, + "reason": "An index error during Deepspeed/Bert init is not the same as a broad torch-version incompatibility report." + }, + { + "left": "issue:44155", + "right": "issue:45584", + "accept": false, + "reason": "Both are audio-related, but one is cross-track embedding leakage and the other is empty-transcription generation failure." + }, + { + "left": "issue:45663", + "right": "issue:45823", + "accept": false, + "reason": "Same model family, but FSDP2 shared_kv_state rebuilding and device_map auto PLE mismatch are different bugs." + }, + { + "left": "issue:43366", + "right": "issue:44589", + "accept": false, + "reason": "These are unrelated failures on different models; one is a support request, the other is a storage-type error." + }, + { + "left": "issue:43638", + "right": "issue:46515", + "accept": false, + "reason": "Both mention training/runtime issues, but the specific Deepspeed/Bert index error is unrelated to torch 2.4.1 incompatibility." + }, + { + "left": "issue:43950", + "right": "issue:45834", + "accept": false, + "reason": "Different regressions: non-persistent buffer corruption versus a Kosmos2.5 OCR length/index problem." + }, + { + "left": "issue:43824", + "right": "issue:43957", + "accept": false, + "reason": "ImportError for a missing class is not the same bug as meta-device loading breakage." + }, + { + "left": "issue:42915", + "right": "issue:43957", + "accept": false, + "reason": "Qwen3-MoE FP8 failure and meta-device loading regression are distinct code paths." + }, + { + "left": "issue:43716", + "right": "issue:44560", + "accept": false, + "reason": "Different multimodal bugs: image-preprocessor dtype mismatch versus Qwen3-VL video StopIteration." + }, + { + "left": "issue:44910", + "right": "issue:46858", + "accept": false, + "reason": "Both touch caching/position handling, but one is a Flash Attention packed-sequence crash and the other is a device-tensor seq-length arithmetic bug." + }, + { + "left": "issue:43645", + "right": "issue:43646", + "accept": true, + "reason": "Same custom-model-initialization regression in Transformers 5.0.0; 43645 is just a Jupyter-notebook repro." + }, + { + "left": "issue:42617", + "right": "issue:43366", + "accept": false, + "reason": "A script execution failure and GGUF support request are not the same underlying issue." + }, + { + "left": "issue:44560", + "right": "issue:45685", + "accept": false, + "reason": "Both are model/runtime failures, but one is Qwen3-VL video handling and the other is an MPS histogram error in MoE code." + }, + { + "left": "issue:43550", + "right": "issue:44910", + "accept": false, + "reason": "Torch.compile+SDPA on Bamba-9B-v2 is a different failure from Qwen3.5 flash-attention position-id handling." + }, + { + "left": "issue:36010", + "right": "issue:42915", + "accept": false, + "reason": "Importing GenerationMixin and Qwen3-MoE FP8 are unrelated regressions." + }, + { + "left": "issue:44683", + "right": "issue:46515", + "accept": false, + "reason": "Compiled flex_attention on torch>=2.9 is a different compatibility break than the torch 2.4.1 report." + }, + { + "left": "issue:44560", + "right": "issue:44918", + "accept": false, + "reason": "Both mention Qwen3.5 VL, but one is a video StopIteration and the other is SFT embedding unpacking." + }, + { + "left": "issue:45812", + "right": "issue:46489", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different model families and likely different mappings." + }, + { + "left": "issue:45823", + "right": "issue:46515", + "accept": false, + "reason": "Gemma4 device-mapping and a generic torch incompatibility are not the same bug." + }, + { + "left": "issue:44661", + "right": "issue:45823", + "accept": false, + "reason": "add-new-model-like mapping logic is unrelated to Gemma4 PLE device placement." + }, + { + "left": "issue:38175", + "right": "issue:43994", + "accept": true, + "reason": "Both reports are about the same SigLIP2 base model producing obviously wrong outputs; the later issue is the broader duplicate target." + }, + { + "left": "issue:44560", + "right": "issue:44805", + "accept": false, + "reason": "Different symptoms and code paths: video StopIteration versus mask shape mismatch." + }, + { + "left": "issue:45672", + "right": "issue:46899", + "accept": false, + "reason": "Both are Gemma4 quantization issues, but one is an audio-module uint8 torch.finfo error and the other is unsafe dtype casting in multimodal embedders." + }, + { + "left": "issue:44877", + "right": "issue:45030", + "accept": false, + "reason": "Config validation blocking granite_speech and tiny-random glm4v loading are separate config regressions." + }, + { + "left": "issue:43632", + "right": "issue:45834", + "accept": false, + "reason": "Both are runtime errors, but one is the `_is_hf_initialized` flag regression and the other is a Kosmos2.5 OCR index error." + }, + { + "left": "issue:43638", + "right": "issue:44661", + "accept": false, + "reason": "Deepspeed zero3/Bert initialization and add-new-model-like mapping validation are not the same failure." + }, + { + "left": "issue:43596", + "right": "issue:44661", + "accept": false, + "reason": "A zero-sized parameter index error is unrelated to tokenizer-mapping validation." + }, + { + "left": "issue:42491", + "right": "issue:43720", + "accept": false, + "reason": "Qwen3-MoE LoRA compatibility and BitNet AutoBitLinear packed-weight loading are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet, and the soft-edge decisions stay conservative overall. The two accepted duplicate pairs are well supported by the issue titles, while the rest are reasonably rejected as similar subsystems/model families but different concrete bugs." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 7, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:38175", + "issue:39692", + "issue:41720", + "issue:42491", + "issue:42915", + "issue:43257", + "issue:43317", + "issue:43644", + "issue:43645", + "issue:43708", + "issue:43854", + "issue:43856", + "issue:43873", + "issue:43931", + "issue:43994", + "issue:44075", + "issue:44164", + "issue:44291", + "issue:44387", + "issue:44451", + "issue:44488", + "issue:44534", + "issue:44661", + "issue:44756", + "issue:44805", + "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:43317|issue:44756", + "issue:44661|issue:44805", + "issue:38175|issue:39692", + "issue:43644|issue:44534", + "issue:41720|issue:42915", + "issue:42915|issue:44756", + "issue:42491|issue:45276", + "issue:44960|issue:44977", + "issue:44451|issue:44488", + "issue:44863|issue:46846", + "issue:46459|issue:46512", + "issue:45313|issue:45357", + "issue:45507|issue:46468", + "issue:43644|issue:43645", + "issue:39692|issue:43994", + "issue:43257|issue:43854", + "issue:44488|issue:44991", + "issue:43856|issue:43873", + "issue:42491|issue:45292", + "issue:43854|issue:43931", + "issue:43856|issue:44387", + "issue:43708|issue:44075", + "issue:42491|issue:45335", + "issue:44960|issue:46050", + "issue:44164|issue:44291", + "issue:46897|issue:46901", + "issue:45488|issue:45812", + "issue:45276|issue:45335", + "issue:45362|issue:45406", + "issue:44910|issue:45588" + ], + "prepared_review_unit_hash": "4eb2cb3666c77204b56937f1e960a0136953eefec77515171e3570d6455129ff", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36657, + "estimated_input_tokens": 9165, + "estimated_eval_tokens": 18586 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly heterogeneous and not one clean duplicate cluster. I only see a few genuine duplicate pairs: the Transformers v5 non-persistent-buffer corruption reports, the `resize_token_embeddings` output-embedding mismatch reports, and the training-loss double-shift reports.", + "confidence": 0.71, + "canonical_issue_reason": "Issue 43644 is the best canonical representative of the clearest duplicate subcluster: 44534 is essentially a reworded report of the same Transformers v5 non-persistent-buffer junk regression.", + "canonical_pr_reason": null, + "best_issue_reason": "43644 is the strongest overall issue to anchor because it is the most explicit, broad, and clearly duplicated report among the true duplicate subgroups.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43317", + "right": "issue:44756", + "accept": false, + "reason": "Different bugs: one is quantized/dequantized loading with device_map/offload, the other is Strix Halo mmap-related OOM mitigation." + }, + { + "left": "issue:44661", + "right": "issue:44805", + "accept": false, + "reason": "Unrelated failures: tokenizer mapping/name handling vs an attention mask shape mismatch." + }, + { + "left": "issue:38175", + "right": "issue:39692", + "accept": false, + "reason": "Same model family, but different concrete issues: unexpected zero probabilities vs docs example model/processor mismatch and quantization errors." + }, + { + "left": "issue:43644", + "right": "issue:44534", + "accept": true, + "reason": "Same regression, just reworded: both report Transformers v5 filling non-persistent buffers with junk." + }, + { + "left": "issue:41720", + "right": "issue:42915", + "accept": false, + "reason": "Both involve Qwen3, but one is auto device mapping on A800 and the other is FineGrainedFP8Config; different code paths and symptoms." + }, + { + "left": "issue:42915", + "right": "issue:44756", + "accept": false, + "reason": "Completely different areas: FP8/Qwen3-MoE failure versus Strix Halo mmap/OOM behavior." + }, + { + "left": "issue:42491", + "right": "issue:45276", + "accept": false, + "reason": "Different bugs and models: LoRA compatibility across hf4.x/hf5.x for Qwen3 MoE versus Gemma4 resize-token-embedding behavior." + }, + { + "left": "issue:44960", + "right": "issue:44977", + "accept": false, + "reason": "Different model-specific generation bugs; no evidence they share the same concrete code path." + }, + { + "left": "issue:44451", + "right": "issue:44488", + "accept": false, + "reason": "Both are model-loading failures, but for different checkpoints with no evidence of the same root cause." + }, + { + "left": "issue:44863", + "right": "issue:46846", + "accept": false, + "reason": "Different problems: NemotronH checkpoint loading versus Qwen3.5 tensor-parallel planning/OOM at TP>1." + }, + { + "left": "issue:46459", + "right": "issue:46512", + "accept": false, + "reason": "Related to dtype handling, but one is composite checkpoint loading and the other is from_config dtype/config staleness; not the same bug." + }, + { + "left": "issue:45313", + "right": "issue:45357", + "accept": false, + "reason": "Different components and code paths: ZeRO-3 weight loading for language_model versus save_pretrained key serialization for the visual encoder." + }, + { + "left": "issue:45507", + "right": "issue:46468", + "accept": false, + "reason": "Unrelated issues in different model/example flows." + }, + { + "left": "issue:43644", + "right": "issue:43645", + "accept": false, + "reason": "Both are Transformers v5 regressions, but one is buffer initialization and the other is Jupyter custom-model initialization; different concrete bugs." + }, + { + "left": "issue:39692", + "right": "issue:43994", + "accept": false, + "reason": "Same model family, but one is a docs example/mismatch report and the other is nonsensical inference outputs; not the same concrete failure." + }, + { + "left": "issue:43257", + "right": "issue:43854", + "accept": false, + "reason": "Different models and symptoms: Qwen3 MoE conversion under accelerate+deepspeed versus GLM-4.7-Flash unit-test loading." + }, + { + "left": "issue:44488", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-load regressions, but on different models with no direct evidence they stem from the same root cause." + }, + { + "left": "issue:43856", + "right": "issue:43873", + "accept": false, + "reason": "Different resource-management problems: Qwen3 MoE training memory inefficiency versus quantization offload behavior." + }, + { + "left": "issue:42491", + "right": "issue:45292", + "accept": false, + "reason": "Different architectures and failure modes; only the shared phrase `resize_token_embeddings` overlaps." + }, + { + "left": "issue:43854", + "right": "issue:43931", + "accept": false, + "reason": "Different loading failures for different models: GLM-4.7-Flash test loading versus Qwen3-VL weight-shape mismatch." + }, + { + "left": "issue:43856", + "right": "issue:44387", + "accept": false, + "reason": "Both are memory-related, but one is training-memory inefficiency and the other is int4 quantization reserved-memory/OOM regression." + }, + { + "left": "issue:43708", + "right": "issue:44075", + "accept": false, + "reason": "Trainer checkpoint-resume math bug is unrelated to optimizer argument handling." + }, + { + "left": "issue:42491", + "right": "issue:45335", + "accept": false, + "reason": "Different model families and different affected embeddings; same API name alone is not enough." + }, + { + "left": "issue:44960", + "right": "issue:46050", + "accept": false, + "reason": "Different bug classes: GLM5 issue versus quantized-model save/load regression." + }, + { + "left": "issue:44164", + "right": "issue:44291", + "accept": false, + "reason": "Different serialization/loading failures: extra_state handling versus unexpected `_is_hf_initialized` argument." + }, + { + "left": "issue:46897", + "right": "issue:46901", + "accept": true, + "reason": "Same concrete bug pattern: both report the training-loss double-shift where labels are trained against a shifted slice instead of the original labels." + }, + { + "left": "issue:45488", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different model families and different tokenizer behaviors." + }, + { + "left": "issue:45276", + "right": "issue:45335", + "accept": true, + "reason": "Same concrete `resize_token_embeddings` failure class: the resized embedding changes are not propagated to the model's output/decoder embeddings." + }, + { + "left": "issue:45362", + "right": "issue:45406", + "accept": false, + "reason": "Different failures: Qwen3.5 chat crashing versus Gemma4Processor missing `_tokenizer` in transformers serve." + }, + { + "left": "issue:44910", + "right": "issue:45588", + "accept": false, + "reason": "Different flash-attention bugs: 3D `position_ids` misinterpreted as packed sequence versus `s_aux=None` attribute error." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays appropriately conservative. The accepted duplicate pairs are all supported by the issue titles and the analyst avoids overclaiming on the many near-matches that lack direct evidence of the same root cause. The main cluster characterization as mostly heterogeneous is reasonable." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 8, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:36010", + "issue:39404", + "issue:39692", + "issue:41720", + "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:44479", + "issue:44488", + "issue:44756", + "issue:44863", + "issue:44877", + "issue:44912", + "issue:44928", + "issue:44960", + "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" + ], + "soft_pairs": [ + "issue:44402|issue:44488", + "issue:39692|issue:44960", + "issue:41720|issue:44155", + "issue:43425|issue:43957", + "issue:45663|issue:45759", + "issue:42947|issue:45704", + "issue:43404|issue:43454", + "issue:44291|issue:44756", + "issue:43873|issue:44387", + "issue:44079|issue:46829", + "issue:45663|issue:45859", + "issue:44292|issue:44912", + "issue:36010|issue:39404", + "issue:42947|issue:43381", + "issue:44387|issue:44756", + "issue:44155|issue:45823", + "issue:43957|issue:44756", + "issue:43335|issue:45663", + "issue:44863|issue:46468", + "issue:45276|issue:45292", + "issue:44479|issue:44877", + "issue:43381|issue:44928", + "issue:44164|issue:44756", + "issue:42796|issue:46326", + "issue:45588|issue:45663", + "issue:45684|issue:45698", + "issue:43854|issue:45216", + "issue:45292|issue:45335", + "issue:43645|issue:44164", + "issue:43828|issue:43957", + "issue:43646|issue:43708" + ], + "prepared_review_unit_hash": "95cd7d0a21672584bab9b5a5579b38c95d76a18d9305062505c896958e90bf4f", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37085, + "estimated_input_tokens": 9272, + "estimated_eval_tokens": 18800 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most items are unrelated bug reports. The only clear duplicate cluster is around `resize_token_embeddings` not updating model-specific embeddings/output heads, with `issue:45292` as the broadest report. `issue:45276` and `issue:45335` look like model-specific variants of that same bug.", + "confidence": 0.79, + "canonical_issue_reason": "`issue:45292` is the cleanest generic statement of the `resize_token_embeddings` bug and best represents the duplicate cluster; the other matched reports are narrower model-specific variants.", + "canonical_pr_reason": null, + "best_issue_reason": "`issue:45292` is the best global issue candidate because it is concise, model-agnostic, and captures the shared underlying failure more generally than the other resize-token reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44402", + "right": "issue:44488", + "accept": false, + "reason": "Different failure modes: one is tokenizer vocab/indexing, the other is a specific model load regression." + }, + { + "left": "issue:39692", + "right": "issue:44960", + "accept": false, + "reason": "Unrelated topics: docs/example errors versus a GLM5 model issue." + }, + { + "left": "issue:41720", + "right": "issue:44155", + "accept": false, + "reason": "Different code paths and symptoms: device-map CUDA assert versus batched inference token leakage." + }, + { + "left": "issue:43425", + "right": "issue:43957", + "accept": false, + "reason": "Both involve compatibility/loading, but they are distinct bugs with different triggers and failure points." + }, + { + "left": "issue:45663", + "right": "issue:45759", + "accept": false, + "reason": "Different concrete problems: FSDP/shared-kv state bug versus config unwrapping for multimodal Qwen." + }, + { + "left": "issue:42947", + "right": "issue:45704", + "accept": false, + "reason": "Training checkpointing/LoRA behavior is unrelated to the T5 RMSNorm memory leak." + }, + { + "left": "issue:43404", + "right": "issue:43454", + "accept": false, + "reason": "Same general theme of untied lm_head weights, but different model implementations and likely different fixes." + }, + { + "left": "issue:44291", + "right": "issue:44756", + "accept": false, + "reason": "Different problems: init_empty_weights argument handling versus Strix Halo mmap/OOM behavior." + }, + { + "left": "issue:43873", + "right": "issue:44387", + "accept": false, + "reason": "Both mention quantization/memory, but one is offloading behavior and the other is int4 reserved-memory growth." + }, + { + "left": "issue:44079", + "right": "issue:46829", + "accept": false, + "reason": "Generic ModelOutput key assignment bug versus a specific model output field mismatch." + }, + { + "left": "issue:45663", + "right": "issue:45859", + "accept": false, + "reason": "Different model-specific offload issues with different failing components." + }, + { + "left": "issue:44292", + "right": "issue:44912", + "accept": false, + "reason": "Different quantization formats and failure modes; same broad area only." + }, + { + "left": "issue:36010", + "right": "issue:39404", + "accept": false, + "reason": "Import path regression and Whisper pipeline behavior are unrelated." + }, + { + "left": "issue:42947", + "right": "issue:43381", + "accept": false, + "reason": "Both concern gradient checkpointing, but one is ineffective LoRA training and the other is eval-mode incompatibility." + }, + { + "left": "issue:44387", + "right": "issue:44756", + "accept": false, + "reason": "Both end in OOM, but the causes are unrelated." + }, + { + "left": "issue:44155", + "right": "issue:45823", + "accept": false, + "reason": "A batched inference leak is not the same bug as a device-mapping mismatch." + }, + { + "left": "issue:43957", + "right": "issue:44756", + "accept": false, + "reason": "Different loading/memory issues with different root causes." + }, + { + "left": "issue:43335", + "right": "issue:45663", + "accept": false, + "reason": "Sparse-layer config bug is unrelated to Gemma4 FSDP attention state handling." + }, + { + "left": "issue:44863", + "right": "issue:46468", + "accept": false, + "reason": "Checkpoint-loading failure and broken example code are related only by model family, not by the same bug." + }, + { + "left": "issue:45276", + "right": "issue:45292", + "accept": true, + "reason": "Same underlying bug: `resize_token_embeddings` does not propagate correctly to the model's output embeddings; `45276` is just a model-specific report." + }, + { + "left": "issue:44479", + "right": "issue:44877", + "accept": false, + "reason": "Video-input regression and strict config loading are different issues." + }, + { + "left": "issue:43381", + "right": "issue:44928", + "accept": false, + "reason": "Gradient checkpointing eval-mode behavior is unrelated to the SDPA/BF16 instability." + }, + { + "left": "issue:44164", + "right": "issue:44756", + "accept": false, + "reason": "Serialization `extra_state` handling is unrelated to mmap/OOM on Strix Halo." + }, + { + "left": "issue:42796", + "right": "issue:46326", + "accept": false, + "reason": "Both tokenizer-related, but one is BPE marker handling and the other is `continue_final_message` behavior." + }, + { + "left": "issue:45588", + "right": "issue:45663", + "accept": false, + "reason": "Flash-attention sink handling is a different code path from Gemma4 FSDP attention state rebuilding." + }, + { + "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." + }, + { + "left": "issue:43854", + "right": "issue:45216", + "accept": false, + "reason": "Different models and different regressions: unit-test loading versus save_pretrained checkpoint correctness." + }, + { + "left": "issue:45292", + "right": "issue:45335", + "accept": true, + "reason": "Same underlying `resize_token_embeddings` bug; `45335` is a model-specific manifestation on `decoder.embed_tokens`." + }, + { + "left": "issue:43645", + "right": "issue:44164", + "accept": false, + "reason": "Custom-model notebook initialization and `extra_state` save/load are separate failures." + }, + { + "left": "issue:43828", + "right": "issue:43957", + "accept": false, + "reason": "Autocast dtype mismatch and meta-device loading are different runtime paths." + }, + { + "left": "issue:43646", + "right": "issue:43708", + "accept": false, + "reason": "Custom model initialization regression and checkpoint-resume step counting are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The resize_token_embeddings cluster is well-supported by the titles: issue:45292 is the most generic report, and issue:45276 / issue:45335 are clearly model-specific variants of the same failure mode. The other soft-edge rejections are phrased cautiously enough based on the available title-level evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 9, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:36032", + "issue:39404", + "issue:43299", + "issue:43425", + "issue:43452", + "issue:43611", + "issue:43645", + "issue:43716", + "issue:43828", + "issue:43856", + "issue:43901", + "issue:43906", + "issue:43931", + "issue:43994", + "issue:44079", + "issue:44164", + "issue:44265", + "issue:44402", + "issue:44451", + "issue:44479", + "issue:44509", + "issue:44530", + "issue:44589", + "issue:44671", + "issue:44683", + "issue:44756", + "issue:44805", + "issue:44898", + "issue:44912", + "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", + "issue:46901" + ], + "soft_pairs": [ + "issue:44402|issue:44451", + "issue:46894|issue:46901", + "issue:44589|issue:44756", + "issue:43425|issue:43828", + "issue:46315|issue:46491", + "issue:45812|issue:46315", + "issue:39404|issue:45701", + "issue:44530|issue:45084", + "issue:43611|issue:43645", + "issue:39404|issue:46665", + "issue:43425|issue:46869", + "issue:44683|issue:45823", + "issue:43425|issue:43716", + "issue:43425|issue:45823", + "issue:44479|issue:44805", + "issue:44589|issue:45084", + "issue:43901|issue:44509", + "issue:43906|issue:44530", + "issue:43994|issue:44079", + "issue:44265|issue:44898", + "issue:46050|issue:46326", + "issue:39404|issue:44912", + "issue:36032|issue:43452", + "issue:44164|issue:45759", + "issue:43906|issue:46326", + "issue:44671|issue:44991", + "issue:45823|issue:46899", + "issue:44991|issue:45081", + "issue:45479|issue:46894", + "issue:43299|issue:43931", + "issue:43856|issue:44756", + "issue:44164|issue:46821", + "issue:45479|issue:46897" + ], + "prepared_review_unit_hash": "de115e4047f76b5e011da26ff98aadaaad256a450ba4ea9549822766bfa74a66", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37761, + "estimated_input_tokens": 9441, + "estimated_eval_tokens": 19138 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Loose similarity cluster of mostly distinct Transformers v5-era regression reports spanning tokenization, model loading, multimodal dtype/device issues, and a few performance bugs. None of the soft pairs look like the same concrete bug/change path, so I reject all soft edges.", + "confidence": 0.83, + "canonical_issue_reason": "Issue 43425 is the broadest, most central compatibility report in the set and is the best single anchor for this loose cluster, even though it is not a true duplicate of the others.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43425 is the most general and triage-friendly issue here: concise, high-level, and closer to an umbrella regression than the more model-specific reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44402", + "right": "issue:44451", + "accept": false, + "reason": "Both are tokenizer-loading complaints, but they point to different failure modes and different models; not enough evidence of one underlying bug." + }, + { + "left": "issue:46894", + "right": "issue:46901", + "accept": false, + "reason": "Same broad \"double-shift\" wording, but different model implementations; likely separate bugs rather than one shared code-path fix." + }, + { + "left": "issue:44589", + "right": "issue:44756", + "accept": false, + "reason": "A missing Float8 storage error and an mmap/OOM issue are unrelated failure modes." + }, + { + "left": "issue:43425", + "right": "issue:43828", + "accept": false, + "reason": "Torch version incompatibility is too broad to equate with a specific autocast dtype mismatch." + }, + { + "left": "issue:46315", + "right": "issue:46491", + "accept": false, + "reason": "One is a performance regression in convert_tokens_to_ids; the other is an encode/decode whitespace regression." + }, + { + "left": "issue:45812", + "right": "issue:46315", + "accept": false, + "reason": "Both involve tokenizers, but one is wrong token IDs and the other is conversion slowness; different bugs." + }, + { + "left": "issue:39404", + "right": "issue:45701", + "accept": false, + "reason": "Whisper pipeline return_language and a general tokenization regression are not the same issue." + }, + { + "left": "issue:44530", + "right": "issue:45084", + "accept": false, + "reason": "Different subsystems and different exceptions; no common concrete code path is evident." + }, + { + "left": "issue:43611", + "right": "issue:43645", + "accept": false, + "reason": "Both are v5 model-loading regressions, but one is about base_model_prefix handling and the other about notebook-defined custom models." + }, + { + "left": "issue:39404", + "right": "issue:46665", + "accept": false, + "reason": "One is a Whisper pipeline regression; the other is from_pretrained behavior inside Python classes." + }, + { + "left": "issue:43425", + "right": "issue:46869", + "accept": false, + "reason": "Torch compatibility and multi-GPU device-map behavior are distinct problems." + }, + { + "left": "issue:44683", + "right": "issue:45823", + "accept": false, + "reason": "Compiled flex_attention on newer torch and Gemma4 device mismatch are unrelated." + }, + { + "left": "issue:43425", + "right": "issue:43716", + "accept": false, + "reason": "A generic torch incompatibility report does not match a specific image-preprocessor/model dtype mismatch." + }, + { + "left": "issue:43425", + "right": "issue:45823", + "accept": false, + "reason": "These are different classes of failures: version incompatibility vs device placement mismatch." + }, + { + "left": "issue:44479", + "right": "issue:44805", + "accept": false, + "reason": "A video-input regression and a mask-shape IndexError are too generic to treat as one bug." + }, + { + "left": "issue:44589", + "right": "issue:45084", + "accept": false, + "reason": "Storage lookup failure and template compilation failure are clearly different." + }, + { + "left": "issue:43901", + "right": "issue:44509", + "accept": false, + "reason": "Both are docs-related, but they cover different removed/changed pipeline APIs." + }, + { + "left": "issue:43906", + "right": "issue:44530", + "accept": false, + "reason": "An isolated reproduction of another issue and a Qwen3.5 cache crash do not share a concrete code path." + }, + { + "left": "issue:43994", + "right": "issue:44079", + "accept": false, + "reason": "A wrong-output multimodal model issue and a ModelOutput key-assignment bug are unrelated." + }, + { + "left": "issue:44265", + "right": "issue:44898", + "accept": false, + "reason": "torch.export failure due to torch_compilable_check is not the same as Perceiver interpolation failure." + }, + { + "left": "issue:46050", + "right": "issue:46326", + "accept": false, + "reason": "Quantized save/load breakage and mistral_common continue_final_message behavior are different features." + }, + { + "left": "issue:39404", + "right": "issue:44912", + "accept": false, + "reason": "Whisper pipeline behavior and MXFP4 quantization fallback are unrelated." + }, + { + "left": "issue:36032", + "right": "issue:43452", + "accept": false, + "reason": "Both are loading errors, but one is a T5 tokenizer method conflict and the other is gguf_file handling." + }, + { + "left": "issue:44164", + "right": "issue:45759", + "accept": false, + "reason": "Extra state persistence and composite multimodal config unwrapping are separate save/load issues." + }, + { + "left": "issue:43906", + "right": "issue:46326", + "accept": false, + "reason": "These point to different tokenizer-related regressions with no shared concrete failure." + }, + { + "left": "issue:44671", + "right": "issue:44991", + "accept": false, + "reason": "Different model families and different symptoms; not the same bug." + }, + { + "left": "issue:45823", + "right": "issue:46899", + "accept": false, + "reason": "Both are Gemma4 multimodal issues, but one is device placement and the other is a 4-bit dtype-cast problem." + }, + { + "left": "issue:44991", + "right": "issue:45081", + "accept": false, + "reason": "Two tokenizer-loading regressions with distinct causes and model-specific patches." + }, + { + "left": "issue:45479", + "right": "issue:46894", + "accept": false, + "reason": "A generic classification-loss issue and an ASR double-shift issue are not the same underlying bug." + }, + { + "left": "issue:43299", + "right": "issue:43931", + "accept": false, + "reason": "Both mention Qwen3-VL loading, but one is a dev0 breakage and the other is a weight-shape mismatch; too different to merge." + }, + { + "left": "issue:43856", + "right": "issue:44756", + "accept": false, + "reason": "Memory-usage inefficiency during training and an mmap OOM workaround are unrelated." + }, + { + "left": "issue:44164", + "right": "issue:46821", + "accept": false, + "reason": "save/from_pretrained extra_state handling and TimesFM window_size AttributeError are different bugs." + }, + { + "left": "issue:45479", + "right": "issue:46897", + "accept": false, + "reason": "Different model families and different loss bugs; no shared code path is evident." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 10, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:34567", + "issue:36032", + "issue:39404", + "issue:40444", + "issue:40990", + "issue:42175", + "issue:42796", + "issue:42915", + "issue:43317", + "issue:43404", + "issue:43421", + "issue:43452", + "issue:43482", + "issue:43541", + "issue:43582", + "issue:43596", + "issue:43611", + "issue:43632", + "issue:43643", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43716", + "issue:43746", + "issue:43883", + "issue:44206", + "issue:44220", + "issue:44451", + "issue:44462", + "issue:44488", + "issue:44530", + "issue:44661", + "issue:44671", + "issue:44683", + "issue:44737", + "issue:44811", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45507", + "issue:45584", + "issue:45859", + "issue:46178", + "issue:46285", + "issue:46326", + "issue:46402", + "issue:46489" + ], + "soft_pairs": [ + "issue:43596|issue:44683", + "issue:44488|issue:44987", + "issue:44488|issue:45081", + "issue:43611|issue:43644", + "issue:42796|issue:43404", + "issue:44462|issue:46489", + "issue:36032|issue:42915", + "issue:46285|issue:46402", + "issue:43716|issue:45084", + "issue:39404|issue:44671", + "issue:44737|issue:45859", + "issue:43643|issue:44530", + "issue:44661|issue:44683", + "issue:42915|issue:44530", + "issue:40990|issue:43421", + "issue:44451|issue:44987", + "issue:43541|issue:45084", + "issue:34567|issue:40444", + "issue:43644|issue:43646", + "issue:43582|issue:43632", + "issue:40444|issue:46326", + "issue:42175|issue:46178", + "issue:44206|issue:44220", + "issue:36032|issue:43645", + "issue:44811|issue:45584", + "issue:43452|issue:43482", + "issue:44987|issue:44991", + "issue:45084|issue:45507", + "issue:43317|issue:43482", + "issue:43883|issue:44488", + "issue:43746|issue:46326" + ], + "prepared_review_unit_hash": "a0d57b71484228aa1a87f8ce6edce04d85703d6920ff7a2e403276394af5c2ba", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37019, + "estimated_input_tokens": 9255, + "estimated_eval_tokens": 18766 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose set of Transformers v5 regression reports rather than one true duplicate family. Only two soft pairs look like strong duplicates: GGUF loading failures and the `_torch_extract_fbank_features()` audio regression.", + "confidence": 0.61, + "canonical_issue_reason": "Issue 43452 is the best canonical anchor: it describes the broadest concrete loading regression in the set, spanning both AutoTokenizer and AutoModelForCausalLM via `gguf_file`, and it naturally covers the more specific Qwen2.5-GGUF failure.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43452 is the strongest global representative because it has the clearest, broad user-facing code-path break and the most general title among the candidates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43596", + "right": "issue:44683", + "accept": false, + "reason": "Different failures: DeepSpeed/zero3 BertModel indexing vs torch>=2.9 flex_attention compilation." + }, + { + "left": "issue:44488", + "right": "issue:44987", + "accept": false, + "reason": "Both are tokenizer load regressions, but on different models and likely different root causes." + }, + { + "left": "issue:44488", + "right": "issue:45081", + "accept": false, + "reason": "Different tokenizer bugs: repo loading failure vs Mistral regex/backend tokenizer crash." + }, + { + "left": "issue:43611", + "right": "issue:43644", + "accept": false, + "reason": "Both are v5 regressions, but one is base_model_prefix loading and the other is non-persistent buffer initialization." + }, + { + "left": "issue:42796", + "right": "issue:43404", + "accept": false, + "reason": "BPE-marker tokenizer bug vs tied lm_head weight bug." + }, + { + "left": "issue:44462", + "right": "issue:46489", + "accept": false, + "reason": "Different tokenizer-loading regressions: ignoring tokenizer.json vs a DeepSeek-Coder v5 output regression." + }, + { + "left": "issue:36032", + "right": "issue:42915", + "accept": false, + "reason": "Unrelated: T5 tokenizer method conflict vs Qwen3Moe FineGrainedFP8Config failure." + }, + { + "left": "issue:46285", + "right": "issue:46402", + "accept": false, + "reason": "Different model families and symptoms: CLIP structure mismatch vs Sam3 text-encoder weights not loading." + }, + { + "left": "issue:43716", + "right": "issue:45084", + "accept": false, + "reason": "Image dtype mismatch vs template compilation TypeError; different code paths." + }, + { + "left": "issue:39404", + "right": "issue:44671", + "accept": false, + "reason": "Whisper return_language pipeline bug vs CamemBERT masked-LM prediction regression." + }, + { + "left": "issue:44737", + "right": "issue:45859", + "accept": false, + "reason": "Both involve device placement, but the concrete bugs are different: XLNet CPU arange vs Qwen3.5 CPU-offload incompatibility." + }, + { + "left": "issue:43643", + "right": "issue:44530", + "accept": false, + "reason": "trust_remote_code config-field loss vs Qwen3.5 PagedAttentionCache linear_attention crash." + }, + { + "left": "issue:44661", + "right": "issue:44683", + "accept": false, + "reason": "TOKENIZER_MAPPING_NAMES registration bug vs flex_attention compilation regression." + }, + { + "left": "issue:42915", + "right": "issue:44530", + "accept": false, + "reason": "Different Qwen3-family failures and different subsystems (FP8 config vs paged-attention cache)." + }, + { + "left": "issue:40990", + "right": "issue:43421", + "accept": false, + "reason": "Perplexity regression on a model vs tokenizer post-processor update behavior." + }, + { + "left": "issue:44451", + "right": "issue:44987", + "accept": false, + "reason": "Both are model-loading failures, but on different repositories and with different failure modes." + }, + { + "left": "issue:43541", + "right": "issue:45084", + "accept": false, + "reason": "Different compile/tracing problems: grouped_mm dtype error vs non-template-node compilation failure." + }, + { + "left": "issue:34567", + "right": "issue:40444", + "accept": false, + "reason": "Trainer token accounting bug vs multimodal finetuning failure with IterableDataset." + }, + { + "left": "issue:43644", + "right": "issue:43646", + "accept": false, + "reason": "Related v5 initialization regressions, but one is a specific buffer-state issue and the other is a broad custom-model init break." + }, + { + "left": "issue:43582", + "right": "issue:43632", + "accept": false, + "reason": "Apple Silicon allocator warmup TypeError vs `_is_hf_initialized` flag regression." + }, + { + "left": "issue:40444", + "right": "issue:46326", + "accept": false, + "reason": "Different domains: multimodal iterable finetuning vs Whisper/mistral_common tokenizer handling." + }, + { + "left": "issue:42175", + "right": "issue:46178", + "accept": false, + "reason": "Backend-install/mapping problems, but one is missing TF backend in extras and the other is a bad BACKENDS_MAPPING entry." + }, + { + "left": "issue:44206", + "right": "issue:44220", + "accept": true, + "reason": "Same audio feature-extraction regression: both point to `_torch_extract_fbank_features()` / the same fbank pipeline bug." + }, + { + "left": "issue:44462", + "right": "issue:44987", + "accept": false, + "reason": "Different tokenizer-loading regressions for different models; no shared concrete bug is shown." + }, + { + "left": "issue:43541", + "right": "issue:45084", + "accept": false, + "reason": "Different compile/tracing problems with unrelated underlying causes." + }, + { + "left": "issue:34567", + "right": "issue:40444", + "accept": false, + "reason": "Duplicate of another unrelated pair; these are different bugs." + }, + { + "left": "issue:43644", + "right": "issue:43646", + "accept": false, + "reason": "Duplicate of another unrelated pair; the failure modes differ." + }, + { + "left": "issue:44811", + "right": "issue:45584", + "accept": false, + "reason": "Both are Whisper special-token issues, but one is decode-time skip_special_tokens handling and the other is empty-transcription generation after align_special_tokens." + }, + { + "left": "issue:43452", + "right": "issue:43482", + "accept": true, + "reason": "Same GGUF load path; 43482 looks like the model-specific manifestation of the broader `gguf_file`/from_pretrained break in 43452." + }, + { + "left": "issue:44987", + "right": "issue:44991", + "accept": false, + "reason": "Both are loading failures after v5, but they affect different models and likely different tokenizer setup bugs." + }, + { + "left": "issue:45084", + "right": "issue:45507", + "accept": false, + "reason": "Template compilation failure vs invalid method call on GraniteMoEHybrid." + }, + { + "left": "issue:43317", + "right": "issue:43482", + "accept": false, + "reason": "Different bugs: device_map/offload dequantization vs GGUF loading." + }, + { + "left": "issue:43883", + "right": "issue:44488", + "accept": false, + "reason": "Missing `all_tied_weights_keys` vs tokenizer loading failure for a different model." + }, + { + "left": "issue:43746", + "right": "issue:46326", + "accept": false, + "reason": "PEFT/local-checkpoint loading issue vs Whisper tokenizer `continue_final_message` bug." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays mostly conservative. The two accepted soft pairs are plausibly supported by the issue titles, especially 44206/44220 and the GGUF loading pair 43452/43482. The canonical anchor choice for 43452 is also reasonable from the title alone. " + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 11, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:28530", + "issue:34567", + "issue:36032", + "issue:41720", + "issue:42915", + "issue:43089", + "issue:43208", + "issue:43262", + "issue:43322", + "issue:43377", + "issue:43404", + "issue:43531", + "issue:43540", + "issue:43632", + "issue:43645", + "issue:43646", + "issue:43742", + "issue:43792", + "issue:43819", + "issue:43883", + "issue:43994", + "issue:44008", + "issue:44062", + "issue:44265", + "issue:44291", + "issue:44403", + "issue:44451", + "issue:44464", + "issue:44466", + "issue:44479", + "issue:44534", + "issue:44589", + "issue:44610", + "issue:44849", + "issue:44877", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45216", + "issue:45325", + "issue:45357", + "issue:45987", + "issue:46661", + "issue:46821", + "issue:46869", + "issue:46899", + "issue:47013" + ], + "soft_pairs": [ + "issue:43262|issue:43377", + "issue:43208|issue:47013", + "issue:43646|issue:44534", + "issue:43645|issue:46661", + "issue:44403|issue:45081", + "issue:43792|issue:46869", + "issue:42915|issue:45084", + "issue:44849|issue:45325", + "issue:43377|issue:43819", + "issue:43404|issue:44466", + "issue:28530|issue:34567", + "issue:44464|issue:45987", + "issue:43994|issue:44610", + "issue:44987|issue:45081", + "issue:42915|issue:44291", + "issue:44008|issue:46899", + "issue:36032|issue:43742", + "issue:44451|issue:45081", + "issue:43883|issue:44991", + "issue:44265|issue:44610", + "issue:44062|issue:44987", + "issue:43632|issue:46821", + "issue:43322|issue:43540", + "issue:42915|issue:44589", + "issue:43089|issue:45987", + "issue:44479|issue:45325", + "issue:36032|issue:41720", + "issue:43531|issue:44877", + "issue:45216|issue:45357" + ], + "prepared_review_unit_hash": "5f4a048d24c49d52c62753ba0fc1a36ddd424d1ff5b007ef1421c063b5488cb1", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36635, + "estimated_input_tokens": 9159, + "estimated_eval_tokens": 18574 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most candidate pairs are related only by broad subsystem or model family, not the same underlying bug. I would accept only the two Qwen3.5 save_pretrained issues as duplicates; the rest are distinct enough to keep separate.", + "confidence": 0.88, + "canonical_issue_reason": "issue:45216 is the broader, earlier regression report about Qwen3.5 save_pretrained producing an incorrect checkpoint; issue:45357 reads like a narrower follow-up symptom focused on the visual encoder keys.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45216 is the best representative issue because it captures the general Qwen3.5 save_pretrained regression without narrowing to one specific malformed key set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43262", + "right": "issue:43377", + "accept": false, + "reason": "Both are audio-related, but one is a processor chat-template sampling-rate default and the other is MIMI batching/padding-mask behavior; different bug and code path." + }, + { + "left": "issue:43208", + "right": "issue:47013", + "accept": false, + "reason": "Shared xLSTM label only; one is a bundle of training blockers for small models, the other is a specific train-mode unpacking crash when return_last_states=False." + }, + { + "left": "issue:43646", + "right": "issue:44534", + "accept": false, + "reason": "One is custom model initialization breaking in v5, the other is non-persistent buffers being filled with junk; different failure modes." + }, + { + "left": "issue:43645", + "right": "issue:46661", + "accept": false, + "reason": "Both involve model initialization/loading, but one is Jupyter/custom model init and the other is remote-code AutoModel config_class handling." + }, + { + "left": "issue:44403", + "right": "issue:45081", + "accept": false, + "reason": "Unnecessary loading noise and a Mistral regex patch crash are not the same bug." + }, + { + "left": "issue:43792", + "right": "issue:46869", + "accept": false, + "reason": "Different models and failures: Whisper runtime failure vs YOLO multi-GPU device-map issue." + }, + { + "left": "issue:42915", + "right": "issue:45084", + "accept": false, + "reason": "Qwen3-MoE auto device mapping CUDA assert is unrelated to a templating compilation TypeError." + }, + { + "left": "issue:44849", + "right": "issue:45325", + "accept": false, + "reason": "Both are Qwen-family regressions, but one is output_hidden_states behavior and the other is still-image rope/position-id scaling; not the same bug." + }, + { + "left": "issue:43377", + "right": "issue:43819", + "accept": false, + "reason": "Both are audio model issues, but one is missing padding-mask support in MIMI and the other is DAC.from_latents vs forward mismatch without STE." + }, + { + "left": "issue:43404", + "right": "issue:44466", + "accept": false, + "reason": "Both mention lm_head/tied weights, but one is a missing tie in a specific model class and the other is save_pretrained serialization varying by device." + }, + { + "left": "issue:28530", + "right": "issue:34567", + "accept": false, + "reason": "Training early-stopping metric selection and TrainerState token counter updating are unrelated." + }, + { + "left": "issue:44464", + "right": "issue:45987", + "accept": false, + "reason": "Both touch generation/cache behavior, but one is chunked-generation inconsistency under compiled forward and the other is a return type contract issue for StaticCache.get_seq_length()." + }, + { + "left": "issue:43994", + "right": "issue:44610", + "accept": false, + "reason": "Different subsystems: SigLIP2 inference/pipeline correctness vs OmDet-Turbo processor image size mismatch." + }, + { + "left": "issue:44987", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer loading for a specific model and a Mistral tokenizer regex crash are separate issues." + }, + { + "left": "issue:42915", + "right": "issue:44291", + "accept": false, + "reason": "Both involve model loading/initialization under v5, but the concrete bugs differ: CUDA assert in Qwen3-MoE vs unexpected _is_hf_initialized argument with init_empty_weights." + }, + { + "left": "issue:44008", + "right": "issue:46899", + "accept": false, + "reason": "Both are Gemma multimodal issues, but one is a variable-name collision in forward and the other is an unsafe dtype cast under 4-bit quantization." + }, + { + "left": "issue:36032", + "right": "issue:43742", + "accept": false, + "reason": "Different tokenizer/model load failures with distinct root causes." + }, + { + "left": "issue:44451", + "right": "issue:45081", + "accept": false, + "reason": "ScandiBERT loading failure and Mistral regex patch crash are unrelated." + }, + { + "left": "issue:43883", + "right": "issue:44991", + "accept": false, + "reason": "An attribute error on tied-weights metadata and a tokenizer load failure for est-roberta are not the same bug." + }, + { + "left": "issue:44265", + "right": "issue:44610", + "accept": false, + "reason": "torch.export/torch_compilable_check failure is unrelated to OmDet-Turbo processor sizing." + }, + { + "left": "issue:44062", + "right": "issue:44987", + "accept": false, + "reason": "Both are tokenizer-related, but one is AddedToken argument duplication and the other is a specific model load failure." + }, + { + "left": "issue:43632", + "right": "issue:46821", + "accept": false, + "reason": "_is_hf_initialized flag breakage and TimesFM window_size AttributeError are different regressions." + }, + { + "left": "issue:43322", + "right": "issue:43540", + "accept": false, + "reason": "Both are multimodal loading/processing failures, but Llava Next segfault and Qwen3OmniMoe video-processing ValueError are not the same code path." + }, + { + "left": "issue:42915", + "right": "issue:44589", + "accept": false, + "reason": "Qwen3-MoE CUDA assert and Float8 storage lookup failure are unrelated." + }, + { + "left": "issue:43089", + "right": "issue:45987", + "accept": false, + "reason": "Generation performance overhead is not the same as StaticCache returning the wrong type." + }, + { + "left": "issue:44479", + "right": "issue:45325", + "accept": false, + "reason": "These are both Qwen vision-language regressions, but one is a broad video-input regression and the other is still-image temporal position-id scaling; too different to merge confidently." + }, + { + "left": "issue:36032", + "right": "issue:41720", + "accept": false, + "reason": "Tokenizer construction error and Qwen3 device-map CUDA assert are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:44877", + "accept": false, + "reason": "Sliding-window behavior in Qwen3-MoE and strict config loading for granite_speech are different issues." + }, + { + "left": "issue:45216", + "right": "issue:45357", + "accept": true, + "reason": "Same underlying Qwen3.5 save_pretrained regression; 45357 is a narrower follow-up describing incorrect visual encoder keys, while 45216 reports the broader broken checkpoint output." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The analysis is grounded in the packet\u2019s titles/body summaries, and the soft-edge calls stay conservative overall. The only accepted duplicate pair, issue:45216 and issue:45357, looks mergeable as the same Qwen3.5 save_pretrained regression. The remaining pairs are plausibly distinct bugs rather than overconfident duplicates." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 12, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:33666", + "issue:35707", + "issue:40444", + "issue:40990", + "issue:41720", + "issue:42175", + "issue:42796", + "issue:43296", + "issue:43317", + "issue:43377", + "issue:43381", + "issue:43421", + "issue:43425", + "issue:43482", + "issue:43525", + "issue:43531", + "issue:43611", + "issue:43632", + "issue:43646", + "issue:43716", + "issue:43746", + "issue:43792", + "issue:43856", + "issue:44079", + "issue:44220", + "issue:44291", + "issue:44403", + "issue:44462", + "issue:44534", + "issue:44568", + "issue:44589", + "issue:44683", + "issue:44756", + "issue:44811", + "issue:44991", + "issue:45081", + "issue:45325", + "issue:45491", + "issue:45538", + "issue:45542", + "issue:45663", + "issue:45812", + "issue:45923", + "issue:46315", + "issue:46491", + "issue:46710", + "issue:46821", + "issue:46899" + ], + "soft_pairs": [ + "issue:44991|issue:46315", + "issue:43425|issue:44403", + "issue:40444|issue:45325", + "issue:43611|issue:44534", + "issue:43716|issue:46899", + "issue:43296|issue:43531", + "issue:44220|issue:44683", + "issue:45812|issue:46491", + "issue:42796|issue:43746", + "issue:44403|issue:44991", + "issue:43381|issue:45923", + "issue:43632|issue:44756", + "issue:33666|issue:40444", + "issue:40990|issue:44811", + "issue:43792|issue:44589", + "issue:44291|issue:44589", + "issue:42796|issue:46710", + "issue:35707|issue:40444", + "issue:44079|issue:45663", + "issue:43482|issue:43525", + "issue:43632|issue:44534", + "issue:43531|issue:44291", + "issue:43421|issue:44568", + "issue:42175|issue:45542", + "issue:43482|issue:45538", + "issue:41720|issue:45081", + "issue:44079|issue:44534", + "issue:43317|issue:43856", + "issue:44462|issue:45812", + "issue:43296|issue:46821", + "issue:43611|issue:43646", + "issue:43377|issue:45491" + ], + "prepared_review_unit_hash": "733f342dc9769c5b77eb1c270f5fd5945eef771fffb742cb8f1308d9cbb3a451", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37301, + "estimated_input_tokens": 9326, + "estimated_eval_tokens": 18908 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is a collection of largely unrelated issues; the soft pairs are mostly superficial matches by subsystem or version. No PRs are present. If a single issue must be treated as the best representative, the broad Granite tokenizer regression is the strongest candidate.", + "confidence": 0.82, + "canonical_issue_reason": "Issue 45812 is the broadest, most centrally relevant tokenizer regression in the set (all Granite models, silent v4\u2192v5 behavior change) and has the strongest reference signal.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45812 is the best overall representative because it affects an entire model family and describes a concrete, user-visible regression with cross-cutting impact.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44991", + "right": "issue:46315", + "accept": false, + "reason": "Both involve tokenizer regressions, but one is model-specific loading and the other is a slow conversion path; different bugs and code paths." + }, + { + "left": "issue:43425", + "right": "issue:44403", + "accept": false, + "reason": "Torch compatibility and loading noise are unrelated problems." + }, + { + "left": "issue:40444", + "right": "issue:45325", + "accept": false, + "reason": "Both are Qwen2.5-VL-related, but one is a multi-image iterable dataset failure and the other is still-image temporal position-id scaling; not the same issue." + }, + { + "left": "issue:43611", + "right": "issue:44534", + "accept": false, + "reason": "One is a loading failure around base_model_prefix, the other is non-persistent buffers being filled with junk; distinct regressions." + }, + { + "left": "issue:43716", + "right": "issue:46899", + "accept": false, + "reason": "Both mention multimodal dtype handling, but they affect different models and different components (image preprocessor vs audio/vision embedders)." + }, + { + "left": "issue:43296", + "right": "issue:43531", + "accept": false, + "reason": "PaddleOCR-VL load failure and Qwen3-MoE sliding_window behavior are unrelated." + }, + { + "left": "issue:44220", + "right": "issue:44683", + "accept": false, + "reason": "Fbank feature extraction and compiled flex_attention on torch>=2.9 are unrelated issues." + }, + { + "left": "issue:45812", + "right": "issue:46491", + "accept": false, + "reason": "Both are tokenizer regressions, but one is Granite token IDs and the other is CodeLlama leading-space round-trip behavior; different models and failure modes." + }, + { + "left": "issue:42796", + "right": "issue:43746", + "accept": false, + "reason": "Tokenizer BPE-marker handling and PEFT checkpoint loading are unrelated." + }, + { + "left": "issue:44403", + "right": "issue:44991", + "accept": false, + "reason": "Generic loading noise is not the same as AutoTokenizer ignoring tokenizer.json." + }, + { + "left": "issue:43381", + "right": "issue:45923", + "accept": false, + "reason": "Gradient checkpointing in eval mode and a missing supports_gradient_checkpointing flag are related only by theme, not the same bug." + }, + { + "left": "issue:43632", + "right": "issue:44756", + "accept": false, + "reason": "_is_hf_initialized regression and Strix Halo mmap OOM are unrelated." + }, + { + "left": "issue:33666", + "right": "issue:40444", + "accept": false, + "reason": "Both are Qwen-VL training issues, but they concern different setups and failures." + }, + { + "left": "issue:40990", + "right": "issue:44811", + "accept": false, + "reason": "Perplexity regression and Whisper batch_decode skip_special_tokens behavior are unrelated." + }, + { + "left": "issue:43792", + "right": "issue:44589", + "accept": false, + "reason": "Whisper runtime failure and missing Float8 storage support are unrelated." + }, + { + "left": "issue:44291", + "right": "issue:44589", + "accept": false, + "reason": "Both are transformers v5 regressions, but they affect different initialization/storage code paths." + }, + { + "left": "issue:42796", + "right": "issue:46710", + "accept": false, + "reason": "Tokenizer BPE-marker handling and an accuracy regression after upgrading Transformers are not the same bug." + }, + { + "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 dataset loading." + }, + { + "left": "issue:44079", + "right": "issue:45663", + "accept": false, + "reason": "ModelOutput key assignment and Gemma4 FSDP2 shared_kv_states rebuilding are unrelated." + }, + { + "left": "issue:43482", + "right": "issue:43525", + "accept": false, + "reason": "Qwen2.5-GGUF loading and missing pad_token_id on Llama4Config are different model-specific failures." + }, + { + "left": "issue:43632", + "right": "issue:44534", + "accept": false, + "reason": "Different v5 regressions: one is initialization flag handling, the other is buffer state corruption." + }, + { + "left": "issue:43531", + "right": "issue:44291", + "accept": false, + "reason": "Sliding-window behavior in Qwen3-MoE and init_empty_weights argument handling are unrelated." + }, + { + "left": "issue:43421", + "right": "issue:44568", + "accept": false, + "reason": "Runtime post-processor updates for special tokens and add_special_tokens BOS/EOS behavior are different tokenizer paths." + }, + { + "left": "issue:42175", + "right": "issue:45542", + "accept": false, + "reason": "Both involve TensorFlow packaging/backend, but one is missing TF in the torch extra while the other is an undefined backend error from tensorboard-only installs." + }, + { + "left": "issue:43482", + "right": "issue:45538", + "accept": false, + "reason": "Qwen2.5-GGUF loading and CLIPTokenizer model_max_length are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:45081", + "accept": false, + "reason": "Qwen3 device-map cudaErrorAssert and Mistral regex patch crashes are unrelated." + }, + { + "left": "issue:44079", + "right": "issue:44534", + "accept": false, + "reason": "ModelOutput key assignment and non-persistent buffer initialization are unrelated." + }, + { + "left": "issue:43317", + "right": "issue:43856", + "accept": false, + "reason": "Dequantized model offload loading and Qwen3 MoE training memory usage are different memory-related issues with different causes." + }, + { + "left": "issue:44462", + "right": "issue:45812", + "accept": false, + "reason": "AutoTokenizer ignoring tokenizer.json could be a contributor, but the Granite token-ID regression is a distinct, broader failure and not proven to be the same bug." + }, + { + "left": "issue:43296", + "right": "issue:46821", + "accept": false, + "reason": "PaddleOCR-VL loading and TimesFM window_size AttributeError are unrelated." + }, + { + "left": "issue:43611", + "right": "issue:43646", + "accept": false, + "reason": "Both are transformers v5 regressions, but base_model_prefix loading and custom model initialization are different code-path 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": 26, + "review_unit_index": 13, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:34634", + "issue:35707", + "issue:39692", + "issue:40444", + "issue:40990", + "issue:41553", + "issue:41720", + "issue:42175", + "issue:42796", + "issue:42915", + "issue:43262", + "issue:43296", + "issue:43454", + "issue:43482", + "issue:43493", + "issue:43550", + "issue:43582", + "issue:43606", + "issue:43644", + "issue:43701", + "issue:43746", + "issue:43818", + "issue:43819", + "issue:43828", + "issue:43873", + "issue:43994", + "issue:44079", + "issue:44186", + "issue:44265", + "issue:44291", + "issue:44466", + "issue:44479", + "issue:44509", + "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", + "issue:46909" + ], + "soft_pairs": [ + "issue:44509|issue:46909", + "issue:42175|issue:43296", + "issue:43644|issue:44079", + "issue:44829|issue:45479", + "issue:41553|issue:43582", + "issue:40444|issue:45381", + "issue:44265|issue:45072", + "issue:43818|issue:46402", + "issue:43493|issue:43994", + "issue:43818|issue:45357", + "issue:45325|issue:45834", + "issue:45325|issue:45381", + "issue:43454|issue:44466", + "issue:42915|issue:43482", + "issue:44291|issue:44877", + "issue:43873|issue:45701", + "issue:43701|issue:45325", + "issue:42796|issue:45381", + "issue:40990|issue:45584", + "issue:43606|issue:44265", + "issue:43262|issue:43746", + "issue:44479|issue:45823", + "issue:35707|issue:41720", + "issue:43828|issue:46821", + "issue:34634|issue:35707", + "issue:39692|issue:40444", + "issue:44186|issue:44265", + "issue:43454|issue:43550", + "issue:41720|issue:45823", + "issue:43606|issue:43819" + ], + "prepared_review_unit_hash": "70a54d523e66d4f95aabe96a9a5d5fa51be95dc4eb8315428d9c5501345aaad4", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36271, + "estimated_input_tokens": 9068, + "estimated_eval_tokens": 18392 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a heterogeneous similarity cluster with two real duplicate-like subclusters: SigLIP2 implementation/output mismatch, and a Qwen2.5-VL rope-index/video-position regression. The rest are different bugs, docs problems, or unrelated model/runtime failures.", + "confidence": 0.69, + "canonical_issue_reason": "issue:45325 is the clearest, most actionable root-cause report in the cluster: it names the specific get_rope_index / still-image temporal position_ids bug, and issue:45381 looks like the same regression from a different reproduction.", + "canonical_pr_reason": null, + "best_issue_reason": "Best representative of the cluster\u2019s strongest duplicate pair: it is specific, code-path focused, and clearly maps to the same regression as 45381.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44509", + "right": "issue:46909", + "accept": false, + "reason": "Both are documentation complaints, but they refer to different docs problems and different examples; not the same underlying bug." + }, + { + "left": "issue:42175", + "right": "issue:43296", + "accept": false, + "reason": "One is missing TensorFlow in the torch extra backend; the other is a vLLM/PaddleOCR-VL load failure. Different issues." + }, + { + "left": "issue:43644", + "right": "issue:44079", + "accept": false, + "reason": "Non-persistent buffer junk and ModelOutput key assignment are unrelated code paths." + }, + { + "left": "issue:44829", + "right": "issue:45479", + "accept": false, + "reason": "Both affect sequence classification, but one is a flash-attention-3 training regression and the other is a zero-loss classification-label bug." + }, + { + "left": "issue:41553", + "right": "issue:43582", + "accept": false, + "reason": "Tokenizer loading error message vs Apple Silicon allocator TypeError are unrelated." + }, + { + "left": "issue:40444", + "right": "issue:45381", + "accept": false, + "reason": "Both involve multimodal/video models, but these are different models and different failures." + }, + { + "left": "issue:44265", + "right": "issue:45072", + "accept": false, + "reason": "torch.export failures from torch_compilable_check are not the same as bfloat16 dtype mismatch bugs." + }, + { + "left": "issue:43818", + "right": "issue:46402", + "accept": false, + "reason": "Video-LLaVA architecture/weights issue vs Sam3Model text encoder loading issue; different bugs." + }, + { + "left": "issue:43493", + "right": "issue:43994", + "accept": true, + "reason": "Both describe the same SigLIP2 HF-vs-reference implementation mismatch leading to incorrect/nonsensical outputs." + }, + { + "left": "issue:43818", + "right": "issue:45357", + "accept": false, + "reason": "Video-LLaVA temporal-attention/weight-sharing bug is unrelated to Qwen3.5 save_pretrained key serialization." + }, + { + "left": "issue:45325", + "right": "issue:45834", + "accept": false, + "reason": "Qwen2.5-VL rope-index position bug and Kosmos2.5 OCR index error are different failures." + }, + { + "left": "issue:45325", + "right": "issue:45381", + "accept": true, + "reason": "Same Qwen2.5-VL video/still-image temporal-position_ids regression, described with the same underlying code path." + }, + { + "left": "issue:43454", + "right": "issue:44466", + "accept": false, + "reason": "One is lm_head not being tied for generation correctness; the other is device-dependent serialization of tied weights. Related theme, different bug." + }, + { + "left": "issue:42915", + "right": "issue:43482", + "accept": false, + "reason": "Qwen3Moe FP8 config failure and Qwen2.5-GGUF loading failure are separate model/config compatibility issues." + }, + { + "left": "issue:44291", + "right": "issue:44877", + "accept": false, + "reason": "init_empty_weights argument handling and strict granite_speech config loading are different loading bugs." + }, + { + "left": "issue:43873", + "right": "issue:45701", + "accept": false, + "reason": "Quantized offloading behavior and tokenization changes across versions are unrelated." + }, + { + "left": "issue:43701", + "right": "issue:45325", + "accept": false, + "reason": "Checkpoint resume key mismatch is not the same as the Qwen2.5-VL rope-index regression." + }, + { + "left": "issue:42796", + "right": "issue:45381", + "accept": false, + "reason": "Tokenizer BPE-marker handling and Qwen2.5-VL video position_ids are unrelated." + }, + { + "left": "issue:40990", + "right": "issue:45584", + "accept": false, + "reason": "High perplexity on gpt-oss and Whisper empty-transcription failure are unrelated model behaviors." + }, + { + "left": "issue:43606", + "right": "issue:44265", + "accept": false, + "reason": "CPU offload device mismatch and torch.export / torch_compilable_check failure are different execution paths." + }, + { + "left": "issue:43262", + "right": "issue:43746", + "accept": false, + "reason": "Audio chat-template sample-rate defaulting and GraniteSpeech PEFT checkpoint loading are unrelated." + }, + { + "left": "issue:44479", + "right": "issue:45823", + "accept": false, + "reason": "Both are device/vision related, but they are different model families and different regressions." + }, + { + "left": "issue:35707", + "right": "issue:41720", + "accept": false, + "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen3 auto device-mapping CUDA asserts." + }, + { + "left": "issue:43828", + "right": "issue:46821", + "accept": false, + "reason": "Autocast dtype mismatch in Phi-tiny-MoE is unrelated to a TimesFM window_size AttributeError." + }, + { + "left": "issue:34634", + "right": "issue:35707", + "accept": false, + "reason": "Bark voice_preset and progressive-generation issues are different bugs." + }, + { + "left": "issue:39692", + "right": "issue:40444", + "accept": false, + "reason": "SigLIP2 docs example errors and Qwen2.5-VL iterable dataset failures are different issues." + }, + { + "left": "issue:44186", + "right": "issue:44265", + "accept": false, + "reason": "Tokenizer batching crashes and torch.export failures are unrelated." + }, + { + "left": "issue:43454", + "right": "issue:43550", + "accept": false, + "reason": "lm_head weight tying and Bamba torch.compile SDPA failure are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:45823", + "accept": false, + "reason": "Both mention device mapping, but the underlying models and failure modes differ." + }, + { + "left": "issue:43606", + "right": "issue:43819", + "accept": false, + "reason": "CPU offload device mismatch and DAC STE mismatch are unrelated code-path problems." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The analysis is grounded in the packet and stays mostly conservative. The two accepted pairs are plausibly the same regressions from the issue titles/body context, while the rejected pairs are generally justified by clearly different models, code paths, or failure modes. The summary does not overclaim beyond the evidence provided." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 14, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "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:43572", + "issue:43720", + "issue:43749", + "issue:43873", + "issue:43881", + "issue:44186", + "issue:44387", + "issue:44466", + "issue:44488", + "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:43257|issue:45923", + "issue:42796|issue:46209", + "issue:44568|issue:45920", + "issue:43454|issue:45072", + "issue:46305|issue:46315", + "issue:45823|issue:46821", + "issue:40990|issue:42947", + "issue:42617|issue:42915", + "issue:43881|issue:44488", + "issue:43377|issue:44898", + "issue:43454|issue:43720", + "issue:43749|issue:44534", + "issue:43317|issue:43873", + "issue:43572|issue:45759", + "issue:43322|issue:46050", + "issue:41720|issue:43322", + "issue:36331|issue:42617", + "issue:44387|issue:45704", + "issue:41720|issue:44977", + "issue:36331|issue:42915", + "issue:40990|issue:45701", + "issue:44877|issue:44991", + "issue:44466|issue:45357", + "issue:43450|issue:45701", + "issue:43065|issue:46402", + "issue:41720|issue:42617", + "issue:26413|issue:28530", + "issue:44186|issue:45072", + "issue:43873|issue:44756" + ], + "prepared_review_unit_hash": "fa9d62d95a5e0aed5280c59ec69609b616de56fffca83db025ee83f612240dc2", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36065, + "estimated_input_tokens": 9017, + "estimated_eval_tokens": 18290 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: most items are distinct tokenizer, loading, quantization, or model-specific bugs. I rejected all soft edges as non-duplicates and used the broad offloading/quantization regression issue as the best representative.", + "confidence": 0.71, + "canonical_issue_reason": "issue:43873 is the broadest representative of the repeated loading/offloading/quantization regression theme in this set, even though the cluster is not a true duplicate group.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43873 best matches the cluster\u2019s widest recurring theme (offloading/quantization behavior) and is the least specific representative among the issues.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43377", + "right": "issue:44610", + "accept": false, + "reason": "Different models and failure modes: padding-mask batched-output bug vs image input shape mismatch." + }, + { + "left": "issue:43257", + "right": "issue:45923", + "accept": false, + "reason": "Different subsystems: MoE weight conversion/loading vs missing gradient-checkpointing flag." + }, + { + "left": "issue:42796", + "right": "issue:46209", + "accept": false, + "reason": "Tokenizer BPE-marker handling is unrelated to FP8 inference failure on B200." + }, + { + "left": "issue:44568", + "right": "issue:45920", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different code paths and model families." + }, + { + "left": "issue:43454", + "right": "issue:45072", + "accept": false, + "reason": "lm_head tying/generation corruption is unrelated to bf16 dtype mismatches." + }, + { + "left": "issue:46305", + "right": "issue:46315", + "accept": false, + "reason": "Different areas: sliding-window masking off-by-one vs slow token-id conversion." + }, + { + "left": "issue:45823", + "right": "issue:46821", + "accept": false, + "reason": "Device-map forward mismatch is unrelated to TimesFM window_size AttributeError." + }, + { + "left": "issue:40990", + "right": "issue:42947", + "accept": false, + "reason": "High perplexity on a model is not the same bug as LoRA gradient-checkpointing behavior." + }, + { + "left": "issue:42617", + "right": "issue:42915", + "accept": false, + "reason": "No shared concrete code path: 3D parallel script failure vs FineGrainedFP8Config failure." + }, + { + "left": "issue:43881", + "right": "issue:44488", + "accept": false, + "reason": "Both are load failures, but for different models and likely different root causes." + }, + { + "left": "issue:43377", + "right": "issue:44898", + "accept": false, + "reason": "Both involve input/shape behavior, but one is batched-vs-single encoder output and the other is image resolution handling." + }, + { + "left": "issue:43454", + "right": "issue:43720", + "accept": false, + "reason": "Weight tying bug is unrelated to packed-weight unpacking during accelerate loading." + }, + { + "left": "issue:43749", + "right": "issue:44534", + "accept": false, + "reason": "Different bugs: FSDP CPU-efficient loading vs non-persistent buffer serialization junk." + }, + { + "left": "issue:43317", + "right": "issue:43873", + "accept": false, + "reason": "Related broad area, but one is dequantized gpu+cpu offload loading and the other is general quantized offloading behavior." + }, + { + "left": "issue:43572", + "right": "issue:45759", + "accept": false, + "reason": "Config missing pad_token_idx is not the same as failing to unwrap text_config for composite models." + }, + { + "left": "issue:43322", + "right": "issue:46050", + "accept": false, + "reason": "Segfault loading Llava Next is unrelated to sinq save/load corruption." + }, + { + "left": "issue:41720", + "right": "issue:43322", + "accept": false, + "reason": "Both are loading failures, but the model families and concrete failure modes differ." + }, + { + "left": "issue:36331", + "right": "issue:42617", + "accept": false, + "reason": "Trainer signature change is unrelated to a parallel script runtime failure." + }, + { + "left": "issue:44387", + "right": "issue:45704", + "accept": false, + "reason": "OOM from int4 quantization is not the same as the T5 apex RMSNorm memory leak." + }, + { + "left": "issue:41720", + "right": "issue:44977", + "accept": false, + "reason": "Same model family only; device-map CUDA assert and flash-attention generation bug are different issues." + }, + { + "left": "issue:36331", + "right": "issue:42915", + "accept": false, + "reason": "Unexpected trainer keyword argument is unrelated to Qwen3MoE FP8 loading." + }, + { + "left": "issue:40990", + "right": "issue:45701", + "accept": false, + "reason": "Perplexity regression and tokenization-version change are not the same concrete bug." + }, + { + "left": "issue:44877", + "right": "issue:44991", + "accept": false, + "reason": "Strict config loading and tokenizer loading failures are separate problems." + }, + { + "left": "issue:44466", + "right": "issue:45357", + "accept": false, + "reason": "Both are save_pretrained serialization regressions, but they affect different tensors/keys and model paths." + }, + { + "left": "issue:43450", + "right": "issue:45701", + "accept": false, + "reason": "Video processor batching shape bug is unrelated to tokenization behavior changes." + }, + { + "left": "issue:43065", + "right": "issue:46402", + "accept": false, + "reason": "Same model family, but pixel-decoder dummy conv and text-encoder weight loading are different code paths." + }, + { + "left": "issue:41720", + "right": "issue:42617", + "accept": false, + "reason": "Different models and failures: device-map CUDA assert vs a separate script execution issue." + }, + { + "left": "issue:26413", + "right": "issue:28530", + "accept": false, + "reason": "Checkpoint resume iterator exhaustion and early-stopping metric lookup are unrelated training issues." + }, + { + "left": "issue:44186", + "right": "issue:45072", + "accept": false, + "reason": "Tokenizer crash on NER/padding is unrelated to bf16 dtype mismatches." + }, + { + "left": "issue:43873", + "right": "issue:44756", + "accept": false, + "reason": "Both mention memory/offloading, but mmap-on-Strix-Halo OOM is not the same as quantized offloading behavior." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 15, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:28530", + "issue:33666", + "issue:36010", + "issue:36331", + "issue:41720", + "issue:42617", + "issue:42915", + "issue:43295", + "issue:43317", + "issue:43404", + "issue:43526", + "issue:43531", + "issue:43550", + "issue:43572", + "issue:43577", + "issue:43606", + "issue:43646", + "issue:43708", + "issue:43720", + "issue:43881", + "issue:43883", + "issue:44008", + "issue:44186", + "issue:44361", + "issue:44423", + "issue:44451", + "issue:44464", + "issue:44610", + "issue:44671", + "issue:44734", + "issue:44743", + "issue:44898", + "issue:44991", + "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", + "issue:47020" + ], + "soft_pairs": [ + "issue:46612|issue:46858", + "issue:43577|issue:43708", + "issue:43526|issue:43883", + "issue:43550|issue:43720", + "issue:43881|issue:44991", + "issue:43295|issue:45701", + "issue:45685|issue:46821", + "issue:43646|issue:46665", + "issue:44186|issue:44610", + "issue:44451|issue:47020", + "issue:43550|issue:43606", + "issue:36010|issue:42617", + "issue:43404|issue:45357", + "issue:44610|issue:44898", + "issue:42915|issue:43317", + "issue:44423|issue:45406", + "issue:44734|issue:46858", + "issue:46821|issue:46869", + "issue:44008|issue:45823", + "issue:41720|issue:45161", + "issue:28530|issue:33666", + "issue:44464|issue:46858", + "issue:43531|issue:43572", + "issue:45632|issue:45684", + "issue:44671|issue:44743", + "issue:42915|issue:46291", + "issue:36331|issue:41720", + "issue:44186|issue:44361" + ], + "prepared_review_unit_hash": "e7e3b11a7b38b8b39b7c31479de1003b3cc8c3e6ca071a59db5076b3de8b798e", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36263, + "estimated_input_tokens": 9066, + "estimated_eval_tokens": 18388 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is a loose collection of unrelated Transformers issues. None of the soft pairs look like the same underlying bug or change, so all soft edges are rejected.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 46858 is the most self-contained and technically precise report, but there is no true duplicate cluster here; the issues span unrelated subsystems and models.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 46858 has the clearest root-cause framing and reproduction details, making it the best single representative of this mixed set, though not a duplicate hub.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46612", + "right": "issue:46858", + "accept": false, + "reason": "Both involve generation/cache behavior, but one is beam-search reorder skipping and the other is a static-cache seq-length arithmetic bug; different code paths." + }, + { + "left": "issue:43577", + "right": "issue:43708", + "accept": false, + "reason": "One is a dtype-loading bug for Blip2, the other is Trainer checkpoint/resume max_steps logic; unrelated problems." + }, + { + "left": "issue:43526", + "right": "issue:43883", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and Molmo tied-weights metadata are separate components with different failures." + }, + { + "left": "issue:43550", + "right": "issue:43720", + "accept": false, + "reason": "Torch.compile + SDPA in Bamba and BitNet packed-weight loading with accelerate are distinct bugs." + }, + { + "left": "issue:43881", + "right": "issue:44991", + "accept": false, + "reason": "Model loading failures, but for different causes: glm-4v-9b model init versus tokenizer loading for est-roberta." + }, + { + "left": "issue:43295", + "right": "issue:45701", + "accept": false, + "reason": "Both mention tokenization regressions, but one is a processor.tokenizer API break and the other is a broad version-related tokenization change." + }, + { + "left": "issue:45685", + "right": "issue:46821", + "accept": false, + "reason": "An MPS histogram kernel limitation is unrelated to a TimesFM window_size attribute error." + }, + { + "left": "issue:43646", + "right": "issue:46665", + "accept": false, + "reason": "CPU offload/device-map loading failure is unrelated to from_pretrained behavior inside Python classes." + }, + { + "left": "issue:44186", + "right": "issue:44610", + "accept": false, + "reason": "LayoutLMv2 NER/padding crashes and OmDet-Turbo input-size mismatch are different tokenizer/processor bugs." + }, + { + "left": "issue:44451", + "right": "issue:47020", + "accept": false, + "reason": "Both are tokenizer load issues, but the affected models and failure modes differ; not enough evidence of the same underlying defect." + }, + { + "left": "issue:43550", + "right": "issue:43606", + "accept": false, + "reason": "Both are runtime failures, but one is SDPA/compile and the other is CPU offload device mismatch; unrelated." + }, + { + "left": "issue:36010", + "right": "issue:42617", + "accept": false, + "reason": "GenerationMixin import failure and 3d_parallel.py runtime failure are unrelated." + }, + { + "left": "issue:43404", + "right": "issue:45357", + "accept": false, + "reason": "Both touch multimodal model internals, but one is weight tying in Mistral3 and the other is incorrect visual encoder save keys in Qwen3.5." + }, + { + "left": "issue:44610", + "right": "issue:44898", + "accept": false, + "reason": "Both are vision/preprocessing issues, but one is a processor output size bug and the other is a model positional-encoding/interpolation bug." + }, + { + "left": "issue:42915", + "right": "issue:43317", + "accept": false, + "reason": "Different loading paths: FineGrainedFP8Config failure versus dequantized model loading with device_map auto/offload." + }, + { + "left": "issue:44423", + "right": "issue:45406", + "accept": false, + "reason": "Both affect serve with multimodal models, but the errors point to different root causes: string .to() misuse versus missing _tokenizer on Gemma4Processor." + }, + { + "left": "issue:44734", + "right": "issue:46858", + "accept": false, + "reason": "Both involve cache/generation internals, but one is KV-cache continuation indexing in serve and the other is static-cache seq-length arithmetic on non-CUDA backends." + }, + { + "left": "issue:46821", + "right": "issue:46869", + "accept": false, + "reason": "TimesFM window_size attribute and YOLO multi-GPU device-map failure are unrelated." + }, + { + "left": "issue:44008", + "right": "issue:45823", + "accept": false, + "reason": "Different Gemma variants and different bugs: variable name collision in forward versus device mismatch under auto device mapping." + }, + { + "left": "issue:41720", + "right": "issue:45161", + "accept": false, + "reason": "Both involve large-model parallelism, but cudaErrorAssert on A800 and GPT-OSS MoE tensor-parallel failure are distinct issues." + }, + { + "left": "issue:28530", + "right": "issue:33666", + "accept": false, + "reason": "Early-stopping metric handling and Qwen2-VL multi-GPU training are unrelated." + }, + { + "left": "issue:44464", + "right": "issue:46858", + "accept": false, + "reason": "Both mention torch.compile/generation, but one is inconsistent chunked generation output and the other is a concrete static-cache seq-length tensor bug." + }, + { + "left": "issue:43531", + "right": "issue:43572", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and StableLmConfig missing pad_token_idx are separate config/model issues." + }, + { + "left": "issue:45632", + "right": "issue:45684", + "accept": false, + "reason": "Both concern save/cache behavior for custom models, but one is cache-path collision and the other is read-only permission propagation." + }, + { + "left": "issue:44671", + "right": "issue:44743", + "accept": false, + "reason": "CamemBERT masked-LM prediction regression and Qwen3.5 recurrent-state reset under cache are unrelated." + }, + { + "left": "issue:42915", + "right": "issue:46291", + "accept": false, + "reason": "FineGrainedFP8Config model failure and pipelines import dependency/version error are unrelated." + }, + { + "left": "issue:36331", + "right": "issue:41720", + "accept": false, + "reason": "Trainer compute_loss signature breakage is unrelated to Qwen3 auto device-mapping cuda assert." + }, + { + "left": "issue:44186", + "right": "issue:44361", + "accept": false, + "reason": "Both are tokenizer crashes, but they affect different tokenizers and different failure sites; 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": 26, + "review_unit_index": 16, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:28530", + "issue:29159", + "issue:30064", + "issue:33357", + "issue:34567", + "issue:34634", + "issue:35707", + "issue:36032", + "issue:41720", + "issue:42490", + "issue:42915", + "issue:43316", + "issue:43322", + "issue:43329", + "issue:43388", + "issue:43425", + "issue:43452", + "issue:43526", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43606", + "issue:43653", + "issue:43720", + "issue:43746", + "issue:43782", + "issue:43873", + "issue:43881", + "issue:44077", + "issue:44387", + "issue:44492", + "issue:44509", + "issue:44534", + "issue:44589", + "issue:44625", + "issue:44913", + "issue:45059", + "issue:45406", + "issue:45464", + "issue:45701", + "issue:45704", + "issue:45759", + "issue:45859", + "issue:46665", + "issue:46821", + "issue:46858", + "issue:47020" + ], + "soft_pairs": [ + "issue:43425|issue:46821", + "issue:43316|issue:44913", + "issue:28530|issue:30064", + "issue:42915|issue:43322", + "issue:33357|issue:35707", + "issue:43540|issue:43782", + "issue:43873|issue:45704", + "issue:29159|issue:36032", + "issue:30064|issue:34634", + "issue:44387|issue:45701", + "issue:43550|issue:43653", + "issue:44534|issue:46665", + "issue:43531|issue:45759", + "issue:43452|issue:47020", + "issue:41720|issue:43746", + "issue:43606|issue:43720", + "issue:44492|issue:44509", + "issue:42490|issue:43329", + "issue:43526|issue:43881", + "issue:45406|issue:45464", + "issue:43388|issue:45059", + "issue:34567|issue:35707", + "issue:43606|issue:43653", + "issue:34634|issue:41720", + "issue:43531|issue:46821", + "issue:43540|issue:46858", + "issue:45759|issue:45859", + "issue:43531|issue:44589", + "issue:44077|issue:44625" + ], + "prepared_review_unit_hash": "36140365305c3c6060abf0069b801283b28396e4d7b074ab8315e40579fba7a2", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35788, + "estimated_input_tokens": 8947, + "estimated_eval_tokens": 18150 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "All soft pairs look like false positives: they mention similar components (tokenizers, configs, loading, offload, Qwen, etc.) but describe different concrete failures, models, or code paths. No duplicate cluster is supported here.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43425", + "right": "issue:46821", + "accept": false, + "reason": "Torch version compatibility and a TimesFM window_size AttributeError are unrelated bugs." + }, + { + "left": "issue:43316", + "right": "issue:44913", + "accept": false, + "reason": "Both are config-related, but they concern different models and different fields with different failure modes." + }, + { + "left": "issue:28530", + "right": "issue:30064", + "accept": false, + "reason": "Early stopping metric lookup and void segmentation map processing are unrelated subsystems." + }, + { + "left": "issue:42915", + "right": "issue:43322", + "accept": false, + "reason": "Qwen3Moe FP8 config failure and Llava Next segfault are different model-loading problems." + }, + { + "left": "issue:33357", + "right": "issue:35707", + "accept": false, + "reason": "MacOS CLIP bus error and progressive generation with inputs_embeds/past_key_values are distinct issues." + }, + { + "left": "issue:43540", + "right": "issue:43782", + "accept": false, + "reason": "Video processing in Qwen3OmniMoe and Qwen3VL weight_only loading error are different code paths." + }, + { + "left": "issue:43873", + "right": "issue:45704", + "accept": false, + "reason": "Quantization/offloading memory behavior is unrelated to T5's apex RMSNorm leak." + }, + { + "left": "issue:29159", + "right": "issue:36032", + "accept": false, + "reason": "Slow/fast tokenizer inconsistency and T5 add_special_tokens load-time conflict are different tokenizer bugs." + }, + { + "left": "issue:30064", + "right": "issue:34634", + "accept": false, + "reason": "Segmentation-map processing and BarkProcessor voice_preset handling do not describe the same defect." + }, + { + "left": "issue:44387", + "right": "issue:45701", + "accept": false, + "reason": "Int4 quantization OOM and tokenization changing across versions are unrelated." + }, + { + "left": "issue:43550", + "right": "issue:43653", + "accept": false, + "reason": "torch.compile/SDPA failure in Bamba and BigBirdTokenizer special-token registration are unrelated." + }, + { + "left": "issue:44534", + "right": "issue:46665", + "accept": false, + "reason": "Non-persistent buffer initialization in v5 is not the same as from_pretrained behavior inside Python classes." + }, + { + "left": "issue:43531", + "right": "issue:45759", + "accept": false, + "reason": "Qwen3-MoE sliding_window and composite config unwrapping are different Qwen config issues." + }, + { + "left": "issue:43452", + "right": "issue:47020", + "accept": false, + "reason": "gguf_file loading support and XLMRobertaTokenizer/Unigram type mismatch are different loading failures." + }, + { + "left": "issue:41720", + "right": "issue:43746", + "accept": false, + "reason": "Qwen3 auto device mapping cudaErrorAssert and GraniteSpeech PEFT checkpoint loading are unrelated." + }, + { + "left": "issue:43606", + "right": "issue:43720", + "accept": false, + "reason": "CPU offload device mismatch and BitNet packed-weight unpacking are different accelerate-loading bugs." + }, + { + "left": "issue:44492", + "right": "issue:44509", + "accept": false, + "reason": "Cache strategy typo and removed pipeline-task docs are both documentation-like but not the same issue." + }, + { + "left": "issue:42490", + "right": "issue:43329", + "accept": false, + "reason": "Trailing-slash local path warning and missing video-branch variables in multimodal token counting are unrelated." + }, + { + "left": "issue:43526", + "right": "issue:43881", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and glm-4v-9b loading failure are different problems." + }, + { + "left": "issue:45406", + "right": "issue:45464", + "accept": false, + "reason": "Gemma4Processor missing _tokenizer and Qwen3.5 streaming chat/completions failure are separate serving bugs." + }, + { + "left": "issue:43388", + "right": "issue:45059", + "accept": false, + "reason": "Last-batch label gathering and SAM3 text/box behavior are different evaluation/model issues." + }, + { + "left": "issue:34567", + "right": "issue:35707", + "accept": false, + "reason": "TrainerState token counter not updating is unrelated to progressive generation cache handling." + }, + { + "left": "issue:43606", + "right": "issue:43653", + "accept": false, + "reason": "CPU offload device mismatch and BigBirdTokenizer mask-token registration are unrelated." + }, + { + "left": "issue:34634", + "right": "issue:41720", + "accept": false, + "reason": "BarkProcessor voice_preset and Qwen3 auto device mapping failures do not share a code path." + }, + { + "left": "issue:43531", + "right": "issue:46821", + "accept": false, + "reason": "Qwen3-MoE sliding_window and TimesFM window_size are superficially similar but different APIs and bugs." + }, + { + "left": "issue:43540", + "right": "issue:46858", + "accept": false, + "reason": "Video-input handling in Qwen3OmniMoe and static-cache slice arithmetic on non-CUDA backends are unrelated." + }, + { + "left": "issue:45759", + "right": "issue:45859", + "accept": false, + "reason": "Composite config unwrapping and CPU-offload rotary embedding compatibility are different Qwen issues." + }, + { + "left": "issue:43531", + "right": "issue:44589", + "accept": false, + "reason": "Sliding-window behavior and missing Float8 storage support are unrelated." + }, + { + "left": "issue:44077", + "right": "issue:44625", + "accept": false, + "reason": "patchtsmixer post_init policy and Qwen3.5 num_labels propagation are different model-config issues." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 17, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:30064", + "issue:36010", + "issue:39692", + "issue:41720", + "issue:42886", + "issue:43381", + "issue:43421", + "issue:43493", + "issue:43644", + "issue:43653", + "issue:43746", + "issue:43782", + "issue:43844", + "issue:43873", + "issue:43927", + "issue:43994", + "issue:44062", + "issue:44186", + "issue:44368", + "issue:44451", + "issue:44514", + "issue:44545", + "issue:44610", + "issue:44625", + "issue:44671", + "issue:44843", + "issue:44898", + "issue:44918", + "issue:44936", + "issue:44987", + "issue:44991", + "issue:45042", + "issue:45072", + "issue:45081", + "issue:45356", + "issue:45357", + "issue:45381", + "issue:45479", + "issue:45704", + "issue:45759", + "issue:45923", + "issue:46032", + "issue:46050", + "issue:46402", + "issue:46463", + "issue:46655", + "issue:46665", + "issue:46693" + ], + "soft_pairs": [ + "issue:45357|issue:46402", + "issue:44451|issue:44991", + "issue:43644|issue:46665", + "issue:44898|issue:45072", + "issue:44987|issue:45042", + "issue:44625|issue:45479", + "issue:44062|issue:44451", + "issue:43381|issue:43844", + "issue:39692|issue:44610", + "issue:41720|issue:45704", + "issue:44368|issue:45357", + "issue:43746|issue:46050", + "issue:46655|issue:46693", + "issue:43381|issue:45704", + "issue:43421|issue:43927", + "issue:44918|issue:44936", + "issue:45704|issue:45923", + "issue:42886|issue:44843", + "issue:45759|issue:45923", + "issue:43746|issue:43873", + "issue:44514|issue:44545", + "issue:43994|issue:44671", + "issue:39692|issue:43493", + "issue:46032|issue:46463", + "issue:43421|issue:43653", + "issue:30064|issue:36010", + "issue:43782|issue:45381", + "issue:45081|issue:45356", + "issue:43421|issue:44186" + ], + "prepared_review_unit_hash": "75dbd990140b8466a8fe64927bc52e10281e51a8a9ed0f807616c1020bd61c41", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36806, + "estimated_input_tokens": 9202, + "estimated_eval_tokens": 18660 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a heterogeneous mix of unrelated Transformers bug reports across tokenizers, multimodal processors, model loading, and training/runtime behavior. The only clear duplicate pair is the Qwen2_5_VLProcessor batched apply_chat_template crash with padding=False.", + "confidence": 0.92, + "canonical_issue_reason": "issue:44545 is the best canonical issue: it is an exact duplicate of 44514, has slightly more discussion/inbound signal, and cleanly names the concrete batched-input padding=False crash.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44545 is the strongest representative of the only true duplicate pair in this set; it captures the specific code-path failure most clearly.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45357", + "right": "issue:46402", + "accept": false, + "reason": "Both are multimodal save/load failures, but one is about incorrect visual encoder keys and the other about missing text encoder weights; different components and failure modes." + }, + { + "left": "issue:44451", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-loading regressions, but they affect different model repos and appear to stem from different issues." + }, + { + "left": "issue:43644", + "right": "issue:46665", + "accept": false, + "reason": "Non-persistent buffer initialization junk is unrelated to from_pretrained behavior inside Python classes." + }, + { + "left": "issue:44898", + "right": "issue:45072", + "accept": false, + "reason": "One is an image-resolution/interpolation bug; the other is a bfloat16 dtype mismatch in inference." + }, + { + "left": "issue:44987", + "right": "issue:45042", + "accept": false, + "reason": "A model-loading failure for a specific checkpoint is not the same as an image processor incorrectly requiring torchvision." + }, + { + "left": "issue:44625", + "right": "issue:45479", + "accept": false, + "reason": "num_labels propagation and single-label classification loss degeneracy are different bugs." + }, + { + "left": "issue:44062", + "right": "issue:44451", + "accept": false, + "reason": "AddedToken keyword conflicts are unrelated to the ScandiBERT loading failure." + }, + { + "left": "issue:43381", + "right": "issue:43844", + "accept": false, + "reason": "Eval-mode gradient checkpointing restriction is unrelated to the ZeRO-3 gradient explosion regression." + }, + { + "left": "issue:39692", + "right": "issue:44610", + "accept": false, + "reason": "A documentation/example mismatch is not the same as an image processor producing the wrong resolution." + }, + { + "left": "issue:41720", + "right": "issue:45704", + "accept": false, + "reason": "CUDA assert under auto device mapping is unrelated to the T5 apex RMSNorm memory leak." + }, + { + "left": "issue:44368", + "right": "issue:45357", + "accept": false, + "reason": "A warning about tie_word_embeddings does not match the Qwen3.5 save_pretrained visual-key bug." + }, + { + "left": "issue:43746", + "right": "issue:46050", + "accept": false, + "reason": "Local PEFT checkpoint loading and sinq quantized save/load breakage are different code paths." + }, + { + "left": "issue:46655", + "right": "issue:46693", + "accept": false, + "reason": "Video processor padding calculation is unrelated to the FA2 long-context performance regression." + }, + { + "left": "issue:43381", + "right": "issue:45704", + "accept": false, + "reason": "Eval-mode checkpointing behavior and a memory leak in T5 normalization are not the same bug." + }, + { + "left": "issue:43421", + "right": "issue:43927", + "accept": false, + "reason": "Runtime post-processor updates for special tokens and DiaConfig custom-token-ID loss are distinct tokenizer/config issues." + }, + { + "left": "issue:44918", + "right": "issue:44936", + "accept": false, + "reason": "TRL SFT embedding unpacking and trainer train/evaluate lifecycle failure are different trainer problems." + }, + { + "left": "issue:45704", + "right": "issue:45923", + "accept": false, + "reason": "The T5 memory leak and the missing gradient_checkpointing flag on a trust_remote_code model are unrelated." + }, + { + "left": "issue:42886", + "right": "issue:44843", + "accept": false, + "reason": "Both involve offline mode, but one is cache loading in general while the other is an unconditional model_info() call in _patch_mistral_regex." + }, + { + "left": "issue:45759", + "right": "issue:45923", + "accept": false, + "reason": "Composite Qwen config unwrapping is unrelated to a missing gradient_checkpointing capability flag." + }, + { + "left": "issue:43746", + "right": "issue:43873", + "accept": false, + "reason": "PEFT checkpoint loading and quantization offloading are different save/load and inference behaviors." + }, + { + "left": "issue:44514", + "right": "issue:44545", + "accept": true, + "reason": "These are the same crash report: Qwen2_5_VLProcessor.apply_chat_template fails on batched input when padding=False." + }, + { + "left": "issue:43994", + "right": "issue:44671", + "accept": false, + "reason": "Different models and different failure modes: nonsensical SigLIP2 predictions vs CamemBERT masked LM errors." + }, + { + "left": "issue:39692", + "right": "issue:43493", + "accept": false, + "reason": "SigLIP2 docs/example errors are not the same as a model-implementation discrepancy versus JAX." + }, + { + "left": "issue:46032", + "right": "issue:46463", + "accept": false, + "reason": "A correctness bug in use_cache for seq_len>1 is unrelated to an error-message issue in T5 decoder input validation." + }, + { + "left": "issue:43421", + "right": "issue:43653", + "accept": false, + "reason": "Special-token post-processor refresh and BigBird mask-token registration are different tokenizer bugs." + }, + { + "left": "issue:30064", + "right": "issue:36010", + "accept": false, + "reason": "Segmentation-map processing and a GenerationMixin import error are unrelated." + }, + { + "left": "issue:43782", + "right": "issue:45381", + "accept": false, + "reason": "Weight-only load failure for Qwen3VL is unrelated to the qwen2.5-vl vision_position_ids bug." + }, + { + "left": "issue:45081", + "right": "issue:45356", + "accept": false, + "reason": "The Mistral regex AttributeError and the Kimi-K2.5 codec/regression warning are related areas but not the same concrete bug." + }, + { + "left": "issue:43421", + "right": "issue:44186", + "accept": false, + "reason": "Runtime special-token update logic is unrelated to LayoutLMv2 NER padding/truncation crashes." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative on the soft-edge calls. The only accepted duplicate pair, issue:44514/issue:44545, is clearly supported by the identical titles and matching crash description, and the rest of the rejections do not overstate similarity." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 18, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:30064", + "issue:36331", + "issue:39692", + "issue:41720", + "issue:43295", + "issue:43335", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43596", + "issue:43611", + "issue:43618", + "issue:43643", + "issue:43653", + "issue:43697", + "issue:43720", + "issue:43756", + "issue:43819", + "issue:43873", + "issue:43927", + "issue:44008", + "issue:44062", + "issue:44186", + "issue:44220", + "issue:44361", + "issue:44488", + "issue:44661", + "issue:44821", + "issue:44843", + "issue:44877", + "issue:44898", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45356", + "issue:45381", + "issue:45507", + "issue:45834", + "issue:45907", + "issue:45987", + "issue:46294", + "issue:46665", + "issue:46693", + "issue:46829", + "issue:46832", + "issue:46858", + "issue:46932", + "issue:47020" + ], + "soft_pairs": [ + "issue:45507|issue:46858", + "issue:45356|issue:45907", + "issue:39692|issue:43873", + "issue:43720|issue:43819", + "issue:43653|issue:43720", + "issue:43295|issue:45834", + "issue:46829|issue:46832", + "issue:45987|issue:46829", + "issue:43550|issue:43927", + "issue:43550|issue:44361", + "issue:43540|issue:46294", + "issue:45381|issue:46693", + "issue:43531|issue:43643", + "issue:44062|issue:44488", + "issue:43596|issue:44220", + "issue:43611|issue:46665", + "issue:44186|issue:44898", + "issue:43653|issue:43819", + "issue:41720|issue:44991", + "issue:30064|issue:36331", + "issue:44008|issue:46858", + "issue:44220|issue:44661", + "issue:45987|issue:46932", + "issue:43618|issue:43697", + "issue:43335|issue:43756", + "issue:44488|issue:47020", + "issue:43643|issue:44877", + "issue:43550|issue:43819", + "issue:44843|issue:45081", + "issue:39692|issue:45834", + "issue:44821|issue:45084" + ], + "prepared_review_unit_hash": "5d739cea5809e06dd1a38ab8eb9430d8e354ff2b6558464b73ac464b000c1e59", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37261, + "estimated_input_tokens": 9316, + "estimated_eval_tokens": 18888 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No duplicate cluster here: the items are mostly unrelated bug reports spanning tokenizers, vision/image processing, cache semantics, quantization/offloading, and model-specific regressions. All soft edges look like superficial topic overlap rather than the same underlying fix. No PR candidates were present.", + "confidence": 0.92, + "canonical_issue_reason": "No clear canonical duplicate hub exists; the issues are too heterogeneous, and the apparent overlaps are only broad subsystem similarity.", + "canonical_pr_reason": null, + "best_issue_reason": "If one issue must anchor the set, 44843 is the most generally actionable library regression report, but it still does not represent the rest of the cluster well.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45507", + "right": "issue:46858", + "accept": false, + "reason": "Different bugs: GraniteMoEHybrid invalid-method call vs static-cache seq-length tensor arithmetic on non-CUDA backends." + }, + { + "left": "issue:45356", + "right": "issue:45907", + "accept": false, + "reason": "Unrelated regressions: Kimi tokenizer codec/fix_mistral_regex behavior vs Hub list_repo_files revision forwarding." + }, + { + "left": "issue:39692", + "right": "issue:43873", + "accept": false, + "reason": "Docs/model mismatch and quantization failure are not the same bug as offloading not working with quantization." + }, + { + "left": "issue:43720", + "right": "issue:43819", + "accept": false, + "reason": "Different model/code paths: BitNet packed-weight loading vs DAC.from_latents missing STE parity." + }, + { + "left": "issue:43653", + "right": "issue:43720", + "accept": false, + "reason": "BigBird special-token registration bug is unrelated to BitNet accelerate loading/unpacking." + }, + { + "left": "issue:43295", + "right": "issue:45834", + "accept": false, + "reason": "Processor/tokenizer regression is unrelated to Kosmos2.5 long-OCR index errors." + }, + { + "left": "issue:46829", + "right": "issue:46832", + "accept": false, + "reason": "Both are vision-model failures, but one is an output-field mixup and the other is an RT-DETR forward TypeError with different triggers." + }, + { + "left": "issue:45987", + "right": "issue:46829", + "accept": false, + "reason": "StaticCache seq-length contract issue is unrelated to VideoPrism output mapping." + }, + { + "left": "issue:43550", + "right": "issue:43927", + "accept": false, + "reason": "Different failures: torch.compile/SDPA in Bamba vs DiaConfig losing custom token IDs on save/load." + }, + { + "left": "issue:43550", + "right": "issue:44361", + "accept": false, + "reason": "Bamba torch.compile/SDPA regression is unrelated to MLukeTokenizer task AttributeError." + }, + { + "left": "issue:43540", + "right": "issue:46294", + "accept": false, + "reason": "Both involve video inputs, but the failures are different layers: Qwen3OmniMoe video processing vs chat-template token/feature mismatch." + }, + { + "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:43531", + "right": "issue:43643", + "accept": false, + "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to AutoConfig trust_remote_code missing fields." + }, + { + "left": "issue:44062", + "right": "issue:44488", + "accept": false, + "reason": "AddedToken special-arg duplication is a tokenizer construction bug, not the same as loading cjvt/sleng-bert." + }, + { + "left": "issue:43596", + "right": "issue:44220", + "accept": false, + "reason": "DeepSpeed ZeRO-3/BertModel index error is unrelated to _torch_extract_fbank_features()." + }, + { + "left": "issue:43611", + "right": "issue:46665", + "accept": false, + "reason": "Both concern loading behavior, but base_model_prefix breakage and from_pretrained-in-classes inconsistency are distinct code paths." + }, + { + "left": "issue:44186", + "right": "issue:44898", + "accept": false, + "reason": "LayoutLMv2 tokenizer NER/padding issues are unrelated to Perceiver image classification interpolation failures." + }, + { + "left": "issue:43653", + "right": "issue:43819", + "accept": false, + "reason": "Tokenizer special-token registration and DAC latent reconstruction mismatch are different bugs." + }, + { + "left": "issue:41720", + "right": "issue:44991", + "accept": false, + "reason": "Qwen3 auto-device-map cudaErrorAssert is unrelated to loading EMBEDDIA/est-roberta tokenizers." + }, + { + "left": "issue:30064", + "right": "issue:36331", + "accept": false, + "reason": "Image processor segmentation-map failure is unrelated to CustomTrainer.compute_loss signature changes." + }, + { + "left": "issue:44008", + "right": "issue:46858", + "accept": false, + "reason": "Gemma3n variable-collision forward bug is unrelated to static-cache seq-length tensor behavior." + }, + { + "left": "issue:44220", + "right": "issue:44661", + "accept": false, + "reason": "Audio fbank extraction and add-new-model-like tokenizer mapping are unrelated subsystems." + }, + { + "left": "issue:45987", + "right": "issue:46932", + "accept": false, + "reason": "StaticCache seq-length tensor return is unrelated to Gemma4 quantized vision/audio embedders dtype casting." + }, + { + "left": "issue:43618", + "right": "issue:43697", + "accept": false, + "reason": "Missing CLIPOutput attentions and RTDetrV2 output divergence are distinct model-output regressions." + }, + { + "left": "issue:43335", + "right": "issue:43756", + "accept": false, + "reason": "SwitchTransformers sparse-layer config bug and Smollm3 RoPE-layer mismatch are different model-configuration problems." + }, + { + "left": "issue:44488", + "right": "issue:47020", + "accept": false, + "reason": "Model-loading failure for cjvt/sleng-bert is not the same as XLMRobertaTokenizer Unigram dict-vs-sequence initialization." + }, + { + "left": "issue:43643", + "right": "issue:44877", + "accept": false, + "reason": "AutoConfig trust_remote_code missing-fields regression is unrelated to strict config blocking granite_speech loading." + }, + { + "left": "issue:43550", + "right": "issue:43819", + "accept": false, + "reason": "Bamba torch.compile/SDPA failure is unrelated to DAC.from_latents forward mismatch." + }, + { + "left": "issue:44843", + "right": "issue:45081", + "accept": false, + "reason": "Both touch _patch_mistral_regex, but one is an offline model_info call bug and the other is a backend_tokenizer attribute assumption." + }, + { + "left": "issue:39692", + "right": "issue:45834", + "accept": false, + "reason": "SigLIP2 docs/quantization errors are unrelated to Kosmos2.5 OCR length/index issues." + }, + { + "left": "issue:44821", + "right": "issue:45084", + "accept": false, + "reason": "AutoImageProcessor URL loading and template-node compilation errors 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": 26, + "review_unit_index": 19, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:28218", + "issue:29159", + "issue:34634", + "issue:36010", + "issue:43322", + "issue:43388", + "issue:43421", + "issue:43540", + "issue:43550", + "issue:43577", + "issue:43582", + "issue:43606", + "issue:43644", + "issue:43716", + "issue:43749", + "issue:43782", + "issue:43818", + "issue:43825", + "issue:43931", + "issue:43994", + "issue:44008", + "issue:44186", + "issue:44442", + "issue:44448", + "issue:44479", + "issue:44561", + "issue:44610", + "issue:44671", + "issue:44805", + "issue:44811", + "issue:44936", + "issue:44977", + "issue:44991", + "issue:45020", + "issue:45072", + "issue:45183", + "issue:45381", + "issue:45479", + "issue:45584", + "issue:45685", + "issue:45834", + "issue:46032", + "issue:46291", + "issue:46294", + "issue:46463", + "issue:46512", + "issue:46846", + "issue:47020" + ], + "soft_pairs": [ + "issue:34634|issue:36010", + "issue:44448|issue:44671", + "issue:43540|issue:44008", + "issue:44186|issue:44442", + "issue:45584|issue:45834", + "issue:45183|issue:45479", + "issue:43322|issue:45685", + "issue:43644|issue:43749", + "issue:43606|issue:45072", + "issue:44991|issue:47020", + "issue:43782|issue:44977", + "issue:43421|issue:44811", + "issue:44479|issue:45834", + "issue:43582|issue:46512", + "issue:43550|issue:45072", + "issue:43825|issue:43994", + "issue:43577|issue:43818", + "issue:44561|issue:45020", + "issue:43388|issue:46032", + "issue:44991|issue:46291", + "issue:46294|issue:46463", + "issue:43716|issue:44008", + "issue:44805|issue:44936", + "issue:43931|issue:46846", + "issue:43540|issue:45381", + "issue:44610|issue:45072", + "issue:28218|issue:29159" + ], + "prepared_review_unit_hash": "e8db899e029a78e4534f6b211d1325eaecad8f1cbb9c2621922839e8a0bca6bf", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 36042, + "estimated_input_tokens": 9011, + "estimated_eval_tokens": 18278 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No soft pair looks like the same concrete bug/change. The set is a loose semantic cluster of tokenizer, loading, dtype, and video-input regressions that mostly share only broad subsystem vocabulary.", + "confidence": 0.91, + "canonical_issue_reason": "Issue 45020 is the broadest, most representative regression report in the set: it describes recent-version breakage for remote_code models, which best matches the cluster\u2019s loose theme without pretending the other reports are duplicates.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45020 is the most general and centrally phrased report among the items, so it is the best umbrella representative of the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:34634", + "right": "issue:36010", + "accept": false, + "reason": "BarkProcessor voice_preset is a model-specific processor bug; GenerationMixin import failure is a separate API/module-path issue." + }, + { + "left": "issue:44448", + "right": "issue:44671", + "accept": false, + "reason": "Different models and symptoms: Pegasus output drift vs CamemBERT masked-LM prediction errors." + }, + { + "left": "issue:43540", + "right": "issue:44008", + "accept": false, + "reason": "Qwen3OmniMoe video-input processing error and Gemma3n audio_mel_mask variable collision are unrelated code paths." + }, + { + "left": "issue:44186", + "right": "issue:44442", + "accept": false, + "reason": "LayoutLMv2Tokenizer NER/padding crash is distinct from AutoTokenizer failing to load FastSpeech2ConformerTokenizer." + }, + { + "left": "issue:45584", + "right": "issue:45834", + "accept": false, + "reason": "Whisper empty-transcription decoding after align_special_tokens is unrelated to Kosmos2.5 long-OCR index errors." + }, + { + "left": "issue:45183", + "right": "issue:45479", + "accept": false, + "reason": "Validation of input_ids/inputs_embeds is a different bug from degenerate zero loss in sequence classification." + }, + { + "left": "issue:43322", + "right": "issue:45685", + "accept": false, + "reason": "Llava Next loading segfault and MPS histogram Int error are different failures in different subsystems." + }, + { + "left": "issue:43644", + "right": "issue:43749", + "accept": false, + "reason": "Non-persistent buffer corruption and FSDP CPU RAM efficient loading are both loading-related, but not the same concrete bug." + }, + { + "left": "issue:43606", + "right": "issue:45072", + "accept": false, + "reason": "Bark CPU-offload device mismatch and SwitchTransformers/TimmWrapperModel dtype mismatch are separate mismatch bugs." + }, + { + "left": "issue:44991", + "right": "issue:47020", + "accept": false, + "reason": "Tokenizer loading failure for EMBEDDIA/est-roberta is not the same as XLMRobertaTokenizer passing a dict to Unigram." + }, + { + "left": "issue:43782", + "right": "issue:44977", + "accept": false, + "reason": "Qwen3VL weight-only loading error and Qwen3.5 flash-attention generation failure are different stages and code paths." + }, + { + "left": "issue:43421", + "right": "issue:44811", + "accept": false, + "reason": "Runtime post-processor updates for special tokens and Whisper batch_decode skip_special_tokens handling are distinct tokenization behaviors." + }, + { + "left": "issue:44479", + "right": "issue:45834", + "accept": false, + "reason": "Broad video-input regression across Qwen variants does not match a Kosmos2.5 long-OCR indexing bug." + }, + { + "left": "issue:43582", + "right": "issue:46512", + "accept": false, + "reason": "Apple Silicon warmup TypeError is unrelated to from_config(dtype=...) leaving config.dtype stale." + }, + { + "left": "issue:43550", + "right": "issue:45072", + "accept": false, + "reason": "Bamba torch.compile SDPA failure and bfloat16 dtype mismatches in other models are not the same bug." + }, + { + "left": "issue:43825", + "right": "issue:43994", + "accept": false, + "reason": "pipeline translation error messaging and SigLIP2 nonsensical outputs are different problems." + }, + { + "left": "issue:43577", + "right": "issue:43818", + "accept": false, + "reason": "BLIP2 dtype reporting mismatch and Video-LLaVA missing temporal attention/shared weights are unrelated model bugs." + }, + { + "left": "issue:44561", + "right": "issue:45020", + "accept": false, + "reason": "Both concern recent-version breakage around remote_code, but one is a specific API removal impact and the other is a broader compatibility report; not the same concrete bug." + }, + { + "left": "issue:43388", + "right": "issue:46032", + "accept": false, + "reason": "gather_for_metrics label truncation and Mamba2Mixer use_cache correctness are separate training/runtime issues." + }, + { + "left": "issue:44991", + "right": "issue:46291", + "accept": false, + "reason": "Tokenizer loading for a specific model and a kernels dependency/version requirement during pipelines import are different failures." + }, + { + "left": "issue:46294", + "right": "issue:46463", + "accept": false, + "reason": "Qwen3.5 video/chat-template mismatch and T5 decoder-input validation errors are unrelated model-family bugs." + }, + { + "left": "issue:43716", + "right": "issue:44008", + "accept": false, + "reason": "Mistral-3 preprocessor/model dtype mismatch is not the same as Gemma3n forward-pass variable collision." + }, + { + "left": "issue:44805", + "right": "issue:44936", + "accept": false, + "reason": "A mask-vs-tensor shape IndexError and trainer.evaluate() failing after trainer.train() are too generic and likely separate causes." + }, + { + "left": "issue:43931", + "right": "issue:46846", + "accept": false, + "reason": "Model weight-shape mismatch on load and missing linear_attn entries in TP plan causing OOM/shape error are different loading/parallelism bugs." + }, + { + "left": "issue:43540", + "right": "issue:45381", + "accept": false, + "reason": "Qwen3OmniMoe video-input processing and Qwen2.5-VL/Qwen3 video_position_ids issues are related only by the broad video domain, not the same bug." + }, + { + "left": "issue:44610", + "right": "issue:45072", + "accept": false, + "reason": "Processor output size mismatch for OmDet-Turbo and dtype mismatches in unrelated models are not duplicates." + }, + { + "left": "issue:28218", + "right": "issue:29159", + "accept": false, + "reason": "Added-token extra-space behavior may be related to tokenizer parity, but 29159 is only a broad slow-vs-fast inconsistency report and not clearly 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": 26, + "review_unit_index": 20, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:34634", + "issue:36010", + "issue:36331", + "issue:42175", + "issue:43257", + "issue:43262", + "issue:43329", + "issue:43452", + "issue:43526", + "issue:43550", + "issue:43577", + "issue:43582", + "issue:43645", + "issue:43653", + "issue:43697", + "issue:43720", + "issue:43742", + "issue:43749", + "issue:43782", + "issue:43819", + "issue:43825", + "issue:43827", + "issue:43906", + "issue:44291", + "issue:44589", + "issue:44671", + "issue:44857", + "issue:44912", + "issue:44991", + "issue:45003", + "issue:45020", + "issue:45042", + "issue:45072", + "issue:45081", + "issue:45325", + "issue:45372", + "issue:45406", + "issue:45600", + "issue:45701", + "issue:46175", + "issue:46209", + "issue:46285", + "issue:46326", + "issue:46665", + "issue:46697", + "issue:46728", + "issue:46932" + ], + "soft_pairs": [ + "issue:43749|issue:44291", + "issue:43720|issue:45072", + "issue:43645|issue:46665", + "issue:43825|issue:43827", + "issue:44671|issue:45701", + "issue:43577|issue:43582", + "issue:34634|issue:36331", + "issue:42175|issue:45081", + "issue:46175|issue:46932", + "issue:46697|issue:46728", + "issue:45003|issue:46285", + "issue:43782|issue:45325", + "issue:43262|issue:43819", + "issue:43653|issue:44912", + "issue:43582|issue:44857", + "issue:45372|issue:46326", + "issue:44589|issue:45042", + "issue:43742|issue:45020", + "issue:43550|issue:44912", + "issue:36010|issue:43645", + "issue:43257|issue:43329", + "issue:36010|issue:36331", + "issue:43526|issue:43697", + "issue:43906|issue:46209", + "issue:45081|issue:45406", + "issue:43452|issue:44991", + "issue:43329|issue:45600" + ], + "prepared_review_unit_hash": "05bdffb7ca6b07d377cd82d6f5010a4d51de4197b7ee36b6d9c3592195ad5c70", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35660, + "estimated_input_tokens": 8915, + "estimated_eval_tokens": 18086 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: the soft pairs are mostly unrelated model-, tokenizer-, loading-, or processor-specific regressions, not true duplicates. I would not merge any of the candidate edges.", + "confidence": 0.91, + "canonical_issue_reason": "Issue 45020 is the best representative umbrella item because it describes the broadest recurring theme in the set: recent Transformers versions breaking model loading/behavior. The other issues are much narrower and model-specific.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45020 is the strongest overall representative for the cluster because it captures a broad, user-facing v5 regression pattern and is more general than the other reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43749", + "right": "issue:44291", + "accept": false, + "reason": "Both are loading-related, but one is FSDP CPU RAM-efficient loading and the other is an init_empty_weights/_is_hf_initialized TypeError. Different code paths and root causes." + }, + { + "left": "issue:43720", + "right": "issue:45072", + "accept": false, + "reason": "Different models and failures: BitNet packed-weight loading vs dtype mismatch in SwitchTransformers/TimmWrapperModel inference. Not the same bug." + }, + { + "left": "issue:43645", + "right": "issue:46665", + "accept": false, + "reason": "Both mention custom usage patterns, but one is about defining/initializing custom models in Jupyter notebooks and the other is from_pretrained inside Python classes. Too different to be the same issue." + }, + { + "left": "issue:43825", + "right": "issue:43827", + "accept": false, + "reason": "One is a runtime error message about translation tasks in pipeline(), the other is docs still referencing pipeline() after removals. Related area, but not the same underlying bug/change." + }, + { + "left": "issue:44671", + "right": "issue:45701", + "accept": false, + "reason": "CamemBERT masked-LM prediction regressions and general tokenization changes are different symptoms with different likely causes." + }, + { + "left": "issue:43577", + "right": "issue:43582", + "accept": false, + "reason": "Blip2 dtype propagation and Apple Silicon caching_allocator_warmup TypeError are unrelated." + }, + { + "left": "issue:34634", + "right": "issue:36331", + "accept": false, + "reason": "BarkProcessor voice_preset and CustomTrainer.compute_loss keyword handling are completely separate bugs." + }, + { + "left": "issue:42175", + "right": "issue:45081", + "accept": false, + "reason": "Package backend dependency selection and a Mistral tokenizer regex patch crash are unrelated." + }, + { + "left": "issue:46175", + "right": "issue:46932", + "accept": false, + "reason": "Gemma4VisionPatchEmbedder one-hot materialization and Gemma4 quantized embedder dtype casting are different issues in different parts of the model stack." + }, + { + "left": "issue:46697", + "right": "issue:46728", + "accept": false, + "reason": "Both are image-processing bugs, but they affect different models and different computations: Swin2SR padding vs Aria patch counting." + }, + { + "left": "issue:45003", + "right": "issue:46285", + "accept": false, + "reason": "Unsafe sys.modules access in modeling_utils is unrelated to CLIP text model structure divergence." + }, + { + "left": "issue:43782", + "right": "issue:45325", + "accept": false, + "reason": "Qwen3VL weight_only loading failure and Qwen2.5-VL rope index scaling are different models and different code paths." + }, + { + "left": "issue:43262", + "right": "issue:43819", + "accept": false, + "reason": "Audio chat-template sampling-rate defaults and DAC.from_latents STE mismatch are separate audio-processing bugs." + }, + { + "left": "issue:43653", + "right": "issue:44912", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and MXFP4 quantization fallback in git-oss-20b are unrelated." + }, + { + "left": "issue:43582", + "right": "issue:44857", + "accept": false, + "reason": "Apple Silicon allocator warmup TypeError and LwDetrImageLoss AMP/CUDA crash are different runtime failures." + }, + { + "left": "issue:45372", + "right": "issue:46326", + "accept": false, + "reason": "Gemma4 processor import failure from mistral_common and continue_final_message behavior for mistral_common tokenizers are not the same bug." + }, + { + "left": "issue:44589", + "right": "issue:45042", + "accept": false, + "reason": "Float8 storage lookup failure and PIL image processors requiring torchvision are unrelated." + }, + { + "left": "issue:43742", + "right": "issue:45020", + "accept": false, + "reason": "KeyError loading a specific model and broad remote_code breakage are different failures." + }, + { + "left": "issue:43550", + "right": "issue:44912", + "accept": false, + "reason": "torch.compile/SDPA failure in Bamba-9B-v2 is unrelated to MXFP4 quantization fallback in git-oss-20b." + }, + { + "left": "issue:36010", + "right": "issue:43645", + "accept": false, + "reason": "GenerationMixin import failure and custom-model Jupyter notebook initialization are unrelated." + }, + { + "left": "issue:43257", + "right": "issue:43329", + "accept": false, + "reason": "Qwen3 MoE weight conversion during accelerate/deepspeed loading and a multimodal token counting video-branch NameError are different bugs." + }, + { + "left": "issue:36010", + "right": "issue:36331", + "accept": false, + "reason": "ImportError for GenerationMixin and CustomTrainer.compute_loss keyword mismatch are separate API regressions." + }, + { + "left": "issue:43526", + "right": "issue:43697", + "accept": false, + "reason": "BeitImageProcessorFast reduce_labels behavior and RTDetrV2 output differences are unrelated." + }, + { + "left": "issue:43906", + "right": "issue:46209", + "accept": false, + "reason": "A reproduction of an older issue and FP8 inference failure on B200 are not the same bug." + }, + { + "left": "issue:45081", + "right": "issue:45406", + "accept": false, + "reason": "Mistral tokenizer regex patch crash and Gemma4Processor missing _tokenizer in transformers serve are different processor/tokenizer issues." + }, + { + "left": "issue:43452", + "right": "issue:44991", + "accept": false, + "reason": "gguf_file loading regressions and EMBEDDIA/est-roberta tokenizer loading failures are different tokenizer/model-load problems." + }, + { + "left": "issue:43329", + "right": "issue:45600", + "accept": false, + "reason": "Undefined multimodal token helper names and removed Sam3LiteText auto-mapping references are unrelated CI/runtime 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": 26, + "review_unit_index": 21, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:29159", + "issue:34567", + "issue:38175", + "issue:39692", + "issue:40990", + "issue:42175", + "issue:42490", + "issue:42915", + "issue:43257", + "issue:43262", + "issue:43329", + "issue:43450", + "issue:43526", + "issue:43531", + "issue:43638", + "issue:43643", + "issue:43650", + "issue:43696", + "issue:43720", + "issue:43742", + "issue:43746", + "issue:43819", + "issue:43854", + "issue:43906", + "issue:43927", + "issue:43931", + "issue:44060", + "issue:44291", + "issue:44368", + "issue:44387", + "issue:44479", + "issue:44554", + "issue:44610", + "issue:44617", + "issue:44811", + "issue:44821", + "issue:44912", + "issue:44936", + "issue:44977", + "issue:45042", + "issue:45084", + "issue:45459", + "issue:45538", + "issue:45759", + "issue:45803", + "issue:46291", + "issue:46421", + "issue:46424" + ], + "soft_pairs": [ + "issue:44060|issue:44368", + "issue:43526|issue:44554", + "issue:43819|issue:43927", + "issue:43531|issue:45042", + "issue:43638|issue:44936", + "issue:45084|issue:45538", + "issue:42175|issue:46291", + "issue:43746|issue:44811", + "issue:42490|issue:43257", + "issue:43854|issue:44977", + "issue:43643|issue:45759", + "issue:39692|issue:44912", + "issue:38175|issue:44610", + "issue:43720|issue:44610", + "issue:43262|issue:44811", + "issue:43257|issue:43696", + "issue:44387|issue:44617", + "issue:43329|issue:43450", + "issue:43257|issue:43931", + "issue:43329|issue:44479", + "issue:42915|issue:43742", + "issue:34567|issue:40990", + "issue:45459|issue:45803", + "issue:44291|issue:45042", + "issue:46421|issue:46424", + "issue:43650|issue:43906", + "issue:42915|issue:44821", + "issue:29159|issue:34567" + ], + "prepared_review_unit_hash": "c82d672d74af73fcf5673cd0978c94dbbdf0ce3caa59e06dc3d1848b2ce27d5c", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36284, + "estimated_input_tokens": 9071, + "estimated_eval_tokens": 18398 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are a heterogeneous set of unrelated issues; none of the proposed soft pairs look like the same underlying bug or change, so I rejected all duplicate links.", + "confidence": 0.92, + "canonical_issue_reason": "No single canonical issue stands out because the set spans unrelated bugs across tokenizers, training, multimodal processing, quantization, and model loading.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45042 is the clearest standalone report to anchor on if one must be chosen: it is specific, reproducible, and has external references, but it is not a duplicate hub for the rest of the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44060", + "right": "issue:44368", + "accept": false, + "reason": "Both mention tied-weights warnings, but they concern different Qwen variants and different causes; not the same fix." + }, + { + "left": "issue:43526", + "right": "issue:44554", + "accept": false, + "reason": "One is a BeitImageProcessorFast label-reduction bug; the other is an MPS attention correctness issue." + }, + { + "left": "issue:43819", + "right": "issue:43927", + "accept": false, + "reason": "Different subsystems and failures: DAC latent reconstruction vs DiaConfig save/load token ID corruption." + }, + { + "left": "issue:43531", + "right": "issue:45042", + "accept": false, + "reason": "Qwen3-MoE sliding-window handling and PIL-backend torchvision dependency are unrelated bugs." + }, + { + "left": "issue:43638", + "right": "issue:44936", + "accept": false, + "reason": "Different training paths: a Zero3/non-pretrained Bert IndexError versus evaluate() failing after train()." + }, + { + "left": "issue:45084", + "right": "issue:45538", + "accept": false, + "reason": "Compilation/template-node error and CLIPTokenizer max_length bug are distinct code paths." + }, + { + "left": "issue:42175", + "right": "issue:46291", + "accept": false, + "reason": "Packaging/backend install issue is unrelated to kernels import-time dependency validation." + }, + { + "left": "issue:43746", + "right": "issue:44811", + "accept": false, + "reason": "Local checkpoint loading for PEFT adapters is unrelated to Whisper batch_decode skip_special_tokens handling." + }, + { + "left": "issue:42490", + "right": "issue:43257", + "accept": false, + "reason": "A trailing-slash path warning is not the same bug as Qwen3 MOE weight conversion with accelerate+deepspeed." + }, + { + "left": "issue:43854", + "right": "issue:44977", + "accept": false, + "reason": "GLM-4.7-Flash test loading failure and Qwen3.5 flash-attention generation regression are different model-specific issues." + }, + { + "left": "issue:43643", + "right": "issue:45759", + "accept": false, + "reason": "AutoConfig trust_remote_code field loss and AutoModelForCausalLM text_config unwrapping are separate config-loading bugs." + }, + { + "left": "issue:39692", + "right": "issue:44912", + "accept": false, + "reason": "SigLIP2 doc-example mistakes are not the same as GPT-oss MXFP4 quantization falling back to bf16." + }, + { + "left": "issue:38175", + "right": "issue:44610", + "accept": false, + "reason": "Unexpected zero probabilities in SigLIP2 is unrelated to OmDet-Turbo's input-size mismatch." + }, + { + "left": "issue:43720", + "right": "issue:44610", + "accept": false, + "reason": "BitNet packed-weight loading and OmDet-Turbo processor sizing are different failure modes." + }, + { + "left": "issue:43262", + "right": "issue:44811", + "accept": false, + "reason": "Audio chat-template sampling-rate default and Whisper batch_decode token skipping are different processor behaviors." + }, + { + "left": "issue:43257", + "right": "issue:43696", + "accept": false, + "reason": "MOE weight conversion under distributed loading is unrelated to a GPT-oss CUDA OOM report." + }, + { + "left": "issue:44387", + "right": "issue:44617", + "accept": false, + "reason": "Quantization-related CUDA reserved-memory regression is not the same as Sam3Video's model-specific OOM." + }, + { + "left": "issue:43329", + "right": "issue:43450", + "accept": false, + "reason": "Undefined video-token-count helpers and batched video output shape are separate multimodal processor bugs." + }, + { + "left": "issue:43257", + "right": "issue:43931", + "accept": false, + "reason": "Qwen3 MOE conversion/loading and Qwen3-VL weight-shape mismatch are different model-loading problems." + }, + { + "left": "issue:43329", + "right": "issue:44479", + "accept": false, + "reason": "The undefined video-branch helper bug is not the same as the v5.3.0 video regression across multiple model families." + }, + { + "left": "issue:42915", + "right": "issue:43742", + "accept": false, + "reason": "FineGrainedFP8Config failure for Qwen3Moe and a MobileLLM key error are unrelated." + }, + { + "left": "issue:34567", + "right": "issue:40990", + "accept": false, + "reason": "TrainerState token counter update and GPT-oss perplexity are not the same bug." + }, + { + "left": "issue:45459", + "right": "issue:45803", + "accept": false, + "reason": "Both concern exception handling, but they affect different functions and different error paths, so they are not duplicates." + }, + { + "left": "issue:44291", + "right": "issue:45042", + "accept": false, + "reason": "init_empty_weights argument incompatibility is unrelated to the PIL-backend torchvision dependency regression." + }, + { + "left": "issue:46421", + "right": "issue:46424", + "accept": false, + "reason": "Chunked-prefill recompilation and max_cache_length only applying to quantized cache are distinct cache issues." + }, + { + "left": "issue:43650", + "right": "issue:43906", + "accept": false, + "reason": "The placeholder 'ADD THE DATA' issue is not the same as an isolated reproduction of issue 38071." + }, + { + "left": "issue:42915", + "right": "issue:44821", + "accept": false, + "reason": "Qwen3Moe FP8 loading and AutoImageProcessor URL loading are unrelated." + }, + { + "left": "issue:29159", + "right": "issue:34567", + "accept": false, + "reason": "Tokenizer slow/fast inconsistency and TrainerState token counting are different subsystems and 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": 26, + "review_unit_index": 22, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:24643", + "issue:28282", + "issue:29127", + "issue:29159", + "issue:29870", + "issue:33357", + "issue:34567", + "issue:34689", + "issue:42175", + "issue:42915", + "issue:43404", + "issue:43425", + "issue:43452", + "issue:43454", + "issue:43525", + "issue:43540", + "issue:43577", + "issue:43582", + "issue:43606", + "issue:43697", + "issue:43742", + "issue:43749", + "issue:43824", + "issue:43927", + "issue:44291", + "issue:44479", + "issue:44568", + "issue:44756", + "issue:44857", + "issue:44912", + "issue:44936", + "issue:44987", + "issue:45081", + "issue:45127", + "issue:45137", + "issue:45356", + "issue:45357", + "issue:45405", + "issue:45538", + "issue:45859", + "issue:46291", + "issue:46512", + "issue:46693", + "issue:46701", + "issue:46821", + "issue:46832", + "issue:46869", + "issue:47020" + ], + "soft_pairs": [ + "issue:43606|issue:45859", + "issue:43749|issue:44756", + "issue:43540|issue:46701", + "issue:28282|issue:29870", + "issue:43454|issue:43577", + "issue:43525|issue:46291", + "issue:43927|issue:44912", + "issue:28282|issue:34689", + "issue:44568|issue:45356", + "issue:44987|issue:45538", + "issue:33357|issue:34567", + "issue:43425|issue:46291", + "issue:44857|issue:46832", + "issue:43697|issue:45357", + "issue:43404|issue:45127", + "issue:45137|issue:45405", + "issue:42175|issue:43824", + "issue:28282|issue:29127", + "issue:45538|issue:46869", + "issue:45081|issue:47020", + "issue:42915|issue:45538", + "issue:44479|issue:46693", + "issue:43742|issue:44291", + "issue:44936|issue:46821", + "issue:43452|issue:43582", + "issue:43577|issue:46512", + "issue:24643|issue:29159", + "issue:42915|issue:43749" + ], + "prepared_review_unit_hash": "03e40d38477441ae49ac8409571db61d3fb68b70b08f46b8693640e9ff2b7a8a", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36318, + "estimated_input_tokens": 9080, + "estimated_eval_tokens": 18416 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is heterogeneous: most items are distinct model-loading, tokenizer, dtype, or training bugs rather than duplicates. I rejected all soft edges because they do not look like the same underlying bug/change.", + "confidence": 0.92, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43606", + "right": "issue:45859", + "accept": false, + "reason": "Both mention CPU offload, but they affect different models/code paths and different failures; not the same bug." + }, + { + "left": "issue:43749", + "right": "issue:44756", + "accept": false, + "reason": "One is FSDP CPU RAM efficient loading, the other is mmap/OOM on Strix Halo; separate runtime issues." + }, + { + "left": "issue:43540", + "right": "issue:46701", + "accept": false, + "reason": "Both are multimodal/pipeline failures, but on different models and distinct error conditions." + }, + { + "left": "issue:28282", + "right": "issue:29870", + "accept": false, + "reason": "ImportError about missing PyTorch is unrelated to Bart generate() unbound local error." + }, + { + "left": "issue:43454", + "right": "issue:43577", + "accept": false, + "reason": "Both involve model loading, but one is untied lm_head weights and the other is dtype propagation in BLIP2; different bugs." + }, + { + "left": "issue:43525", + "right": "issue:46291", + "accept": false, + "reason": "Missing pad_token_id in a config is unrelated to pipelines import failing on kernels version/revision checks." + }, + { + "left": "issue:43927", + "right": "issue:44912", + "accept": false, + "reason": "DiaConfig save/load token ID loss is a different problem from MXFP4 quantization fallback behavior." + }, + { + "left": "issue:28282", + "right": "issue:34689", + "accept": false, + "reason": "Both are loading-related, but one is missing PyTorch and the other is a specific regression loading Llama 3.2 vision." + }, + { + "left": "issue:44568", + "right": "issue:45356", + "accept": false, + "reason": "Both concern tokenizers, but the affected models and the underlying regression are different." + }, + { + "left": "issue:44987", + "right": "issue:45538", + "accept": false, + "reason": "Model loading failure for one repo is not the same as CLIPTokenizer's model_max_length bug." + }, + { + "left": "issue:33357", + "right": "issue:34567", + "accept": false, + "reason": "A MacOS CLIP bus error and TrainerState token counter not updating are unrelated." + }, + { + "left": "issue:43425", + "right": "issue:46291", + "accept": false, + "reason": "Torch compatibility and pipeline import dependency resolution are separate issues." + }, + { + "left": "issue:44857", + "right": "issue:46832", + "accept": false, + "reason": "Different DETR-family crashes with different triggers and code paths." + }, + { + "left": "issue:43697", + "right": "issue:45357", + "accept": false, + "reason": "RT-DETR output mismatch and Qwen3.5 save_pretrained key mapping are unrelated." + }, + { + "left": "issue:43404", + "right": "issue:45127", + "accept": false, + "reason": "Both touch tied embeddings/lm_head, but one is a model init bug and the other is LoRA merge with extended vocab; not the same concrete bug." + }, + { + "left": "issue:45137", + "right": "issue:45405", + "accept": false, + "reason": "DeepSpeed ZeRO3 deque error and an unreleased PEFT version bump are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:43824", + "accept": false, + "reason": "Backend packaging issue and missing Qwen2.5-VL import are different failure modes." + }, + { + "left": "issue:28282", + "right": "issue:29127", + "accept": false, + "reason": "Missing PyTorch import error is unrelated to LayoutLMv3 box-validation messaging." + }, + { + "left": "issue:45538", + "right": "issue:46869", + "accept": false, + "reason": "CLIP tokenizer max_length bug and YOLO multi-GPU device-map failure are different problems." + }, + { + "left": "issue:45081", + "right": "issue:47020", + "accept": false, + "reason": "These are both tokenizer bugs, but with different tokenizer classes and distinct failure mechanisms." + }, + { + "left": "issue:42915", + "right": "issue:45538", + "accept": false, + "reason": "FineGrainedFP8 quantization failure is unrelated to CLIPTokenizer's max_length sentinel issue." + }, + { + "left": "issue:44479", + "right": "issue:46693", + "accept": false, + "reason": "A video-input regression and a long-context FA2 performance regression are different bugs." + }, + { + "left": "issue:43742", + "right": "issue:44291", + "accept": false, + "reason": "KeyError loading one model and init_empty_weights argument mismatch in another are distinct loading paths." + }, + { + "left": "issue:44936", + "right": "issue:46821", + "accept": false, + "reason": "Trainer evaluate/train state bug and TimesFM window_size AttributeError are unrelated." + }, + { + "left": "issue:43452", + "right": "issue:43582", + "accept": false, + "reason": "gguf_file loading issues and Apple Silicon allocator warmup TypeError are separate bugs." + }, + { + "left": "issue:43577", + "right": "issue:46512", + "accept": false, + "reason": "Both are dtype-related, but one is incorrect loaded tensor dtypes and the other is stale config.dtype after from_config." + }, + { + "left": "issue:24643", + "right": "issue:29159", + "accept": false, + "reason": "DeepSpeed training weight-shape error and slow/fast tokenizer inconsistency are unrelated." + }, + { + "left": "issue:42915", + "right": "issue:43749", + "accept": false, + "reason": "FP8 config failure and FSDP CPU RAM efficient loading are different features and failure modes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 23, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:27957", + "issue:28282", + "issue:34567", + "issue:36331", + "issue:38175", + "issue:41720", + "issue:42175", + "issue:43257", + "issue:43531", + "issue:43540", + "issue:43577", + "issue:43582", + "issue:43716", + "issue:43742", + "issue:43756", + "issue:43906", + "issue:44062", + "issue:44291", + "issue:44479", + "issue:44488", + "issue:44560", + "issue:44568", + "issue:44589", + "issue:44737", + "issue:44792", + "issue:44821", + "issue:44849", + "issue:44898", + "issue:44964", + "issue:44987", + "issue:44991", + "issue:45042", + "issue:45072", + "issue:45081", + "issue:45325", + "issue:45381", + "issue:45405", + "issue:45488", + "issue:45600", + "issue:45701", + "issue:45823", + "issue:45907", + "issue:46399", + "issue:46463", + "issue:46491", + "issue:46693", + "issue:46701", + "issue:46821" + ], + "soft_pairs": [ + "issue:44062|issue:44991", + "issue:44849|issue:45701", + "issue:27957|issue:28282", + "issue:41720|issue:43540", + "issue:43540|issue:44560", + "issue:44987|issue:45907", + "issue:42175|issue:44488", + "issue:44479|issue:45381", + "issue:43540|issue:44479", + "issue:34567|issue:36331", + "issue:38175|issue:44898", + "issue:44792|issue:44964", + "issue:44568|issue:45488", + "issue:44291|issue:44821", + "issue:43257|issue:43756", + "issue:44062|issue:45081", + "issue:43582|issue:46821", + "issue:45042|issue:45081", + "issue:43742|issue:44589", + "issue:43756|issue:46399", + "issue:43756|issue:45325", + "issue:43716|issue:46701", + "issue:43531|issue:44821", + "issue:45405|issue:45600", + "issue:44737|issue:46463", + "issue:43577|issue:45072", + "issue:44568|issue:46491", + "issue:45823|issue:46463", + "issue:43906|issue:46693" + ], + "prepared_review_unit_hash": "fb861cc3c8bbf6cb2a6a5e5a723da098c061a23e634d3769783d1888c8932d94", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36512, + "estimated_input_tokens": 9128, + "estimated_eval_tokens": 18512 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly heterogeneous; there is no strong cluster-wide duplicate theme. The only likely true duplicate pair is the Qwen2.5-VL video-position-id regression, where one report is the broader umbrella and the other is a focused reproduction.", + "confidence": 0.61, + "canonical_issue_reason": "issue:44479 is the best available anchor because it is the broader, more general report for the one clear duplicate subcluster (Qwen video-input regression); the rest of the items are too diverse to form a single coherent canonical issue.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44479 is the strongest issue to keep as the representative report for the small Qwen VL video regression cluster, but it is not a good canonical for the full set because most other items are unrelated bugs.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44062", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-related, but one is an AddedToken keyword collision and the other is a tokenizer-loading failure for a specific model; different bugs." + }, + { + "left": "issue:44849", + "right": "issue:45701", + "accept": false, + "reason": "Different symptoms and code paths: Qwen3.5 hidden-state output bug vs a broad tokenization regression report." + }, + { + "left": "issue:27957", + "right": "issue:28282", + "accept": false, + "reason": "Unrelated failures: FlashAttention2 support for XLMRoberta vs PyTorch import detection in AutoModel." + }, + { + "left": "issue:41720", + "right": "issue:43540", + "accept": false, + "reason": "Different model families and failures: cudaErrorAssert during auto device mapping vs a video-input ValueError in Qwen3OmniMoe." + }, + { + "left": "issue:43540", + "right": "issue:44560", + "accept": false, + "reason": "Both mention video inputs, but the affected models and error paths differ; no evidence of the same underlying bug." + }, + { + "left": "issue:44987", + "right": "issue:45907", + "accept": false, + "reason": "One is a model loading regression, the other is a hub API forwarding bug; unrelated." + }, + { + "left": "issue:42175", + "right": "issue:44488", + "accept": false, + "reason": "Packaging/dependency issue vs a model-loading failure for cjvt/sleng-bert; not the same problem." + }, + { + "left": "issue:44479", + "right": "issue:45381", + "accept": true, + "reason": "Same Qwen2.5-VL video-position-id regression family in v5.3.0; one is the broader regression report and the other is a focused description of the same underlying bug." + }, + { + "left": "issue:43540", + "right": "issue:44479", + "accept": false, + "reason": "Both are video-related Qwen issues, but they target different models and different failures; not mergeable as one bug." + }, + { + "left": "issue:34567", + "right": "issue:36331", + "accept": false, + "reason": "TrainerState token-count tracking vs a CustomTrainer compute_loss signature error are separate training bugs." + }, + { + "left": "issue:38175", + "right": "issue:44898", + "accept": false, + "reason": "Both are image-classification-related, but one is zero probabilities on SigLIP2 and the other is a Perceiver non-default-resolution failure; distinct root causes." + }, + { + "left": "issue:44792", + "right": "issue:44964", + "accept": false, + "reason": "Janus image-generation test failure vs Phi-4 multimodal loading failure; same broad area, different issues." + }, + { + "left": "issue:44568", + "right": "issue:45488", + "accept": false, + "reason": "Tokenizer regressions, but one concerns add_special_tokens behavior and the other overrides the pre-tokenizer; not the same bug." + }, + { + "left": "issue:44291", + "right": "issue:44821", + "accept": false, + "reason": "Unexpected _is_hf_initialized argument during init_empty_weights vs loading AutoImageProcessor from a URL; unrelated." + }, + { + "left": "issue:43257", + "right": "issue:43756", + "accept": false, + "reason": "Qwen3 MOE weight conversion under accelerate/deepspeed vs Smollm3 RoPE-layer count mismatch; different models and fixes." + }, + { + "left": "issue:44062", + "right": "issue:45081", + "accept": false, + "reason": "Both involve tokenizer internals, but one is an AddedToken duplicate-kwarg error and the other is a Mistral regex-patching attribute error." + }, + { + "left": "issue:43582", + "right": "issue:46821", + "accept": false, + "reason": "Apple Silicon warmup TypeError vs TimesFM window_size AttributeError; unrelated backend/model bugs." + }, + { + "left": "issue:45042", + "right": "issue:45081", + "accept": false, + "reason": "Image-processor dependency requirements vs Mistral tokenizer regex patching; completely different code paths." + }, + { + "left": "issue:43742", + "right": "issue:44589", + "accept": false, + "reason": "A key-error while loading MobileLLM vs a Float8 storage lookup TypeError; different failures." + }, + { + "left": "issue:43756", + "right": "issue:46399", + "accept": false, + "reason": "Both mention rotary/rope terms, but Smollm3 RoPE-layer dropping and Qwen3.5 kernel registration are unrelated bugs." + }, + { + "left": "issue:43756", + "right": "issue:45325", + "accept": false, + "reason": "Both are RoPE-related, but they affect different model families and different logic paths; not the same underlying issue." + }, + { + "left": "issue:43716", + "right": "issue:46701", + "accept": false, + "reason": "Image-preprocessor dtype mismatch vs a DiffusionGemma pipeline failure; distinct problems." + }, + { + "left": "issue:43531", + "right": "issue:44821", + "accept": false, + "reason": "Sliding-window bug in Qwen3-MoE vs AutoImageProcessor URL loading failure; unrelated." + }, + { + "left": "issue:45405", + "right": "issue:45600", + "accept": false, + "reason": "Dependency-version bump not yet on PyPI vs stale auto_mappings references; one is packaging, the other is CI cleanup." + }, + { + "left": "issue:44737", + "right": "issue:46463", + "accept": false, + "reason": "XLNet CPU placement bug in relative position encoding vs an incorrect T5 error message; different areas." + }, + { + "left": "issue:43577", + "right": "issue:45072", + "accept": false, + "reason": "Both are dtype-related, but they affect different models and different mismatches; not one concrete bug." + }, + { + "left": "issue:44568", + "right": "issue:46491", + "accept": false, + "reason": "Tokenizer behavior regressions, but one is special-token insertion and the other is encode/decode leading-space round-trip; separate issues." + }, + { + "left": "issue:45823", + "right": "issue:46463", + "accept": false, + "reason": "Gemma4 device-map mismatch during forward vs a T5 decoder-input validation message bug; unrelated." + }, + { + "left": "issue:43906", + "right": "issue:46693", + "accept": false, + "reason": "An isolated reproduction of one prior issue vs an FA2 long-context performance regression; no shared bug." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The heterogeneous-cluster conclusion is supported by the issue titles, and the one accepted pair (44479/45381) is a reasonable duplicate/near-duplicate within the Qwen2.5-VL video regression subcluster. No soft-edge verdict appears to overstate evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 24, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:24643", + "issue:28282", + "issue:29870", + "issue:32090", + "issue:38175", + "issue:41762", + "issue:43295", + "issue:43452", + "issue:43526", + "issue:43540", + "issue:43643", + "issue:43653", + "issue:43688", + "issue:43697", + "issue:43720", + "issue:43749", + "issue:43906", + "issue:44361", + "issue:44442", + "issue:44466", + "issue:44488", + "issue:44493", + "issue:44625", + "issue:44737", + "issue:44843", + "issue:44877", + "issue:44898", + "issue:44913", + "issue:44936", + "issue:45005", + "issue:45200", + "issue:45216", + "issue:45357", + "issue:45375", + "issue:45479", + "issue:45588", + "issue:45663", + "issue:45672", + "issue:45674", + "issue:45920", + "issue:46050", + "issue:46121", + "issue:46463", + "issue:46468", + "issue:46693", + "issue:46829", + "issue:46846", + "issue:46899" + ], + "soft_pairs": [ + "issue:45672|issue:45674", + "issue:24643|issue:29870", + "issue:43906|issue:46468", + "issue:44625|issue:44877", + "issue:43540|issue:46693", + "issue:43452|issue:44488", + "issue:45663|issue:46846", + "issue:43697|issue:46829", + "issue:28282|issue:32090", + "issue:44737|issue:45588", + "issue:46463|issue:46899", + "issue:44913|issue:46121", + "issue:45357|issue:45920", + "issue:43526|issue:45479", + "issue:43697|issue:45216", + "issue:43720|issue:44898", + "issue:44466|issue:45005", + "issue:45200|issue:46463", + "issue:43643|issue:45375", + "issue:44493|issue:46050", + "issue:41762|issue:44936", + "issue:43688|issue:43749", + "issue:44361|issue:44442", + "issue:38175|issue:43295", + "issue:43653|issue:44361", + "issue:43452|issue:44843" + ], + "prepared_review_unit_hash": "0dc7f04f8bb14aa216de80a131d6a95bfc82a3c1c91b821af2e3fa851eb1f4b3", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 26, + "serialized_chars": 36123, + "estimated_input_tokens": 9031, + "estimated_eval_tokens": 18318 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most pairs are clearly different bug reports across unrelated models or code paths. The only plausible duplicate pair is the Qwen3.5 `save_pretrained` regression reports (45216/45357); everything else should stay separate.", + "confidence": 0.62, + "canonical_issue_reason": "Issue 45216 is the earliest, broadest report of the Qwen3.5 `save_pretrained` regression; 45357 reads like a later, narrower follow-up on the same serialization path.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45216 is the best representative issue here because it captures the core checkpoint corruption problem with the broadest framing and strongest standalone signal.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45672", + "right": "issue:45674", + "accept": false, + "reason": "Both involve 4-bit inference, but the root causes differ: unsafe dtype casting in Gemma4 audio modules vs. BitsAndBytes skip-module handling." + }, + { + "left": "issue:24643", + "right": "issue:29870", + "accept": false, + "reason": "DeepSpeed weight-shape error during training is unrelated to Bart generate() using `decoder_attention_mask` before assignment." + }, + { + "left": "issue:43906", + "right": "issue:46468", + "accept": false, + "reason": "An isolated reproduction of an old issue is not the same as a broken Nemotron example; no shared concrete bug is evident." + }, + { + "left": "issue:44625", + "right": "issue:44877", + "accept": false, + "reason": "Different config-loading failures on different models: `num_labels` propagation in Qwen3.5 vs strict config blocking `granite_speech`." + }, + { + "left": "issue:43540", + "right": "issue:46693", + "accept": false, + "reason": "Different multimodal/video-generation failures and different likely code paths; only broad feature similarity." + }, + { + "left": "issue:43452", + "right": "issue:44488", + "accept": false, + "reason": "Both are load failures, but one is about `gguf_file` handling for tokenizer/model loading and the other is a specific model that currently won't load." + }, + { + "left": "issue:45663", + "right": "issue:46846", + "accept": false, + "reason": "Both involve large-model parallelism, but one is an FSDP2 `shared_kv_states` bug and the other is missing `linear_attn` entries in tensor-parallel planning." + }, + { + "left": "issue:43697", + "right": "issue:46829", + "accept": false, + "reason": "Different models and symptoms: RTDetrV2 output drift in v5 vs. VideoPrism returning the wrong tensor in `hidden_states`." + }, + { + "left": "issue:28282", + "right": "issue:32090", + "accept": false, + "reason": "One is a missing PyTorch import environment error; the other is a Trainer broadcast failure on `NoneType`." + }, + { + "left": "issue:44737", + "right": "issue:45588", + "accept": false, + "reason": "XLNet CPU-device placement bug and flash-attention `s_aux=None` crash are unrelated code paths." + }, + { + "left": "issue:46463", + "right": "issue:46899", + "accept": false, + "reason": "T5 decoder-input validation messaging is unrelated to Gemma4 4-bit dtype-cast failures." + }, + { + "left": "issue:44913", + "right": "issue:46121", + "accept": false, + "reason": "GPTNeoX config reload behavior is unrelated to rope-validation JSON list type handling." + }, + { + "left": "issue:45357", + "right": "issue:45920", + "accept": false, + "reason": "Different regressions in different APIs: Qwen3.5 `save_pretrained` vs. broad AutoTokenizer token-ID errors across many models." + }, + { + "left": "issue:43526", + "right": "issue:45479", + "accept": false, + "reason": "`reduce_labels` truncation in BeitImageProcessorFast is not the same as zero-loss behavior for `num_labels=1` classification." + }, + { + "left": "issue:45200", + "right": "issue:46463", + "accept": false, + "reason": "Gemma 4 missing `mm_token_type_ids` defaults is unrelated to T5 decoder-input validation messaging." + }, + { + "left": "issue:43643", + "right": "issue:45375", + "accept": false, + "reason": "Both mention missing fields, but they affect different configs/models and different strictness behaviors." + }, + { + "left": "issue:44493", + "right": "issue:46050", + "accept": false, + "reason": "Unexpected position-id keys and sinq quantization save/load breakage are different issues." + }, + { + "left": "issue:41762", + "right": "issue:44936", + "accept": false, + "reason": "Gemma3 ZeRO-3 loading failure is not the same as `trainer.evaluate()` breaking after `trainer.train()`." + }, + { + "left": "issue:43688", + "right": "issue:43749", + "accept": false, + "reason": "Aux-loss normalization in MoE models and `FSDP_CPU_RAM_EFFICIENT_LOADING` are unrelated." + }, + { + "left": "issue:44361", + "right": "issue:44442", + "accept": false, + "reason": "Different tokenizer-loading failures for different models; same broad symptom, not the same bug." + }, + { + "left": "issue:38175", + "right": "issue:43295", + "accept": false, + "reason": "Siglip2 zero probabilities and the v4.57.5 processor/tokenizer regression are distinct multimodal issues." + }, + { + "left": "issue:43653", + "right": "issue:44361", + "accept": false, + "reason": "BigBird special-token registration/decode bug is unrelated to MLukeTokenizer task failures." + }, + { + "left": "issue:43452", + "right": "issue:44843", + "accept": false, + "reason": "Both touch tokenizer loading, but one is a `gguf_file` loader break and the other is an offline-mode regression from an unconditional hub query." + }, + { + "left": "issue:44466", + "right": "issue:45005", + "accept": false, + "reason": "Both mention tied weights, but one is device-dependent `lm_head.weight` serialization while the other is a broader translation-model tied-weights issue." + }, + { + "left": "issue:45672", + "right": "issue:46899", + "accept": false, + "reason": "Both concern Gemma4 and quantization, but one is audio/vision embedders failing on `torch.finfo(uint8)` and the other is a different unsafe cast path." + }, + { + "left": "issue:45216", + "right": "issue:45357", + "accept": true, + "reason": "Same Qwen3.5 `save_pretrained` regression: the second report is a later, narrower manifestation of the same incorrect checkpoint serialization 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": 26, + "review_unit_index": 25, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:29159", + "issue:33357", + "issue:42898", + "issue:43295", + "issue:43452", + "issue:43525", + "issue:43531", + "issue:43577", + "issue:43646", + "issue:43653", + "issue:43688", + "issue:43742", + "issue:43749", + "issue:43756", + "issue:43761", + "issue:43819", + "issue:43927", + "issue:43994", + "issue:44464", + "issue:44479", + "issue:44568", + "issue:44625", + "issue:44743", + "issue:44811", + "issue:44821", + "issue:44843", + "issue:44857", + "issue:44898", + "issue:44991", + "issue:45356", + "issue:45372", + "issue:45491", + "issue:45542", + "issue:45701", + "issue:45704", + "issue:46002", + "issue:46170", + "issue:46178", + "issue:46402", + "issue:46491", + "issue:46701", + "issue:46821", + "issue:46829", + "issue:46832", + "issue:46932", + "issue:47000", + "issue:47013" + ], + "soft_pairs": [ + "issue:43819|issue:44811", + "issue:43994|issue:44898", + "issue:43761|issue:46829", + "issue:45372|issue:46170", + "issue:44857|issue:47000", + "issue:46832|issue:46932", + "issue:43525|issue:46701", + "issue:43531|issue:43742", + "issue:43295|issue:45356", + "issue:45542|issue:46178", + "issue:29159|issue:33357", + "issue:43688|issue:45704", + "issue:43653|issue:44568", + "issue:43688|issue:45701", + "issue:43646|issue:43749", + "issue:43452|issue:43577", + "issue:43756|issue:46402", + "issue:44464|issue:44743", + "issue:42898|issue:46491", + "issue:44843|issue:44991", + "issue:43761|issue:47013", + "issue:43653|issue:43927", + "issue:44479|issue:44625", + "issue:44821|issue:46821", + "issue:45491|issue:46002" + ], + "prepared_review_unit_hash": "03ee525e486dfac96a03486cd2d2bd1f92fc6ebb1c0def1fd05515a65a6e108e", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 25, + "serialized_chars": 34888, + "estimated_input_tokens": 8722, + "estimated_eval_tokens": 17700 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "All soft links look like false positives: the paired issues usually share only a broad theme (tokenizers, regressions, backend/import errors, or model-specific failures) but not the same concrete bug or code path.", + "confidence": 0.9, + "canonical_issue_reason": "issue:43646 is the broadest, most general regression title in the set, but the cluster is heterogeneous so it is only a weak umbrella candidate.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43646 is the closest thing to a cluster representative because it covers a broad Transformers 5.0 initialization regression, but the set is too mixed for a true duplicate canonical.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43819", + "right": "issue:44811", + "accept": false, + "reason": "Different subsystems and failures: DAC latent reconstruction mismatch vs Whisper batch_decode ignoring skip_special_tokens." + }, + { + "left": "issue:43994", + "right": "issue:44898", + "accept": false, + "reason": "Different models and root causes: SigLIP2 pipeline outputs vs Perceiver non-default resolution interpolation." + }, + { + "left": "issue:43761", + "right": "issue:46829", + "accept": false, + "reason": "Both involve hidden-state outputs, but they are different models with different output bugs and code paths." + }, + { + "left": "issue:45372", + "right": "issue:46170", + "accept": false, + "reason": "Unrelated import failures: Gemma 4 processor loading vs Windows ROCm torch.distributed dependency issue." + }, + { + "left": "issue:44857", + "right": "issue:47000", + "accept": false, + "reason": "Different DETR losses and distinct crash causes: float16 AMP overflow vs infeasible cost matrix." + }, + { + "left": "issue:46832", + "right": "issue:46932", + "accept": false, + "reason": "Different models and problems: RT-DETR feature-level mismatch vs Gemma4 quantized dtype casting." + }, + { + "left": "issue:43525", + "right": "issue:46701", + "accept": false, + "reason": "DiaConfig pad_token_id loss and DiffusionGemma pipeline failure are different bugs in different components." + }, + { + "left": "issue:43531", + "right": "issue:43742", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and MobileLLM-125M load key error are unrelated." + }, + { + "left": "issue:43295", + "right": "issue:45356", + "accept": false, + "reason": "Both are regressions, but one is processor.tokenizer/image handling and the other is Kimi-K2.5 codec/regex handling." + }, + { + "left": "issue:45542", + "right": "issue:46178", + "accept": false, + "reason": "Both touch backend handling, but one is undefined tf backend setup and the other is an invalid tensorflow_text backend entry." + }, + { + "left": "issue:29159", + "right": "issue:33357", + "accept": false, + "reason": "Tokenizer inconsistency vs MacOS bus error on a CLIP community model are not the same bug." + }, + { + "left": "issue:43688", + "right": "issue:45704", + "accept": false, + "reason": "Auxiliary loss normalization and T5 apex.FusedRMSNorm memory leak are unrelated model bugs." + }, + { + "left": "issue:43653", + "right": "issue:44568", + "accept": false, + "reason": "Both concern special tokens, but they affect different tokenizers and different symptoms." + }, + { + "left": "issue:43688", + "right": "issue:45701", + "accept": false, + "reason": "Aux loss scaling and version-dependent tokenization changes do not share a concrete code path." + }, + { + "left": "issue:43646", + "right": "issue:43749", + "accept": false, + "reason": "Custom model initialization regression vs FSDP CPU RAM efficient loading are different features." + }, + { + "left": "issue:43452", + "right": "issue:43577", + "accept": false, + "reason": "GGUF loading breaks and BLIP2 dtype staying float32 are unrelated issues." + }, + { + "left": "issue:43756", + "right": "issue:46402", + "accept": false, + "reason": "SmolLM3 RoPE-layer mismatch and Sam3 text-encoder weight loading are different model-specific bugs." + }, + { + "left": "issue:44464", + "right": "issue:44743", + "accept": false, + "reason": "Chunked generation with compiled forward and xLSTM state unpacking are separate state-handling failures." + }, + { + "left": "issue:42898", + "right": "issue:46491", + "accept": false, + "reason": "Both are tokenizer regressions, but one is clean_up_tokenization_spaces behavior and the other is CodeLlama leading-space round-trip loss." + }, + { + "left": "issue:44843", + "right": "issue:44991", + "accept": false, + "reason": "Offline model_info() patching and est-roberta tokenizer loading failure are different tokenizer-loading regressions." + }, + { + "left": "issue:43761", + "right": "issue:47013", + "accept": false, + "reason": "CLIPVisionModel hidden_states regression and xLSTM return_last_states unpacking bug are unrelated models and code paths." + }, + { + "left": "issue:43653", + "right": "issue:43927", + "accept": false, + "reason": "Tokenizer special-token registration and DiaConfig custom token ID persistence are different bug classes." + }, + { + "left": "issue:44479", + "right": "issue:44625", + "accept": false, + "reason": "Video-input regression in Qwen models vs num_labels propagation in Qwen3.5 are distinct issues." + }, + { + "left": "issue:44821", + "right": "issue:46821", + "accept": false, + "reason": "AutoImageProcessor URL loading and TimesFM window_size AttributeError are unrelated." + }, + { + "left": "issue:45491", + "right": "issue:46002", + "accept": false, + "reason": "Both mention NaNs, but the triggering conditions and affected modules are different." + } + ] + }, + "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": 26, + "review_unit_index": 26, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:28218", + "issue:32090", + "issue:33357", + "issue:33666", + "issue:36010", + "issue:36331", + "issue:41720", + "issue:43262", + "issue:43278", + "issue:43295", + "issue:43493", + "issue:43504", + "issue:43525", + "issue:43550", + "issue:43572", + "issue:43577", + "issue:43606", + "issue:43611", + "issue:43643", + "issue:43688", + "issue:43720", + "issue:43782", + "issue:43825", + "issue:43927", + "issue:44361", + "issue:44464", + "issue:44466", + "issue:44484", + "issue:44514", + "issue:44554", + "issue:44561", + "issue:44610", + "issue:44625", + "issue:44704", + "issue:44734", + "issue:44964", + "issue:44977", + "issue:45290", + "issue:45405", + "issue:45440", + "issue:45584", + "issue:45823", + "issue:46050", + "issue:46209", + "issue:46365", + "issue:46612", + "issue:46693", + "issue:46846" + ], + "soft_pairs": [ + "issue:33666|issue:36010", + "issue:44484|issue:44704", + "issue:32090|issue:33666", + "issue:44466|issue:46050", + "issue:44464|issue:44977", + "issue:43688|issue:45440", + "issue:44977|issue:46693", + "issue:41720|issue:44464", + "issue:44734|issue:46612", + "issue:43504|issue:43550", + "issue:43504|issue:43720", + "issue:43278|issue:43688", + "issue:33666|issue:36331", + "issue:43782|issue:46693", + "issue:43577|issue:44554", + "issue:43611|issue:46846", + "issue:43262|issue:43295", + "issue:43720|issue:43927", + "issue:43606|issue:43927", + "issue:43504|issue:44610", + "issue:44514|issue:45290", + "issue:43643|issue:44561", + "issue:44964|issue:45405", + "issue:43493|issue:43825", + "issue:43720|issue:44361", + "issue:43525|issue:43572", + "issue:44625|issue:45584", + "issue:44554|issue:46365", + "issue:44977|issue:46209", + "issue:45823|issue:46050", + "issue:28218|issue:33357" + ], + "prepared_review_unit_hash": "0fe1ff3493c45be1cbebefdab3623215c715eb922c882ebce8431a39dbfac8b4", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37100, + "estimated_input_tokens": 9275, + "estimated_eval_tokens": 18806 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is heterogeneous: the soft-similarity pairs mostly share broad Transformers themes (generation, processors, configs, loading) but not the same concrete bug. I would not merge any of the soft edges.", + "confidence": 0.86, + "canonical_issue_reason": "issue:46693 is the best representative anchor because it has the highest discussion/inbound activity and is a broad, user-visible generation regression, but the surrounding items are not true duplicates of it.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:46693 is the strongest overall triage anchor for this cluster: it is active, recent, and central enough to represent the general generation/attention-regression theme better than the rest.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:33666", + "right": "issue:36010", + "accept": false, + "reason": "Different failures: multi-GPU training support vs an import error for GenerationMixin." + }, + { + "left": "issue:44484", + "right": "issue:44704", + "accept": false, + "reason": "One is about save_pretrained sharding size, the other about AutoProcessor dropping kwargs to cached_file." + }, + { + "left": "issue:32090", + "right": "issue:33666", + "accept": false, + "reason": "Unsupported NoneType broadcast in Trainer is unrelated to Qwen2-VL multi-GPU training." + }, + { + "left": "issue:44466", + "right": "issue:46050", + "accept": false, + "reason": "Both involve model serialization, but one is tied-weight serialization and the other is sinq quantization reload breakage." + }, + { + "left": "issue:44464", + "right": "issue:44977", + "accept": false, + "reason": "Compiled-forward chunked generation mismatch is a different code path from flash-attention generation failure." + }, + { + "left": "issue:43688", + "right": "issue:45440", + "accept": false, + "reason": "Both concern MoE models, but they describe different implementation discrepancies and likely different fixes." + }, + { + "left": "issue:44977", + "right": "issue:46693", + "accept": false, + "reason": "One is a correctness bug in Qwen3.5 generation with flash-attention, the other is a performance regression in FA2 long-context generation." + }, + { + "left": "issue:41720", + "right": "issue:44464", + "accept": false, + "reason": "cudaErrorAssert on auto device mapping is unrelated to compiled-forward chunked generation inconsistency." + }, + { + "left": "issue:44734", + "right": "issue:46612", + "accept": false, + "reason": "Both are generation/cache bugs, but one is KV-cache continuation indexing in serve and the other is beam-search cache reorder skipping in several models." + }, + { + "left": "issue:43504", + "right": "issue:43550", + "accept": false, + "reason": "Legacy-field load failure for Beit is unrelated to Bamba torch.compile SDPA failure." + }, + { + "left": "issue:43504", + "right": "issue:43720", + "accept": false, + "reason": "Different model-loading bugs: a legacy preset field versus packed weights not being unpacked during accelerate loading." + }, + { + "left": "issue:43278", + "right": "issue:43688", + "accept": false, + "reason": "Embedding dtype drift in evaluation is unrelated to auxiliary-loss normalization in MoE models." + }, + { + "left": "issue:33666", + "right": "issue:36331", + "accept": false, + "reason": "Multi-GPU training for Qwen2-VL is not the same bug as CustomTrainer.compute_loss() rejecting num_items_in_batch." + }, + { + "left": "issue:43782", + "right": "issue:46693", + "accept": false, + "reason": "Qwen3VL weight_only load errors are a model-loading problem, not the FA2 performance regression described here." + }, + { + "left": "issue:43577", + "right": "issue:44554", + "accept": false, + "reason": "Blip2 dtype loading and an MPS attention correctness issue are different models, devices, and failure modes." + }, + { + "left": "issue:43611", + "right": "issue:46846", + "accept": false, + "reason": "Base_model_prefix loading regression is unrelated to missing linear_attn entries in tensor-parallel planning." + }, + { + "left": "issue:43262", + "right": "issue:43295", + "accept": false, + "reason": "Audio sampling-rate defaulting and a v4.57.5 processor.tokenizer regression affect different processor paths." + }, + { + "left": "issue:43720", + "right": "issue:43927", + "accept": false, + "reason": "Accelerate packed-weight loading and lost custom token IDs are separate persistence bugs." + }, + { + "left": "issue:43606", + "right": "issue:43927", + "accept": false, + "reason": "CPU offload device mismatch is unrelated to config token-ID loss on save/load." + }, + { + "left": "issue:43504", + "right": "issue:44610", + "accept": false, + "reason": "Beit legacy-field loading and OmDet processor output-size mismatch are different model families and bugs." + }, + { + "left": "issue:44514", + "right": "issue:45290", + "accept": false, + "reason": "Both hit apply_chat_template, but the crash triggers differ: batched padding=False versus assistant tool-call messages with no content." + }, + { + "left": "issue:43643", + "right": "issue:44561", + "accept": false, + "reason": "trust_remote_code missing fields is unrelated to the removal of is_torch_fx_available." + }, + { + "left": "issue:44964", + "right": "issue:45405", + "accept": false, + "reason": "Phi-4 multimodal loading failure is not the same as an out-of-sync MIN_PEFT_VERSION bump." + }, + { + "left": "issue:43493", + "right": "issue:43825", + "accept": false, + "reason": "SigLIP2 implementation discrepancy and a pipeline error message about translation tasks are unrelated." + }, + { + "left": "issue:43720", + "right": "issue:44361", + "accept": false, + "reason": "Packed-weight loading during accelerate is unrelated to MLukeTokenizer task AttributeError." + }, + { + "left": "issue:43525", + "right": "issue:43572", + "accept": false, + "reason": "Missing pad_token_id in Llama4Config and missing pad_token_idx in StableLmConfig are similar in spirit but distinct model-specific config bugs." + }, + { + "left": "issue:44625", + "right": "issue:45584", + "accept": false, + "reason": "num_labels propagation in Qwen3.5 is unrelated to Whisper empty-transcription failure after align_special_tokens." + }, + { + "left": "issue:44554", + "right": "issue:46365", + "accept": false, + "reason": "MPS attention correctness is unrelated to the Trainer import problem caused by datasets." + }, + { + "left": "issue:44977", + "right": "issue:46209", + "accept": false, + "reason": "Qwen3.5 flash-attention generation issues differ from FP8 inference failure on B200." + }, + { + "left": "issue:45823", + "right": "issue:46050", + "accept": false, + "reason": "Gemma4 device-mapping mismatch is a runtime placement issue, not a save/load quantization regression." + }, + { + "left": "issue:28218", + "right": "issue:33357", + "accept": false, + "reason": "Tokenizer added-space behavior and a macOS bus error in CLIP are completely 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": 26, + "review_unit_index": 27, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:33357", + "issue:33666", + "issue:39692", + "issue:42796", + "issue:43257", + "issue:43278", + "issue:43295", + "issue:43299", + "issue:43334", + "issue:43452", + "issue:43493", + "issue:43504", + "issue:43572", + "issue:43575", + "issue:43606", + "issue:43646", + "issue:43701", + "issue:43824", + "issue:43844", + "issue:43927", + "issue:43994", + "issue:44291", + "issue:44361", + "issue:44451", + "issue:44466", + "issue:44488", + "issue:44493", + "issue:44568", + "issue:44589", + "issue:44792", + "issue:44843", + "issue:44849", + "issue:44977", + "issue:45245", + "issue:45357", + "issue:45405", + "issue:45520", + "issue:45759", + "issue:45812", + "issue:46192", + "issue:46209", + "issue:46285", + "issue:46399", + "issue:46489", + "issue:46710", + "issue:46821", + "issue:46832", + "issue:46846" + ], + "soft_pairs": [ + "issue:43452|issue:44451", + "issue:43994|issue:46192", + "issue:42796|issue:44977", + "issue:39692|issue:43295", + "issue:43493|issue:43844", + "issue:43278|issue:43844", + "issue:43646|issue:44466", + "issue:46209|issue:46710", + "issue:44568|issue:46489", + "issue:43278|issue:43701", + "issue:44849|issue:46209", + "issue:33357|issue:33666", + "issue:44792|issue:45405", + "issue:44568|issue:45812", + "issue:43927|issue:44361", + "issue:43575|issue:45405", + "issue:44488|issue:44843", + "issue:43824|issue:44589", + "issue:44849|issue:45357", + "issue:46821|issue:46832", + "issue:43504|issue:43606", + "issue:45245|issue:46710", + "issue:43334|issue:45759", + "issue:44493|issue:46285", + "issue:43334|issue:43572", + "issue:43572|issue:44291", + "issue:44589|issue:45520", + "issue:43257|issue:43299", + "issue:46399|issue:46846" + ], + "prepared_review_unit_hash": "566aae3d7a194dfaea55229c0841a69c67e985f13c4eb3428a3d3b889852f2bd", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36601, + "estimated_input_tokens": 9151, + "estimated_eval_tokens": 18558 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is not a duplicate cluster: the issues span many unrelated regressions and model families (tokenizers, multimodal loading, serialization, DeepSpeed/TP, docs, and runtime/device bugs). None of the soft pairs look like the same underlying bug or change.", + "confidence": 0.93, + "canonical_issue_reason": "No single canonical issue; the items are heterogeneous and do not describe one shared bug path.", + "canonical_pr_reason": null, + "best_issue_reason": "No best issue for deduplication; none of the issues can represent the rest of the set as a common root cause.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43452", + "right": "issue:44451", + "accept": false, + "reason": "Both are load failures, but one is gguf/from_pretrained handling and the other is a specific model load regression; no shared root cause is evident." + }, + { + "left": "issue:43994", + "right": "issue:46192", + "accept": false, + "reason": "SigLIP2 wrong outputs vs return_tensors being ignored are different API behaviors and code paths." + }, + { + "left": "issue:42796", + "right": "issue:44977", + "accept": false, + "reason": "Tokenizer BPE-marker handling and Qwen3.5 flash-attention generation are unrelated bugs." + }, + { + "left": "issue:39692", + "right": "issue:43295", + "accept": false, + "reason": "Docs example/model-processor mismatch and a processor.tokenizer regression are distinct issues." + }, + { + "left": "issue:43493", + "right": "issue:43844", + "accept": false, + "reason": "SigLIP2 implementation mismatch vs DeepSpeed ZeRO-3 gradient growth are unrelated." + }, + { + "left": "issue:43278", + "right": "issue:43844", + "accept": false, + "reason": "Embedding dtype regression and gradient instability under ZeRO-3 are different failures." + }, + { + "left": "issue:43646", + "right": "issue:44466", + "accept": false, + "reason": "Custom model init breakage and lm_head serialization inconsistency are separate v5 regressions." + }, + { + "left": "issue:46209", + "right": "issue:46710", + "accept": false, + "reason": "FP8 inference failure on B200 and a DeepSeek accuracy regression are not the same bug." + }, + { + "left": "issue:44568", + "right": "issue:46489", + "accept": false, + "reason": "Different tokenizer bugs affecting different models and symptoms." + }, + { + "left": "issue:43278", + "right": "issue:43701", + "accept": false, + "reason": "Embedding dtype change vs resume_from_checkpoint key mismatch are unrelated." + }, + { + "left": "issue:44849", + "right": "issue:46209", + "accept": false, + "reason": "Qwen3.5 hidden-state behavior and FP8/B200 inference are different code paths." + }, + { + "left": "issue:33357", + "right": "issue:33666", + "accept": false, + "reason": "MacOS bus error with CLIP and Qwen2-VL multi-GPU training are unrelated." + }, + { + "left": "issue:44792", + "right": "issue:45405", + "accept": false, + "reason": "A failing Janus test and a PEFT version bump are not the same underlying change." + }, + { + "left": "issue:44568", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different models with different observed failures." + }, + { + "left": "issue:43927", + "right": "issue:44361", + "accept": false, + "reason": "Custom token ID persistence and MLukeTokenizer task errors are distinct bugs." + }, + { + "left": "issue:43575", + "right": "issue:45405", + "accept": false, + "reason": "TP OOM for Qwen2-57B and an unreleased dependency version bump are unrelated." + }, + { + "left": "issue:44488", + "right": "issue:44843", + "accept": false, + "reason": "Model load regression for a specific model and HF_HUB_OFFLINE network access breakage are different causes." + }, + { + "left": "issue:43824", + "right": "issue:44589", + "accept": false, + "reason": "Missing Qwen2.5-VL import and Float8 storage lookup failure are unrelated." + }, + { + "left": "issue:44849", + "right": "issue:45357", + "accept": false, + "reason": "Qwen3.5 hidden-states behavior and incorrect save_pretrained visual keys are different regressions." + }, + { + "left": "issue:46821", + "right": "issue:46832", + "accept": false, + "reason": "TimesFM window_size AttributeError and RT-DETR feature-level mismatch are unrelated." + }, + { + "left": "issue:43504", + "right": "issue:43606", + "accept": false, + "reason": "BEiT legacy-field loading and Bark CPU-offload device mismatch are different issues." + }, + { + "left": "issue:45245", + "right": "issue:46710", + "accept": false, + "reason": "Category-count limit error and DeepSeek accuracy regression are not the same underlying bug." + }, + { + "left": "issue:43334", + "right": "issue:45759", + "accept": false, + "reason": "Qwen3-VL missing pad_token_id and multimodal text_config unwrapping are different loading problems." + }, + { + "left": "issue:44493", + "right": "issue:46285", + "accept": false, + "reason": "Unexpected position_id keys and CLIP text-model structure divergence are unrelated." + }, + { + "left": "issue:43334", + "right": "issue:43572", + "accept": false, + "reason": "Qwen3-VL pad_token_id missing and StableLmConfig pad_token_idx missing are separate config regressions." + }, + { + "left": "issue:43572", + "right": "issue:44291", + "accept": false, + "reason": "Missing config field and init_empty_weights unexpected argument are unrelated version-breakages." + }, + { + "left": "issue:44589", + "right": "issue:45520", + "accept": false, + "reason": "Float8 storage lookup failure and flash_attn import_utils KeyError are different compatibility issues." + }, + { + "left": "issue:43257", + "right": "issue:43299", + "accept": false, + "reason": "Qwen3 MOE weight conversion and Qwen3VL loading failure are distinct bugs." + }, + { + "left": "issue:46399", + "right": "issue:46846", + "accept": false, + "reason": "Missing apply_rotary_pos_emb decorator and missing linear_attn TP plan entries are different Qwen3.5 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": 26, + "review_unit_index": 28, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:15354", + "issue:28218", + "issue:29159", + "issue:29942", + "issue:30333", + "issue:32090", + "issue:34567", + "issue:38175", + "issue:41720", + "issue:42490", + "issue:43065", + "issue:43493", + "issue:43504", + "issue:43606", + "issue:43653", + "issue:43696", + "issue:43701", + "issue:43844", + "issue:43873", + "issue:43937", + "issue:43994", + "issue:44112", + "issue:44423", + "issue:44464", + "issue:44493", + "issue:44554", + "issue:44734", + "issue:44869", + "issue:44964", + "issue:44977", + "issue:45072", + "issue:45084", + "issue:45200", + "issue:45229", + "issue:45406", + "issue:45446", + "issue:45520", + "issue:45588", + "issue:45823", + "issue:46050", + "issue:46121", + "issue:46369", + "issue:46424", + "issue:46701", + "issue:46710", + "issue:46752", + "issue:46832" + ], + "soft_pairs": [ + "issue:43606|issue:44112", + "issue:45072|issue:46832", + "issue:43504|issue:43653", + "issue:44464|issue:46424", + "issue:43937|issue:46701", + "issue:43873|issue:46050", + "issue:43844|issue:44964", + "issue:43994|issue:44493", + "issue:45200|issue:45229", + "issue:44869|issue:46752", + "issue:45588|issue:46050", + "issue:45229|issue:45823", + "issue:29159|issue:32090", + "issue:29942|issue:30333", + "issue:42490|issue:43696", + "issue:44977|issue:46701", + "issue:28218|issue:34567", + "issue:45084|issue:45520", + "issue:15354|issue:38175", + "issue:43065|issue:46369", + "issue:44554|issue:45446", + "issue:43493|issue:43701", + "issue:44423|issue:44734", + "issue:41720|issue:46710", + "issue:45406|issue:46121", + "issue:44977|issue:46710" + ], + "prepared_review_unit_hash": "45d6591e12e3116cbb448db0ac5ad90a74514effc752b4190da2853b63af8813", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 26, + "serialized_chars": 34893, + "estimated_input_tokens": 8724, + "estimated_eval_tokens": 17704 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "All proposed soft edges look like false positives: they span unrelated models, subsystems, and failure modes rather than the same concrete bug or change.", + "confidence": 0.98, + "canonical_issue_reason": "issue:45229 is the most substantial and best-documented issue in the set, with the highest discussion activity and a concrete reproducible failure; the rest of the cluster is too heterogeneous to form a true duplicate group.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45229 is the strongest representative candidate by engagement and specificity, but only as a loose anchor since the surrounding items do not describe the same underlying defect.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43606", + "right": "issue:44112", + "accept": false, + "reason": "Different subsystems and bugs: CPU offload/device mismatch for a model vs a stale CI test failure." + }, + { + "left": "issue:45072", + "right": "issue:46832", + "accept": false, + "reason": "Unrelated model failures with different root causes: dtype mismatch in inference vs RT-DETR feature-level forward validation." + }, + { + "left": "issue:43504", + "right": "issue:43653", + "accept": false, + "reason": "Different components and symptoms: BEiT model loading with a legacy field vs BigBird tokenizer special-token decoding." + }, + { + "left": "issue:44464", + "right": "issue:46424", + "accept": false, + "reason": "Both mention cache/generation, but one is compiled chunked generation inconsistency and the other is a cache-length config limitation; not the same bug." + }, + { + "left": "issue:43937", + "right": "issue:46701", + "accept": false, + "reason": "Different models and code paths: invalid GenerationConfig vs a diffusion/generation output pipeline failure." + }, + { + "left": "issue:43873", + "right": "issue:46050", + "accept": false, + "reason": "Quantization is the shared theme, but offloading behavior and save/load persistence are distinct problems." + }, + { + "left": "issue:43844", + "right": "issue:44964", + "accept": false, + "reason": "Completely different issues: DeepSpeed/ZeRO gradient behavior vs loading a multimodal model." + }, + { + "left": "issue:43994", + "right": "issue:44493", + "accept": false, + "reason": "Different symptoms and likely code paths: SigLIP2 output quality vs a broad unexpected position-id key regression." + }, + { + "left": "issue:45200", + "right": "issue:45229", + "accept": false, + "reason": "Both involve Gemma4, but one is missing/default mm_token_type_ids and the other is a multi-GPU CUDA OOM; not the same defect." + }, + { + "left": "issue:44869", + "right": "issue:46752", + "accept": false, + "reason": "Both are runtime crashes, but in unrelated text-processing paths: Whisper timestamp decoding vs apply_chat_template on empty input." + }, + { + "left": "issue:45588", + "right": "issue:46050", + "accept": false, + "reason": "Different failure classes: flash-attention integration crashes on None aux state vs quantized model save/load breakage." + }, + { + "left": "issue:45229", + "right": "issue:45823", + "accept": false, + "reason": "Same model family, but different concrete bugs: multi-GPU OOM vs device mismatch during auto device mapping." + }, + { + "left": "issue:29159", + "right": "issue:32090", + "accept": false, + "reason": "Tokenizer inconsistency and GPU broadcast NoneType errors are unrelated." + }, + { + "left": "issue:29942", + "right": "issue:30333", + "accept": false, + "reason": "Flash Attention 2 test failures and MLflow job-reporting issues are unrelated." + }, + { + "left": "issue:42490", + "right": "issue:43696", + "accept": false, + "reason": "Loading-path warning and GPT-oss OOM are unrelated." + }, + { + "left": "issue:44977", + "right": "issue:46701", + "accept": false, + "reason": "Different model families and bugs: Qwen3.5 flash-attention generation issue vs DiffusionGemma pipeline failure." + }, + { + "left": "issue:28218", + "right": "issue:34567", + "accept": false, + "reason": "Tokenizer spacing behavior and TrainerState token counting are unrelated." + }, + { + "left": "issue:45084", + "right": "issue:45520", + "accept": false, + "reason": "Different failure areas: template-node compilation vs import-utils flash_attn lookup on Python 3.13." + }, + { + "left": "issue:15354", + "right": "issue:38175", + "accept": false, + "reason": "TorchScript export on ViT and SigLIP2 zero-probability outputs are unrelated." + }, + { + "left": "issue:43065", + "right": "issue:46369", + "accept": false, + "reason": "Same model family, but one is a dummy Conv2d implementation detail and the other is a memory overflow; not clearly the same bug." + }, + { + "left": "issue:44554", + "right": "issue:45446", + "accept": false, + "reason": "Different concerns: MPS attention correctness vs a PyTorch version check for an import." + }, + { + "left": "issue:43493", + "right": "issue:43701", + "accept": false, + "reason": "SigLIP2 implementation discrepancy and resume-from-checkpoint key mismatch are unrelated." + }, + { + "left": "issue:44423", + "right": "issue:44734", + "accept": false, + "reason": "Both are transformers serve crashes, but they hit different endpoints and different code paths; not safely mergeable as one fix." + }, + { + "left": "issue:41720", + "right": "issue:46710", + "accept": false, + "reason": "Different models and symptoms: Qwen3 auto-device-map cuda assert vs DeepSeek accuracy regression after upgrade." + }, + { + "left": "issue:45406", + "right": "issue:46121", + "accept": false, + "reason": "Gemma4Processor missing _tokenizer and rope-parameter type handling are unrelated." + }, + { + "left": "issue:44977", + "right": "issue:46710", + "accept": false, + "reason": "Qwen3.5 flash-attention generation failure and DeepSeek accuracy 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": 26, + "review_unit_index": 29, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:28218", + "issue:29870", + "issue:29942", + "issue:35141", + "issue:37428", + "issue:39692", + "issue:42490", + "issue:42915", + "issue:43232", + "issue:43278", + "issue:43295", + "issue:43450", + "issue:43493", + "issue:43540", + "issue:43550", + "issue:43572", + "issue:43575", + "issue:43824", + "issue:43844", + "issue:43856", + "issue:43873", + "issue:43957", + "issue:43975", + "issue:44186", + "issue:44466", + "issue:44521", + "issue:44554", + "issue:44610", + "issue:44737", + "issue:44849", + "issue:44987", + "issue:45072", + "issue:45127", + "issue:45198", + "issue:45200", + "issue:45290", + "issue:45356", + "issue:45381", + "issue:45520", + "issue:45588", + "issue:45759", + "issue:45823", + "issue:46178", + "issue:46192", + "issue:46209", + "issue:46821", + "issue:46894" + ], + "soft_pairs": [ + "issue:43295|issue:43450", + "issue:37428|issue:43957", + "issue:45072|issue:45198", + "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:43550|issue:43844", + "issue:43450|issue:43575", + "issue:45200|issue:45823", + "issue:29870|issue:29942", + "issue:45520|issue:46178", + "issue:43572|issue:44987", + "issue:37428|issue:43824", + "issue:42490|issue:43493", + "issue:39692|issue:46192", + "issue:43232|issue:44849", + "issue:44554|issue:45356", + "issue:43295|issue:43493", + "issue:44521|issue:45290", + "issue:45381|issue:46209", + "issue:44466|issue:45127", + "issue:28218|issue:35141", + "issue:44737|issue:46894", + "issue:43278|issue:43856", + "issue:42915|issue:43824", + "issue:43824|issue:43975" + ], + "prepared_review_unit_hash": "d83429f495168e60c1d9359c05a0eb950bf5f007b46885f36dda54368711effc", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36337, + "estimated_input_tokens": 9085, + "estimated_eval_tokens": 18426 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These are broad similarity matches, but the paired issues describe different bugs/regressions across unrelated code paths. I would not collapse any of the soft pairs into duplicates, and there is no clear canonical issue for the whole set.", + "confidence": 0.84, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43295", + "right": "issue:43450", + "accept": false, + "reason": "Both are multimodal-related, but one is a custom tokenizer/processor regression and the other is a batched video-shape bug; different failure modes and code paths." + }, + { + "left": "issue:37428", + "right": "issue:43957", + "accept": false, + "reason": "An import error for a missing flash-attention helper is unrelated to meta-device loading failures in model init." + }, + { + "left": "issue:45072", + "right": "issue:45198", + "accept": false, + "reason": "Bfloat16 inference dtype mismatches and Wav2Vec2 save/tokenization failures are unrelated." + }, + { + "left": "issue:45588", + "right": "issue:45759", + "accept": false, + "reason": "A flash-attention sinkless-model crash is not the same as unwrapping text_config for composite multimodal configs." + }, + { + "left": "issue:43540", + "right": "issue:45200", + "accept": false, + "reason": "Video-processing errors in Qwen3OmniMoe are unrelated to Gemma 4's missing mm_token_type_ids default." + }, + { + "left": "issue:43295", + "right": "issue:43873", + "accept": false, + "reason": "Processor/tokenizer regression and quantization offloading failure are different issues despite both touching model loading." + }, + { + "left": "issue:44186", + "right": "issue:46192", + "accept": false, + "reason": "LayoutLMv2 NER/batching crashes are not the same as return_tensors being ignored when text_kwargs is passed." + }, + { + "left": "issue:43572", + "right": "issue:46821", + "accept": false, + "reason": "Missing pad_token_idx in StableLmConfig is unrelated to TimesFM's window_size AttributeError." + }, + { + "left": "issue:43295", + "right": "issue:44610", + "accept": false, + "reason": "These are different processor bugs: one about tokenizer images/custom code, the other about OmDet-Turbo input resolution mismatch." + }, + { + "left": "issue:45823", + "right": "issue:46192", + "accept": false, + "reason": "A device_map auto forward mismatch is unrelated to return_tensors being silently ignored." + }, + { + "left": "issue:43550", + "right": "issue:43844", + "accept": false, + "reason": "torch.compile + SDPA failure in Bamba-9B-v2 is not the same as a DeepSpeed ZeRO-3 gradient growth issue." + }, + { + "left": "issue:43450", + "right": "issue:43575", + "accept": false, + "reason": "Video batch shape handling and tensor-parallel OOM during Qwen2 loading are unrelated." + }, + { + "left": "issue:45200", + "right": "issue:45823", + "accept": false, + "reason": "Defaulting mm_token_type_ids to zeros and a forward-time device mismatch are distinct bugs." + }, + { + "left": "issue:29870", + "right": "issue:29942", + "accept": false, + "reason": "A BART generation local-variable bug and Flash Attention 2 test failures do not describe the same code-path problem." + }, + { + "left": "issue:45520", + "right": "issue:46178", + "accept": false, + "reason": "Both are import/backend issues, but they concern different missing backends and different error mechanisms." + }, + { + "left": "issue:43572", + "right": "issue:44987", + "accept": false, + "reason": "A StableLmConfig field regression is unrelated to a model-loading failure for physical-intelligence/fast." + }, + { + "left": "issue:37428", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but they involve different missing symbols from different modules and likely different fixes." + }, + { + "left": "issue:42490", + "right": "issue:43493", + "accept": false, + "reason": "A trailing-slash local path warning is unrelated to a SigLIP2 implementation discrepancy." + }, + { + "left": "issue:39692", + "right": "issue:46192", + "accept": false, + "reason": "SigLIP2 documentation errors and return_tensors handling are unrelated." + }, + { + "left": "issue:43232", + "right": "issue:44849", + "accept": false, + "reason": "A generation sync_gpus kwargs bug and a Qwen3.5 output_hidden_states bug are different model/runtime issues." + }, + { + "left": "issue:44554", + "right": "issue:45356", + "accept": false, + "reason": "An MPS attention correctness issue is unrelated to a tokenizer codec regression in Kimi-K2.5." + }, + { + "left": "issue:43295", + "right": "issue:43493", + "accept": false, + "reason": "Custom processor/tokenizer regression and SigLIP2 HF-vs-JAX discrepancy are not the same underlying bug." + }, + { + "left": "issue:44521", + "right": "issue:45290", + "accept": false, + "reason": "Both hit apply_chat_template, but one is all-zero assistant masks for multimodal inputs while the other is a crash on tool-call assistant messages." + }, + { + "left": "issue:45381", + "right": "issue:46209", + "accept": false, + "reason": "Wrong vision_position_ids for Qwen2.5-VL video inputs is unrelated to FP8 inference failing on B200." + }, + { + "left": "issue:44466", + "right": "issue:45127", + "accept": false, + "reason": "Inconsistent lm_head.weight serialization and LoRA merge collapse with extended vocab are related to weights, but they are different failure modes and not a single bug." + }, + { + "left": "issue:28218", + "right": "issue:35141", + "accept": false, + "reason": "Tokenizer spacing after added tokens and embedding reinitialization after resize are distinct issues." + }, + { + "left": "issue:44737", + "right": "issue:46894", + "accept": false, + "reason": "CPU placement in XLNet positional encoding is unrelated to CohereASR's double-shift training-loss bug." + }, + { + "left": "issue:43278", + "right": "issue:43856", + "accept": false, + "reason": "BF16-to-FP32 dtype drift during evaluation and MoE training memory inefficiency are not the same defect." + }, + { + "left": "issue:42915", + "right": "issue:43824", + "accept": false, + "reason": "A FineGrainedFP8Config failure in Qwen3Moe is unrelated to an import error for Qwen2_5_VLForConditionalGeneration." + }, + { + "left": "issue:43824", + "right": "issue:43975", + "accept": false, + "reason": "An import failure for a Qwen VL class and DeepSeek detokenization regression are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 30, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:37428", + "issue:38175", + "issue:41720", + "issue:43262", + "issue:43334", + "issue:43335", + "issue:43388", + "issue:43425", + "issue:43450", + "issue:43452", + "issue:43525", + "issue:43526", + "issue:43550", + "issue:43606", + "issue:43653", + "issue:43819", + "issue:43827", + "issue:43901", + "issue:43950", + "issue:44112", + "issue:44206", + "issue:44442", + "issue:44561", + "issue:44610", + "issue:44792", + "issue:44857", + "issue:44898", + "issue:44995", + "issue:45083", + "issue:45198", + "issue:45200", + "issue:45325", + "issue:45412", + "issue:45563", + "issue:45584", + "issue:45593", + "issue:45759", + "issue:46192", + "issue:46209", + "issue:46234", + "issue:46285", + "issue:46424", + "issue:46512", + "issue:46821", + "issue:46858", + "issue:46899", + "issue:46932" + ], + "soft_pairs": [ + "issue:43653|issue:44112", + "issue:43452|issue:46512", + "issue:43950|issue:46285", + "issue:43262|issue:43450", + "issue:43334|issue:43525", + "issue:45200|issue:46899", + "issue:44442|issue:45198", + "issue:45200|issue:46932", + "issue:44792|issue:45563", + "issue:44898|issue:45412", + "issue:43526|issue:46192", + "issue:45200|issue:45325", + "issue:43819|issue:44112", + "issue:43335|issue:43550", + "issue:41720|issue:46209", + "issue:44561|issue:46234", + "issue:43827|issue:43901", + "issue:44610|issue:44857", + "issue:43819|issue:45083", + "issue:43388|issue:43526", + "issue:46512|issue:46821", + "issue:45759|issue:46285", + "issue:44857|issue:45593", + "issue:43606|issue:45083", + "issue:44995|issue:46858", + "issue:44206|issue:45584", + "issue:38175|issue:45412", + "issue:46424|issue:46858", + "issue:43335|issue:43653", + "issue:37428|issue:43425" + ], + "prepared_review_unit_hash": "71f34a51e14a6bd1b474f34ce47a08fc76f79d0c05ab5664828c2b650102200f", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36502, + "estimated_input_tokens": 9126, + "estimated_eval_tokens": 18508 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: the soft pairs mostly share only broad subsystem language (tokenizers, processors, cache, config, multimodal models) but not the same concrete bug. I would not merge any of the proposed soft edges as duplicates.", + "confidence": 0.74, + "canonical_issue_reason": "Issue #43950 is the most representative broad regression report in the set: it describes a core from_pretrained behavior break, has substantial detail, and is more general than the many model-specific edge cases here.", + "canonical_pr_reason": null, + "best_issue_reason": "#43950 is the best single issue to use as a representative canonical reference because it is well-scoped, framework-level, and clearly reproducible.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43653", + "right": "issue:44112", + "accept": false, + "reason": "Different problems: tokenizer special-token registration vs a GraniteSpeech CI device-override test failure." + }, + { + "left": "issue:43452", + "right": "issue:46512", + "accept": false, + "reason": "Unrelated bugs: gguf loading path breakage vs stale config.dtype after from_config(dtype=...)." + }, + { + "left": "issue:43950", + "right": "issue:46285", + "accept": false, + "reason": "Different code paths: non-persistent buffer loading regression vs CLIP text model structural inconsistency." + }, + { + "left": "issue:43262", + "right": "issue:43450", + "accept": false, + "reason": "Audio chat-template sampling-rate default and batched video output shape are unrelated." + }, + { + "left": "issue:43334", + "right": "issue:43525", + "accept": false, + "reason": "Both are missing pad_token_id symptoms, but in different model families and likely different fixes." + }, + { + "left": "issue:45200", + "right": "issue:46899", + "accept": false, + "reason": "Same Gemma4 area, but one is mm_token_type_ids handling and the other is quantized embedder dtype casting." + }, + { + "left": "issue:44442", + "right": "issue:45198", + "accept": false, + "reason": "Different audio/tokenizer failures affecting different model classes and code paths." + }, + { + "left": "issue:45200", + "right": "issue:46932", + "accept": false, + "reason": "Both Gemma4-related, but they describe distinct issues: token-type defaults vs quantized vision/audio dtype casting." + }, + { + "left": "issue:44792", + "right": "issue:45563", + "accept": false, + "reason": "Janus image generation test failure vs stale paged-generate warning; no shared underlying bug." + }, + { + "left": "issue:44898", + "right": "issue:45412", + "accept": false, + "reason": "Perceiver resize/interpolation bug and RT-DETR memory release issue are unrelated." + }, + { + "left": "issue:43526", + "right": "issue:46192", + "accept": false, + "reason": "Beit fast image label reduction vs text_kwargs/return_tensors precedence are different behaviors." + }, + { + "left": "issue:45200", + "right": "issue:45325", + "accept": false, + "reason": "Different multimodal bugs: default mm_token_type_ids vs Qwen2.5-VL rope position scaling." + }, + { + "left": "issue:43819", + "right": "issue:44112", + "accept": false, + "reason": "DAC latent reconstruction mismatch is unrelated to the GraniteSpeech CI failure." + }, + { + "left": "issue:43335", + "right": "issue:43550", + "accept": false, + "reason": "SwitchTransformers sparse-layer construction and Bamba torch.compile SDPA failure are different model bugs." + }, + { + "left": "issue:41720", + "right": "issue:46209", + "accept": false, + "reason": "Different hardware/runtime failures: A800 cudaErrorAssert vs B200 FP8 inference incompatibility." + }, + { + "left": "issue:44561", + "right": "issue:46234", + "accept": false, + "reason": "Both are missing import-utils symbols, but they are different helpers affecting different model families." + }, + { + "left": "issue:43827", + "right": "issue:43901", + "accept": false, + "reason": "Both are docs issues, but they concern different pipeline APIs and different deprecated mentions." + }, + { + "left": "issue:44610", + "right": "issue:44857", + "accept": false, + "reason": "Different detection/image-processing failures: wrong input size vs AMP crash in a loss function." + }, + { + "left": "issue:43819", + "right": "issue:45083", + "accept": false, + "reason": "DAC latent forward mismatch and qwen3_omni_moe feature-length helper behavior are unrelated." + }, + { + "left": "issue:43388", + "right": "issue:43526", + "accept": false, + "reason": "Both mention labels, but one is metrics gathering on tuples and the other is BeitImageProcessorFast reduce_labels." + }, + { + "left": "issue:46512", + "right": "issue:46821", + "accept": false, + "reason": "Stale dtype metadata after from_config and TimesFM window_size AttributeError are unrelated." + }, + { + "left": "issue:45759", + "right": "issue:46285", + "accept": false, + "reason": "Composite Qwen causal-LM config unwrapping and CLIP text-model structure divergence are different issues." + }, + { + "left": "issue:44857", + "right": "issue:45593", + "accept": false, + "reason": "AMP crash in LwDetrImageLoss and missing auxiliary losses in D-FINE are separate detection bugs." + }, + { + "left": "issue:43606", + "right": "issue:45083", + "accept": false, + "reason": "CPU offload device mismatch and qwen3_omni_moe feature-length helper behavior do not match." + }, + { + "left": "issue:44995", + "right": "issue:46858", + "accept": false, + "reason": "Stale indexer cache crash in one model and static-cache sequence-length tensor arithmetic in prefill are different cache bugs." + }, + { + "left": "issue:44206", + "right": "issue:45584", + "accept": false, + "reason": "Unsupported feature-extractor arg and Whisper empty-transcription failure are unrelated." + }, + { + "left": "issue:38175", + "right": "issue:45412", + "accept": false, + "reason": "Zero probabilities in SigLIP2 and RT-DETR memory not released are unrelated." + }, + { + "left": "issue:46424", + "right": "issue:46858", + "accept": false, + "reason": "Both are cache-related, but one is max_cache_length semantics and the other is device-tensor sequence length in prefill." + }, + { + "left": "issue:43335", + "right": "issue:43653", + "accept": false, + "reason": "SwitchTransformers sparse-layer construction and BigBirdTokenizer special-token registration are unrelated." + }, + { + "left": "issue:37428", + "right": "issue:43425", + "accept": false, + "reason": "Flash-attention import error and Torch 2.10 incompatibility 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": 26, + "review_unit_index": 31, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:28218", + "issue:28282", + "issue:29127", + "issue:29942", + "issue:32090", + "issue:33357", + "issue:34689", + "issue:35141", + "issue:37428", + "issue:39692", + "issue:42490", + "issue:43278", + "issue:43335", + "issue:43504", + "issue:43526", + "issue:43550", + "issue:43606", + "issue:43611", + "issue:43720", + "issue:43819", + "issue:43825", + "issue:43854", + "issue:44112", + "issue:44351", + "issue:44448", + "issue:44466", + "issue:44493", + "issue:44514", + "issue:44521", + "issue:44743", + "issue:44779", + "issue:44855", + "issue:44857", + "issue:45083", + "issue:45290", + "issue:45310", + "issue:45459", + "issue:45478", + "issue:45520", + "issue:45588", + "issue:45593", + "issue:45859", + "issue:46170", + "issue:46234", + "issue:46326", + "issue:46431", + "issue:46566" + ], + "soft_pairs": [ + "issue:28282|issue:33357", + "issue:29942|issue:32090", + "issue:43611|issue:44466", + "issue:43550|issue:44112", + "issue:43335|issue:43504", + "issue:44855|issue:45083", + "issue:45520|issue:46170", + "issue:44448|issue:44743", + "issue:43335|issue:45083", + "issue:45290|issue:46326", + "issue:37428|issue:44351", + "issue:44514|issue:44521", + "issue:43335|issue:45593", + "issue:43335|issue:44112", + "issue:43278|issue:43825", + "issue:29127|issue:32090", + "issue:45588|issue:45859", + "issue:42490|issue:43854", + "issue:34689|issue:35141", + "issue:43819|issue:44493", + "issue:43720|issue:44112", + "issue:39692|issue:43526", + "issue:46431|issue:46566", + "issue:43606|issue:44857", + "issue:44448|issue:44779", + "issue:39692|issue:44493", + "issue:28218|issue:32090", + "issue:45310|issue:45478", + "issue:45459|issue:46234" + ], + "prepared_review_unit_hash": "34ee9b3beb67d6d33ea58622f8813e517960a5e9feb1ec3bbf25317239c5a194", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 36139, + "estimated_input_tokens": 9035, + "estimated_eval_tokens": 18326 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous; only issue 45310 and 45478 are a true duplicate pair (same Qwen3.5 MoE from_pretrained regression). All other soft links look like related-in-theme but different bugs.", + "confidence": 0.94, + "canonical_issue_reason": "45310 is the best canonical anchor because it is an exact duplicate of 45478 with the same title and failure mode, but the surrounding cluster is too mixed for it to represent the whole set strongly.", + "canonical_pr_reason": null, + "best_issue_reason": "45310 is the best issue overall only for the confirmed duplicate pair; it is not a good global representative because most other items are unrelated regressions.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:28282", + "right": "issue:33357", + "accept": false, + "reason": "Different failures: missing PyTorch import vs a MacOS bus error in a CLIP model." + }, + { + "left": "issue:29942", + "right": "issue:32090", + "accept": false, + "reason": "Flash Attention 2 test failures and Trainer `_gpu_broadcast_one` NoneType errors are unrelated code paths." + }, + { + "left": "issue:43611", + "right": "issue:44466", + "accept": false, + "reason": "Both are loading-related, but one is a `base_model_prefix` regression and the other is tied-weight serialization behavior." + }, + { + "left": "issue:43550", + "right": "issue:44112", + "accept": false, + "reason": "Torch.compile/SDPA failure in Bamba is not the same bug as a stale GraniteSpeech device-override CI test." + }, + { + "left": "issue:43335", + "right": "issue:43504", + "accept": false, + "reason": "Config incorrectly creating sparse layers and loading a legacy field are different model bugs." + }, + { + "left": "issue:44855", + "right": "issue:45083", + "accept": false, + "reason": "Python 3.13 import parsing failure and qwen3_omni_moe feature-length behavior are unrelated." + }, + { + "left": "issue:45520", + "right": "issue:46170", + "accept": false, + "reason": "Different platform/dependency issues: flash_attn import key error vs Windows ROCm torch.distributed dependency problems." + }, + { + "left": "issue:44448", + "right": "issue:44743", + "accept": false, + "reason": "Pegasus v4/v5 output drift and Qwen modular recurrent-state reset are different underlying regressions." + }, + { + "left": "issue:43335", + "right": "issue:45083", + "accept": false, + "reason": "SwitchTransformers sparse-layer creation and qwen3_omni_moe length helper behavior do not share the same bug." + }, + { + "left": "issue:45290", + "right": "issue:46326", + "accept": false, + "reason": "Both involve chat templates, but one crashes on tool-call assistant messages and the other concerns `continue_final_message` for mistral_common." + }, + { + "left": "issue:37428", + "right": "issue:44351", + "accept": false, + "reason": "Missing `_flash_supports_window_size` and missing `HybridCache` are distinct import errors." + }, + { + "left": "issue:44514", + "right": "issue:44521", + "accept": false, + "reason": "Same area, but one is a batched-input crash and the other is an all-zero assistant_mask bug; not the same concrete defect." + }, + { + "left": "issue:43335", + "right": "issue:45593", + "accept": false, + "reason": "SwitchTransformers sparse-layer config and D-FINE auxiliary-loss behavior are unrelated." + }, + { + "left": "issue:43335", + "right": "issue:44112", + "accept": false, + "reason": "These are different model/CI issues with no shared code path." + }, + { + "left": "issue:43278", + "right": "issue:43825", + "accept": false, + "reason": "Embedding dtype regression and pipeline translation error-message regression are unrelated." + }, + { + "left": "issue:29127", + "right": "issue:32090", + "accept": false, + "reason": "LayoutLMv3 box-info messaging and Trainer GPU broadcast errors are different bugs." + }, + { + "left": "issue:45588", + "right": "issue:45859", + "accept": false, + "reason": "Flash-attention sink-less `s_aux=None` crash and Qwen3_5MoeTextRotaryEmbedding CPU-offload incompatibility are separate failures." + }, + { + "left": "issue:42490", + "right": "issue:43854", + "accept": false, + "reason": "Local-path import warning and GLM-4.7-Flash test loading failure are not the same issue." + }, + { + "left": "issue:34689", + "right": "issue:35141", + "accept": false, + "reason": "Llama 3.2 vision model loading breakage and embedding reinitialization after resize are different regressions." + }, + { + "left": "issue:43819", + "right": "issue:44493", + "accept": false, + "reason": "DAC latent/forward mismatch and unexpected position-id keys are unrelated bugs." + }, + { + "left": "issue:43720", + "right": "issue:44112", + "accept": false, + "reason": "Packed-weight unloading under accelerate and a GraniteSpeech stale-device CI test are different problems." + }, + { + "left": "issue:39692", + "right": "issue:43526", + "accept": false, + "reason": "Docs example mistakes and BeitImageProcessorFast label-reduction behavior are not the same underlying bug." + }, + { + "left": "issue:46431", + "right": "issue:46566", + "accept": false, + "reason": "Missing chat template error and DiffusionGemma meta-device loading are unrelated." + }, + { + "left": "issue:43606", + "right": "issue:44857", + "accept": false, + "reason": "CPU offload device mismatch in bark-small and float16 AMP crashes in LwDetrImageLoss are different failures." + }, + { + "left": "issue:44448", + "right": "issue:44779", + "accept": false, + "reason": "Pegasus output differences and Deepseek tokenizer regressions are separate model-specific issues." + }, + { + "left": "issue:39692", + "right": "issue:44493", + "accept": false, + "reason": "SigLIP2 docs example errors and unexpected position-id keys have no shared root cause." + }, + { + "left": "issue:28218", + "right": "issue:32090", + "accept": false, + "reason": "Tokenizer extra-space behavior and Trainer broadcast NoneType errors are unrelated." + }, + { + "left": "issue:45310", + "right": "issue:45478", + "accept": true, + "reason": "Exact duplicate titles and same Qwen3.5 MoE from_pretrained failure on transformers>=5.4.0." + }, + { + "left": "issue:45459", + "right": "issue:46234", + "accept": false, + "reason": "Tokenizer error masking when protobuf is missing and a missing `is_falcon_mamba_ssm_available` helper are different import/tokenizer issues." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: the only clearly duplicate-looking pair is 45310/45478, which share the exact title and same Qwen3.5 MoE from_pretrained regression. The remaining links are appropriately described as theme-level similarities rather than duplicates, and the soft-edge rejections stay conservative." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 32, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:10000", + "issue:15354", + "issue:26413", + "issue:27957", + "issue:29870", + "issue:29942", + "issue:31501", + "issue:32090", + "issue:33290", + "issue:33357", + "issue:34689", + "issue:35141", + "issue:38175", + "issue:40444", + "issue:42490", + "issue:43278", + "issue:43329", + "issue:43381", + "issue:43526", + "issue:43550", + "issue:43575", + "issue:43701", + "issue:43927", + "issue:44206", + "issue:44360", + "issue:44448", + "issue:44493", + "issue:44561", + "issue:44617", + "issue:44855", + "issue:44857", + "issue:44977", + "issue:45072", + "issue:45083", + "issue:45127", + "issue:45229", + "issue:45245", + "issue:45290", + "issue:45381", + "issue:45593", + "issue:45698", + "issue:45823", + "issue:45920", + "issue:46192", + "issue:46399", + "issue:46655", + "issue:46697", + "issue:47000" + ], + "soft_pairs": [ + "issue:26413|issue:31501", + "issue:44617|issue:46697", + "issue:45127|issue:45245", + "issue:43701|issue:45083", + "issue:44206|issue:45823", + "issue:32090|issue:33357", + "issue:44857|issue:45072", + "issue:42490|issue:43575", + "issue:45593|issue:47000", + "issue:43550|issue:45083", + "issue:27957|issue:34689", + "issue:38175|issue:44493", + "issue:29942|issue:35141", + "issue:29870|issue:32090", + "issue:43278|issue:43329", + "issue:45290|issue:45381", + "issue:10000|issue:15354", + "issue:44448|issue:44977", + "issue:45920|issue:46192", + "issue:44360|issue:44561", + "issue:29870|issue:33290", + "issue:45290|issue:46655", + "issue:44855|issue:46399", + "issue:43278|issue:43381", + "issue:44977|issue:45229", + "issue:40444|issue:43526", + "issue:43927|issue:45698" + ], + "prepared_review_unit_hash": "9ef6d98eb75310625d98c2d3d2337e276c133d5155a5f08419ad19da7ac4befd", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35831, + "estimated_input_tokens": 8958, + "estimated_eval_tokens": 18172 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a grab bag of unrelated closed issues spanning tokenization, generation, multimodal processing, training, and platform-specific crashes. None of the soft pairs look like the same underlying bug or change, so no soft edge should be merged.", + "confidence": 0.84, + "canonical_issue_reason": "issue:45920 is the strongest canonical issue candidate because it describes a broad v5 regression affecting multiple model families, making it more representative and reusable than the single-model, single-path reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45920 is the best issue to anchor this cluster: it is high-signal, broadly scoped, and clearly framed as a regression impacting several models rather than a narrow edge case.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:26413", + "right": "issue:31501", + "accept": false, + "reason": "Both mention training startup problems, but one is a resume/epoch_iterator failure and the other is FineWeb SLM not starting; the underlying causes and code paths are different." + }, + { + "left": "issue:44617", + "right": "issue:46697", + "accept": false, + "reason": "One is a CUDA OOM in Sam3Video, the other is over-padding in Swin2SR image processing; different models and failure modes." + }, + { + "left": "issue:45127", + "right": "issue:45245", + "accept": false, + "reason": "LoRA/vocab merge collapse and a category-count limit error are unrelated bugs with no shared concrete code path." + }, + { + "left": "issue:43701", + "right": "issue:45083", + "accept": false, + "reason": "Resume/checkpoint key mismatch is unrelated to qwen3_omni_moe feature-length calculation behavior." + }, + { + "left": "issue:44206", + "right": "issue:45823", + "accept": false, + "reason": "A feature extractor regression and a device-map forward mismatch are distinct issues affecting different subsystems." + }, + { + "left": "issue:32090", + "right": "issue:33357", + "accept": false, + "reason": "A Trainer broadcast NoneType error and a MacOS bus error in CLIP are not the same bug." + }, + { + "left": "issue:44857", + "right": "issue:45072", + "accept": false, + "reason": "Both involve dtype/AMP wording, but they are separate model-specific crashes with different execution paths and symptoms." + }, + { + "left": "issue:42490", + "right": "issue:43575", + "accept": false, + "reason": "A trailing-slash local-path import warning is unrelated to tensor-parallel OOM when loading Qwen2-57B-A14B-Instruct." + }, + { + "left": "issue:45593", + "right": "issue:47000", + "accept": false, + "reason": "Both are detection-loss related, but denoising/aux-loss behavior and an infeasible cost matrix are different defects." + }, + { + "left": "issue:43550", + "right": "issue:45083", + "accept": false, + "reason": "A torch.compile/SDPA failure in Bamba-9B-v2 is unrelated to qwen3_omni_moe feature-length helper behavior." + }, + { + "left": "issue:27957", + "right": "issue:34689", + "accept": false, + "reason": "Flash Attention 2 support issues for XLMRoberta and a Llama 3.2 Vision loading regression are different model-path problems." + }, + { + "left": "issue:38175", + "right": "issue:44493", + "accept": false, + "reason": "Unexpected zero probabilities in SigLIP2 and unexpected position-id keys are not the same underlying bug." + }, + { + "left": "issue:29942", + "right": "issue:35141", + "accept": false, + "reason": "Failing Flash Attention 2 tests and output embedding reinitialization after resizing vocab are unrelated changes." + }, + { + "left": "issue:29870", + "right": "issue:32090", + "accept": false, + "reason": "An UnboundLocalError in BART generation and a NoneType GPU broadcast error occur in different parts of the stack." + }, + { + "left": "issue:43278", + "right": "issue:43329", + "accept": false, + "reason": "A dtype drift between train/eval and an undefined-function bug in a multimodal token counter are unrelated." + }, + { + "left": "issue:45290", + "right": "issue:45381", + "accept": false, + "reason": "Chat-template tokenization with tool calls and qwen2.5-vl video vision_position_ids are separate multimodal bugs." + }, + { + "left": "issue:10000", + "right": "issue:15354", + "accept": false, + "reason": "German DistilBertModel and torch.jit.script GeneratorExp support are different historical issues with no shared code path." + }, + { + "left": "issue:44448", + "right": "issue:44977", + "accept": false, + "reason": "A cross-version Pegasus output difference and a Qwen3.5 flash-attention generation bug are not the same issue." + }, + { + "left": "issue:45920", + "right": "issue:46192", + "accept": false, + "reason": "Wrong token IDs in AutoTokenizer and return_tensors being ignored are distinct tokenizer API bugs." + }, + { + "left": "issue:44360", + "right": "issue:44561", + "accept": false, + "reason": "A DSA indexer ReLU discussion and removal of is_torch_fx_available breaking trust_remote_code models are unrelated." + }, + { + "left": "issue:29870", + "right": "issue:33290", + "accept": false, + "reason": "BART generate-time masking error and deepspeed adafactor OOM are different failure classes." + }, + { + "left": "issue:45290", + "right": "issue:46655", + "accept": false, + "reason": "Assistant tool-call chat templating and SmolVLM video padding are both multimodal but not the same concrete bug." + }, + { + "left": "issue:44855", + "right": "issue:46399", + "accept": false, + "reason": "Python 3.13 import parsing failure in DebertaV2Model and a missing decorator crash in Qwen3.5 are different defects." + }, + { + "left": "issue:43278", + "right": "issue:43381", + "accept": false, + "reason": "Embedding dtype mismatch in evaluation is unrelated to the inability to use gradient checkpointing in eval mode." + }, + { + "left": "issue:44977", + "right": "issue:45229", + "accept": false, + "reason": "Flash-attention generation issues for Qwen3.5 and multi-GPU OOM for Gemma4 are different problems." + }, + { + "left": "issue:40444", + "right": "issue:43526", + "accept": false, + "reason": "IterableDataset multi-image finetuning failure and BeitImageProcessorFast reduce_labels behavior are unrelated." + }, + { + "left": "issue:43927", + "right": "issue:45698", + "accept": false, + "reason": "Losing custom token IDs after save/load and loading the wrong custom module after save_pretrained are both persistence-related, but they are different bugs and fixes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 33, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:25251", + "issue:29127", + "issue:29870", + "issue:29942", + "issue:30064", + "issue:32090", + "issue:33290", + "issue:33357", + "issue:34567", + "issue:34689", + "issue:37428", + "issue:39692", + "issue:41720", + "issue:42175", + "issue:42490", + "issue:42915", + "issue:42994", + "issue:43278", + "issue:43381", + "issue:43575", + "issue:43611", + "issue:43653", + "issue:43701", + "issue:43873", + "issue:43976", + "issue:44186", + "issue:44206", + "issue:44361", + "issue:44448", + "issue:44466", + "issue:44479", + "issue:44561", + "issue:44568", + "issue:44743", + "issue:44779", + "issue:44871", + "issue:44977", + "issue:45198", + "issue:45229", + "issue:45440", + "issue:45491", + "issue:45563", + "issue:45588", + "issue:45803", + "issue:45823", + "issue:46121", + "issue:46421", + "issue:46858" + ], + "soft_pairs": [ + "issue:37428|issue:42175", + "issue:32090|issue:33290", + "issue:46421|issue:46858", + "issue:43873|issue:45229", + "issue:42490|issue:43873", + "issue:25251|issue:32090", + "issue:43976|issue:45440", + "issue:44448|issue:44466", + "issue:29127|issue:34567", + "issue:33357|issue:39692", + "issue:43381|issue:43701", + "issue:44568|issue:45491", + "issue:44561|issue:44779", + "issue:44871|issue:44977", + "issue:43278|issue:43575", + "issue:45588|issue:45823", + "issue:45198|issue:45803", + "issue:44977|issue:45563", + "issue:41720|issue:45229", + "issue:44186|issue:45198", + "issue:44361|issue:45198", + "issue:43611|issue:46121", + "issue:29870|issue:34689", + "issue:43653|issue:44743", + "issue:44206|issue:44479", + "issue:29942|issue:30064", + "issue:42915|issue:42994" + ], + "prepared_review_unit_hash": "a24b4bc627e0d5181291b8c3b49becf6fd9d73df05a747f54f6e33c7bdf48afa", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35932, + "estimated_input_tokens": 8983, + "estimated_eval_tokens": 18222 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No duplicate cluster here: the issues span unrelated bugs across generation, tokenization, training, quantization, cache logic, docs, and packaging. All soft edges look like similarity-only false positives.", + "confidence": 0.96, + "canonical_issue_reason": "No canonical issue: this cluster is heterogeneous and does not share one underlying bug or change.", + "canonical_pr_reason": null, + "best_issue_reason": "No strong global best issue; none of these issues safely represents the rest as a duplicate target.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:37428", + "right": "issue:42175", + "accept": false, + "reason": "Both are import/backend-related, but one is a missing flash-attention helper and the other is a pip extra omitting TensorFlow; different failures." + }, + { + "left": "issue:32090", + "right": "issue:33290", + "accept": false, + "reason": "Trainer broadcast NoneType vs Adafactor/DeepSpeed OOM are different code paths and error modes." + }, + { + "left": "issue:46421", + "right": "issue:46858", + "accept": false, + "reason": "Both touch cache behavior, but one is lazy cache init causing recompilation and the other is device-tensor seq-length arithmetic on non-CUDA backends." + }, + { + "left": "issue:43873", + "right": "issue:45229", + "accept": false, + "reason": "Offloading/quantization behavior is unrelated to Gemma4 multi-GPU CUDA OOM." + }, + { + "left": "issue:42490", + "right": "issue:43873", + "accept": false, + "reason": "A trailing-slash import warning is not the same bug as quantization offloading malfunction." + }, + { + "left": "issue:25251", + "right": "issue:32090", + "accept": false, + "reason": "Pipeline top_k nesting changes output shape; the other is a distributed trainer broadcast TypeError." + }, + { + "left": "issue:43976", + "right": "issue:45440", + "accept": false, + "reason": "Python version compatibility and DeepSeekV3MoE implementation divergence are unrelated." + }, + { + "left": "issue:44448", + "right": "issue:44466", + "accept": false, + "reason": "Both involve v5-vs-v4 differences, but one is generation output drift and the other is serialization of tied weights." + }, + { + "left": "issue:29127", + "right": "issue:34567", + "accept": false, + "reason": "LayoutLMv3 error messaging for boxes and TrainerState token counting are unrelated." + }, + { + "left": "issue:33357", + "right": "issue:39692", + "accept": false, + "reason": "MacOS bus error on CLIP load and SigLIP2 doc/model mismatch are different issues despite both involving vision models." + }, + { + "left": "issue:43381", + "right": "issue:43701", + "accept": false, + "reason": "Gradient checkpointing in eval mode is not the same as a resume_from_checkpoint key mismatch." + }, + { + "left": "issue:44568", + "right": "issue:45491", + "accept": false, + "reason": "Tokenizer special-token insertion and Gemma3 NaN embeddings come from different subsystems and failure mechanisms." + }, + { + "left": "issue:44561", + "right": "issue:44779", + "accept": false, + "reason": "Removal of an internal helper breaks trust_remote_code models; Deepseek tokenizer regression is a separate tokenizer bug." + }, + { + "left": "issue:44871", + "right": "issue:44977", + "accept": false, + "reason": "Gemma eos_token_id config inconsistency and Qwen3.5 flash-attention generation failure are not the same underlying bug." + }, + { + "left": "issue:43278", + "right": "issue:43575", + "accept": false, + "reason": "Embedding dtype drift in eval is unrelated to Qwen2-57B-A14B-Instruct tensor-parallel OOM." + }, + { + "left": "issue:45588", + "right": "issue:45823", + "accept": false, + "reason": "A flash_attention.py None handling crash and a Gemma4 PLE device mismatch are different code-path problems." + }, + { + "left": "issue:45198", + "right": "issue:45803", + "accept": false, + "reason": "Wav2Vec2 save/tokenization failure and FuyuBatchFeature's bare except swallowing errors are unrelated." + }, + { + "left": "issue:44977", + "right": "issue:45563", + "accept": false, + "reason": "A generation regression and a stale warning in paged generate() are distinct issues." + }, + { + "left": "issue:41720", + "right": "issue:45229", + "accept": false, + "reason": "Both concern large-model inference, but Qwen3 auto device mapping cuda asserts and Gemma4 multi-GPU OOM are different bugs." + }, + { + "left": "issue:44186", + "right": "issue:45198", + "accept": false, + "reason": "LayoutLMv2 tokenizer crashes on NER/padding and Wav2Vec2 save/tokenization fail for different reasons." + }, + { + "left": "issue:44361", + "right": "issue:45198", + "accept": false, + "reason": "MLukeTokenizer AttributeError is not the same as Wav2Vec2 serialization/tokenization failure." + }, + { + "left": "issue:43611", + "right": "issue:46121", + "accept": false, + "reason": "Model loading with base_model_prefix and rope validation TypeError are separate configuration/loading issues." + }, + { + "left": "issue:29870", + "right": "issue:34689", + "accept": false, + "reason": "BART generate() decoder mask bug and Llama 3.2 Vision loading breakage are unrelated regressions." + }, + { + "left": "issue:43653", + "right": "issue:44743", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and Qwen3.5 recurrent-state reset with cache are different bugs." + }, + { + "left": "issue:44206", + "right": "issue:44479", + "accept": false, + "reason": "Unsupported center arg in a feature extractor and a video-input regression in Qwen models are not the same issue." + }, + { + "left": "issue:29942", + "right": "issue:30064", + "accept": false, + "reason": "Failing Flash Attention 2 tests and void segmentation-map processing are unrelated." + }, + { + "left": "issue:42915", + "right": "issue:42994", + "accept": false, + "reason": "Both mention quantization, but one is a FineGrainedFP8Config model failure and the other is quantized model saving." + } + ] + }, + "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": 26, + "review_unit_index": 34, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:27957", + "issue:29127", + "issue:33290", + "issue:33357", + "issue:34689", + "issue:35141", + "issue:42490", + "issue:42915", + "issue:43065", + "issue:43066", + "issue:43381", + "issue:43412", + "issue:43421", + "issue:43526", + "issue:43575", + "issue:43653", + "issue:43701", + "issue:43756", + "issue:43824", + "issue:43931", + "issue:43994", + "issue:44186", + "issue:44265", + "issue:44279", + "issue:44387", + "issue:44442", + "issue:44462", + "issue:44464", + "issue:44857", + "issue:45072", + "issue:45198", + "issue:45245", + "issue:45372", + "issue:45406", + "issue:45412", + "issue:45440", + "issue:45459", + "issue:45491", + "issue:45563", + "issue:45584", + "issue:45588", + "issue:45593", + "issue:45712", + "issue:45834", + "issue:46369", + "issue:46693", + "issue:46701" + ], + "soft_pairs": [ + "issue:43756|issue:45440", + "issue:45372|issue:45588", + "issue:43653|issue:44442", + "issue:43065|issue:43994", + "issue:43412|issue:45588", + "issue:44265|issue:45198", + "issue:45459|issue:45712", + "issue:45491|issue:46693", + "issue:29127|issue:33357", + "issue:43065|issue:45412", + "issue:43575|issue:43701", + "issue:45245|issue:45563", + "issue:45372|issue:45406", + "issue:42490|issue:43381", + "issue:29127|issue:35141", + "issue:45440|issue:45593", + "issue:43994|issue:45412", + "issue:43421|issue:45491", + "issue:45372|issue:45459", + "issue:45406|issue:46701", + "issue:45412|issue:46369", + "issue:43575|issue:44387", + "issue:43066|issue:44462", + "issue:44442|issue:45072", + "issue:44464|issue:45584", + "issue:43526|issue:45834", + "issue:43824|issue:44279", + "issue:43526|issue:46693", + "issue:42915|issue:43931", + "issue:27957|issue:33290", + "issue:27957|issue:29127", + "issue:44857|issue:45588", + "issue:43994|issue:46369", + "issue:33290|issue:34689", + "issue:33290|issue:35141", + "issue:43526|issue:44186" + ], + "prepared_review_unit_hash": "2cf55ff4837b7c0abb74d86dcaa7823cf97219c2cd01cdcb1ef37564de248f7b", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 36, + "serialized_chars": 37825, + "estimated_input_tokens": 9457, + "estimated_eval_tokens": 19170 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These soft pairs are not true duplicates: they span unrelated issues across tokenizers, model loading, memory/OOM, export, and model-specific regressions. The similarity is mostly lexical or subsystem-level, not the same concrete bug.", + "confidence": 0.98, + "canonical_issue_reason": "There is no single canonical issue for this cluster because the items describe different bugs and features rather than one underlying defect.", + "canonical_pr_reason": null, + "best_issue_reason": "If an anchor is needed, issue:44462 is the most generic user-facing loading bug, but it still does not represent the cluster as a whole.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43756", + "right": "issue:45440", + "accept": false, + "reason": "Different models and failure modes: RoPE-layer count vs MoE divergence." + }, + { + "left": "issue:45372", + "right": "issue:45588", + "accept": false, + "reason": "Gemma processor import error vs flash-attention auxiliary-state crash." + }, + { + "left": "issue:43653", + "right": "issue:44442", + "accept": false, + "reason": "BigBird special-token registration bug vs tokenizer class loading failure." + }, + { + "left": "issue:43065", + "right": "issue:43994", + "accept": false, + "reason": "Dummy Conv2d in Sam3 decoder vs SigLIP2 inference corruption." + }, + { + "left": "issue:43412", + "right": "issue:45588", + "accept": false, + "reason": "ExecuTorch export guard/erfinv issue vs flash-attention None-handling bug." + }, + { + "left": "issue:44265", + "right": "issue:45198", + "accept": false, + "reason": "torch.export failure with torch_compilable_check vs Wav2Vec2 save/tokenization bug." + }, + { + "left": "issue:45459", + "right": "issue:45712", + "accept": false, + "reason": "Tokenizer error masking vs leftover dummy class cleanup." + }, + { + "left": "issue:45491", + "right": "issue:46693", + "accept": false, + "reason": "NaN embeddings bug vs FA2 performance regression." + }, + { + "left": "issue:29127", + "right": "issue:33357", + "accept": false, + "reason": "LayoutLMv3 error messaging vs CLIP bus error on macOS." + }, + { + "left": "issue:43065", + "right": "issue:45412", + "accept": false, + "reason": "Dummy decoder layer bug vs RT-DETR memory release issue." + }, + { + "left": "issue:43575", + "right": "issue:43701", + "accept": false, + "reason": "Tensor-parallel OOM vs checkpoint resume key mismatch." + }, + { + "left": "issue:45245", + "right": "issue:45563", + "accept": false, + "reason": "Category-count limit error vs stale generation warning." + }, + { + "left": "issue:45372", + "right": "issue:45406", + "accept": false, + "reason": "Processor import dependency failure vs missing _tokenizer attribute in serve." + }, + { + "left": "issue:42490", + "right": "issue:43381", + "accept": false, + "reason": "Trailing-slash import warning vs gradient checkpointing eval-mode restriction." + }, + { + "left": "issue:29127", + "right": "issue:35141", + "accept": false, + "reason": "LayoutLMv3 box-validation messaging vs embedding reinitialization bug." + }, + { + "left": "issue:45440", + "right": "issue:45593", + "accept": false, + "reason": "DeepSeekV3 remote/native divergence vs D-FINE auxiliary-loss issue." + }, + { + "left": "issue:43994", + "right": "issue:45412", + "accept": false, + "reason": "SigLIP2 bad outputs vs RT-DETR memory leak." + }, + { + "left": "issue:43421", + "right": "issue:45491", + "accept": false, + "reason": "Runtime post-processor update bug vs Gemma3 NaN embeddings." + }, + { + "left": "issue:45372", + "right": "issue:45459", + "accept": false, + "reason": "Import dependency breakage vs overly broad tokenizer exception handling." + }, + { + "left": "issue:45406", + "right": "issue:46701", + "accept": false, + "reason": "Gemma4 processor attribute error vs DiffusionGemma pipeline failure." + }, + { + "left": "issue:45412", + "right": "issue:46369", + "accept": false, + "reason": "RT-DETR cleanup leak vs SAM3 memory overflow." + }, + { + "left": "issue:43575", + "right": "issue:44387", + "accept": false, + "reason": "Tensor-parallel load OOM vs int4 quantization reserved-memory regression." + }, + { + "left": "issue:43066", + "right": "issue:44462", + "accept": false, + "reason": "Wrong tokenizer decoder type vs AutoTokenizer ignoring tokenizer.json." + }, + { + "left": "issue:44442", + "right": "issue:45072", + "accept": false, + "reason": "Tokenizer load failure vs bfloat16 dtype mismatch in inference." + }, + { + "left": "issue:44464", + "right": "issue:45584", + "accept": false, + "reason": "Compiled-forward generation inconsistency vs Whisper empty-transcription failure." + }, + { + "left": "issue:43526", + "right": "issue:45834", + "accept": false, + "reason": "BeitImageProcessorFast label reduction bug vs Kosmos2.5 OCR index error." + }, + { + "left": "issue:43824", + "right": "issue:44279", + "accept": false, + "reason": "Missing Qwen2.5-VL import vs generic dependency issue." + }, + { + "left": "issue:43526", + "right": "issue:46693", + "accept": false, + "reason": "Beit label reduction bug vs long-context FA2 performance regression." + }, + { + "left": "issue:42915", + "right": "issue:43931", + "accept": false, + "reason": "Qwen3Moe FP8 config failure vs Qwen3-VL weight-shape mismatch." + }, + { + "left": "issue:27957", + "right": "issue:33290", + "accept": false, + "reason": "XLM-R FlashAttention2 support vs Adafactor/deepspeed OOM." + }, + { + "left": "issue:27957", + "right": "issue:29127", + "accept": false, + "reason": "FlashAttention2 model support vs LayoutLMv3 error clarity." + }, + { + "left": "issue:44857", + "right": "issue:45588", + "accept": false, + "reason": "AMP float16 crash in loss vs flash-attention aux-state None crash." + }, + { + "left": "issue:43994", + "right": "issue:46369", + "accept": false, + "reason": "SigLIP2 inference bug vs SAM3 lite text memory overflow." + }, + { + "left": "issue:33290", + "right": "issue:34689", + "accept": false, + "reason": "Optimizer OOM vs model-loading breakage for Llama 3.2 Vision." + }, + { + "left": "issue:33290", + "right": "issue:35141", + "accept": false, + "reason": "Adafactor OOM vs post-init embedding reinitialization." + }, + { + "left": "issue:43526", + "right": "issue:44186", + "accept": false, + "reason": "Image label reduction bug vs LayoutLMv2 tokenizer padding/truncation crash." + } + ] + }, + "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": 26, + "review_unit_index": 35, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:27957", + "issue:29870", + "issue:42175", + "issue:42490", + "issue:43262", + "issue:43316", + "issue:43334", + "issue:43381", + "issue:43493", + "issue:43525", + "issue:43531", + "issue:43643", + "issue:43646", + "issue:43720", + "issue:43749", + "issue:43761", + "issue:43827", + "issue:43873", + "issue:43931", + "issue:44062", + "issue:44077", + "issue:44206", + "issue:44442", + "issue:44479", + "issue:44514", + "issue:44743", + "issue:44811", + "issue:44855", + "issue:44871", + "issue:44898", + "issue:44913", + "issue:44977", + "issue:45003", + "issue:45137", + "issue:45245", + "issue:45250", + "issue:45325", + "issue:45381", + "issue:45440", + "issue:45588", + "issue:45859", + "issue:45907", + "issue:46121", + "issue:46234", + "issue:46399", + "issue:46640", + "issue:46697", + "issue:46821" + ], + "soft_pairs": [ + "issue:43334|issue:43531", + "issue:46121|issue:46399", + "issue:43381|issue:43873", + "issue:45250|issue:46640", + "issue:45588|issue:46821", + "issue:44514|issue:45381", + "issue:43646|issue:44077", + "issue:43761|issue:45859", + "issue:44977|issue:45245", + "issue:44743|issue:44811", + "issue:43262|issue:44206", + "issue:44442|issue:44898", + "issue:42490|issue:44743", + "issue:43720|issue:44442", + "issue:42175|issue:46234", + "issue:45003|issue:45440", + "issue:27957|issue:29870", + "issue:44077|issue:44479", + "issue:43493|issue:43827", + "issue:43316|issue:43525", + "issue:43873|issue:44871", + "issue:45325|issue:46697", + "issue:44062|issue:45907", + "issue:43643|issue:44913", + "issue:44855|issue:45137", + "issue:43749|issue:43931" + ], + "prepared_review_unit_hash": "561d478d06a328b0c4f80017b029a9a3260021d61b0a81640a059828b61200cd", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 26, + "serialized_chars": 35551, + "estimated_input_tokens": 8888, + "estimated_eval_tokens": 18032 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: the paired items share only broad topical overlap (Flash Attention, Qwen, loading/config regressions, audio/vision processors), but they describe different code paths and failure modes. None of the soft edges look like true duplicates or mergeable into one fix.", + "confidence": 0.94, + "canonical_issue_reason": "No single canonical issue; the items span many unrelated bugs and regressions rather than one underlying defect.", + "canonical_pr_reason": null, + "best_issue_reason": "No single best issue; none of these issues subsumes the others or represents the whole cluster cleanly.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43334", + "right": "issue:43531", + "accept": false, + "reason": "Both involve Qwen text configs, but one is a missing pad_token_id attribute and the other is a sliding_window behavior bug; different failures and fixes." + }, + { + "left": "issue:46121", + "right": "issue:46399", + "accept": false, + "reason": "Rope-validation TypeError and Qwen3.5 kernel registration crash are unrelated code paths; only a loose Qwen/rope overlap." + }, + { + "left": "issue:43381", + "right": "issue:43873", + "accept": false, + "reason": "Eval-mode gradient checkpointing and quantization offloading are distinct runtime behaviors with different root causes." + }, + { + "left": "issue:45250", + "right": "issue:46640", + "accept": false, + "reason": "Both mention Flash Attention 2, but the titles are too generic to show the same concrete bug or fix." + }, + { + "left": "issue:45588", + "right": "issue:46821", + "accept": false, + "reason": "Flash-attention sink handling and TimesFM window_size AttributeError are unrelated subsystems and errors." + }, + { + "left": "issue:44514", + "right": "issue:45381", + "accept": false, + "reason": "These are separate Qwen2.5-VL bugs: batched chat-template padding versus incorrect video position IDs." + }, + { + "left": "issue:43646", + "right": "issue:44077", + "accept": false, + "reason": "A config-loading discrepancy is not the same as patchtsmixer post_init API enforcement; different model-init issues." + }, + { + "left": "issue:43761", + "right": "issue:45859", + "accept": false, + "reason": "CLIPVisionModel hidden_states regression and Qwen3_5Moe CPU offload incompatibility affect different models and paths." + }, + { + "left": "issue:44977", + "right": "issue:45245", + "accept": false, + "reason": "Qwen3.5 flash-attention generation failure and a categories-limit runtime error are unrelated." + }, + { + "left": "issue:44743", + "right": "issue:44811", + "accept": false, + "reason": "Recurrent state reset with cache is unrelated to Whisper batch_decode ignoring skip_special_tokens." + }, + { + "left": "issue:43262", + "right": "issue:44206", + "accept": false, + "reason": "Audio chat-template sampling-rate defaulting and a feature-extractor center-arg crash are different preprocessing bugs." + }, + { + "left": "issue:44442", + "right": "issue:44898", + "accept": false, + "reason": "Tokenizer loading failure and Perceiver image-classification size handling are separate issues in different components." + }, + { + "left": "issue:42490", + "right": "issue:44743", + "accept": false, + "reason": "Local-path import warnings and Qwen recurrent-state resetting do not share a concrete code path." + }, + { + "left": "issue:43720", + "right": "issue:44442", + "accept": false, + "reason": "BitNet packed-weight loading and tokenizer loading for FastSpeech2Conformer are unrelated failures." + }, + { + "left": "issue:42175", + "right": "issue:46234", + "accept": false, + "reason": "pip extras dependency inclusion and a missing import_utils capability flag are different packaging/import problems." + }, + { + "left": "issue:45003", + "right": "issue:45440", + "accept": false, + "reason": "Unsafe sys.modules access in modeling_utils and DeepseekV3MoE divergence are not the same bug." + }, + { + "left": "issue:27957", + "right": "issue:29870", + "accept": false, + "reason": "XLMRoberta Flash Attention 2 support and BART generate() decoder_attention_mask assignment are unrelated issues." + }, + { + "left": "issue:44077", + "right": "issue:44479", + "accept": false, + "reason": "patchtsmixer post_init policy and Qwen video-input regression are different model families and fixes." + }, + { + "left": "issue:43493", + "right": "issue:43827", + "accept": false, + "reason": "SigLIP2 implementation discrepancy and docs still referencing pipeline() are not the same problem." + }, + { + "left": "issue:43316", + "right": "issue:43525", + "accept": false, + "reason": "Gemma3TextConfig API mismatch and Llama4Config missing pad_token_id are different config-field issues." + }, + { + "left": "issue:43873", + "right": "issue:44871", + "accept": false, + "reason": "Quantization offloading behavior and Gemma-3 eos_token_id inconsistency involve different mechanisms and symptoms." + }, + { + "left": "issue:45325", + "right": "issue:46697", + "accept": false, + "reason": "Qwen2.5-VL rope indexing and Swin2SR image padding are unrelated vision preprocessing bugs." + }, + { + "left": "issue:44062", + "right": "issue:45907", + "accept": false, + "reason": "AddedToken special-argument collision and list_repo_files revision forwarding are separate API bugs." + }, + { + "left": "issue:43643", + "right": "issue:44913", + "accept": false, + "reason": "Missing fields from trust_remote_code loading and GPTNeoX rotary_pct reload behavior are different config-serialization issues." + }, + { + "left": "issue:44855", + "right": "issue:45137", + "accept": false, + "reason": "Python 3.13 torch.jit parsing/import failure and DeepSpeed ZeRO3 deque underflow are unrelated runtime errors." + }, + { + "left": "issue:43749", + "right": "issue:43931", + "accept": false, + "reason": "FSDP CPU-efficient loading and Qwen3-VL weight-shape mismatch are distinct model-loading failures." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 36, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:16998", + "issue:22355", + "issue:25251", + "issue:27957", + "issue:28282", + "issue:32090", + "issue:42617", + "issue:42907", + "issue:43232", + "issue:43295", + "issue:43316", + "issue:43388", + "issue:43452", + "issue:43550", + "issue:43575", + "issue:43606", + "issue:43646", + "issue:43701", + "issue:43716", + "issue:43742", + "issue:43761", + "issue:43825", + "issue:43931", + "issue:44077", + "issue:44079", + "issue:44188", + "issue:44190", + "issue:44442", + "issue:44464", + "issue:44496", + "issue:44534", + "issue:44871", + "issue:44913", + "issue:44987", + "issue:45216", + "issue:45584", + "issue:45698", + "issue:45759", + "issue:45803", + "issue:46050", + "issue:46178", + "issue:46234", + "issue:46291", + "issue:46421", + "issue:46506", + "issue:46821", + "issue:46829", + "issue:46932" + ], + "soft_pairs": [ + "issue:43316|issue:45759", + "issue:44987|issue:46178", + "issue:44442|issue:45803", + "issue:44190|issue:44496", + "issue:43232|issue:43388", + "issue:43931|issue:45216", + "issue:43761|issue:44079", + "issue:43761|issue:44188", + "issue:42617|issue:43575", + "issue:43701|issue:46506", + "issue:46829|issue:46932", + "issue:43575|issue:43742", + "issue:43550|issue:43701", + "issue:27957|issue:32090", + "issue:25251|issue:28282", + "issue:42907|issue:45698", + "issue:43606|issue:43701", + "issue:16998|issue:22355", + "issue:43295|issue:43825", + "issue:43295|issue:44871", + "issue:44871|issue:45584", + "issue:46234|issue:46291", + "issue:44464|issue:46421", + "issue:44077|issue:44534", + "issue:43316|issue:43716", + "issue:43316|issue:46821", + "issue:44871|issue:46050", + "issue:43646|issue:44913", + "issue:43452|issue:44496" + ], + "prepared_review_unit_hash": "865a85f3d6622af55f73c75ebfe7f40f9b1a3539a86b465089efc75e0b3c86cb", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36342, + "estimated_input_tokens": 9086, + "estimated_eval_tokens": 18428 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These are mostly thematically similar but distinct bugs. I would not merge any of the proposed soft pairs: they span different models, different APIs, and different failure modes rather than the same concrete code-path issue.", + "confidence": 0.94, + "canonical_issue_reason": "No single canonical issue stands out, because none of the soft pairs are strong duplicate matches; they only share broad subsystem similarity.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue is clearly a global best representative for a duplicate cluster, since the candidates are largely unrelated beyond being Transformers regressions/load failures.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43316", + "right": "issue:45759", + "accept": false, + "reason": "Both involve config/model construction, but they target different composite-model APIs and different bugs." + }, + { + "left": "issue:44987", + "right": "issue:46178", + "accept": false, + "reason": "One is a model load regression for a specific repo; the other is an import-utils backend mapping error." + }, + { + "left": "issue:44442", + "right": "issue:45803", + "accept": false, + "reason": "Tokenizer loading failure and a bare-except tensor conversion bug are unrelated code paths." + }, + { + "left": "issue:44190", + "right": "issue:44496", + "accept": false, + "reason": "A dataset/script issue and an unrecognized-model/config-json issue are not the same bug." + }, + { + "left": "issue:43232", + "right": "issue:43388", + "accept": false, + "reason": "Generation kwargs sync and metric gathering last-batch handling are separate trainer behaviors." + }, + { + "left": "issue:43931", + "right": "issue:45216", + "accept": false, + "reason": "Both mention Qwen VL, but one is weight-shape loading failure and the other is save_pretrained checkpoint corruption." + }, + { + "left": "issue:43761", + "right": "issue:44079", + "accept": false, + "reason": "Hidden states not returned is different from ModelOutput key assignment semantics." + }, + { + "left": "issue:43761", + "right": "issue:44188", + "accept": false, + "reason": "A CLIP output regression and an attention-kernel branching issue are unrelated." + }, + { + "left": "issue:42617", + "right": "issue:43575", + "accept": false, + "reason": "One is a parallel script failure; the other is a tensor-parallel OOM on a specific model." + }, + { + "left": "issue:43701", + "right": "issue:46506", + "accept": false, + "reason": "Checkpoint key mismatch and CPU/CUDA output divergence are different failure modes." + }, + { + "left": "issue:46829", + "right": "issue:46932", + "accept": false, + "reason": "Image classification output misassignment and Gemma4 dtype casting are unrelated." + }, + { + "left": "issue:43575", + "right": "issue:43742", + "accept": false, + "reason": "OOM under tensor parallelism is not the same as a key error while loading another model." + }, + { + "left": "issue:43550", + "right": "issue:43701", + "accept": false, + "reason": "torch.compile SDPA failure and resume_from_checkpoint key mismatch are different code paths." + }, + { + "left": "issue:27957", + "right": "issue:32090", + "accept": false, + "reason": "Flash Attention behavior and a Trainer broadcast TypeError are unrelated." + }, + { + "left": "issue:25251", + "right": "issue:28282", + "accept": false, + "reason": "Pipeline top_k output-shape change is unrelated to missing PyTorch import error." + }, + { + "left": "issue:42907", + "right": "issue:45698", + "accept": false, + "reason": "Both concern save/load, but one is dequantized model saving and the other is custom-module resolution after reload." + }, + { + "left": "issue:43606", + "right": "issue:43701", + "accept": false, + "reason": "CPU offload device mismatch is unrelated to checkpoint key mismatch." + }, + { + "left": "issue:16998", + "right": "issue:22355", + "accept": false, + "reason": "A model_max_length question and a missing transformers.onnx module are unrelated." + }, + { + "left": "issue:43295", + "right": "issue:43825", + "accept": false, + "reason": "Custom processor/tokenizer regression and a pipeline error-message regression are different bugs." + }, + { + "left": "issue:43295", + "right": "issue:44871", + "accept": false, + "reason": "Processor/tokenizer API breakage is unrelated to eos_token_id config inconsistency." + }, + { + "left": "issue:44871", + "right": "issue:45584", + "accept": false, + "reason": "Gemma eos-token config mismatch and Whisper empty-transcription generation failure are separate issues." + }, + { + "left": "issue:46234", + "right": "issue:46291", + "accept": false, + "reason": "Missing availability helper and kernels version/revision import failure are different import problems." + }, + { + "left": "issue:44464", + "right": "issue:46421", + "accept": false, + "reason": "Compiled-forward output inconsistency and chunked-prefill recompilation are related to generation but not the same defect." + }, + { + "left": "issue:44077", + "right": "issue:44534", + "accept": false, + "reason": "Model init validation and non-persistent buffer junk values are not the same bug." + }, + { + "left": "issue:43316", + "right": "issue:43716", + "accept": false, + "reason": "Gemma3TextConfig API discrepancy and Mistral-3 dtype mismatch are different multimodal model issues." + }, + { + "left": "issue:43316", + "right": "issue:46821", + "accept": false, + "reason": "A config API mismatch and a TimesFM window_size AttributeError are unrelated." + }, + { + "left": "issue:44871", + "right": "issue:46050", + "accept": false, + "reason": "These are both save/load-adjacent, but one is config inconsistency and the other is quantized model reload breakage." + }, + { + "left": "issue:43646", + "right": "issue:44913", + "accept": false, + "reason": "Custom model initialization regression and GPTNeoXConfig reload-default loss are different configuration problems." + }, + { + "left": "issue:43452", + "right": "issue:44496", + "accept": false, + "reason": "gguf_file handling and a missing model_type in config.json are not the same loading 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": 26, + "review_unit_index": 37, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:35141", + "issue:41950", + "issue:42490", + "issue:43065", + "issue:43352", + "issue:43388", + "issue:43404", + "issue:43441", + "issue:43475", + "issue:43504", + "issue:43525", + "issue:43531", + "issue:43540", + "issue:43572", + "issue:43650", + "issue:43673", + "issue:43688", + "issue:44162", + "issue:44190", + "issue:44242", + "issue:44514", + "issue:44534", + "issue:44610", + "issue:44617", + "issue:44734", + "issue:44743", + "issue:44792", + "issue:44913", + "issue:44964", + "issue:44977", + "issue:45127", + "issue:45200", + "issue:45406", + "issue:45542", + "issue:45600", + "issue:45632", + "issue:45698", + "issue:45803", + "issue:46170", + "issue:46178", + "issue:46202", + "issue:46234", + "issue:46365", + "issue:46369", + "issue:46424", + "issue:46640", + "issue:46894" + ], + "soft_pairs": [ + "issue:45803|issue:46365", + "issue:44162|issue:44792", + "issue:43404|issue:43441", + "issue:44242|issue:46894", + "issue:45632|issue:46424", + "issue:44610|issue:46369", + "issue:45803|issue:46170", + "issue:43065|issue:44617", + "issue:43540|issue:44514", + "issue:45542|issue:46234", + "issue:43650|issue:46202", + "issue:43572|issue:46178", + "issue:44792|issue:44977", + "issue:43504|issue:45698", + "issue:44792|issue:45600", + "issue:45127|issue:45200", + "issue:43388|issue:43688", + "issue:43673|issue:44743", + "issue:43352|issue:46640", + "issue:43525|issue:43531", + "issue:41950|issue:43475", + "issue:44534|issue:44913", + "issue:44190|issue:44964", + "issue:35141|issue:42490", + "issue:44734|issue:45406" + ], + "prepared_review_unit_hash": "54687ca62d7f706146a334fc1b9ff11a336817c0b64275c76ece8f416a242dc8", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 25, + "serialized_chars": 34480, + "estimated_input_tokens": 8620, + "estimated_eval_tokens": 17496 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is a bag of unrelated issue reports; none of the soft pairs describe the same underlying bug closely enough to merge.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue is safe here: the titles span unrelated subsystems and failure modes (model config, processors, caching, serving, import/backends, and test failures).", + "canonical_pr_reason": null, + "best_issue_reason": "No issue in this set is globally representative enough to serve as a canonical duplicate target; the overlaps are mostly superficial (same model family or same broad area) rather than the same defect.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45803", + "right": "issue:46365", + "accept": false, + "reason": "Different bugs: bare `except` swallowing errors in Fuyu tensor conversion vs Trainer import failure due to datasets on import path." + }, + { + "left": "issue:44162", + "right": "issue:44792", + "accept": false, + "reason": "Unrelated: ESM2 breakage vs a Janus image-generation test failure; no shared code-path or defect." + }, + { + "left": "issue:43404", + "right": "issue:43441", + "accept": false, + "reason": "Both are Mistral-family, but one is missing lm_head tying and the other is FlashAttention support; distinct fixes." + }, + { + "left": "issue:44242", + "right": "issue:46894", + "accept": false, + "reason": "Both involve training loss, but the concrete bugs differ: router-logits load balancing vs a double-shift loss issue in CohereASR." + }, + { + "left": "issue:45632", + "right": "issue:46424", + "accept": false, + "reason": "Both mention cache, but one is remote-code cache path collisions and the other is max_cache_length handling; separate problems." + }, + { + "left": "issue:44610", + "right": "issue:46369", + "accept": false, + "reason": "Different models and symptoms: OmDet-Turbo processor image size mismatch vs Sam3-LiteText memory overflow." + }, + { + "left": "issue:45803", + "right": "issue:46170", + "accept": false, + "reason": "Unrelated import/runtime issues: Fuyu conversion swallowing exceptions vs Windows ROCm import dependency failure." + }, + { + "left": "issue:43065", + "right": "issue:44617", + "accept": false, + "reason": "Same SAM3 area, but dummy Conv2d in pixel decoder is not the same bug as a Sam3Video OOM." + }, + { + "left": "issue:43540", + "right": "issue:44514", + "accept": false, + "reason": "Both are Qwen-related input issues, but one is video input validation and the other is batched chat-template padding behavior." + }, + { + "left": "issue:45542", + "right": "issue:46234", + "accept": false, + "reason": "Different dependency problems: TensorBoard/TensorFlow backend detection vs a missing import_utils symbol." + }, + { + "left": "issue:43650", + "right": "issue:46202", + "accept": false, + "reason": "No meaningful overlap; one is a vague request, the other is a Gemma pipeline roundtrip tokenizer issue." + }, + { + "left": "issue:43572", + "right": "issue:46178", + "accept": false, + "reason": "Different backend/config regressions: StableLmConfig pad_token field vs tensorflow_text backend mapping." + }, + { + "left": "issue:44792", + "right": "issue:44977", + "accept": false, + "reason": "Different failures: Janus image-generation test vs Qwen3.5 flash-attention generation bug." + }, + { + "left": "issue:43504", + "right": "issue:45698", + "accept": false, + "reason": "Both concern model loading, but one is a legacy config field on Beit and the other is wrong custom module selection after save/load." + }, + { + "left": "issue:44792", + "right": "issue:45600", + "accept": false, + "reason": "Unrelated: Janus test failure vs auto_mappings referencing removed Sam3LiteText configs." + }, + { + "left": "issue:45127", + "right": "issue:45200", + "accept": false, + "reason": "Distinct bugs: LoRA merge with extended vocab on tied embeddings vs Gemma4 mm_token_type_ids defaulting." + }, + { + "left": "issue:43388", + "right": "issue:43688", + "accept": false, + "reason": "Different algorithmic bugs in different training/eval code paths; only a superficial metrics/loss similarity." + }, + { + "left": "issue:43673", + "right": "issue:44743", + "accept": false, + "reason": "Both touch cache/state handling, but one is GenerationMixin chunked_prefill cache and the other is Qwen3 recurrent states resetting." + }, + { + "left": "issue:43352", + "right": "issue:46640", + "accept": false, + "reason": "One is a specific model not supporting Flash Attention 2.0 yet; the other is a generic Flash Attention 2 problem with no same code-path evidence." + }, + { + "left": "issue:43525", + "right": "issue:43531", + "accept": false, + "reason": "Unrelated config issues: missing pad_token_id in Llama4Config vs sliding_window behavior in Qwen3-MoE." + }, + { + "left": "issue:41950", + "right": "issue:43475", + "accept": false, + "reason": "Different SAM/video bugs: processor picking image processors vs missing fpn_position_embeddings attribute." + }, + { + "left": "issue:44534", + "right": "issue:44913", + "accept": false, + "reason": "No same defect: non-persistent buffer initialization junk vs GPTNeoX rotary_pct not persisting on reload." + }, + { + "left": "issue:44190", + "right": "issue:44964", + "accept": false, + "reason": "Different loading failures for different entry points/models; no shared underlying bug." + }, + { + "left": "issue:35141", + "right": "issue:42490", + "accept": false, + "reason": "Unrelated: embedding reinitialization after resize vs import warning on local paths with trailing slash." + }, + { + "left": "issue:44734", + "right": "issue:45406", + "accept": false, + "reason": "Both are transformers serve crashes, but the concrete failures differ: KV-cache tensor indexing vs missing `_tokenizer` on Gemma4Processor." + } + ] + }, + "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": 26, + "review_unit_index": 38, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:25251", + "issue:27957", + "issue:29870", + "issue:35141", + "issue:39692", + "issue:41720", + "issue:43065", + "issue:43335", + "issue:43550", + "issue:43644", + "issue:43673", + "issue:43827", + "issue:43854", + "issue:43931", + "issue:43994", + "issue:44060", + "issue:44077", + "issue:44190", + "issue:44242", + "issue:44464", + "issue:44479", + "issue:44610", + "issue:44617", + "issue:44734", + "issue:44792", + "issue:44855", + "issue:44869", + "issue:44877", + "issue:44913", + "issue:44933", + "issue:45072", + "issue:45245", + "issue:45362", + "issue:45381", + "issue:45563", + "issue:45584", + "issue:45588", + "issue:45632", + "issue:46082", + "issue:46369", + "issue:46421", + "issue:46617", + "issue:46697", + "issue:46832", + "issue:46858", + "issue:46897", + "issue:46932", + "issue:47013" + ], + "soft_pairs": [ + "issue:45072|issue:46932", + "issue:44855|issue:45588", + "issue:45632|issue:46617", + "issue:44869|issue:46832", + "issue:45245|issue:46858", + "issue:43827|issue:43994", + "issue:46082|issue:46697", + "issue:29870|issue:35141", + "issue:44617|issue:45584", + "issue:43673|issue:46421", + "issue:43931|issue:44060", + "issue:43644|issue:44913", + "issue:44242|issue:46897", + "issue:25251|issue:27957", + "issue:43335|issue:44855", + "issue:45563|issue:45584", + "issue:44077|issue:44877", + "issue:44933|issue:45362", + "issue:45381|issue:46369", + "issue:43550|issue:44190", + "issue:41720|issue:45245", + "issue:44479|issue:44734", + "issue:43065|issue:44610", + "issue:43335|issue:44190", + "issue:45588|issue:47013", + "issue:43854|issue:44792", + "issue:44617|issue:46369", + "issue:39692|issue:44190", + "issue:44464|issue:45563" + ], + "prepared_review_unit_hash": "cc45e23249b465f5188c809ef3626fffb48a24edc2ced5645ae7255c74fc37f3", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36526, + "estimated_input_tokens": 9132, + "estimated_eval_tokens": 18520 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose semantic neighborhood rather than a duplicate set. The soft edges are mostly broad topical similarities (cache, vision, dtype, generation), but the concrete failure modes differ, so none should be merged.", + "confidence": 0.87, + "canonical_issue_reason": "No single canonical issue fits well: the issues span unrelated models and failure modes, and the similarity links are mostly shallow topical overlap.", + "canonical_pr_reason": null, + "best_issue_reason": "No global best issue for consolidation; the closest pairwise matches still describe distinct bugs, so keeping them separate is safer.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45072", + "right": "issue:46932", + "accept": false, + "reason": "Both are dtype-casting problems, but they affect different models/components and describe different concrete failures." + }, + { + "left": "issue:44855", + "right": "issue:45588", + "accept": false, + "reason": "One is a Python 3.13 import/parser issue, the other is a flash-attention AttributeError; same general area only at a very high level." + }, + { + "left": "issue:45632", + "right": "issue:46617", + "accept": false, + "reason": "Both involve trust_remote_code/cache loading, but one is a leaf-directory cache collision and the other is a FileNotFoundError from cache lookup; not clearly the same bug." + }, + { + "left": "issue:44869", + "right": "issue:46832", + "accept": false, + "reason": "Whisper timestamp decoding and RT-DETR feature-level handling are unrelated code paths." + }, + { + "left": "issue:45245", + "right": "issue:46858", + "accept": false, + "reason": "A category-cardinality runtime error is unrelated to static-cache slice arithmetic under torch.compile." + }, + { + "left": "issue:43827", + "right": "issue:43994", + "accept": false, + "reason": "Docs stale pipeline references are not the same as a SigLIP2 model/processor inference correctness bug." + }, + { + "left": "issue:46082", + "right": "issue:46697", + "accept": false, + "reason": "Llama head-dim validation and Swin2SR padding behavior are unrelated." + }, + { + "left": "issue:29870", + "right": "issue:35141", + "accept": false, + "reason": "BART decoder-mask generation error and embedding reinitialization after resize are different bugs." + }, + { + "left": "issue:44617", + "right": "issue:45584", + "accept": false, + "reason": "CUDA OOM in Sam3Video and Whisper empty-transcription failure are unrelated failures." + }, + { + "left": "issue:43673", + "right": "issue:46421", + "accept": false, + "reason": "Both mention chunked prefill/cache, but one is missing cache state in v5.0.0 and the other is recompilation from lazy cache init; not the same defect." + }, + { + "left": "issue:43931", + "right": "issue:44060", + "accept": false, + "reason": "Model weight-shape mismatch and a tied-weights warning bug are different Qwen3 issues." + }, + { + "left": "issue:43644", + "right": "issue:44913", + "accept": false, + "reason": "Non-persistent buffer junk values and GPTNeoX config reload defaulting are unrelated." + }, + { + "left": "issue:44242", + "right": "issue:46897", + "accept": false, + "reason": "Load-balancing loss omission and Florence2 double-shift training loss are different training bugs." + }, + { + "left": "issue:25251", + "right": "issue:27957", + "accept": false, + "reason": "Pipeline top_k nesting behavior and XLMRoberta Flash Attention 2 are unrelated." + }, + { + "left": "issue:43335", + "right": "issue:44855", + "accept": false, + "reason": "SwitchTransformers sparse-layer config and DebertaV2 import parsing are unrelated." + }, + { + "left": "issue:45563", + "right": "issue:45584", + "accept": false, + "reason": "A stale warning in paged generate is not the same as Whisper empty-transcription generation failure." + }, + { + "left": "issue:44077", + "right": "issue:44877", + "accept": false, + "reason": "PatchTSMixer post_init policy and granite_speech strict-config loading are different configuration issues." + }, + { + "left": "issue:44933", + "right": "issue:45362", + "accept": false, + "reason": "A missing import and Qwen3.5 chat crashes are unrelated." + }, + { + "left": "issue:45381", + "right": "issue:46369", + "accept": false, + "reason": "Qwen2.5-VL video position IDs and Sam3-litetext memory overflow are different model bugs." + }, + { + "left": "issue:43550", + "right": "issue:44190", + "accept": false, + "reason": "torch.compile SDPA failure and local dataset loading are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:45245", + "accept": false, + "reason": "Auto device mapping cudaErrorAssert and category-count overflow are unrelated." + }, + { + "left": "issue:44479", + "right": "issue:44734", + "accept": false, + "reason": "Video-input regression for Qwen VL models and KV-cache continuation indexing in serve are different code paths." + }, + { + "left": "issue:43065", + "right": "issue:44610", + "accept": false, + "reason": "Sam3PixelDecoder dummy conv and OmDet-Turbo processor size mismatch are not the same bug." + }, + { + "left": "issue:43335", + "right": "issue:44190", + "accept": false, + "reason": "SwitchTransformers config creation and dataset loading are unrelated." + }, + { + "left": "issue:45588", + "right": "issue:47013", + "accept": false, + "reason": "Flash-attention sink handling and xLSTM state unpacking are unrelated failures." + }, + { + "left": "issue:43854", + "right": "issue:44792", + "accept": false, + "reason": "GLM-4.7-Flash unit-test loading and janus image-generation test failure are unrelated." + }, + { + "left": "issue:44617", + "right": "issue:46369", + "accept": false, + "reason": "OOM in Sam3Video and memory overflow in sam3-litetext are different models and failure modes." + }, + { + "left": "issue:39692", + "right": "issue:44190", + "accept": false, + "reason": "SigLIP2 doc/example errors and local dataset loading are unrelated." + }, + { + "left": "issue:44464", + "right": "issue:45563", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency and a stale generate warning 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": 26, + "review_unit_index": 39, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:25251", + "issue:27957", + "issue:29870", + "issue:32090", + "issue:35141", + "issue:36246", + "issue:41628", + "issue:42175", + "issue:42796", + "issue:43295", + "issue:43404", + "issue:43526", + "issue:43540", + "issue:43643", + "issue:43673", + "issue:43854", + "issue:44112", + "issue:44246", + "issue:44279", + "issue:44462", + "issue:44496", + "issue:44521", + "issue:44734", + "issue:44737", + "issue:44779", + "issue:44938", + "issue:44964", + "issue:45003", + "issue:45127", + "issue:45216", + "issue:45276", + "issue:45290", + "issue:45357", + "issue:45381", + "issue:45542", + "issue:45584", + "issue:45600", + "issue:45632", + "issue:45803", + "issue:45907", + "issue:46082", + "issue:46121", + "issue:46234", + "issue:46489", + "issue:46655", + "issue:46697", + "issue:46752", + "issue:46858" + ], + "soft_pairs": [ + "issue:45381|issue:46655", + "issue:45127|issue:45357", + "issue:43295|issue:45276", + "issue:44112|issue:45600", + "issue:44779|issue:46489", + "issue:43540|issue:44734", + "issue:45290|issue:46752", + "issue:45276|issue:45584", + "issue:43526|issue:46697", + "issue:44521|issue:45381", + "issue:44737|issue:45803", + "issue:25251|issue:29870", + "issue:44462|issue:45907", + "issue:44938|issue:45542", + "issue:43854|issue:44964", + "issue:43673|issue:46858", + "issue:43643|issue:44496", + "issue:44246|issue:45003", + "issue:44964|issue:45600", + "issue:45632|issue:46234", + "issue:27957|issue:35141", + "issue:42175|issue:44279", + "issue:44112|issue:44964", + "issue:42796|issue:44521", + "issue:32090|issue:35141", + "issue:36246|issue:41628", + "issue:43404|issue:45216", + "issue:46082|issue:46121" + ], + "prepared_review_unit_hash": "6eaeb5aca066ec99c15b11b2660f3c419483b1122721c2bd467dc310e774036f", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36570, + "estimated_input_tokens": 9143, + "estimated_eval_tokens": 18542 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster does not look like a true duplicate set: the issues span many unrelated Transformers bugs, and the soft links are mostly broad topical similarity rather than the same concrete code path.", + "confidence": 0.96, + "canonical_issue_reason": "issue:44521 is the strongest single representative by signal and specificity (multiple inbound refs, active discussion, concrete multimodal failure), although the cluster itself does not collapse into one duplicate bug.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44521 has the clearest user-facing regression and the highest community signal in this set, so it is the best issue to keep as the representative anchor.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45381", + "right": "issue:46655", + "accept": false, + "reason": "Both are multimodal/video-processing bugs, but one is wrong vision_position_ids in Qwen2.5-VL and the other is over-padding in SmolVLM; different underlying failures." + }, + { + "left": "issue:45127", + "right": "issue:45357", + "accept": false, + "reason": "Both involve save_pretrained/tied-weight behavior, but one is LoRA merge collapse with extended vocab and the other is incorrect visual encoder keys; not the same bug." + }, + { + "left": "issue:43295", + "right": "issue:45276", + "accept": false, + "reason": "These are both embedding/serialization regressions, but they affect different mechanisms: processor/tokenizer access vs resize_token_embeddings not updating Gemma4 layers." + }, + { + "left": "issue:44112", + "right": "issue:45600", + "accept": false, + "reason": "One is a stale GraniteSpeech CI test, the other is an auto_mappings reference cleanup issue; they are unrelated CI failures." + }, + { + "left": "issue:44779", + "right": "issue:46489", + "accept": false, + "reason": "Same model family, but 44779 is a broad v5 tokenizer regression while 46489 is a specific DeepSeek-Coder v1 gap in a prior fix; not one duplicate bug." + }, + { + "left": "issue:43540", + "right": "issue:44734", + "accept": false, + "reason": "Both are runtime generation errors, but one is Qwen3OmniMoe video-input validation and the other is KV-cache continuation indexing in serve." + }, + { + "left": "issue:45290", + "right": "issue:46752", + "accept": false, + "reason": "Both crash in apply_chat_template, but one is assistant tool-call messages with no content and the other is an empty-list IndexError; distinct edge cases." + }, + { + "left": "issue:45276", + "right": "issue:45584", + "accept": false, + "reason": "These touch different subsystems: Gemma4 embedding resizing versus Whisper empty-transcription handling after align_special_tokens." + }, + { + "left": "issue:43526", + "right": "issue:46697", + "accept": false, + "reason": "One is a label-reduction bug in BeitImageProcessorFast; the other is over-padding in Swin2SR image processing." + }, + { + "left": "issue:44521", + "right": "issue:45381", + "accept": false, + "reason": "Both are multimodal issues, but one is assistant_masks being all zero and the other is incorrect vision_position_ids; different preprocessing stages." + }, + { + "left": "issue:44737", + "right": "issue:45803", + "accept": false, + "reason": "XLNet CPU tensor creation overhead and a bare-except swallowing errors in FuyuBatchFeature are unrelated bugs." + }, + { + "left": "issue:25251", + "right": "issue:29870", + "accept": false, + "reason": "Top-k nesting in pipeline output and Bart generate() unbound decoder_attention_mask are different code paths and symptoms." + }, + { + "left": "issue:44462", + "right": "issue:45907", + "accept": false, + "reason": "AutoTokenizer ignoring tokenizer.json and list_repo_files not forwarding revision are unrelated hub/tokenizer behaviors." + }, + { + "left": "issue:44938", + "right": "issue:45542", + "accept": false, + "reason": "Python 3.14 import failure and missing TensorFlow backend with the torch extra are different packaging/import problems." + }, + { + "left": "issue:43854", + "right": "issue:44964", + "accept": false, + "reason": "Different models and different load failures: GLM-4.7-Flash test loading versus Phi-4 multimodal loading." + }, + { + "left": "issue:43673", + "right": "issue:46858", + "accept": false, + "reason": "Both mention generation cache, but one is missing cache in chunked_prefill and the other is a device-tensor seq_length arithmetic bug in static cache." + }, + { + "left": "issue:43643", + "right": "issue:44496", + "accept": false, + "reason": "trust_remote_code missing fields is a different config-return issue from unrecognized model/config lacking model_type." + }, + { + "left": "issue:44246", + "right": "issue:45003", + "accept": false, + "reason": "Import slowdown and unsafe sys.modules access are not the same defect, even if both affect import-time behavior." + }, + { + "left": "issue:44964", + "right": "issue:45600", + "accept": false, + "reason": "A model-loading failure and a stale auto_mappings reference cleanup are separate problems." + }, + { + "left": "issue:45632", + "right": "issue:46234", + "accept": false, + "reason": "trust_remote_code cache-path collision and a missing import_utils helper are unrelated." + }, + { + "left": "issue:27957", + "right": "issue:35141", + "accept": false, + "reason": "XLMRoberta FlashAttention 2 support and resize-token-embeddings reinitialization are different model/runtime issues." + }, + { + "left": "issue:42175", + "right": "issue:44279", + "accept": false, + "reason": "Both are dependency-related at a high level, but one is a torch extra missing TensorFlow backend while the other is a generic dependency complaint." + }, + { + "left": "issue:44112", + "right": "issue:44964", + "accept": false, + "reason": "A flaky GraniteSpeech CI test and a GLM model-load error are unrelated." + }, + { + "left": "issue:42796", + "right": "issue:44521", + "accept": false, + "reason": "Tokenizer BPE marker handling and multimodal assistant-mask generation are different layers of the stack." + }, + { + "left": "issue:32090", + "right": "issue:35141", + "accept": false, + "reason": "GPU broadcast rejecting NoneType and embedding reinitialization after resize are unrelated bugs." + }, + { + "left": "issue:36246", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors around image processors, but one is a missing model-specific symbol and the other is a top-level AutoImageProcessor import failure; not the same fix." + }, + { + "left": "issue:43404", + "right": "issue:45216", + "accept": false, + "reason": "Mistral3 lm_head tying and Qwen3.5 incorrect save_pretrained checkpoints are distinct model-family regressions." + }, + { + "left": "issue:46082", + "right": "issue:46121", + "accept": false, + "reason": "LlamaConfig head_dim validation and JSON-loaded rope ignore_keys TypeError are separate config-validation 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": 26, + "review_unit_index": 40, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:36246", + "issue:38617", + "issue:41950", + "issue:42907", + "issue:43295", + "issue:43381", + "issue:43441", + "issue:43479", + "issue:43526", + "issue:43531", + "issue:43582", + "issue:43650", + "issue:43673", + "issue:43824", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44062", + "issue:44220", + "issue:44279", + "issue:44295", + "issue:44327", + "issue:44336", + "issue:44496", + "issue:44521", + "issue:44683", + "issue:44792", + "issue:44933", + "issue:44987", + "issue:45042", + "issue:45083", + "issue:45335", + "issue:45375", + "issue:45446", + "issue:45584", + "issue:45588", + "issue:45593", + "issue:45600", + "issue:45759", + "issue:46123", + "issue:46369", + "issue:46424", + "issue:46506", + "issue:46515", + "issue:46693", + "issue:46752", + "issue:46932" + ], + "soft_pairs": [ + "issue:36246|issue:38617", + "issue:43531|issue:45600", + "issue:44295|issue:46506", + "issue:41950|issue:44016", + "issue:43673|issue:46424", + "issue:43582|issue:46932", + "issue:43650|issue:44220", + "issue:43295|issue:45335", + "issue:43976|issue:44987", + "issue:45083|issue:46506", + "issue:44496|issue:45042", + "issue:44016|issue:44220", + "issue:43526|issue:43901", + "issue:44683|issue:45446", + "issue:43901|issue:44496", + "issue:43673|issue:44336", + "issue:42907|issue:43441", + "issue:43381|issue:45593", + "issue:44496|issue:45759", + "issue:43976|issue:44279", + "issue:44792|issue:46424", + "issue:43441|issue:45588", + "issue:44016|issue:44062", + "issue:43824|issue:44933", + "issue:44521|issue:46693", + "issue:44279|issue:46515", + "issue:43479|issue:45375", + "issue:45335|issue:45584", + "issue:46123|issue:46369", + "issue:44327|issue:46752" + ], + "prepared_review_unit_hash": "324be5a0072fe6c88b53a270c45e89ff4f6dc71b5326b12b81bec272db1b2e70", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36023, + "estimated_input_tokens": 9006, + "estimated_eval_tokens": 18268 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The soft pairs are mostly superficial text matches across unrelated bugs; none look like the same underlying issue or change, so all soft edges should be rejected.", + "confidence": 0.91, + "canonical_issue_reason": "issue:44521 is the strongest anchor in this mixed set: it has active discussion, inbound references, and a concrete reproducible multimodal bug report.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44521 is the most substantive and broadly relevant bug report here, with the clearest evidence and triage value.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:36246", + "right": "issue:38617", + "accept": false, + "reason": "Both are ImportErrors, but for different missing symbols in different modules and unrelated code paths." + }, + { + "left": "issue:43531", + "right": "issue:45600", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and Sam3LiteText auto_mappings CI breakage are unrelated failures." + }, + { + "left": "issue:44295", + "right": "issue:46506", + "accept": false, + "reason": "One is a position_ids buffer read error; the other is a CPU/CUDA output divergence. Different bugs." + }, + { + "left": "issue:41950", + "right": "issue:44016", + "accept": false, + "reason": "Pipeline image-processor lookup and a notebook syntax error are not the same bug." + }, + { + "left": "issue:43673", + "right": "issue:46424", + "accept": false, + "reason": "Both mention cache, but one is missing GenerationMixin cache in chunked_prefill while the other is max_cache_length being ignored except for quantized cache." + }, + { + "left": "issue:43582", + "right": "issue:46932", + "accept": false, + "reason": "Apple Silicon allocator TypeError and Gemma4 quantized embedder dtype casting are unrelated issues." + }, + { + "left": "issue:43650", + "right": "issue:44220", + "accept": false, + "reason": "\"ADD THE DATA\" is too vague to match the specific _torch_extract_fbank_features() bug." + }, + { + "left": "issue:43295", + "right": "issue:45335", + "accept": false, + "reason": "Processor/tokenizer regression and t5gemma embedding resize behavior are different model behaviors." + }, + { + "left": "issue:43976", + "right": "issue:44987", + "accept": false, + "reason": "Python version compatibility and a model-loading failure for physical-intelligence/fast are unrelated." + }, + { + "left": "issue:45083", + "right": "issue:46506", + "accept": false, + "reason": "Qwen3_omni_moe feature-length calculation and PPDocLayoutV3 CPU/CUDA mismatch are different code paths." + }, + { + "left": "issue:44496", + "right": "issue:45042", + "accept": false, + "reason": "Unrecognized model config and PIL backend torchvision dependency are unrelated." + }, + { + "left": "issue:44016", + "right": "issue:44220", + "accept": false, + "reason": "Notebook syntax error and audio feature extraction bug are not the same change." + }, + { + "left": "issue:43526", + "right": "issue:43901", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and TextClassificationPipeline docs mismatch are unrelated." + }, + { + "left": "issue:44683", + "right": "issue:45446", + "accept": false, + "reason": "Both touch flex_attention, but one is a compiled-kernel failure on torch>=2.9 and the other is an AuxRequest import/version-check bug." + }, + { + "left": "issue:43901", + "right": "issue:44496", + "accept": false, + "reason": "Docs mismatch for return_all_scores and unrecognized model config are unrelated." + }, + { + "left": "issue:43673", + "right": "issue:44336", + "accept": false, + "reason": "Cache missing in generation and ANSI codes in loading_report are unrelated." + }, + { + "left": "issue:42907", + "right": "issue:43441", + "accept": false, + "reason": "Ministral save/dequantization and Ministral-3 FlashAttention failure are different bugs." + }, + { + "left": "issue:43381", + "right": "issue:45593", + "accept": false, + "reason": "Gradient checkpointing in eval mode and D-FINE auxiliary loss behavior are unrelated." + }, + { + "left": "issue:44496", + "right": "issue:45759", + "accept": false, + "reason": "Unrecognized model config and composite Qwen text_config unwrapping are different failures." + }, + { + "left": "issue:43976", + "right": "issue:44279", + "accept": false, + "reason": "Python version support and a generic dependency issue are not the same bug." + }, + { + "left": "issue:44792", + "right": "issue:46424", + "accept": false, + "reason": "Janus test failure on image generation and cache max length semantics are unrelated." + }, + { + "left": "issue:43441", + "right": "issue:45588", + "accept": false, + "reason": "Both involve flash_attention, but one is model-specific Ministral failure and the other is an s_aux=None crash for sink-less models." + }, + { + "left": "issue:44016", + "right": "issue:44062", + "accept": false, + "reason": "Notebook syntax error and AddedToken special-argument duplication are unrelated." + }, + { + "left": "issue:43824", + "right": "issue:44933", + "accept": false, + "reason": "Missing Qwen2_5_VLForConditionalGeneration import and a different missing import from image_utils are not the same symbol or fix." + }, + { + "left": "issue:44521", + "right": "issue:46693", + "accept": false, + "reason": "All-zero assistant masks and a long-context FA2 performance regression are different multimodal generation issues." + }, + { + "left": "issue:44279", + "right": "issue:46515", + "accept": false, + "reason": "Generic transformers dependency issue and torch 2.4.1 incompatibility are different environment problems." + }, + { + "left": "issue:43479", + "right": "issue:45375", + "accept": false, + "reason": "Default config initialization with None and missing deepstack_visual_indexes under strict config handling are different multimodal config bugs." + }, + { + "left": "issue:45335", + "right": "issue:45584", + "accept": false, + "reason": "Embedding resize propagation and Whisper empty-transcription generation failure are unrelated." + }, + { + "left": "issue:46123", + "right": "issue:46369", + "accept": false, + "reason": "Partial-batch mask generation dropping results and sam3-litetext memory overflow are unrelated." + }, + { + "left": "issue:44327", + "right": "issue:46752", + "accept": false, + "reason": "QA decode_spans top_k bounds error and empty-list apply_chat_template crash are different input-shape 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": 26, + "review_unit_index": 41, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:33290", + "issue:39692", + "issue:41950", + "issue:42796", + "issue:43295", + "issue:43299", + "issue:43381", + "issue:43441", + "issue:43479", + "issue:43525", + "issue:43577", + "issue:43650", + "issue:43653", + "issue:43854", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44075", + "issue:44242", + "issue:44279", + "issue:44291", + "issue:44351", + "issue:44368", + "issue:44462", + "issue:44466", + "issue:44683", + "issue:44779", + "issue:45020", + "issue:45081", + "issue:45200", + "issue:45216", + "issue:45276", + "issue:45278", + "issue:45357", + "issue:45405", + "issue:45542", + "issue:45588", + "issue:45593", + "issue:45656", + "issue:45812", + "issue:45823", + "issue:45834", + "issue:45907", + "issue:46082", + "issue:46209", + "issue:46399", + "issue:46512", + "issue:47000" + ], + "soft_pairs": [ + "issue:43299|issue:45216", + "issue:44279|issue:45405", + "issue:43295|issue:44779", + "issue:39692|issue:43901", + "issue:42796|issue:43650", + "issue:44075|issue:44368", + "issue:46082|issue:46399", + "issue:43577|issue:45216", + "issue:44351|issue:45278", + "issue:43976|issue:45405", + "issue:41950|issue:45020", + "issue:46082|issue:46512", + "issue:43381|issue:47000", + "issue:44466|issue:45276", + "issue:44779|issue:45812", + "issue:43299|issue:45357", + "issue:43653|issue:45593", + "issue:44242|issue:45200", + "issue:43650|issue:46209", + "issue:41950|issue:45823", + "issue:43299|issue:43854", + "issue:44462|issue:45081", + "issue:45588|issue:47000", + "issue:44016|issue:44291", + "issue:44683|issue:45542", + "issue:43479|issue:43525", + "issue:43441|issue:45834", + "issue:33290|issue:45656", + "issue:45081|issue:45907" + ], + "prepared_review_unit_hash": "8a08a6f8e895ca5685f651cde1f7ba12a87b155549d7b2875c05b46e9f40e5d0", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36481, + "estimated_input_tokens": 9121, + "estimated_eval_tokens": 18498 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "A loose set of mostly unrelated Transformer issues spanning tokenizer regressions, multimodal/model-loading bugs, docs problems, and optimizer/DeepSpeed behavior; none of the soft pairs looks like the same underlying defect.", + "confidence": 0.84, + "canonical_issue_reason": "No strong canonical issue: these reports are only loosely related at a high level and cover different code paths and model families.", + "canonical_pr_reason": null, + "best_issue_reason": "If one issue must represent the set, 45020 is the broadest regression umbrella, but it is still too generic to subsume the rest.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43299", + "right": "issue:45216", + "accept": false, + "reason": "Both are regressions, but one is Qwen3VL MoE loading and the other is Qwen3.5 save_pretrained serialization; different operations and failure modes." + }, + { + "left": "issue:44279", + "right": "issue:45405", + "accept": false, + "reason": "One is a generic dependency issue; the other is a specific MIN_PEFT_VERSION bump problem. Same packaging theme, not the same bug." + }, + { + "left": "issue:43295", + "right": "issue:44779", + "accept": false, + "reason": "Processor/tokenizer API breakage vs incorrect Deepseek tokenization logic; different models and root causes." + }, + { + "left": "issue:39692", + "right": "issue:43901", + "accept": false, + "reason": "Both mention docs, but one is a SigLIP2 example bug and the other is stale textclassification docs; unrelated changes." + }, + { + "left": "issue:42796", + "right": "issue:43650", + "accept": false, + "reason": "Video-classification processor lookup and a trivial 'ADD THE DATA' issue are clearly unrelated." + }, + { + "left": "issue:44075", + "right": "issue:44368", + "accept": false, + "reason": "SGD args not used is an optimizer bug; tie_word_embeddings warnings are a model config warning, not the same issue." + }, + { + "left": "issue:46082", + "right": "issue:46399", + "accept": false, + "reason": "LlamaConfig head_dim validation and Qwen3.5 kernel registration failure are different config/runtime bugs." + }, + { + "left": "issue:43577", + "right": "issue:45216", + "accept": false, + "reason": "Blip2 dtype propagation and Qwen3.5 save_pretrained key serialization are separate code paths." + }, + { + "left": "issue:44351", + "right": "issue:45278", + "accept": false, + "reason": "Both are import problems after upgrades, but one is a missing HybridCache symbol and the other is broad import breakage; no concrete shared bug." + }, + { + "left": "issue:43976", + "right": "issue:45405", + "accept": false, + "reason": "Python version compatibility and an unreleased PEFT minimum version are different dependency issues." + }, + { + "left": "issue:41950", + "right": "issue:45020", + "accept": false, + "reason": "A specific video-classification processor bug is not the same as the broad remote_code regression report." + }, + { + "left": "issue:46082", + "right": "issue:46512", + "accept": false, + "reason": "Explicit head_dim rejection and stale config.dtype after from_config are different configuration bugs." + }, + { + "left": "issue:43381", + "right": "issue:47000", + "accept": false, + "reason": "Gradient checkpointing in eval mode and RTDetrLoss infeasible cost matrix are unrelated runtime failures." + }, + { + "left": "issue:44466", + "right": "issue:45276", + "accept": false, + "reason": "Inconsistent lm_head serialization and Gemma4 resize_token_embeddings behavior are both weight-related, but not the same defect." + }, + { + "left": "issue:44779", + "right": "issue:45812", + "accept": false, + "reason": "Deepseek tokenization regression and Granite token ID regression are separate tokenizer bugs for different model families." + }, + { + "left": "issue:43299", + "right": "issue:45357", + "accept": false, + "reason": "Both involve Qwen3.5, but one is model loading and the other is save_pretrained visual encoder keys; different failures." + }, + { + "left": "issue:43653", + "right": "issue:45593", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and D-FINE auxiliary-loss handling are unrelated." + }, + { + "left": "issue:44242", + "right": "issue:45200", + "accept": false, + "reason": "MoE load-balancing loss and Gemma4 mm_token_type_ids defaults are different multimodal/MoE issues." + }, + { + "left": "issue:43650", + "right": "issue:46209", + "accept": false, + "reason": "A placeholder issue cannot be the same bug as an FP8 inference failure on B200." + }, + { + "left": "issue:41950", + "right": "issue:45823", + "accept": false, + "reason": "Video-classification pipeline processor selection and Gemma4 PLE device placement are unrelated." + }, + { + "left": "issue:43299", + "right": "issue:43854", + "accept": false, + "reason": "Qwen3VL MoE loading and GLM-4.7-Flash unit-test loading are different model-loading bugs." + }, + { + "left": "issue:44462", + "right": "issue:45081", + "accept": false, + "reason": "AutoTokenizer ignoring tokenizer.json and Mistral regex patch crash are both tokenizer-related, but they fail in different places with different causes." + }, + { + "left": "issue:45588", + "right": "issue:47000", + "accept": false, + "reason": "Flash-attention s_aux handling and RTDetr loss infeasibility are unrelated model/runtime errors." + }, + { + "left": "issue:44016", + "right": "issue:44291", + "accept": false, + "reason": "Notebook syntax error and init_empty_weights argument handling are completely different problems." + }, + { + "left": "issue:44683", + "right": "issue:45542", + "accept": false, + "reason": "Compiled flex_attention on torch>=2.9 and tensorboard-without-TF backend errors are unrelated dependency/runtime issues." + }, + { + "left": "issue:43479", + "right": "issue:43525", + "accept": false, + "reason": "Phi4Multimodal default vision/audio config initialization and missing Llama4Config.pad_token_id are different config bugs." + }, + { + "left": "issue:43441", + "right": "issue:45834", + "accept": false, + "reason": "Ministral FlashAttention failure and Kosmos2.5 long-OCR index error are unrelated model-specific bugs." + }, + { + "left": "issue:33290", + "right": "issue:45656", + "accept": false, + "reason": "OOM with Adafactor in DeepSpeed and optimizer.step being called twice are distinct optimizer/engine issues." + }, + { + "left": "issue:45081", + "right": "issue:45907", + "accept": false, + "reason": "Tokenizer regex crash and list_repo_files revision forwarding are unrelated utility 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": 26, + "review_unit_index": 42, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:41628", + "issue:42175", + "issue:42796", + "issue:42994", + "issue:43317", + "issue:43441", + "issue:43525", + "issue:43643", + "issue:43673", + "issue:43723", + "issue:43761", + "issue:43824", + "issue:43881", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44162", + "issue:44186", + "issue:44188", + "issue:44246", + "issue:44261", + "issue:44263", + "issue:44336", + "issue:44373", + "issue:44496", + "issue:44589", + "issue:44623", + "issue:44671", + "issue:44749", + "issue:44779", + "issue:45083", + "issue:45103", + "issue:45325", + "issue:45362", + "issue:45372", + "issue:45397", + "issue:45542", + "issue:45600", + "issue:45712", + "issue:45800", + "issue:45823", + "issue:45859", + "issue:46291", + "issue:46399", + "issue:46655", + "issue:46726", + "issue:46851", + "issue:47000" + ], + "soft_pairs": [ + "issue:42994|issue:43317", + "issue:45103|issue:45372", + "issue:44336|issue:46851", + "issue:44623|issue:44749", + "issue:45859|issue:46851", + "issue:42175|issue:43976", + "issue:44261|issue:44263", + "issue:45600|issue:45712", + "issue:44623|issue:45397", + "issue:44671|issue:44779", + "issue:43901|issue:44246", + "issue:43673|issue:43761", + "issue:43723|issue:44162", + "issue:43525|issue:43881", + "issue:44186|issue:47000", + "issue:43824|issue:43976", + "issue:44016|issue:45823", + "issue:42796|issue:43441", + "issue:45542|issue:45800", + "issue:44188|issue:44336", + "issue:43525|issue:43643", + "issue:44589|issue:45362", + "issue:44336|issue:44373", + "issue:45325|issue:46851", + "issue:46655|issue:46726", + "issue:46291|issue:46399", + "issue:44246|issue:44496", + "issue:41628|issue:43976", + "issue:45083|issue:46851" + ], + "prepared_review_unit_hash": "efc3410b05e145810921f976ab95ab0a335d4aead2b285828d50e39ecff84d9b", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36372, + "estimated_input_tokens": 9093, + "estimated_eval_tokens": 18442 + }, + "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 v5 regressions, import/load failures, tokenizer/processor bugs, and performance issues. None of the soft pairs look like the same concrete underlying bug, so I would not merge any of them as duplicates.", + "confidence": 0.87, + "canonical_issue_reason": "No single canonical issue fits well: the items span multiple distinct bugs and subsystems rather than one duplicate report.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43723 is the broadest user-facing loading regression in the set (AutoTokenizer.from_pretrained in v5) and is the best representative of the loading-related subset, though the cluster is not a true duplicate group.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42994", + "right": "issue:43317", + "accept": false, + "reason": "Both involve quantization/offload, but one is saving and the other is dequantized model loading; different code paths and failures." + }, + { + "left": "issue:45103", + "right": "issue:45372", + "accept": false, + "reason": "Unrelated bugs: auto-docstring annotation handling vs Gemma4 processor loading/import failure." + }, + { + "left": "issue:44336", + "right": "issue:46851", + "accept": false, + "reason": "ANSI log formatting and Qwen position_ids behavior are unrelated." + }, + { + "left": "issue:44623", + "right": "issue:44749", + "accept": false, + "reason": "Processor save_pretrained file emission is unrelated to a data-filtering slowdown regression." + }, + { + "left": "issue:45859", + "right": "issue:46851", + "accept": false, + "reason": "Both are Qwen3.5-related, but one is CPU-offload incompatibility and the other is variable-length position_ids; not the same bug." + }, + { + "left": "issue:42175", + "right": "issue:43976", + "accept": false, + "reason": "Different environment/setup issues: optional backend extras vs Python version compatibility." + }, + { + "left": "issue:44261", + "right": "issue:44263", + "accept": false, + "reason": "Different model helpers and failures; they do not share a concrete bug." + }, + { + "left": "issue:45600", + "right": "issue:45712", + "accept": false, + "reason": "Both touch repo maintenance/mapping, but one is stale auto-mapping entries and the other is dummy-class leakage; not one change." + }, + { + "left": "issue:44623", + "right": "issue:45397", + "accept": false, + "reason": "Saving processor files and ZeRO3 from_pretrained loading are separate problems." + }, + { + "left": "issue:44671", + "right": "issue:44779", + "accept": false, + "reason": "Different regressions affecting different models and paths: CamemBERT MLM vs Deepseek tokenization." + }, + { + "left": "issue:43901", + "right": "issue:44246", + "accept": false, + "reason": "Docs mismatch and import latency are unrelated." + }, + { + "left": "issue:43673", + "right": "issue:43761", + "accept": false, + "reason": "Different forward/output regressions on different models; too broad to be duplicates." + }, + { + "left": "issue:43723", + "right": "issue:44162", + "accept": false, + "reason": "Tokenizer loading in v5 and ESM2 breakage are distinct model-specific failures." + }, + { + "left": "issue:43525", + "right": "issue:43881", + "accept": false, + "reason": "Both are model loading issues, but the missing pad_token_id AttributeError and glm-4v-9b loading failure are not the same bug." + }, + { + "left": "issue:44186", + "right": "issue:47000", + "accept": false, + "reason": "Tokenizer padding/truncation crash is unrelated to RTDetrLoss infeasible cost-matrix failure." + }, + { + "left": "issue:43824", + "right": "issue:43976", + "accept": false, + "reason": "Import error for a specific class and Python compatibility are unrelated." + }, + { + "left": "issue:44016", + "right": "issue:45823", + "accept": false, + "reason": "Notebook syntax error and Gemma4 device mismatch are unrelated." + }, + { + "left": "issue:42796", + "right": "issue:43441", + "accept": false, + "reason": "Same family name, but different concrete failures: BPE marker handling vs FlashAttention loading." + }, + { + "left": "issue:45542", + "right": "issue:45800", + "accept": false, + "reason": "TensorFlow backend installation issue and torch/transformers incompatibility are separate environment problems." + }, + { + "left": "issue:44188", + "right": "issue:44336", + "accept": false, + "reason": "Attention-kernel divergence and ANSI code emission have nothing in common beyond being bugs." + }, + { + "left": "issue:43525", + "right": "issue:43643", + "accept": false, + "reason": "The latter is a generic trust_remote_code missing-fields issue; the former is a specific Llama4 pad_token_id AttributeError. Related symptom class, but not the same concrete bug." + }, + { + "left": "issue:44589", + "right": "issue:45362", + "accept": false, + "reason": "Float8 storage import failure and Qwen3.5 chat crash are unrelated." + }, + { + "left": "issue:44336", + "right": "issue:44373", + "accept": false, + "reason": "Loading-report ANSI formatting and a docstring typo are unrelated." + }, + { + "left": "issue:45325", + "right": "issue:46851", + "accept": false, + "reason": "Both mention position_ids, but they target different Qwen models and different temporal/variable-length bugs." + }, + { + "left": "issue:46655", + "right": "issue:46726", + "accept": false, + "reason": "Video padding calculation and missing vision backend handling are different processor issues." + }, + { + "left": "issue:46291", + "right": "issue:46399", + "accept": false, + "reason": "Kernel dependency/version gating is unrelated to a Qwen3.5 rotary-embedding registration crash." + }, + { + "left": "issue:44246", + "right": "issue:44496", + "accept": false, + "reason": "Import slowness and unrecognized model/config loading are unrelated." + }, + { + "left": "issue:41628", + "right": "issue:43976", + "accept": false, + "reason": "AutoImageProcessor import failure is unrelated to Python 3.9/3.10 compatibility." + }, + { + "left": "issue:45083", + "right": "issue:46851", + "accept": false, + "reason": "Different Qwen helper behavior on feature lengths vs position_ids; 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": 26, + "review_unit_index": 43, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:36246", + "issue:41950", + "issue:42491", + "issue:42757", + "issue:42886", + "issue:43257", + "issue:43262", + "issue:43299", + "issue:43526", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43784", + "issue:43901", + "issue:44220", + "issue:44242", + "issue:44246", + "issue:44295", + "issue:44297", + "issue:44336", + "issue:44351", + "issue:44451", + "issue:44485", + "issue:44496", + "issue:44779", + "issue:44871", + "issue:45003", + "issue:45072", + "issue:45103", + "issue:45137", + "issue:45292", + "issue:45362", + "issue:45397", + "issue:45542", + "issue:45561", + "issue:45563", + "issue:45588", + "issue:45593", + "issue:46132", + "issue:46399", + "issue:46506", + "issue:46614", + "issue:46655", + "issue:46665", + "issue:46752", + "issue:46821", + "issue:46829", + "issue:46851" + ], + "soft_pairs": [ + "issue:43784|issue:45542", + "issue:44297|issue:45397", + "issue:46399|issue:46851", + "issue:45561|issue:45563", + "issue:43526|issue:45003", + "issue:45103|issue:46399", + "issue:45072|issue:46829", + "issue:43540|issue:46752", + "issue:43531|issue:45362", + "issue:44246|issue:44871", + "issue:46614|issue:46665", + "issue:45588|issue:46851", + "issue:42491|issue:43299", + "issue:44295|issue:45397", + "issue:44242|issue:45593", + "issue:45003|issue:46821", + "issue:44336|issue:44485", + "issue:42886|issue:46132", + "issue:43550|issue:46506", + "issue:44779|issue:45292", + "issue:36246|issue:42757", + "issue:36246|issue:44351", + "issue:43901|issue:45003", + "issue:43257|issue:45137", + "issue:44451|issue:46752", + "issue:41950|issue:44220", + "issue:43540|issue:46655", + "issue:44496|issue:45003", + "issue:43262|issue:44246", + "issue:45003|issue:45561" + ], + "prepared_review_unit_hash": "c56d02123736878dadf633f72e6d52ce260c31e4d30f6bea2ddfa56cc43fcef7", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36689, + "estimated_input_tokens": 9173, + "estimated_eval_tokens": 18602 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are mostly unrelated issues across different models, processors, and infrastructure paths. None of the soft-edge pairs look like the same underlying bug/change closely enough to merge.", + "confidence": 0.86, + "canonical_issue_reason": "No single canonical issue stands out: the set spans many distinct regressions (imports, model loading, tokenizers, audio/video processors, deepspeed, docs) with no clear duplicate center.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue is a strong global representative for the cluster because the topics are heterogeneous and the pairwise overlaps are only superficial.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43784", + "right": "issue:45542", + "accept": false, + "reason": "Both are import/backend errors, but they involve different symbols and subsystems: sentence-transformers import vs TensorBoard/TF backend detection." + }, + { + "left": "issue:44297", + "right": "issue:45397", + "accept": false, + "reason": "Different failures: tokenizer save metadata mismatch vs gemma-4 zero3 from_pretrained loading." + }, + { + "left": "issue:46399", + "right": "issue:46851", + "accept": false, + "reason": "Same model family, but different bugs: registration/decorator crash vs variable-length position_ids handling." + }, + { + "left": "issue:45561", + "right": "issue:45563", + "accept": false, + "reason": "Completely different code paths: xdist captured_info.txt race vs stale generate() warning." + }, + { + "left": "issue:43526", + "right": "issue:45003", + "accept": false, + "reason": "Unrelated: BEiT image-label reduction bug vs unsafe sys.modules access in modeling_utils." + }, + { + "left": "issue:45103", + "right": "issue:46399", + "accept": false, + "reason": "Auto-docstring annotation crash is unrelated to Qwen3.5 kernel registration." + }, + { + "left": "issue:45072", + "right": "issue:46829", + "accept": false, + "reason": "Different model paths and symptoms: dtype mismatch in inference vs incorrect hidden_states field in classification output." + }, + { + "left": "issue:43540", + "right": "issue:46752", + "accept": false, + "reason": "Different areas: Qwen3OmniMoe video processing error vs empty apply_chat_template input handling." + }, + { + "left": "issue:43531", + "right": "issue:45362", + "accept": false, + "reason": "Both mention Qwen3 variants, but one is sliding_window behavior and the other is chat loading crash." + }, + { + "left": "issue:44246", + "right": "issue:44871", + "accept": false, + "reason": "MoE load-balancing loss issue is unrelated to Gemma-3 eos_token_id configuration." + }, + { + "left": "issue:46614", + "right": "issue:46665", + "accept": false, + "reason": "Stale phonemizer language backend is a different bug from inconsistent from_pretrained behavior in Python classes." + }, + { + "left": "issue:45588", + "right": "issue:46851", + "accept": false, + "reason": "Flash-attention sinkless-model crash is unrelated to variable-length position_ids." + }, + { + "left": "issue:42491", + "right": "issue:43299", + "accept": false, + "reason": "Both concern Qwen MoE loading regressions, but they target different model variants and failure modes; not the same concrete bug." + }, + { + "left": "issue:44295", + "right": "issue:45397", + "accept": false, + "reason": "Position_ids buffer access and gemma-4 zero3 loading are distinct issues." + }, + { + "left": "issue:44242", + "right": "issue:45593", + "accept": false, + "reason": "Different training losses and different models; only superficial similarity around missing auxiliary loss." + }, + { + "left": "issue:45003", + "right": "issue:46821", + "accept": false, + "reason": "sys.modules access bug is unrelated to TimesFM window_size attribute handling." + }, + { + "left": "issue:44336", + "right": "issue:44485", + "accept": false, + "reason": "ANSI-code reporting bug has nothing to do with GLM-5 RoPE implementation." + }, + { + "left": "issue:42886", + "right": "issue:46132", + "accept": false, + "reason": "Offline cache loading and AttentionInterface.register behavior are unrelated." + }, + { + "left": "issue:43550", + "right": "issue:46506", + "accept": false, + "reason": "torch.compile/SDPA failure in Bamba is unrelated to CPU/CUDA output differences in PPDocLayoutV3." + }, + { + "left": "issue:44779", + "right": "issue:45292", + "accept": false, + "reason": "DeepSeek tokenizer correctness regression is not the same as resize_token_embeddings not updating output embeddings." + }, + { + "left": "issue:36246", + "right": "issue:42757", + "accept": false, + "reason": "Both are import errors, but for different missing symbols from different packages." + }, + { + "left": "issue:36246", + "right": "issue:44351", + "accept": false, + "reason": "Qwen2_5_VLImageProcessor import failure is unrelated to HybridCache export failure." + }, + { + "left": "issue:43901", + "right": "issue:45003", + "accept": false, + "reason": "Documentation mismatch vs unsafe sys.modules access are unrelated." + }, + { + "left": "issue:43257", + "right": "issue:45137", + "accept": false, + "reason": "Both involve distributed training stacks, but one is weights conversion with accelerate/deepspeed and the other is a ZeRO3 deque pop error." + }, + { + "left": "issue:44451", + "right": "issue:46752", + "accept": false, + "reason": "Model loading failure for ScandiBERT is unrelated to empty-list chat template crashes." + }, + { + "left": "issue:41950", + "right": "issue:44220", + "accept": false, + "reason": "Video-classification pipeline using image processors and torch_extract_fbank_features are different pipeline bugs." + }, + { + "left": "issue:43540", + "right": "issue:46655", + "accept": false, + "reason": "Both touch video processing, but they affect different models and different logic: input validation vs padding calculation." + }, + { + "left": "issue:44496", + "right": "issue:45003", + "accept": false, + "reason": "Unrecognized model config is unrelated to sys.modules import safety." + }, + { + "left": "issue:43262", + "right": "issue:44246", + "accept": false, + "reason": "Audio chat-template sample-rate default and MoE load-balancing loss are unrelated." + }, + { + "left": "issue:45003", + "right": "issue:45561", + "accept": false, + "reason": "Different subsystems: modeling_utils import lookup vs pytest-xdist file race." + } + ] + }, + "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": 26, + "review_unit_index": 44, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:29870", + "issue:30333", + "issue:36246", + "issue:42199", + "issue:43295", + "issue:43352", + "issue:43475", + "issue:43526", + "issue:43596", + "issue:43653", + "issue:44038", + "issue:44060", + "issue:44077", + "issue:44162", + "issue:44242", + "issue:44295", + "issue:44368", + "issue:44466", + "issue:44479", + "issue:44623", + "issue:44749", + "issue:44857", + "issue:44877", + "issue:44977", + "issue:44987", + "issue:45137", + "issue:45200", + "issue:45250", + "issue:45292", + "issue:45335", + "issue:45375", + "issue:45412", + "issue:45459", + "issue:45600", + "issue:45759", + "issue:45859", + "issue:46123", + "issue:46170", + "issue:46369", + "issue:46399", + "issue:46506", + "issue:46655", + "issue:46726", + "issue:46851", + "issue:46894", + "issue:46897", + "issue:47000" + ], + "soft_pairs": [ + "issue:45375|issue:45759", + "issue:44479|issue:46655", + "issue:44162|issue:45200", + "issue:45292|issue:47000", + "issue:44295|issue:44749", + "issue:45859|issue:46399", + "issue:45137|issue:45600", + "issue:43295|issue:45292", + "issue:43475|issue:46369", + "issue:36246|issue:43596", + "issue:44466|issue:45335", + "issue:44242|issue:44857", + "issue:44623|issue:46726", + "issue:44877|issue:45375", + "issue:43526|issue:46897", + "issue:44038|issue:46506", + "issue:44987|issue:45459", + "issue:29870|issue:30333", + "issue:44479|issue:46170", + "issue:43526|issue:46894", + "issue:42199|issue:46851", + "issue:44060|issue:44077", + "issue:43352|issue:45250", + "issue:44368|issue:44977", + "issue:43653|issue:44162", + "issue:45412|issue:46123" + ], + "prepared_review_unit_hash": "fdb539227d30ab2ebbea8ed8190da3673f233328186a5a1c9aac990bf57951de", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 26, + "serialized_chars": 35257, + "estimated_input_tokens": 8815, + "estimated_eval_tokens": 17886 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These soft pairs are all false positives: they share broad model families or subsystems, but the underlying bugs/change requests are different. No PRs are present.", + "confidence": 0.95, + "canonical_issue_reason": "issue:44479 is a concrete, high-signal regression report with multiple affected variants and active discussion, making it a reasonable representative issue among this set.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44479 is the strongest standalone issue candidate because it is specific, user-impacting, and broad enough to represent a real regression class.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45375", + "right": "issue:45759", + "accept": false, + "reason": "Both are Qwen multimodal config problems, but one is a missing config field being dropped by strict parsing and the other is AutoModelForCausalLM.from_config not unwrapping text_config. Different failure modes and fixes." + }, + { + "left": "issue:44479", + "right": "issue:46655", + "accept": false, + "reason": "Both mention video inputs, but one is a Qwen-VL regression across several models while the other is a SmolVLM padding bug in the video processor. Not the same bug." + }, + { + "left": "issue:44162", + "right": "issue:45200", + "accept": false, + "reason": "Unrelated models and symptoms: ESM2 is broadly broken, while Gemma 4 needs mm_token_type_ids to default to zeros for text-only fine-tuning." + }, + { + "left": "issue:45292", + "right": "issue:47000", + "accept": false, + "reason": "resize_token_embeddings/output_embeddings is unrelated to RT-DETR loss infeasibility. Different subsystems and root causes." + }, + { + "left": "issue:44295", + "right": "issue:44749", + "accept": false, + "reason": "One is a position_ids buffer access error; the other is a major performance regression after upgrade. No shared bug path." + }, + { + "left": "issue:45859", + "right": "issue:46399", + "accept": false, + "reason": "Both are Qwen3.5 rotary-embedding-related, but one is CPU offload incompatibility and the other is a kernel registration decorator crash. Different concrete issues." + }, + { + "left": "issue:45137", + "right": "issue:45600", + "accept": false, + "reason": "DeepSpeed ZeRO3 deque failure and stale auto_mappings references to removed configs are unrelated." + }, + { + "left": "issue:43295", + "right": "issue:45292", + "accept": false, + "reason": "Processor/tokenizer regression has nothing to do with resize_token_embeddings not updating output_embeddings." + }, + { + "left": "issue:43475", + "right": "issue:46369", + "accept": false, + "reason": "Both involve SAM3, but one is a missing attribute on a vision encoder output and the other is a memory overflow in sam3-litetext. Different problems." + }, + { + "left": "issue:36246", + "right": "issue:43596", + "accept": false, + "reason": "ImportError for Qwen2.5-VL image processor and a DeepSpeed/BertModel IndexError are unrelated." + }, + { + "left": "issue:44466", + "right": "issue:45335", + "accept": false, + "reason": "Both mention tied/resized embeddings, but one is inconsistent serialization across devices and the other is t5gemma resize_token_embeddings not propagating to decoder.embed_tokens." + }, + { + "left": "issue:44242", + "right": "issue:44857", + "accept": false, + "reason": "Load-balancing loss omission and LwDetrImageLoss float16 CUDA crash are different code paths." + }, + { + "left": "issue:44623", + "right": "issue:46726", + "accept": false, + "reason": "Missing files in processor.save_pretrained and AutoProcessor complaining about a missing vision backend are not the same issue." + }, + { + "left": "issue:44877", + "right": "issue:45375", + "accept": false, + "reason": "Both involve strict config handling, but they affect different model families and different missing fields; not the same underlying bug." + }, + { + "left": "issue:43526", + "right": "issue:46897", + "accept": false, + "reason": "BeitImageProcessorFast reduce_labels truncation and CohereASR training-loss double-shift are unrelated." + }, + { + "left": "issue:44038", + "right": "issue:46506", + "accept": false, + "reason": "Qwen3-VL-Moe/transformers 5.0 compatibility and PPDocLayoutV3 CPU-vs-CUDA divergence are unrelated." + }, + { + "left": "issue:44987", + "right": "issue:45459", + "accept": false, + "reason": "Package loading failure and exception masking when protobuf is missing are different bugs." + }, + { + "left": "issue:29870", + "right": "issue:30333", + "accept": false, + "reason": "Bart generation UnboundLocalError and MLflow reporting after failed training jobs are unrelated." + }, + { + "left": "issue:44479", + "right": "issue:46170", + "accept": false, + "reason": "A video-input regression in Qwen models is unrelated to Windows/ROCm import and torch.distributed dependency errors." + }, + { + "left": "issue:43526", + "right": "issue:46894", + "accept": false, + "reason": "Different models and symptoms: BeitImageProcessorFast label reduction bug versus CohereASR double-shift training loss bug." + }, + { + "left": "issue:42199", + "right": "issue:46851", + "accept": false, + "reason": "DETR cardinality error and Qwen3.5 variable-length position_ids behavior are unrelated." + }, + { + "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 change or bug." + }, + { + "left": "issue:43352", + "right": "issue:45250", + "accept": false, + "reason": "A model-specific Flash Attention 2.0 unsupported error is not the same as a generic Flash Attention 2.0 issue." + }, + { + "left": "issue:44368", + "right": "issue:44977", + "accept": false, + "reason": "A tie_word_embeddings warning during LoRA fine-tuning is unrelated to a Qwen3.5 flash-attention generation failure." + }, + { + "left": "issue:43653", + "right": "issue:44162", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and ESM2 breakage are unrelated." + }, + { + "left": "issue:45412", + "right": "issue:46123", + "accept": false, + "reason": "RT-DETR memory cleanup and MaskGenerationPipeline final-batch dropping are different bugs in different components." + } + ] + }, + "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": 26, + "review_unit_index": 45, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:42886", + "issue:43065", + "issue:43408", + "issue:43475", + "issue:43540", + "issue:43643", + "issue:43646", + "issue:43701", + "issue:43723", + "issue:43906", + "issue:43931", + "issue:43994", + "issue:44162", + "issue:44188", + "issue:44297", + "issue:44361", + "issue:44368", + "issue:44483", + "issue:44623", + "issue:44749", + "issue:44871", + "issue:44877", + "issue:44995", + "issue:45103", + "issue:45200", + "issue:45229", + "issue:45375", + "issue:45397", + "issue:45478", + "issue:45563", + "issue:45588", + "issue:45656", + "issue:45685", + "issue:45987", + "issue:46082", + "issue:46123", + "issue:46132", + "issue:46170", + "issue:46202", + "issue:46421", + "issue:46431", + "issue:46710", + "issue:46726", + "issue:46752", + "issue:46832", + "issue:46858", + "issue:47020" + ], + "soft_pairs": [ + "issue:42886|issue:45229", + "issue:45397|issue:45656", + "issue:44877|issue:46170", + "issue:43475|issue:43540", + "issue:43065|issue:43408", + "issue:46170|issue:46832", + "issue:43723|issue:46710", + "issue:44368|issue:45563", + "issue:43931|issue:44368", + "issue:44995|issue:45200", + "issue:44188|issue:46202", + "issue:44483|issue:46202", + "issue:43906|issue:46202", + "issue:45987|issue:46082", + "issue:44188|issue:46421", + "issue:44995|issue:45588", + "issue:45685|issue:46431", + "issue:45103|issue:46858", + "issue:44361|issue:44749", + "issue:43065|issue:46123", + "issue:44297|issue:44623", + "issue:43994|issue:46726", + "issue:46752|issue:47020", + "issue:44749|issue:45656", + "issue:43701|issue:44749", + "issue:43646|issue:44162", + "issue:44623|issue:45656", + "issue:44871|issue:46132", + "issue:43475|issue:43643", + "issue:45375|issue:45478" + ], + "prepared_review_unit_hash": "33ea79e6639d7e3f1da946d1cfc84dfaf9d425c3fc3e7c6d1c1cea0e69c6d619", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36281, + "estimated_input_tokens": 9071, + "estimated_eval_tokens": 18398 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose grab-bag of mostly unrelated bug reports across tokenizers, config loading, vision models, chat templates, cache behavior, and performance regressions. None of the soft pairs look like the same concrete underlying bug/change, so I would not merge any of them as duplicates.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42886", + "right": "issue:45229", + "accept": false, + "reason": "Offline tokenizer cache loading and Gemma4 CUDA OOM are unrelated failures in different code paths." + }, + { + "left": "issue:45397", + "right": "issue:45656", + "accept": false, + "reason": "Gemma4 zero3 from_pretrained breakage and double optimizer-step with deepspeed are different bugs." + }, + { + "left": "issue:44877", + "right": "issue:46170", + "accept": false, + "reason": "Strict config loading for granite_speech and Windows/ROCm torch.distributed import issues do not share a concrete root cause." + }, + { + "left": "issue:43475", + "right": "issue:43540", + "accept": false, + "reason": "SAM3 video missing field and Qwen3OmniMoe video ValueError are separate model-specific loading/processing bugs." + }, + { + "left": "issue:43065", + "right": "issue:43408", + "accept": false, + "reason": "Dummy Conv2d in Sam3PixelDecoder and sam3_video vs sam3_tracker warning are different SAM3 issues." + }, + { + "left": "issue:46170", + "right": "issue:46832", + "accept": false, + "reason": "Windows ROCm import/dependency failure is unrelated to RT-DETR forward TypeError." + }, + { + "left": "issue:43723", + "right": "issue:46710", + "accept": false, + "reason": "Tokenizer loading in v5 and accuracy regression after version upgrade are not the same bug." + }, + { + "left": "issue:44368", + "right": "issue:45563", + "accept": false, + "reason": "A warning about tie_word_embeddings and a stale num_return_sequences warning affect different generation paths." + }, + { + "left": "issue:43931", + "right": "issue:44368", + "accept": false, + "reason": "Qwen3-VL weight-shape mismatch is unrelated to a Qwen3.5 LoRA warning." + }, + { + "left": "issue:44995", + "right": "issue:45200", + "accept": false, + "reason": "Stale indexer cache on second forward and Gemma4 mm_token_type_ids defaulting are separate issues." + }, + { + "left": "issue:44188", + "right": "issue:46202", + "accept": false, + "reason": "Torch.compile attention-kernel branching and Gemma tokenizer roundtrip on \u2581 are unrelated." + }, + { + "left": "issue:44483", + "right": "issue:46202", + "accept": false, + "reason": "Chat/completions request rejection and Gemma pipeline roundtrip failure are different request-processing bugs." + }, + { + "left": "issue:43906", + "right": "issue:46202", + "accept": false, + "reason": "An isolated reproduction of another issue and a Gemma roundtrip failure are not the same defect." + }, + { + "left": "issue:45987", + "right": "issue:46082", + "accept": false, + "reason": "StaticCache seq-length return type and LlamaConfig head_dim validation are unrelated API problems." + }, + { + "left": "issue:44188", + "right": "issue:46421", + "accept": false, + "reason": "Diverging attention kernels under torch.compile and cache lazy-init recompilation are different cache/compile behaviors." + }, + { + "left": "issue:44995", + "right": "issue:45588", + "accept": false, + "reason": "Stale cache on second forward and flash_attention s_aux=None crash do not share the same code path." + }, + { + "left": "issue:45685", + "right": "issue:46431", + "accept": false, + "reason": "MPS histogram type error and missing chat template error are unrelated." + }, + { + "left": "issue:45103", + "right": "issue:46858", + "accept": false, + "reason": "auto_docstring annotation handling and StaticCache seq-length tensor arithmetic under torch.compile are separate subsystems." + }, + { + "left": "issue:44361", + "right": "issue:44749", + "accept": false, + "reason": "MLukeTokenizer AttributeError and a performance regression after upgrade are not duplicates." + }, + { + "left": "issue:43065", + "right": "issue:46123", + "accept": false, + "reason": "SAM3 pixel-decoder placeholder and MaskGenerationPipeline partial-batch dropping are unrelated." + }, + { + "left": "issue:44297", + "right": "issue:44623", + "accept": false, + "reason": "Both involve save_pretrained, but tokenizer_config mismatch and processor file omission are different serialization bugs." + }, + { + "left": "issue:43994", + "right": "issue:46726", + "accept": false, + "reason": "Siglip2 nonsense outputs and AutoProcessor missing-backend error are not the same issue." + }, + { + "left": "issue:46752", + "right": "issue:47020", + "accept": false, + "reason": "Empty chat-template input crash and XLMRobertaTokenizer Unigram init type error are unrelated tokenizer bugs." + }, + { + "left": "issue:44749", + "right": "issue:45656", + "accept": false, + "reason": "Training slowdown and double optimizer-step behavior are different deepspeed-related problems." + }, + { + "left": "issue:43701", + "right": "issue:44749", + "accept": false, + "reason": "Checkpoint key mismatch and performance regression after upgrade do not match." + }, + { + "left": "issue:43646", + "right": "issue:44162", + "accept": false, + "reason": "AutoConfig missing fields with trust_remote_code and ESM2 breakage are separate model/config issues." + }, + { + "left": "issue:44623", + "right": "issue:45656", + "accept": false, + "reason": "Missing processor save files and duplicate optimizer stepping are unrelated." + }, + { + "left": "issue:44871", + "right": "issue:46132", + "accept": false, + "reason": "Gemma eos_token_id inconsistency and AttentionInterface.register behavior change are different API problems." + }, + { + "left": "issue:43475", + "right": "issue:43643", + "accept": false, + "reason": "SAM3 missing attribute and AutoConfig trust_remote_code field loss are unrelated loading bugs." + }, + { + "left": "issue:45375", + "right": "issue:45478", + "accept": false, + "reason": "Missing deepstack_visual_indexes in config and generic Qwen3.5 MoE from_pretrained error are related only by family, 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": 26, + "review_unit_index": 46, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:28282", + "issue:30990", + "issue:36010", + "issue:39692", + "issue:41628", + "issue:42371", + "issue:42994", + "issue:43232", + "issue:43475", + "issue:43482", + "issue:43525", + "issue:43673", + "issue:43723", + "issue:43784", + "issue:43873", + "issue:43986", + "issue:44038", + "issue:44188", + "issue:44351", + "issue:44373", + "issue:44623", + "issue:44857", + "issue:44869", + "issue:44933", + "issue:44945", + "issue:44964", + "issue:45081", + "issue:45092", + "issue:45103", + "issue:45137", + "issue:45230", + "issue:45310", + "issue:45356", + "issue:45375", + "issue:45397", + "issue:45406", + "issue:45593", + "issue:45753", + "issue:46202", + "issue:46326", + "issue:46399", + "issue:46463", + "issue:46701", + "issue:46726", + "issue:46821", + "issue:46851", + "issue:46877", + "issue:47000" + ], + "soft_pairs": [ + "issue:44964|issue:45137", + "issue:42994|issue:43482", + "issue:43475|issue:46821", + "issue:28282|issue:30990", + "issue:44188|issue:46851", + "issue:39692|issue:46726", + "issue:43232|issue:43673", + "issue:43986|issue:46726", + "issue:41628|issue:44933", + "issue:46877|issue:47000", + "issue:45310|issue:45375", + "issue:46463|issue:46877", + "issue:45230|issue:45397", + "issue:44869|issue:45356", + "issue:44373|issue:46851", + "issue:44038|issue:44623", + "issue:42994|issue:43525", + "issue:43723|issue:46326", + "issue:42371|issue:45092", + "issue:43873|issue:44945", + "issue:43784|issue:46399", + "issue:45406|issue:46202", + "issue:45137|issue:47000", + "issue:44945|issue:45753", + "issue:41628|issue:44351", + "issue:36010|issue:41628", + "issue:44857|issue:46877", + "issue:45081|issue:45103", + "issue:46202|issue:46701", + "issue:45593|issue:46877", + "issue:44038|issue:45310" + ], + "prepared_review_unit_hash": "41bd05897c0ee73432360e332bfb42d6132de587648a1bc8f54ac498fea2b22b", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36720, + "estimated_input_tokens": 9180, + "estimated_eval_tokens": 18616 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is heterogeneous: it mixes import/API regressions, model-loading issues, tokenizer/pipeline bugs, docs problems, and training/runtime crashes. None of the soft pairs is safely the same concrete bug, so I would not merge any of them as duplicates.", + "confidence": 0.59, + "canonical_issue_reason": "Issue 41628 is the cleanest representative of the recurring import/API-breakage theme in this cluster and is broader/more canonical than the more niche model- or docs-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 41628 is the most self-contained and broadly representative bug report here, making it the best anchor for triage even though the cluster is not a true duplicate set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44964", + "right": "issue:45137", + "accept": false, + "reason": "Different failures: Phi-4 multimodal loading versus a DeepSpeed ZeRO3 deque underflow." + }, + { + "left": "issue:42994", + "right": "issue:43482", + "accept": false, + "reason": "Quantized-model saving and GGUF loading are opposite directions and use different code paths." + }, + { + "left": "issue:43475", + "right": "issue:46821", + "accept": false, + "reason": "Different model families and different attribute errors; no shared concrete bug." + }, + { + "left": "issue:28282", + "right": "issue:30990", + "accept": false, + "reason": "Missing PyTorch at import time versus Sentence Transformers hanging while loading are unrelated." + }, + { + "left": "issue:44188", + "right": "issue:46851", + "accept": false, + "reason": "torch.compile attention-kernel divergence is not the same as Qwen3.5 variable-length position_ids handling." + }, + { + "left": "issue:39692", + "right": "issue:46726", + "accept": false, + "reason": "Both involve vision processors, but one is a broken docs example and the other is missing-backend handling; not the same bug." + }, + { + "left": "issue:43232", + "right": "issue:43673", + "accept": false, + "reason": "Generation kwargs/update logic and missing GenerationMixin cache in chunked prefill are distinct generation internals." + }, + { + "left": "issue:43986", + "right": "issue:46726", + "accept": false, + "reason": "Both mention missing vision dependencies, but one is a video-model AutoProcessor/torchvision crash and the other is an unrecognized image-processor error." + }, + { + "left": "issue:41628", + "right": "issue:44933", + "accept": false, + "reason": "Top-level AutoImageProcessor export failure and a missing import from image_utils are different missing symbols." + }, + { + "left": "issue:46877", + "right": "issue:47000", + "accept": false, + "reason": "Doc-lint stdout noise at import time is unrelated to an RTDetrLoss infeasible-cost-matrix crash." + }, + { + "left": "issue:45310", + "right": "issue:45375", + "accept": false, + "reason": "Same model family, but one is a broad from_pretrained failure and the other is a specific missing config field; not enough evidence of one bug." + }, + { + "left": "issue:46463", + "right": "issue:46877", + "accept": false, + "reason": "T5 decoder-input validation messaging and auto_docstring stdout diagnostics are unrelated." + }, + { + "left": "issue:45230", + "right": "issue:45397", + "accept": false, + "reason": "Generic bug report versus a Gemma-4 ZeRO3 loading issue; no overlap in concrete behavior." + }, + { + "left": "issue:44869", + "right": "issue:45356", + "accept": false, + "reason": "Whisper timestamp decoding and Kimi tokenizer codec/regression handling are different tokenizer bugs." + }, + { + "left": "issue:44373", + "right": "issue:46851", + "accept": false, + "reason": "A wrong docstring is not the same as a runtime position_ids bug." + }, + { + "left": "issue:44038", + "right": "issue:44623", + "accept": false, + "reason": "Qwen3-VL-Moe loading and processor.save_pretrained missing files are separate issues." + }, + { + "left": "issue:42994", + "right": "issue:43525", + "accept": false, + "reason": "Quantized saving and a missing pad_token_id on Llama4Config are unrelated." + }, + { + "left": "issue:43723", + "right": "issue:46326", + "accept": false, + "reason": "AutoTokenizer loading in v5 and continue_final_message for mistral_common tokenizers are different tokenizer paths." + }, + { + "left": "issue:42371", + "right": "issue:45092", + "accept": false, + "reason": "TF32 API guidance and InternVL2 meta-initialization incompatibility are unrelated." + }, + { + "left": "issue:43873", + "right": "issue:44945", + "accept": false, + "reason": "Quantization offloading and pipeline-parallel output corruption are different execution bugs." + }, + { + "left": "issue:43784", + "right": "issue:46399", + "accept": false, + "reason": "sentence-transformers import failure and Qwen3.5 kernel-registration errors are unrelated." + }, + { + "left": "issue:45406", + "right": "issue:46202", + "accept": false, + "reason": "Gemma4Processor missing _tokenizer and Gemma pipeline roundtrip on '\u2581' are different bugs." + }, + { + "left": "issue:45137", + "right": "issue:47000", + "accept": false, + "reason": "DeepSpeed ZeRO3 deque underflow and RTDetrLoss infeasible-cost-matrix crash are unrelated." + }, + { + "left": "issue:44945", + "right": "issue:45753", + "accept": false, + "reason": "Pipeline-parallel output corruption is not the same as an image-triggered infinite loop." + }, + { + "left": "issue:41628", + "right": "issue:44351", + "accept": false, + "reason": "Both are import/export regressions, but they involve different missing APIs and different fixes." + }, + { + "left": "issue:36010", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors, but GenerationMixin import-path breakage and AutoImageProcessor export failure are different symbols/code paths." + }, + { + "left": "issue:44857", + "right": "issue:46877", + "accept": false, + "reason": "AMP/CUDA loss crashing and doc-lint stdout noise are unrelated." + }, + { + "left": "issue:45081", + "right": "issue:45103", + "accept": false, + "reason": "Mistral regex patch backend-tokenizer access and auto_docstring annotation handling are unrelated AttributeErrors." + }, + { + "left": "issue:46202", + "right": "issue:46701", + "accept": false, + "reason": "Gemma tokenizer roundtrip failure and DiffusionGemma image-text pipeline failure are different bugs." + }, + { + "left": "issue:45593", + "right": "issue:46877", + "accept": false, + "reason": "D-FINE auxiliary-loss wiring and auto_docstring logging are unrelated." + }, + { + "left": "issue:44038", + "right": "issue:45310", + "accept": false, + "reason": "Related model family, but Qwen3-VL-Moe loading and Qwen3.5 MoE from_pretrained errors are not the same concrete regression." + } + ] + }, + "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": 26, + "review_unit_index": 47, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:27957", + "issue:30333", + "issue:30990", + "issue:33290", + "issue:34689", + "issue:35141", + "issue:39692", + "issue:42175", + "issue:42222", + "issue:42673", + "issue:42831", + "issue:42898", + "issue:42994", + "issue:43232", + "issue:43295", + "issue:43479", + "issue:43504", + "issue:43704", + "issue:43723", + "issue:43784", + "issue:43824", + "issue:43976", + "issue:44016", + "issue:44079", + "issue:44222", + "issue:44496", + "issue:44568", + "issue:44623", + "issue:44734", + "issue:44779", + "issue:44792", + "issue:44869", + "issue:44933", + "issue:45103", + "issue:45200", + "issue:45310", + "issue:45405", + "issue:45561", + "issue:45600", + "issue:45712", + "issue:45812", + "issue:46077", + "issue:46123", + "issue:46231", + "issue:46693", + "issue:46829", + "issue:46877", + "issue:46909" + ], + "soft_pairs": [ + "issue:42994|issue:43824", + "issue:44222|issue:45310", + "issue:42673|issue:43704", + "issue:43232|issue:44792", + "issue:42831|issue:44222", + "issue:42175|issue:44933", + "issue:43504|issue:45561", + "issue:45561|issue:45600", + "issue:44496|issue:45561", + "issue:45405|issue:46877", + "issue:43976|issue:44933", + "issue:30333|issue:33290", + "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", + "issue:43479|issue:44869", + "issue:42898|issue:45812", + "issue:43723|issue:44568", + "issue:44016|issue:46231", + "issue:39692|issue:46909", + "issue:43479|issue:44079", + "issue:42898|issue:44779", + "issue:43784|issue:43824", + "issue:27957|issue:30333", + "issue:30333|issue:34689", + "issue:42898|issue:43295", + "issue:44734|issue:45103", + "issue:42222|issue:45200" + ], + "prepared_review_unit_hash": "6abad843d8304235e82a1b679f017968d80e717f0c023fc60ac760e6f44cfbca", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37305, + "estimated_input_tokens": 9327, + "estimated_eval_tokens": 18910 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly a grab bag of unrelated closed issues across different bugs, regressions, docs, and model-specific failures. None of the soft-similarity pairs are close enough to be treated as the same underlying issue or change.", + "confidence": 0.92, + "canonical_issue_reason": "Issue 42898 is the clearest self-contained report: it has a specific regression boundary (v5), an explicit behavior change, and a focused symptom around tokenization cleanup. If a single representative issue must be chosen, this one is the most canonical-style bug report in the set.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 42898 is the strongest representative issue overall because it is concrete, reproducible, and narrowly scoped to one API behavior change. Several others are also well-written, but this one best fits a canonical regression report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42994", + "right": "issue:43824", + "accept": false, + "reason": "Different problem classes: quantized model saving failure vs missing Qwen2.5-VL import. No shared code-path." + }, + { + "left": "issue:44222", + "right": "issue:45310", + "accept": false, + "reason": "Both mention FP8/MoE/Qwen3.5, but one is a save-pretrained issue and the other is a from_pretrained load regression. Not the same bug." + }, + { + "left": "issue:42673", + "right": "issue:43704", + "accept": false, + "reason": "These are the same title about VRAM leaks, but the second appears to be a duplicate report, not evidence of a mergeable distinct same-change pair; still, as soft-duplicate triage they look like the same underlying bug and would normally be rejected as separate items." + }, + { + "left": "issue:43232", + "right": "issue:44792", + "accept": false, + "reason": "Generation kwargs update after sync_gpus is unrelated to a janus image generation test failure. Different code paths." + }, + { + "left": "issue:42831", + "right": "issue:44222", + "accept": false, + "reason": "FineGrainedFP8 accuracy regression is distinct from FP8 save_pretrained/moe serialization. Same theme, different failure mode." + }, + { + "left": "issue:42175", + "right": "issue:44933", + "accept": false, + "reason": "TensorFlow backend packaging issue vs a nonexistent image_utils import. No overlap beyond packaging/import context." + }, + { + "left": "issue:43504", + "right": "issue:45561", + "accept": false, + "reason": "Beit pretrained preset loading bug is unrelated to the pytest-xdist captured_info.txt race. Different subsystem and symptom." + }, + { + "left": "issue:45561", + "right": "issue:45600", + "accept": false, + "reason": "Both touch testing infrastructure, but one is a file race in patched testing utils and the other is auto_mappings referencing removed configs. Not the same bug." + }, + { + "left": "issue:44496", + "right": "issue:45561", + "accept": false, + "reason": "Unrecognized model/config loading failure is unrelated to a test-worker file race. Different paths and fixes." + }, + { + "left": "issue:45405", + "right": "issue:46877", + "accept": false, + "reason": "PEFT version pin issue is a dependency/versioning problem; auto_docstring stdout diagnostics is import-time logging noise. Not the same change." + }, + { + "left": "issue:43976", + "right": "issue:44933", + "accept": false, + "reason": "Python version compatibility regression is unrelated to a missing import from image_utils. No shared code path." + }, + { + "left": "issue:30333", + "right": "issue:33290", + "accept": false, + "reason": "MLflow reporting on training failure and Adafactor OOM in DeepSpeed are different failure modes with different fixes." + }, + { + "left": "issue:30990", + "right": "issue:35141", + "accept": false, + "reason": "Sentence-transformers loading stall is unrelated to embedding resize/post_init reinitialization. Different subsystem." + }, + { + "left": "issue:42898", + "right": "issue:44568", + "accept": false, + "reason": "Tokenization cleanup behavior change and add_special_tokens/BOS-EOS regression both concern tokenizers, but they affect different APIs and behaviors." + }, + { + "left": "issue:42175", + "right": "issue:45712", + "accept": false, + "reason": "Backend dependency omission is unrelated to leftover dummy classes and dir(transformers) leakage. Different maintenance areas." + }, + { + "left": "issue:43479", + "right": "issue:46829", + "accept": false, + "reason": "Phi4Multimodal config initialization bug is unrelated to VideoPrism output field misassignment. Different model families and code paths." + }, + { + "left": "issue:46077", + "right": "issue:46123", + "accept": false, + "reason": "Gemma4 attention-mask construction and mask generation pipeline batch handling are separate generation bugs with different symptoms." + }, + { + "left": "issue:44623", + "right": "issue:45310", + "accept": false, + "reason": "Processor.save_pretrained missing files is a serialization bug, while Qwen3.5 MoE from_pretrained error is a loading/config issue. Not the same bug." + }, + { + "left": "issue:42673", + "right": "issue:46693", + "accept": false, + "reason": "Both involve VRAM/performance around FA2-like workloads, but one is a thread-related leak in Qwen3ForCausalLM and the other is a long-context performance regression after 5.2.0. Too broad to merge." + }, + { + "left": "issue:43479", + "right": "issue:44869", + "accept": false, + "reason": "Phi4Multimodal config defaults are unrelated to Whisper timestamp decoding on trailing replacement characters." + }, + { + "left": "issue:42898", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions, but Granite token IDs and clean_up_tokenization_spaces are distinct behaviors with different fixes." + }, + { + "left": "issue:43723", + "right": "issue:44568", + "accept": false, + "reason": "AutoTokenizer.from_pretrained loading issue in v5 is not the same as add_special_tokens not inserting BOS/EOS for mdeberta-v3-base." + }, + { + "left": "issue:44016", + "right": "issue:46231", + "accept": false, + "reason": "Notebook syntax error and documentation parameter-name mismatch are unrelated doc issues." + }, + { + "left": "issue:39692", + "right": "issue:46909", + "accept": false, + "reason": "SigLIP2 doc example errors and DIA doc example issues are different examples with different failure modes." + }, + { + "left": "issue:43479", + "right": "issue:44079", + "accept": false, + "reason": "Phi4Multimodal config initialization and ModelOutput key assignment are unrelated framework behaviors." + }, + { + "left": "issue:42898", + "right": "issue:44779", + "accept": false, + "reason": "Clean-up-tokenization behavior regression and DeepSeek tokenizer incorrect results are separate tokenizer bugs, not one underlying change." + }, + { + "left": "issue:43784", + "right": "issue:43824", + "accept": false, + "reason": "Sentence-transformers import NameError and Qwen2.5-VL import error are different missing-symbol/import problems." + }, + { + "left": "issue:27957", + "right": "issue:30333", + "accept": false, + "reason": "Flash Attention 2 support for XLM-Roberta is unrelated to MLflow not reporting failed jobs." + }, + { + "left": "issue:30333", + "right": "issue:34689", + "accept": false, + "reason": "MLflow job-status reporting and Llama 3.2 vision model loading regression are different bugs." + }, + { + "left": "issue:42898", + "right": "issue:43295", + "accept": false, + "reason": "Tokenization cleanup behavior change and processor.tokenizer/image-passing regression are separate v5 issues with different APIs." + }, + { + "left": "issue:44734", + "right": "issue:45103", + "accept": false, + "reason": "KV-cache continuation crash in serve is unrelated to auto_docstring crashing on future annotations." + }, + { + "left": "issue:42222", + "right": "issue:45200", + "accept": false, + "reason": "ViTPose breakage and Gemma4 mm_token_type_ids defaulting are unrelated model-specific bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 48, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:22355", + "issue:28282", + "issue:30333", + "issue:36331", + "issue:41628", + "issue:42757", + "issue:42796", + "issue:42831", + "issue:42898", + "issue:43116", + "issue:43122", + "issue:43232", + "issue:43479", + "issue:43644", + "issue:43723", + "issue:43874", + "issue:44060", + "issue:44075", + "issue:44190", + "issue:44222", + "issue:44295", + "issue:44297", + "issue:44360", + "issue:44373", + "issue:44448", + "issue:44484", + "issue:44485", + "issue:44623", + "issue:44933", + "issue:44977", + "issue:45071", + "issue:45092", + "issue:45125", + "issue:45245", + "issue:45310", + "issue:45356", + "issue:45405", + "issue:45656", + "issue:45712", + "issue:46082", + "issue:46132", + "issue:46209", + "issue:46231", + "issue:46424", + "issue:46614", + "issue:46710", + "issue:46762", + "issue:46932" + ], + "soft_pairs": [ + "issue:42898|issue:45356", + "issue:43874|issue:45092", + "issue:44295|issue:45656", + "issue:30333|issue:36331", + "issue:22355|issue:28282", + "issue:45405|issue:45712", + "issue:43116|issue:44190", + "issue:44222|issue:44297", + "issue:44075|issue:46231", + "issue:44060|issue:45125", + "issue:44933|issue:45712", + "issue:42831|issue:44623", + "issue:44448|issue:46132", + "issue:42796|issue:43723", + "issue:44297|issue:45656", + "issue:43479|issue:43644", + "issue:45245|issue:46132", + "issue:44222|issue:44623", + "issue:44360|issue:46762", + "issue:46132|issue:46710", + "issue:41628|issue:42757", + "issue:44373|issue:44485", + "issue:44297|issue:45310", + "issue:43723|issue:44977", + "issue:45071|issue:46082", + "issue:46614|issue:46932", + "issue:44222|issue:44484", + "issue:43122|issue:46209", + "issue:44190|issue:46082", + "issue:43232|issue:46424" + ], + "prepared_review_unit_hash": "7b90f4c9f99d4d7e278ece44d4b648b12c4fcda308437ddc7e484eaa07cbad9f", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36462, + "estimated_input_tokens": 9116, + "estimated_eval_tokens": 18488 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is broadly heterogeneous; the high-similarity pairs mostly share only surface wording (tokenizer, save_pretrained, import, config, or regression language) and do not look like the same underlying bug or change. I reject all soft edges. If forced to name a representative issue, issue:44222 is the densest concrete save_pretrained-related bug, but the cluster is not a true duplicate set.", + "confidence": 0.73, + "canonical_issue_reason": "issue:44222 is the most central concrete bug report in the set, with the densest save_pretrained-adjacent similarity, but this is only a weak representative because the cluster is heterogeneous.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44222 is the best single representative if one must be chosen: it is a concrete implementation bug and sits in the most connected part of the similarity graph.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42898", + "right": "issue:45356", + "accept": false, + "reason": "Both are tokenizer regressions, but one is about clean_up_tokenization_spaces behavior and the other about codec handling/warning text; different bugs." + }, + { + "left": "issue:43874", + "right": "issue:45092", + "accept": false, + "reason": "Both concern multimodal/v5 breakage, but one is a missing image-patch API and the other is remote-code/meta-init compatibility; not the same issue." + }, + { + "left": "issue:44295", + "right": "issue:45656", + "accept": false, + "reason": "Position_ids buffer access and double optimizer stepping with deepspeed are unrelated failures." + }, + { + "left": "issue:30333", + "right": "issue:36331", + "accept": false, + "reason": "MLflow reporting on failed jobs and a CustomTrainer compute_loss signature error are different training-path bugs." + }, + { + "left": "issue:22355", + "right": "issue:28282", + "accept": false, + "reason": "These are different import errors: missing transformers.onnx versus missing PyTorch dependency for AutoModel." + }, + { + "left": "issue:45405", + "right": "issue:45712", + "accept": false, + "reason": "PEFT version pinning and leftover dummy classes in dummy_pt_objects.py are unrelated maintenance issues." + }, + { + "left": "issue:43116", + "right": "issue:44190", + "accept": false, + "reason": "Different example scripts and different failure modes; one is empty multi-label outputs, the other is local dataset loading." + }, + { + "left": "issue:44222", + "right": "issue:44297", + "accept": false, + "reason": "Both mention save_pretrained, but one is FP8 MoE serialization and the other is tokenizer_config class mismatch; different code paths." + }, + { + "left": "issue:44075", + "right": "issue:46231", + "accept": false, + "reason": "Optimizer SGD arguments not being used is unrelated to a documentation parameter-name mismatch." + }, + { + "left": "issue:44060", + "right": "issue:45125", + "accept": false, + "reason": "Same model family only; one is a tied-weights warning, the other is a missing tensor-parallel plan." + }, + { + "left": "issue:44933", + "right": "issue:45712", + "accept": false, + "reason": "Different import/dummy-class problems with different root causes." + }, + { + "left": "issue:42831", + "right": "issue:44623", + "accept": false, + "reason": "FP8 accuracy regression and processor.save_pretrained missing files are distinct bugs." + }, + { + "left": "issue:44448", + "right": "issue:46132", + "accept": false, + "reason": "Pegasus output differences and AttentionInterface.register behavior changes are different code paths." + }, + { + "left": "issue:42796", + "right": "issue:43723", + "accept": false, + "reason": "Both are tokenizer-related, but BPE marker handling and AutoTokenizer loading are distinct bugs." + }, + { + "left": "issue:44297", + "right": "issue:45656", + "accept": false, + "reason": "Tokenizer serialization mismatch and deepspeed optimizer-step behavior are unrelated." + }, + { + "left": "issue:43479", + "right": "issue:43644", + "accept": false, + "reason": "Config default initialization for multimodal subconfigs and non-persistent buffer junk are different issues." + }, + { + "left": "issue:45245", + "right": "issue:46132", + "accept": false, + "reason": "Category-count runtime error and AttentionInterface.register behavior change are unrelated." + }, + { + "left": "issue:44222", + "right": "issue:44623", + "accept": false, + "reason": "Both involve save_pretrained, but FP8 MoE saving and processor file emission are different failures." + }, + { + "left": "issue:44360", + "right": "issue:46762", + "accept": false, + "reason": "DSA indexer missing ReLU and Minimax attention numerical drift are not the same bug." + }, + { + "left": "issue:46132", + "right": "issue:46710", + "accept": false, + "reason": "AttentionInterface.register behavior changes and a DeepSeek accuracy regression are different code-path problems." + }, + { + "left": "issue:41628", + "right": "issue:42757", + "accept": false, + "reason": "Both are import errors, but they arise from different packages and missing symbols." + }, + { + "left": "issue:44373", + "right": "issue:44485", + "accept": false, + "reason": "A wrong docstring and a GLM-5 RoPE implementation issue are unrelated." + }, + { + "left": "issue:44297", + "right": "issue:45310", + "accept": false, + "reason": "Tokenizer class metadata on save is unrelated to Qwen3.5 Moe from_pretrained failures." + }, + { + "left": "issue:43723", + "right": "issue:44977", + "accept": false, + "reason": "Tokenizer loading regression and flash-attention generation failure are different bugs." + }, + { + "left": "issue:45071", + "right": "issue:46082", + "accept": false, + "reason": "PretrainedConfig typing regression and LlamaConfig head_dim validation are distinct config issues." + }, + { + "left": "issue:46614", + "right": "issue:46932", + "accept": false, + "reason": "Stale phonemizer backend state and Gemma4 quantized embedder dtype casting are unrelated." + }, + { + "left": "issue:44222", + "right": "issue:44484", + "accept": false, + "reason": "Both mention save_pretrained, but shard-size defaults and FP8 MoE saving are not the same problem." + }, + { + "left": "issue:43122", + "right": "issue:46209", + "accept": false, + "reason": "Tokenization regression across versions and B200 FP8 inference failure are unrelated." + }, + { + "left": "issue:44190", + "right": "issue:46082", + "accept": false, + "reason": "Local dataset loading in an example script and a LlamaConfig validation bug are different issues." + }, + { + "left": "issue:43232", + "right": "issue:46424", + "accept": false, + "reason": "Generation sync_gpus handling and cache max_length enforcement are different generation/cache 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": 26, + "review_unit_index": 49, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:22355", + "issue:27957", + "issue:28218", + "issue:30990", + "issue:38175", + "issue:39401", + "issue:41628", + "issue:42757", + "issue:42831", + "issue:42886", + "issue:42907", + "issue:42994", + "issue:43066", + "issue:43232", + "issue:43408", + "issue:43475", + "issue:43644", + "issue:43824", + "issue:44038", + "issue:44188", + "issue:44261", + "issue:44263", + "issue:44279", + "issue:44295", + "issue:44360", + "issue:44373", + "issue:44483", + "issue:44568", + "issue:44623", + "issue:44829", + "issue:45092", + "issue:45278", + "issue:45310", + "issue:45357", + "issue:45381", + "issue:45406", + "issue:45753", + "issue:45834", + "issue:46002", + "issue:46050", + "issue:46209", + "issue:46305", + "issue:46328", + "issue:46369", + "issue:46402", + "issue:46431", + "issue:46655", + "issue:46909" + ], + "soft_pairs": [ + "issue:43232|issue:45834", + "issue:44263|issue:45092", + "issue:46209|issue:46431", + "issue:30990|issue:38175", + "issue:44295|issue:44623", + "issue:43475|issue:46305", + "issue:46305|issue:46655", + "issue:44263|issue:44373", + "issue:46369|issue:46909", + "issue:44263|issue:44360", + "issue:42907|issue:46050", + "issue:42831|issue:45310", + "issue:44279|issue:46328", + "issue:22355|issue:27957", + "issue:45092|issue:45357", + "issue:45381|issue:46305", + "issue:44483|issue:45406", + "issue:44829|issue:46002", + "issue:42831|issue:44038", + "issue:41628|issue:43824", + "issue:43644|issue:46305", + "issue:43408|issue:46402", + "issue:44261|issue:45092", + "issue:22355|issue:28218", + "issue:42831|issue:45753", + "issue:39401|issue:44568", + "issue:42886|issue:42994", + "issue:42886|issue:43066", + "issue:42757|issue:45278", + "issue:44188|issue:44373", + "issue:27957|issue:30990", + "issue:43408|issue:46369" + ], + "prepared_review_unit_hash": "916310409de5ecc25f7c086d2b904c65ff04314ed47030ab2f9ba41a015f292d", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 36792, + "estimated_input_tokens": 9198, + "estimated_eval_tokens": 18652 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly a grab-bag of unrelated Transformers issues. None of the soft edges look like true duplicates or the same concrete fix, so all candidate links are rejected.", + "confidence": 0.86, + "canonical_issue_reason": "issue:45278 is the broadest umbrella-style report here: a version-upgrade import regression with multiple missing symbols, so it is the least specific and most plausible anchor if one issue must represent the set.", + "canonical_pr_reason": null, + "best_issue_reason": "It is the most general, user-facing regression report among the items and could plausibly subsume some narrower import-error reports, but it is still not a duplicate of them.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43232", + "right": "issue:45834", + "accept": false, + "reason": "Different bug classes: generation kwargs handling vs OCR long-input indexing." + }, + { + "left": "issue:44263", + "right": "issue:45092", + "accept": false, + "reason": "Unrelated code paths: torch.split/indexer behavior vs remote-code checkpoint meta-init compatibility." + }, + { + "left": "issue:46209", + "right": "issue:46431", + "accept": false, + "reason": "FP8 inference failure and chat-template absence are unrelated issues." + }, + { + "left": "issue:30990", + "right": "issue:38175", + "accept": false, + "reason": "Different symptoms and models: loading stall vs zero-probability outputs." + }, + { + "left": "issue:44295", + "right": "issue:44623", + "accept": false, + "reason": "Buffer registration/read error is not the same as missing files in processor.save_pretrained." + }, + { + "left": "issue:43475", + "right": "issue:46305", + "accept": false, + "reason": "Different models and bugs: missing attribute in SAM3 vs sliding-window off-by-one." + }, + { + "left": "issue:46305", + "right": "issue:46655", + "accept": false, + "reason": "Both concern video processing, but the off-by-one mask bug is not the same as SmolVLM padding calculation." + }, + { + "left": "issue:44263", + "right": "issue:44373", + "accept": false, + "reason": "A code bug in DSA indexing is not the same as a docstring issue." + }, + { + "left": "issue:46369", + "right": "issue:46909", + "accept": false, + "reason": "Memory overflow in SAM3-lite-text is unrelated to a broken documentation example." + }, + { + "left": "issue:44263", + "right": "issue:44360", + "accept": false, + "reason": "Same subsystem, but torch.split return handling and missing ReLU are different fixes." + }, + { + "left": "issue:42907", + "right": "issue:46050", + "accept": false, + "reason": "Both touch quantized saving/loading, but they describe different quantization paths and failures." + }, + { + "left": "issue:42831", + "right": "issue:45310", + "accept": false, + "reason": "FineGrainedFP8 accuracy problems and Qwen3.5-MoE from_pretrained errors are different issues." + }, + { + "left": "issue:44279", + "right": "issue:46328", + "accept": false, + "reason": "A dependency problem is unrelated to a SequenceFeatureExtractor.pad performance regression." + }, + { + "left": "issue:22355", + "right": "issue:27957", + "accept": false, + "reason": "Importing transformers.onnx and XLMRoberta with Flash Attention 2 are unrelated." + }, + { + "left": "issue:45092", + "right": "issue:45357", + "accept": false, + "reason": "Both are multimodal model regressions, but the checkpoint/meta-init issue is not the same as incorrect visual-encoder keys on save_pretrained." + }, + { + "left": "issue:45381", + "right": "issue:46305", + "accept": false, + "reason": "Different models and code paths: Qwen2.5-VL vision_position_ids vs MoonshineStreaming mask off-by-one." + }, + { + "left": "issue:44483", + "right": "issue:45406", + "accept": false, + "reason": "Serve API request rejection and Gemma4Processor _tokenizer crash are different failures." + }, + { + "left": "issue:44829", + "right": "issue:46002", + "accept": false, + "reason": "Training divergence with flash_attention_3 and NaNs after checkout/init are not the same bug." + }, + { + "left": "issue:42831", + "right": "issue:44038", + "accept": false, + "reason": "Generic FP8 accuracy issue does not match a Qwen3-VL-Moe loading bug." + }, + { + "left": "issue:41628", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but they involve different missing symbols and likely different causes." + }, + { + "left": "issue:43644", + "right": "issue:46305", + "accept": false, + "reason": "Non-persistent buffer junk in v5.0 is unrelated to the MoonshineStreaming window bug." + }, + { + "left": "issue:43408", + "right": "issue:46402", + "accept": false, + "reason": "Both involve SAM3, but the model-type warning is not the same as missing text encoder weights." + }, + { + "left": "issue:44261", + "right": "issue:45092", + "accept": false, + "reason": "Layernorm epsilon precision mismatch and checkpoint meta-init incompatibility are unrelated." + }, + { + "left": "issue:22355", + "right": "issue:28218", + "accept": false, + "reason": "No module named transformers.onnx is unrelated to tokenizer spacing after added tokens." + }, + { + "left": "issue:42831", + "right": "issue:45753", + "accept": false, + "reason": "FP8 accuracy issues are not the same as a Qwen3_5 infinite-loop bug on a specific image." + }, + { + "left": "issue:39401", + "right": "issue:44568", + "accept": false, + "reason": "Different tokenizer bugs on different models: offset mapping vs BOS/EOS not added." + }, + { + "left": "issue:42886", + "right": "issue:42994", + "accept": false, + "reason": "Offline cache loading failure is unrelated to quantized model save failures." + }, + { + "left": "issue:42886", + "right": "issue:43066", + "accept": false, + "reason": "Cache/offline loading and decoder-type serialization are different tokenizer problems." + }, + { + "left": "issue:42757", + "right": "issue:45278", + "accept": false, + "reason": "Missing huggingface_hub import is a dependency-layer issue, not the same as Transformers import regressions after upgrade." + }, + { + "left": "issue:44188", + "right": "issue:44373", + "accept": false, + "reason": "torch.compile attention-kernel divergence is unrelated to a docstring correction." + }, + { + "left": "issue:27957", + "right": "issue:30990", + "accept": false, + "reason": "Flash Attention 2 on XLMRoberta and Sentence Transformers loading stall are unrelated." + }, + { + "left": "issue:43408", + "right": "issue:46369", + "accept": false, + "reason": "Both mention SAM3, but a config/model-type mismatch is not the same as memory overflow in sam3-litetext." + } + ] + }, + "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": 26, + "review_unit_index": 50, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:22355", + "issue:29159", + "issue:30990", + "issue:33357", + "issue:39401", + "issue:41628", + "issue:42371", + "issue:42796", + "issue:42907", + "issue:43122", + "issue:43232", + "issue:43408", + "issue:43452", + "issue:43475", + "issue:43540", + "issue:43576", + "issue:43646", + "issue:43756", + "issue:43784", + "issue:43937", + "issue:43976", + "issue:44038", + "issue:44060", + "issue:44117", + "issue:44263", + "issue:44295", + "issue:44297", + "issue:44360", + "issue:44485", + "issue:44488", + "issue:44521", + "issue:44625", + "issue:44704", + "issue:44829", + "issue:44945", + "issue:44964", + "issue:45216", + "issue:45278", + "issue:45290", + "issue:45381", + "issue:45478", + "issue:45593", + "issue:45600", + "issue:46002", + "issue:46172", + "issue:46326", + "issue:46762", + "issue:46909" + ], + "soft_pairs": [ + "issue:42796|issue:43122", + "issue:43232|issue:45593", + "issue:45381|issue:46909", + "issue:43540|issue:46909", + "issue:43646|issue:46002", + "issue:30990|issue:33357", + "issue:42907|issue:44945", + "issue:42907|issue:44964", + "issue:43976|issue:44945", + "issue:44295|issue:44297", + "issue:43756|issue:44060", + "issue:43576|issue:43784", + "issue:39401|issue:44488", + "issue:41628|issue:45278", + "issue:42907|issue:43937", + "issue:42371|issue:44263", + "issue:43784|issue:45600", + "issue:44117|issue:44704", + "issue:46172|issue:46762", + "issue:39401|issue:44521", + "issue:44625|issue:44829", + "issue:44704|issue:45290", + "issue:43408|issue:43475", + "issue:43122|issue:44521", + "issue:44360|issue:44485", + "issue:43122|issue:46326", + "issue:42907|issue:43452", + "issue:44038|issue:44297", + "issue:22355|issue:29159", + "issue:44038|issue:44295", + "issue:45216|issue:45478" + ], + "prepared_review_unit_hash": "5a94b7d6f128609db70b27d0a115a2b208a307cb672e127d4d1cf3836439163e", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36520, + "estimated_input_tokens": 9130, + "estimated_eval_tokens": 18516 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These issues are mostly unrelated standalone bugs/regressions. A few are in adjacent areas (tokenizers, imports, Qwen/SAM families), but none of the soft pairs look like the same concrete underlying defect.", + "confidence": 0.91, + "canonical_issue_reason": "Issue 45278 is the broadest representative regression report in the set (many import errors after upgrading), so it is the best anchor if one issue must serve as the cluster representative.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45278 is the most general, user-facing catch-all among these reports and the likeliest hub for related import-breakage complaints.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42796", + "right": "issue:43122", + "accept": false, + "reason": "Both are tokenizer-related, but one is a model-specific BPE-marker problem and the other is a version-wide tokenization regression." + }, + { + "left": "issue:43232", + "right": "issue:45593", + "accept": false, + "reason": "Different subsystems and failure modes: generation kwargs after sync_gpus versus missing auxiliary losses in D-FINE." + }, + { + "left": "issue:45381", + "right": "issue:46909", + "accept": false, + "reason": "Qwen2.5-VL video position IDs and a broken DIA docs example are unrelated." + }, + { + "left": "issue:43540", + "right": "issue:46909", + "accept": false, + "reason": "A video-input processing ValueError and a documentation example bug are not the same issue." + }, + { + "left": "issue:43646", + "right": "issue:46002", + "accept": false, + "reason": "Both mention initialization, but one is custom model init breakage and the other is NaNs during classification-head init." + }, + { + "left": "issue:30990", + "right": "issue:33357", + "accept": false, + "reason": "Different failures: SentenceTransformers hanging on load versus a Mac bus error loading a CLIP model." + }, + { + "left": "issue:42907", + "right": "issue:44945", + "accept": false, + "reason": "Saving dequantized Ministrals/Devstrals and incorrect pipeline-parallel LLM output are unrelated." + }, + { + "left": "issue:42907", + "right": "issue:44964", + "accept": false, + "reason": "A dequantized-save regression and a multimodal model load failure are different bugs." + }, + { + "left": "issue:43976", + "right": "issue:44945", + "accept": false, + "reason": "Python-version incompatibility and pipeline-parallel output corruption are unrelated." + }, + { + "left": "issue:44295", + "right": "issue:44297", + "accept": false, + "reason": "One is a model buffer access problem; the other is tokenizer metadata written by save_pretrained." + }, + { + "left": "issue:43756", + "right": "issue:44060", + "accept": false, + "reason": "Different models and different bugs: RoPE layer dropping versus tied-weights warnings." + }, + { + "left": "issue:43576", + "right": "issue:43784", + "accept": false, + "reason": "A broken CLI env command and a NameError on importing sentence-transformers are not the same code path." + }, + { + "left": "issue:39401", + "right": "issue:44488", + "accept": false, + "reason": "Wrong offset_mapping in Qwen3 tokenizer versus failure to load cjvt/sleng-bert are unrelated." + }, + { + "left": "issue:41628", + "right": "issue:45278", + "accept": false, + "reason": "Both are import regressions, but one is a specific missing symbol and the other is a broad upgrade-wide import breakage; not enough evidence of the same code path." + }, + { + "left": "issue:42907", + "right": "issue:43937", + "accept": false, + "reason": "Different model families and failure modes: save/dequantization versus invalid GenerationConfig." + }, + { + "left": "issue:42371", + "right": "issue:44263", + "accept": false, + "reason": "TF32 API guidance and a torch.split/indexer bug are unrelated." + }, + { + "left": "issue:43784", + "right": "issue:45600", + "accept": false, + "reason": "One is an import-time NameError in sentence-transformers; the other is a CI break from removed Sam3LiteText configs." + }, + { + "left": "issue:44117", + "right": "issue:44704", + "accept": false, + "reason": "Both touch from_pretrained helpers, but they fail in different places: TOKENIZER_MAPPING_NAMES versus cached_file kwargs propagation." + }, + { + "left": "issue:46172", + "right": "issue:46762", + "accept": false, + "reason": "Different model implementations and different numerical discrepancies; not the same bug." + }, + { + "left": "issue:39401", + "right": "issue:44521", + "accept": false, + "reason": "Tokenizer offset mapping and multimodal assistant masks are separate issues." + }, + { + "left": "issue:44625", + "right": "issue:44829", + "accept": false, + "reason": "Qwen3.5 num_labels propagation and flash_attention_3 training degeneration are unrelated." + }, + { + "left": "issue:44704", + "right": "issue:45290", + "accept": false, + "reason": "Both involve processing helpers, but AutoProcessor kwargs forwarding and chat-template tool-call crashes are different code paths." + }, + { + "left": "issue:43408", + "right": "issue:43475", + "accept": false, + "reason": "Both are SAM3 video-related, but one is a model-type warning and the other is a missing attribute in vision encoder output." + }, + { + "left": "issue:43122", + "right": "issue:44521", + "accept": false, + "reason": "A tokenizer version regression and all-zero assistant masks in multimodal chat templating are different defects." + }, + { + "left": "issue:44360", + "right": "issue:44485", + "accept": false, + "reason": "A DSA indexer activation question and a GLM-5 RoPE implementation issue are unrelated." + }, + { + "left": "issue:43122", + "right": "issue:46326", + "accept": false, + "reason": "Both involve tokenization, but one is a cross-version regression and the other is continue_final_message behavior for mistral_common tokenizers." + }, + { + "left": "issue:42907", + "right": "issue:43452", + "accept": false, + "reason": "Dequantized-saving failures and gguf_file loading failures are different pathways." + }, + { + "left": "issue:44038", + "right": "issue:44297", + "accept": false, + "reason": "Qwen3-VL-Moe loading issues and tokenizer save_pretrained metadata mismatches are unrelated." + }, + { + "left": "issue:22355", + "right": "issue:29159", + "accept": false, + "reason": "Missing transformers.onnx and slow/fast tokenizer inconsistency are different problems." + }, + { + "left": "issue:44038", + "right": "issue:44295", + "accept": false, + "reason": "Qwen3-VL-Moe loading and position_ids buffer access are unrelated." + }, + { + "left": "issue:45216", + "right": "issue:45478", + "accept": false, + "reason": "Same model family, but one is save_pretrained checkpoint correctness and the other is from_pretrained loading; too different to treat as one duplicate." + } + ] + }, + "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": 26, + "review_unit_index": 51, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:22355", + "issue:29127", + "issue:36296", + "issue:38617", + "issue:39401", + "issue:42175", + "issue:42673", + "issue:42757", + "issue:42898", + "issue:43066", + "issue:43116", + "issue:43122", + "issue:43278", + "issue:43493", + "issue:43611", + "issue:43824", + "issue:43937", + "issue:43976", + "issue:44038", + "issue:44075", + "issue:44188", + "issue:44222", + "issue:44351", + "issue:44355", + "issue:44485", + "issue:44610", + "issue:44625", + "issue:44779", + "issue:44936", + "issue:44938", + "issue:44945", + "issue:45072", + "issue:45083", + "issue:45200", + "issue:45278", + "issue:45357", + "issue:45446", + "issue:45478", + "issue:45753", + "issue:46002", + "issue:46231", + "issue:46328", + "issue:46614", + "issue:46726", + "issue:46877", + "issue:46909", + "issue:46981" + ], + "soft_pairs": [ + "issue:39401|issue:46614", + "issue:43824|issue:44351", + "issue:44038|issue:44222", + "issue:43116|issue:45200", + "issue:43122|issue:44779", + "issue:42757|issue:44351", + "issue:43278|issue:44945", + "issue:36296|issue:43116", + "issue:22355|issue:29127", + "issue:42175|issue:44938", + "issue:43493|issue:45753", + "issue:44188|issue:46328", + "issue:46231|issue:46877", + "issue:42898|issue:46614", + "issue:38617|issue:45278", + "issue:43937|issue:44485", + "issue:43976|issue:45446", + "issue:43066|issue:44779", + "issue:42673|issue:45072", + "issue:43611|issue:46002", + "issue:44038|issue:45478", + "issue:43116|issue:43493", + "issue:44355|issue:46231", + "issue:45357|issue:45478", + "issue:44625|issue:46002", + "issue:44610|issue:46909", + "issue:45083|issue:45446", + "issue:46726|issue:46981", + "issue:44075|issue:44936" + ], + "prepared_review_unit_hash": "88edf5d5cd6860171c000b9dbff6e97fd5975e23387fb428553ad8b49fd446bb", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35355, + "estimated_input_tokens": 8839, + "estimated_eval_tokens": 17934 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is highly heterogeneous: it mixes unrelated tokenizer, import, training, model, documentation, and environment issues. None of the soft pairs look like the same underlying bug/change, so I would not merge any of them.", + "confidence": 0.93, + "canonical_issue_reason": "No single issue cleanly explains the cluster; issue:45278 is only the broadest umbrella for import-regression reports, but the rest of the items are largely unrelated.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the least-specific, broadest report in the set, but it is still not a good global canonical because the cluster does not share one root cause.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:39401", + "right": "issue:46614", + "accept": false, + "reason": "Both are tokenizer-related, but one is a Qwen3 offset_mapping regression and the other is a Wav2Vec2 phonemizer backend state bug; different code paths." + }, + { + "left": "issue:43824", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they involve different missing symbols from different modules and likely unrelated fixes." + }, + { + "left": "issue:44038", + "right": "issue:44222", + "accept": false, + "reason": "Same model family only at a very high level; one is a Qwen3-VL-Moe bug and the other is FP8 save_pretrained behavior, so they are not the same change." + }, + { + "left": "issue:43116", + "right": "issue:45200", + "accept": false, + "reason": "Different problems: example script multi-label output handling versus Gemma 4 mm_token_type_ids defaults." + }, + { + "left": "issue:43122", + "right": "issue:44779", + "accept": false, + "reason": "Both mention tokenizer regressions in v5, but they affect different models and likely different implementation bugs." + }, + { + "left": "issue:42757", + "right": "issue:44351", + "accept": false, + "reason": "Two separate import failures from different packages/modules; no shared underlying code path." + }, + { + "left": "issue:43278", + "right": "issue:44945", + "accept": false, + "reason": "A dtype mismatch during evaluation is unrelated to incorrect outputs under pipeline parallelism." + }, + { + "left": "issue:36296", + "right": "issue:43116", + "accept": false, + "reason": "Tensor parallel training bug and example-script classification output bug are different failures." + }, + { + "left": "issue:22355", + "right": "issue:29127", + "accept": false, + "reason": "One is a missing module import and the other is an error-message clarity issue in LayoutLMv3; not duplicates." + }, + { + "left": "issue:42175", + "right": "issue:44938", + "accept": false, + "reason": "Dependency/backend packaging issue versus a Python 3.14 loading failure; different root causes." + }, + { + "left": "issue:43493", + "right": "issue:45753", + "accept": false, + "reason": "Different models and symptoms: SigLIP2 HF/JAX discrepancy versus Qwen3_5 infinite-loop behavior." + }, + { + "left": "issue:44188", + "right": "issue:46328", + "accept": false, + "reason": "One is a torch.compile attention-kernel branching bug; the other is a padding performance issue." + }, + { + "left": "issue:46231", + "right": "issue:46877", + "accept": false, + "reason": "Documentation parameter mismatch versus import-time diagnostics printing; not the same bug." + }, + { + "left": "issue:42898", + "right": "issue:46614", + "accept": false, + "reason": "Tokenizer cleanup behavior change and stale phonemizer language state are unrelated." + }, + { + "left": "issue:38617", + "right": "issue:45278", + "accept": false, + "reason": "General import-regression reports, but the concrete missing symbol and the broader update breakage are not the same underlying fix." + }, + { + "left": "issue:43937", + "right": "issue:44485", + "accept": false, + "reason": "Both are GLM-5 related, but one concerns GenerationConfig validation and the other RoPE implementation; distinct code paths." + }, + { + "left": "issue:43976", + "right": "issue:45446", + "accept": false, + "reason": "Python-version compatibility issue and a PyTorch version check bug are different environment guards." + }, + { + "left": "issue:43066", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer regressions, but for different models and with different failure modes." + }, + { + "left": "issue:42673", + "right": "issue:45072", + "accept": false, + "reason": "VRAM leak in dataloader threads is unrelated to bfloat16 dtype mismatch in inference." + }, + { + "left": "issue:43611", + "right": "issue:46002", + "accept": false, + "reason": "Model loading via base_model_prefix and NaNs in classification heads are separate issues." + }, + { + "left": "issue:44038", + "right": "issue:45478", + "accept": false, + "reason": "Both touch Qwen3.5/Qwen3-VL-Moe, but one is a 5.0-era bug and the other is a from_pretrained regression in newer versions; not one concrete fix." + }, + { + "left": "issue:43116", + "right": "issue:43493", + "accept": false, + "reason": "Example-script classification behavior and SigLIP2 implementation discrepancy are unrelated." + }, + { + "left": "issue:44355", + "right": "issue:46231", + "accept": false, + "reason": "Compiled Python file loading errors are unrelated to a documentation typo." + }, + { + "left": "issue:45357", + "right": "issue:45478", + "accept": false, + "reason": "Both involve Qwen3.5 Moe, but one is save_pretrained key handling and the other is from_pretrained loading; different bugs." + }, + { + "left": "issue:44625", + "right": "issue:46002", + "accept": false, + "reason": "Config propagation of num_labels and NaNs on init/checkout are different failures." + }, + { + "left": "issue:44610", + "right": "issue:46909", + "accept": false, + "reason": "Processor input-size mismatch and a broken documentation example are not the same issue." + }, + { + "left": "issue:45083", + "right": "issue:45446", + "accept": false, + "reason": "Helper-function length calculation in qwen3_omni_moe is unrelated to the flex_attention version-check bug." + }, + { + "left": "issue:46726", + "right": "issue:46981", + "accept": false, + "reason": "Both are AutoProcessor-related, but one is missing-vision-backend handling and the other is retry behavior on HTTP 429; different code paths." + }, + { + "left": "issue:44075", + "right": "issue:44936", + "accept": false, + "reason": "Optimizer argument handling and train-then-evaluate failure are separate trainer/optimizer 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": 26, + "review_unit_index": 52, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:28348", + "issue:33453", + "issue:39401", + "issue:42222", + "issue:42371", + "issue:42673", + "issue:42796", + "issue:43064", + "issue:43066", + "issue:43116", + "issue:43441", + "issue:43618", + "issue:43673", + "issue:43824", + "issue:43825", + "issue:43827", + "issue:43874", + "issue:43937", + "issue:43992", + "issue:44117", + "issue:44263", + "issue:44291", + "issue:44380", + "issue:44466", + "issue:44704", + "issue:44779", + "issue:44977", + "issue:44995", + "issue:45071", + "issue:45092", + "issue:45405", + "issue:45446", + "issue:45563", + "issue:45588", + "issue:46002", + "issue:46077", + "issue:46328", + "issue:46614", + "issue:46640", + "issue:46655", + "issue:46697", + "issue:46710", + "issue:46752", + "issue:46762", + "issue:46829", + "issue:46981", + "issue:47000" + ], + "soft_pairs": [ + "issue:42371|issue:43874", + "issue:43618|issue:46829", + "issue:43116|issue:43825", + "issue:43992|issue:44704", + "issue:43066|issue:46710", + "issue:46614|issue:46752", + "issue:28348|issue:46640", + "issue:44995|issue:46697", + "issue:42673|issue:42796", + "issue:44263|issue:46762", + "issue:43827|issue:45446", + "issue:45092|issue:46762", + "issue:33453|issue:39401", + "issue:43064|issue:43673", + "issue:43937|issue:45563", + "issue:42222|issue:43441", + "issue:43824|issue:46981", + "issue:43673|issue:46328", + "issue:42222|issue:47000", + "issue:42673|issue:46655", + "issue:45405|issue:45446", + "issue:42222|issue:46752", + "issue:44291|issue:45071", + "issue:44117|issue:45588", + "issue:44380|issue:46077", + "issue:43937|issue:44977", + "issue:44466|issue:46002", + "issue:39401|issue:44779" + ], + "prepared_review_unit_hash": "e40515dcd41bf390f90aef3ba9047115e31fb9e8601fb706ae3953b05bf8ef81", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35473, + "estimated_input_tokens": 8869, + "estimated_eval_tokens": 17994 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is very heterogeneous: the soft pairs span unrelated tokenizer, generation, attention, training, processor, and model-export bugs. None of the candidate edges look like true duplicates of the same concrete fix path, so I would not collapse the cluster into a single canonical duplicate.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "issue:46640 is the broadest umbrella item in the set (Flash Attention 2), but it is still too vague to serve as a true duplicate canonical for this mostly unrelated cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42371", + "right": "issue:43874", + "accept": false, + "reason": "TF32 control behavior and a missing image-processor method are different bugs in different subsystems." + }, + { + "left": "issue:43618", + "right": "issue:46829", + "accept": false, + "reason": "Both are output-field issues, but they affect different models and different attributes, so they are not the same bug." + }, + { + "left": "issue:43116", + "right": "issue:43825", + "accept": false, + "reason": "Example-script multi-label behavior and a pipeline error-message regression are unrelated." + }, + { + "left": "issue:43992", + "right": "issue:44704", + "accept": false, + "reason": "Both involve from_pretrained flows, but one misses model weights and the other forwards kwargs to cached_file; different code paths." + }, + { + "left": "issue:43066", + "right": "issue:46710", + "accept": false, + "reason": "Tokenizer decoder typing and a model accuracy regression are not the same underlying issue." + }, + { + "left": "issue:46614", + "right": "issue:46752", + "accept": false, + "reason": "A stale phonemizer backend state bug is unrelated to apply_chat_template crashing on an empty list." + }, + { + "left": "issue:28348", + "right": "issue:46640", + "accept": false, + "reason": "One is a GPT-2 flash-attention support request; the other is only a vague flash-attention issue, not the same concrete change." + }, + { + "left": "issue:44995", + "right": "issue:46697", + "accept": false, + "reason": "Stale indexer cache on second forward pass and image over-padding are unrelated." + }, + { + "left": "issue:42673", + "right": "issue:42796", + "accept": false, + "reason": "VRAM leakage in threaded loading and tokenizer BPE-marker handling are different bugs." + }, + { + "left": "issue:44263", + "right": "issue:46762", + "accept": false, + "reason": "A torch.split return-value issue in one indexer is not the same as Minimax attention numerical drift." + }, + { + "left": "issue:43827", + "right": "issue:45446", + "accept": false, + "reason": "Docs still referencing pipeline() and a bad PyTorch version check in flex_attention are unrelated." + }, + { + "left": "issue:45092", + "right": "issue:46762", + "accept": false, + "reason": "Meta-initialization incompatibility for old checkpoints is unrelated to attention numerical differences." + }, + { + "left": "issue:33453", + "right": "issue:39401", + "accept": false, + "reason": "Both are tokenizer problems, but one is a loading regression and the other is wrong offset mappings for a specific model." + }, + { + "left": "issue:43064", + "right": "issue:43673", + "accept": false, + "reason": "Training-time optimizer state corruption and generation cache loss are distinct failures." + }, + { + "left": "issue:43937", + "right": "issue:45563", + "accept": false, + "reason": "Invalid GenerationConfig and a stale warning in paged generate() are different issues." + }, + { + "left": "issue:42222", + "right": "issue:43441", + "accept": false, + "reason": "VitPose model breakage and Ministral flash-attention failure do not share the same bug." + }, + { + "left": "issue:43824", + "right": "issue:46981", + "accept": false, + "reason": "An import error for a missing class and retry behavior on HTTP 429 are unrelated." + }, + { + "left": "issue:43673", + "right": "issue:46328", + "accept": false, + "reason": "Generation cache handling and an inefficient pad conversion path are different problems." + }, + { + "left": "issue:42222", + "right": "issue:47000", + "accept": false, + "reason": "VitPose breakage and RTDetrLoss infeasible cost matrix are unrelated." + }, + { + "left": "issue:42673", + "right": "issue:46655", + "accept": false, + "reason": "A VRAM leak in threaded loading and a video-processor padding bug are not the same bug." + }, + { + "left": "issue:45405", + "right": "issue:45446", + "accept": false, + "reason": "A dependency version bump to unreleased PEFT and an incorrect version check in flex_attention are unrelated." + }, + { + "left": "issue:42222", + "right": "issue:46752", + "accept": false, + "reason": "VitPose model breakage and apply_chat_template empty-list crash are unrelated." + }, + { + "left": "issue:44291", + "right": "issue:45071", + "accept": false, + "reason": "Both are v5 compatibility problems, but one is an unexpected init argument and the other is a PretrainedConfig type-checking break; not the same fix." + }, + { + "left": "issue:44117", + "right": "issue:45588", + "accept": false, + "reason": "TOKENIZER_MAPPING_NAMES returning None and a flash_attention AttributeError on sink-less models are unrelated." + }, + { + "left": "issue:44380", + "right": "issue:46077", + "accept": false, + "reason": "Both touch attention behavior, but one is backend scaling config being ignored and the other is causal-mask construction for a different model." + }, + { + "left": "issue:43937", + "right": "issue:44977", + "accept": false, + "reason": "Invalid generation config and DeepSeek tokenizer regression are different code paths." + }, + { + "left": "issue:44466", + "right": "issue:46002", + "accept": false, + "reason": "Serialization of tied lm_head weights and NaNs after checkout/init are unrelated." + }, + { + "left": "issue:39401", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer correctness bugs, but they affect different models and failure modes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 53, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:38617", + "issue:39401", + "issue:41628", + "issue:42222", + "issue:42371", + "issue:42617", + "issue:43066", + "issue:43122", + "issue:43208", + "issue:43408", + "issue:43450", + "issue:43742", + "issue:43937", + "issue:43986", + "issue:44075", + "issue:44117", + "issue:44220", + "issue:44261", + "issue:44263", + "issue:44410", + "issue:44484", + "issue:44485", + "issue:44496", + "issue:44829", + "issue:44877", + "issue:44933", + "issue:45071", + "issue:45081", + "issue:45216", + "issue:45278", + "issue:45310", + "issue:45341", + "issue:45362", + "issue:45478", + "issue:45561", + "issue:45632", + "issue:45656", + "issue:46178", + "issue:46209", + "issue:46231", + "issue:46298", + "issue:46328", + "issue:46512", + "issue:46617", + "issue:46762", + "issue:46894", + "issue:46981" + ], + "soft_pairs": [ + "issue:44410|issue:45310", + "issue:43450|issue:43986", + "issue:45071|issue:46512", + "issue:45341|issue:46328", + "issue:45341|issue:45561", + "issue:45632|issue:46981", + "issue:44829|issue:45216", + "issue:44117|issue:45081", + "issue:42617|issue:45341", + "issue:43408|issue:43742", + "issue:38617|issue:41628", + "issue:43122|issue:46298", + "issue:45362|issue:46328", + "issue:42222|issue:46178", + "issue:39401|issue:46209", + "issue:44877|issue:46231", + "issue:44484|issue:45310", + "issue:43937|issue:44496", + "issue:42371|issue:44261", + "issue:44261|issue:46762", + "issue:44933|issue:45278", + "issue:44410|issue:45478", + "issue:42371|issue:46617", + "issue:44075|issue:45656", + "issue:38617|issue:44220", + "issue:43208|issue:46894", + "issue:43066|issue:43122", + "issue:44263|issue:44485" + ], + "prepared_review_unit_hash": "29ad13aee0b056c59646c7d25edd2efe17524df10026f11db3558b863f554e10", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35151, + "estimated_input_tokens": 8788, + "estimated_eval_tokens": 17832 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose mix of unrelated Transformers bugs across tokenizers, model loading, config validation, training, testing utilities, and docs. None of the soft pairs look like the same underlying fix, so all are rejected.", + "confidence": 0.97, + "canonical_issue_reason": "No canonical issue: the set is too heterogeneous, with no clear duplicate chain across the items.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the broadest user-facing regression in the set (many import errors after upgrading), so it is the closest thing to a representative issue, though it is still not a true duplicate anchor.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44410", + "right": "issue:45310", + "accept": false, + "reason": "Different Qwen3 bugs: missing projections in qwen3next vs a Qwen3.5 MoE from_pretrained failure." + }, + { + "left": "issue:43450", + "right": "issue:43986", + "accept": false, + "reason": "One is a batched video processor shape bug; the other is an AutoProcessor crash without torchvision." + }, + { + "left": "issue:45071", + "right": "issue:46512", + "accept": false, + "reason": "Type-checking regression in PreTrainedConfig is unrelated to stale config.dtype after from_config()." + }, + { + "left": "issue:45341", + "right": "issue:46328", + "accept": false, + "reason": "testing_utils bug vs SequenceFeatureExtractor.pad performance issue; different code paths and symptoms." + }, + { + "left": "issue:45341", + "right": "issue:45561", + "accept": false, + "reason": "Both touch testing utils, but one is a small bug and the other is an xdist file race." + }, + { + "left": "issue:45632", + "right": "issue:46981", + "accept": false, + "reason": "Cache path collision for trust_remote_code models is unrelated to repeated 429 retries in AutoProcessor." + }, + { + "left": "issue:44829", + "right": "issue:45216", + "accept": false, + "reason": "Flash-attention training degeneration and a Qwen3.5 save_pretrained regression are different problems." + }, + { + "left": "issue:44117", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer mapping None handling is unrelated to the Mistral regex patch crash." + }, + { + "left": "issue:42617", + "right": "issue:45341", + "accept": false, + "reason": "3d_parallel.py execution failure is unrelated to the testing_utils bug." + }, + { + "left": "issue:43408", + "right": "issue:43742", + "accept": false, + "reason": "Model-type warning for sam3 is unrelated to the MobileLLM-125M loading KeyError." + }, + { + "left": "issue:38617", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors, but they concern different symbols and different underlying causes." + }, + { + "left": "issue:43122", + "right": "issue:46298", + "accept": false, + "reason": "Tokenizer behavior regression is unrelated to a noisy logging message." + }, + { + "left": "issue:45362", + "right": "issue:46328", + "accept": false, + "reason": "Qwen chat crash and padding inefficiency are different issues in different subsystems." + }, + { + "left": "issue:42222", + "right": "issue:46178", + "accept": false, + "reason": "Vitpose breakage and BACKENDS_MAPPING validation are unrelated backend/model-loading failures." + }, + { + "left": "issue:39401", + "right": "issue:46209", + "accept": false, + "reason": "Qwen3 offset_mapping bug and FP8 inference failure on B200 are unrelated." + }, + { + "left": "issue:44877", + "right": "issue:46231", + "accept": false, + "reason": "Strict config loading for granite_speech is unrelated to a docs parameter-name mismatch." + }, + { + "left": "issue:44484", + "right": "issue:45310", + "accept": false, + "reason": "save_pretrained shard-size behavior is unrelated to the Qwen3.5 MoE from_pretrained error." + }, + { + "left": "issue:43937", + "right": "issue:44496", + "accept": false, + "reason": "Invalid GenerationConfig in GLM-5 and unrecognized Olmo-Hybrid model config are distinct failures." + }, + { + "left": "issue:42371", + "right": "issue:44261", + "accept": false, + "reason": "TF32 API guidance and MLA layernorm precision error do not share the same bug." + }, + { + "left": "issue:44261", + "right": "issue:46762", + "accept": false, + "reason": "Both are numerical issues, but they affect different model components and likely different fixes." + }, + { + "left": "issue:44933", + "right": "issue:45278", + "accept": false, + "reason": "A specific image_utils import issue is not the same as a broad multi-import regression." + }, + { + "left": "issue:44410", + "right": "issue:45478", + "accept": false, + "reason": "Missing qwen3next projections is unrelated to the Qwen3.5 MoE from_pretrained error." + }, + { + "left": "issue:42371", + "right": "issue:46617", + "accept": false, + "reason": "TF32 behavior settings and trust_remote_code cache FileNotFoundError are unrelated." + }, + { + "left": "issue:44075", + "right": "issue:45656", + "accept": false, + "reason": "SGD args ignored and optimizer stepping twice with deepspeed are different optimizer bugs." + }, + { + "left": "issue:38617", + "right": "issue:44220", + "accept": false, + "reason": "ImportError for layer_type_validation is unrelated to _torch_extract_fbank_features()." + }, + { + "left": "issue:43208", + "right": "issue:46894", + "accept": false, + "reason": "xLSTM training blockers and CohereASR double-shift loss bug are different model-specific issues." + }, + { + "left": "issue:43066", + "right": "issue:43122", + "accept": false, + "reason": "Tokenizer decoder-type regression and version-to-version tokenization differences are related broadly, but not the same concrete bug." + }, + { + "left": "issue:44263", + "right": "issue:44485", + "accept": false, + "reason": "torch.split() return handling in GlmMoeDsaIndexer is unrelated to GLM-5 RoPE implementation." + } + ] + }, + "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": 26, + "review_unit_index": 54, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:27957", + "issue:38617", + "issue:41628", + "issue:42757", + "issue:42886", + "issue:43012", + "issue:43064", + "issue:43097", + "issue:43352", + "issue:43441", + "issue:43452", + "issue:43531", + "issue:43577", + "issue:43673", + "issue:43874", + "issue:43931", + "issue:43986", + "issue:44038", + "issue:44077", + "issue:44117", + "issue:44261", + "issue:44263", + "issue:44315", + "issue:44360", + "issue:44410", + "issue:44483", + "issue:44484", + "issue:44485", + "issue:44493", + "issue:44829", + "issue:44987", + "issue:45003", + "issue:45071", + "issue:45341", + "issue:45372", + "issue:45375", + "issue:45440", + "issue:45464", + "issue:45478", + "issue:45542", + "issue:45588", + "issue:45753", + "issue:46050", + "issue:46123", + "issue:46172", + "issue:46369", + "issue:46981" + ], + "soft_pairs": [ + "issue:43097|issue:44493", + "issue:44077|issue:46981", + "issue:43673|issue:45341", + "issue:44483|issue:45464", + "issue:43931|issue:44410", + "issue:41628|issue:45071", + "issue:45071|issue:45542", + "issue:44484|issue:45478", + "issue:27957|issue:43352", + "issue:43441|issue:44829", + "issue:43012|issue:43577", + "issue:44263|issue:46172", + "issue:44360|issue:46172", + "issue:43874|issue:44261", + "issue:43931|issue:44038", + "issue:43097|issue:43452", + "issue:43986|issue:45341", + "issue:45003|issue:45341", + "issue:42886|issue:46981", + "issue:44261|issue:44485", + "issue:45753|issue:46369", + "issue:43531|issue:45341", + "issue:44117|issue:44987", + "issue:43097|issue:45588", + "issue:43673|issue:46123", + "issue:45440|issue:46172", + "issue:43874|issue:44263", + "issue:44410|issue:45375", + "issue:43986|issue:45372", + "issue:43097|issue:46050", + "issue:44038|issue:44410", + "issue:38617|issue:42757", + "issue:43064|issue:44315" + ], + "prepared_review_unit_hash": "d8628484731d552c21eb48707477ba52927b86a67303beb06fc78d2b48708e40", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 33, + "serialized_chars": 36857, + "estimated_input_tokens": 9215, + "estimated_eval_tokens": 18686 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: it mixes unrelated loading, import, FlashAttention, tokenizer, config, and training bugs. All soft duplicate suggestions are false positives.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43097", + "right": "issue:44493", + "accept": false, + "reason": "Different bugs: API/removal of tie_embeddings vs position-id key warnings." + }, + { + "left": "issue:44077", + "right": "issue:46981", + "accept": false, + "reason": "Unrelated areas: patchtsmixer post_init validation vs AutoProcessor 429 retry behavior." + }, + { + "left": "issue:43673", + "right": "issue:45341", + "accept": false, + "reason": "Generation cache/chunked_prefill bug vs a testing_utils bug." + }, + { + "left": "issue:44483", + "right": "issue:45464", + "accept": false, + "reason": "Both mention chat/completions, but one is general request rejection and the other is a Qwen3.5 streaming-specific failure." + }, + { + "left": "issue:43931", + "right": "issue:44410", + "accept": false, + "reason": "Different Qwen3 problems: weight-shape mismatch vs missing attention projections." + }, + { + "left": "issue:41628", + "right": "issue:45071", + "accept": false, + "reason": "AutoImageProcessor import failure is unrelated to PretrainedConfig type checking." + }, + { + "left": "issue:45071", + "right": "issue:45542", + "accept": false, + "reason": "Type-checking regression is unrelated to the TensorFlow/tensorboard backend error." + }, + { + "left": "issue:44484", + "right": "issue:45478", + "accept": false, + "reason": "max_shard_size question is unrelated to a Qwen3.5-Moe from_pretrained error." + }, + { + "left": "issue:27957", + "right": "issue:43352", + "accept": false, + "reason": "Different models and failures: XLM-R flash attention support vs NemotronH FA2 support." + }, + { + "left": "issue:43441", + "right": "issue:44829", + "accept": false, + "reason": "Both involve flash attention, but one is model loading and the other is degenerate training with FA3." + }, + { + "left": "issue:43012", + "right": "issue:43577", + "accept": false, + "reason": "Precision warning during compile is not the same as Blip2 submodules staying float32 on load." + }, + { + "left": "issue:44263", + "right": "issue:46172", + "accept": false, + "reason": "Distinct code paths: torch.split behavior in a DSA indexer vs DeepSeek-V4 comb dtype/transpose." + }, + { + "left": "issue:44360", + "right": "issue:46172", + "accept": false, + "reason": "ReLU omission in the DSA indexer is unrelated to DeepSeek-V4 tensor conversion." + }, + { + "left": "issue:43874", + "right": "issue:44261", + "accept": false, + "reason": "Image-patch count method missing vs rms_norm_eps/config precision issue." + }, + { + "left": "issue:43931", + "right": "issue:44038", + "accept": false, + "reason": "Related model family, but different concrete failures and no shared code-path evidence." + }, + { + "left": "issue:43097", + "right": "issue:43452", + "accept": false, + "reason": "Removed tie_embeddings API is unrelated to gguf_file loading breakage." + }, + { + "left": "issue:43986", + "right": "issue:45341", + "accept": false, + "reason": "Video AutoProcessor crash without torchvision is unrelated to testing_utils." + }, + { + "left": "issue:45003", + "right": "issue:45341", + "accept": false, + "reason": "sys.modules access bug in modeling_utils is unrelated to testing_utils." + }, + { + "left": "issue:42886", + "right": "issue:46981", + "accept": false, + "reason": "Both are loading/caching/network-related, but offline cache failure and 429 retry behavior are different bugs." + }, + { + "left": "issue:44261", + "right": "issue:44485", + "accept": false, + "reason": "rms_norm_eps precision problem is not the same as GLM-5 RoPE implementation." + }, + { + "left": "issue:45753", + "right": "issue:46369", + "accept": false, + "reason": "Infinite loop on a specific image vs memory overflow in sam3-litetext." + }, + { + "left": "issue:43531", + "right": "issue:45341", + "accept": false, + "reason": "Qwen3-MoE sliding_window issue is unrelated to testing_utils." + }, + { + "left": "issue:44117", + "right": "issue:44987", + "accept": false, + "reason": "Tokenizer mapping None handling is not the same as physical-intelligence/fast loading failure." + }, + { + "left": "issue:43097", + "right": "issue:45588", + "accept": false, + "reason": "tie_embeddings removal is unrelated to flash_attention.py crashing on s_aux=None." + }, + { + "left": "issue:43673", + "right": "issue:46123", + "accept": false, + "reason": "Generation cache missing in chunked_prefill is unrelated to partial-batch MaskGenerationPipeline dropping results." + }, + { + "left": "issue:45440", + "right": "issue:46172", + "accept": false, + "reason": "Different model families and failures: DeepseekV3MoE divergence vs DeepSeek-V4 tensor handling." + }, + { + "left": "issue:43874", + "right": "issue:44263", + "accept": false, + "reason": "ImageProcessor patch counting bug is unrelated to torch.split return handling." + }, + { + "left": "issue:44410", + "right": "issue:45375", + "accept": false, + "reason": "Qwen3next missing projections and Qwen3_5MoeVisionConfig field dropping are different config/model issues." + }, + { + "left": "issue:43986", + "right": "issue:45372", + "accept": false, + "reason": "Missing torchvision crash is unrelated to mistral_common ReasoningEffort import failure." + }, + { + "left": "issue:43097", + "right": "issue:46050", + "accept": false, + "reason": "API removal in config handling is unrelated to sinq quantization save/load breakage." + }, + { + "left": "issue:44038", + "right": "issue:44410", + "accept": false, + "reason": "Both are Qwen3-family, but they describe different missing-parameter bugs with no shared concrete code path." + }, + { + "left": "issue:38617", + "right": "issue:42757", + "accept": false, + "reason": "Separate import errors from different modules: transformers.configuration_utils vs huggingface_hub." + }, + { + "left": "issue:43064", + "right": "issue:44315", + "accept": false, + "reason": "FSDP2/PEFT optimizer-state corruption is unrelated to Liger Kernel not being applied with model_init." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 55, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:38617", + "issue:42890", + "issue:42907", + "issue:43012", + "issue:43066", + "issue:43097", + "issue:43208", + "issue:43352", + "issue:43502", + "issue:43550", + "issue:43825", + "issue:43873", + "issue:43906", + "issue:43976", + "issue:43986", + "issue:43992", + "issue:44016", + "issue:44062", + "issue:44077", + "issue:44117", + "issue:44162", + "issue:44222", + "issue:44351", + "issue:44368", + "issue:44380", + "issue:44484", + "issue:44655", + "issue:44704", + "issue:44936", + "issue:44938", + "issue:45103", + "issue:45478", + "issue:45584", + "issue:45712", + "issue:45753", + "issue:46132", + "issue:46172", + "issue:46178", + "issue:46298", + "issue:46365", + "issue:46421", + "issue:46655", + "issue:46664", + "issue:46701", + "issue:46710", + "issue:46752", + "issue:46821", + "issue:46897" + ], + "soft_pairs": [ + "issue:43208|issue:46897", + "issue:43986|issue:44484", + "issue:46655|issue:46664", + "issue:46298|issue:46752", + "issue:42890|issue:46298", + "issue:43097|issue:43873", + "issue:38617|issue:44351", + "issue:44062|issue:44117", + "issue:43012|issue:46821", + "issue:44117|issue:46701", + "issue:46172|issue:46710", + "issue:43502|issue:43992", + "issue:44936|issue:46365", + "issue:44704|issue:45103", + "issue:44380|issue:46421", + "issue:44938|issue:46178", + "issue:43986|issue:44016", + "issue:42907|issue:45753", + "issue:42907|issue:44655", + "issue:43012|issue:43550", + "issue:45712|issue:46298", + "issue:43825|issue:44655", + "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:43097|issue:46298", + "issue:45584|issue:46365", + "issue:44162|issue:46172", + "issue:44380|issue:46132", + "issue:44077|issue:46365", + "issue:44222|issue:45478" + ], + "prepared_review_unit_hash": "12dc6ef4826ecc4fc3a1fa92286324f5b163c2e03645fedca2efa5228085569e", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37640, + "estimated_input_tokens": 9410, + "estimated_eval_tokens": 19076 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a heterogeneous set of mostly unrelated issues; none of the soft pairs look like the same underlying bug or change. I\u2019d treat the open quantization/offloading report as the best canonical issue in the set, but only as a representative, not a duplicate anchor.", + "confidence": 0.63, + "canonical_issue_reason": "issue:43873 is the most substantial and active report here (highest discussion/inbound refs) and is a clear, actionable bug report.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43873 is the strongest representative issue overall because it has the broadest engagement and a well-scoped problem statement.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43208", + "right": "issue:46897", + "accept": false, + "reason": "Different models and failures: xLSTM training bugs vs Florence2 training-loss shifting." + }, + { + "left": "issue:43986", + "right": "issue:44484", + "accept": false, + "reason": "Video-model/AutoProcessor import failure is unrelated to a save_pretrained shard-size question." + }, + { + "left": "issue:46655", + "right": "issue:46664", + "accept": false, + "reason": "Same component, but padding calculation and num_frames semantics are distinct bugs." + }, + { + "left": "issue:46298", + "right": "issue:46752", + "accept": false, + "reason": "Noisy logging and empty-list chat-template crash are unrelated." + }, + { + "left": "issue:42890", + "right": "issue:46298", + "accept": false, + "reason": "Missing set_seed in a test is unrelated to a logging message issue." + }, + { + "left": "issue:43097", + "right": "issue:43873", + "accept": false, + "reason": "Removed config field guidance and quantization/offloading behavior are different problems." + }, + { + "left": "issue:38617", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they involve different missing symbols and code paths." + }, + { + "left": "issue:44062", + "right": "issue:44117", + "accept": false, + "reason": "Tokenizer AddedToken construction bug is separate from TOKENIZER_MAPPING_NAMES returning None." + }, + { + "left": "issue:43012", + "right": "issue:46821", + "accept": false, + "reason": "bfloat16 compile warning is unrelated to a TimesFM window_size AttributeError." + }, + { + "left": "issue:44117", + "right": "issue:46701", + "accept": false, + "reason": "Tokenizer mapping fallback bug does not match the pipeline failure described here." + }, + { + "left": "issue:46172", + "right": "issue:46710", + "accept": false, + "reason": "A DeepSeek tensor operation bug is not the same as an accuracy regression after upgrade." + }, + { + "left": "issue:43502", + "right": "issue:43992", + "accept": false, + "reason": "Both involve loading, but one is network access control and the other is missing encoder weights." + }, + { + "left": "issue:44936", + "right": "issue:46365", + "accept": false, + "reason": "Trainer evaluate failure after train is unrelated to importing Trainer with datasets installed." + }, + { + "left": "issue:44704", + "right": "issue:45103", + "accept": false, + "reason": "AutoProcessor kwargs forwarding and auto_docstring annotation parsing are unrelated." + }, + { + "left": "issue:44380", + "right": "issue:46421", + "accept": false, + "reason": "GPT2 attention scaling under SDPA/FlashAttention is different from cache lazy-init recompilation." + }, + { + "left": "issue:44938", + "right": "issue:46178", + "accept": false, + "reason": "Python 3.14 import failure is not the same as an invalid BACKENDS_MAPPING backend." + }, + { + "left": "issue:43986", + "right": "issue:44016", + "accept": false, + "reason": "Video-model import handling and a notebook syntax error are unrelated." + }, + { + "left": "issue:42907", + "right": "issue:45753", + "accept": false, + "reason": "Saving dequantized models and an infinite-loop image bug are different failures." + }, + { + "left": "issue:42907", + "right": "issue:44655", + "accept": false, + "reason": "Both mention saving, but dequantized model save failure and pipeline object save failure are different code paths." + }, + { + "left": "issue:43012", + "right": "issue:43550", + "accept": false, + "reason": "A PyTorch precision warning is not the same as a torch.compile SDPA model failure." + }, + { + "left": "issue:45712", + "right": "issue:46298", + "accept": false, + "reason": "Dummy class leakage into dir(transformers) is unrelated to a logging message." + }, + { + "left": "issue:43825", + "right": "issue:44655", + "accept": false, + "reason": "Pipeline task support messaging and pipeline save_pretrained are unrelated." + }, + { + "left": "issue:43208", + "right": "issue:43550", + "accept": false, + "reason": "xLSTM training-size bugs are unrelated to Bamba compile/SDPA failures." + }, + { + "left": "issue:43012", + "right": "issue:44368", + "accept": false, + "reason": "Compile-time bfloat16 warning is unrelated to tie_word_embeddings config guidance." + }, + { + "left": "issue:43992", + "right": "issue:44484", + "accept": false, + "reason": "Missing embed_tokens weight on load is unrelated to max_shard_size defaults." + }, + { + "left": "issue:43352", + "right": "issue:43976", + "accept": false, + "reason": "Model-specific Flash Attention support and Python version incompatibility are different issues." + }, + { + "left": "issue:43352", + "right": "issue:44380", + "accept": false, + "reason": "Unsupported FlashAttention 2.0 for a model is not the same as attention scaling being ignored." + }, + { + "left": "issue:43066", + "right": "issue:43906", + "accept": false, + "reason": "Tokenizer decoder type regression and an isolated reproduction of another issue are not duplicates." + }, + { + "left": "issue:43097", + "right": "issue:46298", + "accept": false, + "reason": "Removed config API and a noisy logging message are unrelated." + }, + { + "left": "issue:45584", + "right": "issue:46365", + "accept": false, + "reason": "Whisper empty transcription handling is unrelated to importing Trainer." + }, + { + "left": "issue:44162", + "right": "issue:46172", + "accept": false, + "reason": "ESM2 breakage and a DeepSeek tensor operation bug are unrelated." + }, + { + "left": "issue:44380", + "right": "issue:46132", + "accept": false, + "reason": "Attention scaling config loss under backend selection is not the same as register behavior changes." + }, + { + "left": "issue:44077", + "right": "issue:46365", + "accept": false, + "reason": "patchtsmixer post_init validation is unrelated to the Trainer import issue." + }, + { + "left": "issue:44222", + "right": "issue:45478", + "accept": false, + "reason": "Both involve Qwen/MoE, but save_pretrained FP8 and from_pretrained loading errors are different concrete 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": 26, + "review_unit_index": 56, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:29870", + "issue:36010", + "issue:36296", + "issue:36683", + "issue:38617", + "issue:42175", + "issue:42490", + "issue:42757", + "issue:42994", + "issue:43012", + "issue:43064", + "issue:43208", + "issue:43299", + "issue:43352", + "issue:43475", + "issue:43502", + "issue:43525", + "issue:43526", + "issue:43582", + "issue:43611", + "issue:43618", + "issue:43756", + "issue:43761", + "issue:43824", + "issue:43867", + "issue:43937", + "issue:43976", + "issue:44038", + "issue:44077", + "issue:44230", + "issue:44355", + "issue:44380", + "issue:44483", + "issue:44485", + "issue:44492", + "issue:44514", + "issue:44625", + "issue:44655", + "issue:44938", + "issue:45216", + "issue:45276", + "issue:45290", + "issue:45479", + "issue:46050", + "issue:46077", + "issue:46082", + "issue:46365", + "issue:46877" + ], + "soft_pairs": [ + "issue:36683|issue:43582", + "issue:44380|issue:45479", + "issue:42757|issue:43824", + "issue:42175|issue:46365", + "issue:43064|issue:44380", + "issue:42490|issue:43012", + "issue:44483|issue:44514", + "issue:43618|issue:44380", + "issue:38617|issue:43502", + "issue:43208|issue:43526", + "issue:43756|issue:44485", + "issue:43618|issue:43761", + "issue:42994|issue:44655", + "issue:45276|issue:46077", + "issue:36010|issue:42757", + "issue:43299|issue:43867", + "issue:44625|issue:46365", + "issue:44483|issue:45290", + "issue:44038|issue:44230", + "issue:36683|issue:43475", + "issue:43611|issue:44655", + "issue:43352|issue:43937", + "issue:44355|issue:46877", + "issue:44655|issue:46050", + "issue:44625|issue:46077", + "issue:36683|issue:44077", + "issue:43976|issue:44938", + "issue:43867|issue:45216", + "issue:46077|issue:46082", + "issue:43618|issue:45216", + "issue:36683|issue:43525", + "issue:36683|issue:44625", + "issue:29870|issue:36296", + "issue:43064|issue:44492" + ], + "prepared_review_unit_hash": "475fe931104149db7aa60afce29b9ceb9481b5817c820533bb0804e9bea463ca", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37689, + "estimated_input_tokens": 9423, + "estimated_eval_tokens": 19102 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is a grab bag of unrelated Transformers issues that mostly share superficial keywords like import errors, save/load, or model/config regressions. None of the soft pairs look like the same underlying bug or change, so all candidate edges are rejected.", + "confidence": 0.86, + "canonical_issue_reason": "issue:45216 is the closest thing to a representative item because it is a concrete, version-bounded regression around save_pretrained; but the overall set is too heterogeneous to have a strong canonical issue.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45216 is the best representative issue: it has a specific failure mode, a clear regression boundary, and a focused code path, unlike most other items in the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:36683", + "right": "issue:43582", + "accept": false, + "reason": "Different bugs: Gemma3 config missing vocab_size vs Apple Silicon allocator warmup TypeError." + }, + { + "left": "issue:44380", + "right": "issue:45479", + "accept": false, + "reason": "Different areas: attention scaling ignored in GPT-2 backends vs degenerate zero loss in sequence classification." + }, + { + "left": "issue:42757", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but they involve different missing symbols from different packages." + }, + { + "left": "issue:42175", + "right": "issue:46365", + "accept": false, + "reason": "Unrelated packaging/import problems: extras backend missing TensorFlow vs Trainer import breaking with datasets installed." + }, + { + "left": "issue:43064", + "right": "issue:44380", + "accept": false, + "reason": "Different code paths: distributed optimizer state corruption vs GPT-2 attention scaling backend behavior." + }, + { + "left": "issue:42490", + "right": "issue:43012", + "accept": false, + "reason": "Different warnings with different triggers: local path import warning vs bfloat16 compilation precision warning." + }, + { + "left": "issue:44483", + "right": "issue:44514", + "accept": false, + "reason": "API request validation problem vs chat template crash on batched input; not the same bug." + }, + { + "left": "issue:43618", + "right": "issue:44380", + "accept": false, + "reason": "Both mention attention, but one is a CLIP output field regression and the other is backend-specific scaling behavior." + }, + { + "left": "issue:38617", + "right": "issue:43502", + "accept": false, + "reason": "ImportError for a removed utility vs unwanted API calls despite local_files_only=True." + }, + { + "left": "issue:43208", + "right": "issue:43526", + "accept": false, + "reason": "Different subsystems entirely: xLSTM training bugs vs Beit image processor label reduction." + }, + { + "left": "issue:43756", + "right": "issue:44485", + "accept": false, + "reason": "RoPE-related titles only; one is a Smollm3 model bug and the other is a GLM-5 discussion, not the same concrete issue." + }, + { + "left": "issue:43618", + "right": "issue:43761", + "accept": false, + "reason": "Both are CLIP regressions, but one drops attentions and the other drops hidden_states; not enough evidence of one underlying bug." + }, + { + "left": "issue:42994", + "right": "issue:44655", + "accept": false, + "reason": "Both involve saving, but quantized model save failure and Pipeline save_pretrained failure are distinct problems." + }, + { + "left": "issue:45276", + "right": "issue:46077", + "accept": false, + "reason": "Different Gemma4 bugs: embedding resize propagation vs bidirectional attention mask construction." + }, + { + "left": "issue:36010", + "right": "issue:42757", + "accept": false, + "reason": "Different import failures from different modules and symbols." + }, + { + "left": "issue:43299", + "right": "issue:43867", + "accept": false, + "reason": "Different model loading failures: Qwen3VL MoE loading regression vs state_dict ordering issue." + }, + { + "left": "issue:44625", + "right": "issue:46365", + "accept": false, + "reason": "Qwen3.5 num_labels propagation bug is unrelated to the Trainer import issue." + }, + { + "left": "issue:44483", + "right": "issue:45290", + "accept": false, + "reason": "Both are chat-related, but one is a server API request issue and the other is a tokenizer/template crash." + }, + { + "left": "issue:44038", + "right": "issue:44230", + "accept": false, + "reason": "Same model family, but one is a loading bug and the other is an fp8 inference support request." + }, + { + "left": "issue:36683", + "right": "issue:43475", + "accept": false, + "reason": "Different missing-attribute bugs on different model families and different config/output objects." + }, + { + "left": "issue:43611", + "right": "issue:44655", + "accept": false, + "reason": "Model loading with base_model_prefix and Pipeline saving are separate code paths." + }, + { + "left": "issue:43352", + "right": "issue:43937", + "accept": false, + "reason": "Different failures: Flash Attention support gate vs invalid GenerationConfig." + }, + { + "left": "issue:44355", + "right": "issue:46877", + "accept": false, + "reason": "Both touch import-time behavior, but compiled-file execution errors and docstring lint output are unrelated." + }, + { + "left": "issue:44655", + "right": "issue:46050", + "accept": false, + "reason": "Both mention save_pretrained, but one is Pipeline serialization and the other is sinq quantization persistence." + }, + { + "left": "issue:44625", + "right": "issue:46077", + "accept": false, + "reason": "Different config bugs in different models: Qwen3.5 labels vs Gemma4 bidirectional attention." + }, + { + "left": "issue:36683", + "right": "issue:44077", + "accept": false, + "reason": "Different config issues on unrelated model families." + }, + { + "left": "issue:43976", + "right": "issue:44938", + "accept": false, + "reason": "Both are Python-version compatibility complaints, but they concern different versions and likely different causes." + }, + { + "left": "issue:43867", + "right": "issue:45216", + "accept": false, + "reason": "Different checkpoint issues: sorted state_dict loading vs incorrect saved checkpoint content." + }, + { + "left": "issue:46077", + "right": "issue:46082", + "accept": false, + "reason": "Unrelated model config problems: Gemma4 attention masking vs LlamaConfig head_dim validation." + }, + { + "left": "issue:43618", + "right": "issue:45216", + "accept": false, + "reason": "Different models and symptoms: CLIP output regression vs Qwen3.5 save_pretrained regression." + }, + { + "left": "issue:36683", + "right": "issue:43525", + "accept": false, + "reason": "Both are missing config attributes, but on different models and different fields." + }, + { + "left": "issue:36683", + "right": "issue:44625", + "accept": false, + "reason": "Different config propagation bugs on different model families." + }, + { + "left": "issue:29870", + "right": "issue:36296", + "accept": false, + "reason": "Bart generate() decoder_attention_mask bug is unrelated to the tensor-parallel training issue." + }, + { + "left": "issue:43064", + "right": "issue:44492", + "accept": false, + "reason": "Distributed optimizer-state corruption and a cache-strategy typo are 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": 26, + "review_unit_index": 57, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:28282", + "issue:29870", + "issue:33453", + "issue:36296", + "issue:36683", + "issue:41669", + "issue:42890", + "issue:42898", + "issue:43010", + "issue:43064", + "issue:43502", + "issue:43575", + "issue:43618", + "issue:43723", + "issue:43867", + "issue:43976", + "issue:43992", + "issue:44062", + "issue:44075", + "issue:44077", + "issue:44112", + "issue:44246", + "issue:44355", + "issue:44360", + "issue:44393", + "issue:44556", + "issue:44734", + "issue:44855", + "issue:44908", + "issue:44938", + "issue:44964", + "issue:45071", + "issue:45278", + "issue:45310", + "issue:45357", + "issue:45375", + "issue:45440", + "issue:45478", + "issue:45561", + "issue:45600", + "issue:45850", + "issue:46202", + "issue:46285", + "issue:46365", + "issue:46617", + "issue:46664", + "issue:46851", + "issue:47000" + ], + "soft_pairs": [ + "issue:43010|issue:45278", + "issue:44075|issue:46365", + "issue:44908|issue:46851", + "issue:43992|issue:45310", + "issue:44393|issue:47000", + "issue:44938|issue:45071", + "issue:29870|issue:33453", + "issue:44355|issue:44855", + "issue:43867|issue:46202", + "issue:36683|issue:43064", + "issue:42890|issue:45561", + "issue:43867|issue:45375", + "issue:43010|issue:45850", + "issue:36296|issue:43575", + "issue:41669|issue:44246", + "issue:43618|issue:44077", + "issue:43010|issue:43976", + "issue:43867|issue:45357", + "issue:43992|issue:45478", + "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:43502|issue:44062", + "issue:42898|issue:46664", + "issue:44734|issue:46617", + "issue:43723|issue:44556" + ], + "prepared_review_unit_hash": "29868d1f9c53eff1735127a4b9f9a4e1937f3f20987d9571a9d215845e41bc07", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36058, + "estimated_input_tokens": 9015, + "estimated_eval_tokens": 18286 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly unrelated issue reports with one strong duplicate pair around slow `import transformers`. I would keep the import-slow regression issue as the canonical item.", + "confidence": 0.81, + "canonical_issue_reason": "issue:41669 is the clearest, most specific import-performance report: it identifies the regression source (`import *` usage in models), quantifies the impact, and already has inbound reference activity.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:41669 is the best representative issue because it states the concrete root cause and symptom for the import-time slowdown, making it the most actionable anchor for the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43010", + "right": "issue:45278", + "accept": false, + "reason": "Different problems: cache/update no-grad decoration vs broad import errors after upgrading." + }, + { + "left": "issue:44075", + "right": "issue:46365", + "accept": false, + "reason": "Unrelated: optimizer arguments being ignored vs Trainer import failing due to `datasets`." + }, + { + "left": "issue:44908", + "right": "issue:46851", + "accept": false, + "reason": "Different subsystems: scheduler kwargs vs Qwen3.5 variable-length position ids." + }, + { + "left": "issue:43992", + "right": "issue:45310", + "accept": false, + "reason": "Different model-loading failures with no clear shared code path." + }, + { + "left": "issue:44393", + "right": "issue:47000", + "accept": false, + "reason": "Both are vision/detection-related, but they describe different concrete bugs in different models/losses." + }, + { + "left": "issue:44938", + "right": "issue:45071", + "accept": false, + "reason": "Python-version load failure vs `PretrainedConfig` type-checking regression; not the same bug." + }, + { + "left": "issue:29870", + "right": "issue:33453", + "accept": false, + "reason": "Bart `generate()` mask bug is unrelated to tokenizer loading regression." + }, + { + "left": "issue:44355", + "right": "issue:44855", + "accept": false, + "reason": "Both involve import/runtime errors, but one is compiled-file execution and the other is a specific `torch.jit.script` parsing failure." + }, + { + "left": "issue:43867", + "right": "issue:46202", + "accept": false, + "reason": "State-dict sorting/load error and Gemma pipeline roundtrip issue are different failures." + }, + { + "left": "issue:36683", + "right": "issue:43064", + "accept": false, + "reason": "Gemma3 config attribute error is unrelated to FSDP2/PEFT optimizer-state corruption." + }, + { + "left": "issue:42890", + "right": "issue:45561", + "accept": false, + "reason": "Test flakiness from missing seeds is not the same as xdist file-write races." + }, + { + "left": "issue:43867", + "right": "issue:45375", + "accept": false, + "reason": "Different Qwen/Gemma loading and config-schema problems." + }, + { + "left": "issue:43010", + "right": "issue:45850", + "accept": false, + "reason": "Specific cache-layer no-grad bug vs generic post-v5 breakage complaint." + }, + { + "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 while loading a specific model." + }, + { + "left": "issue:41669", + "right": "issue:44246", + "accept": true, + "reason": "Same underlying issue: `import transformers` is slow; 41669 identifies the likely root cause and 44246 reports the same symptom." + }, + { + "left": "issue:43867", + "right": "issue:45357", + "accept": false, + "reason": "Different save/load key mismatches for different models." + }, + { + "left": "issue:43992", + "right": "issue:45478", + "accept": false, + "reason": "Different model families and different from_pretrained failures." + }, + { + "left": "issue:28282", + "right": "issue:33453", + "accept": false, + "reason": "Missing PyTorch dependency error is unrelated to tokenizer-loading regression." + }, + { + "left": "issue:43618", + "right": "issue:46285", + "accept": false, + "reason": "Both are CLIP-related, but one is a missing `attentions` assignment and the other is a class-structure inconsistency." + }, + { + "left": "issue:44360", + "right": "issue:45440", + "accept": false, + "reason": "Different model implementation concerns: DSA indexer activation vs DeepSeekV3MoE parity." + }, + { + "left": "issue:42890", + "right": "issue:44964", + "accept": false, + "reason": "Unrelated test-seeding issue vs model-loading failure for Phi-4 multimodal." + }, + { + "left": "issue:42890", + "right": "issue:44112", + "accept": false, + "reason": "Both are CI/test failures, but the causes are different." + }, + { + "left": "issue:42890", + "right": "issue:45600", + "accept": false, + "reason": "Missing seeds in tests is unrelated to stale auto-mappings referencing removed configs." + }, + { + "left": "issue:43502", + "right": "issue:44062", + "accept": false, + "reason": "Offline-loading API requests vs tokenizer `AddedToken` keyword collision are different bugs." + }, + { + "left": "issue:42898", + "right": "issue:46664", + "accept": false, + "reason": "Tokenizer cleanup behavior change is unrelated to video processor frame-count semantics." + }, + { + "left": "issue:44734", + "right": "issue:46617", + "accept": false, + "reason": "KV-cache continuation indexing crash is unrelated to cache loading `FileNotFoundError`." + }, + { + "left": "issue:43723", + "right": "issue:44556", + "accept": false, + "reason": "Both involve loading after v5, but one is tokenizer loading and the other is checkpoint reload compatibility; not enough evidence of the same bug." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The lone accepted duplicate pair, issue:41669 with issue:44246, is well supported by the titles and shared slowdown symptom. The other soft-edge rejections are phrased cautiously and do not overclaim shared root causes." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 58, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:29870", + "issue:30333", + "issue:30990", + "issue:31795", + "issue:33453", + "issue:34689", + "issue:36296", + "issue:38617", + "issue:41628", + "issue:42548", + "issue:42757", + "issue:43010", + "issue:43065", + "issue:43232", + "issue:43299", + "issue:43352", + "issue:43475", + "issue:43502", + "issue:43576", + "issue:43824", + "issue:43827", + "issue:43874", + "issue:43901", + "issue:44222", + "issue:44230", + "issue:44315", + "issue:44351", + "issue:44355", + "issue:44360", + "issue:44387", + "issue:44485", + "issue:44556", + "issue:44749", + "issue:44938", + "issue:44961", + "issue:45071", + "issue:45125", + "issue:45375", + "issue:45468", + "issue:45478", + "issue:46132", + "issue:46143", + "issue:46144", + "issue:46209", + "issue:46617", + "issue:46664", + "issue:46726" + ], + "soft_pairs": [ + "issue:44222|issue:45125", + "issue:43824|issue:46617", + "issue:43502|issue:44485", + "issue:43010|issue:43065", + "issue:43874|issue:46617", + "issue:43576|issue:46132", + "issue:44387|issue:44556", + "issue:30333|issue:33453", + "issue:45125|issue:45478", + "issue:44355|issue:44749", + "issue:42548|issue:44351", + "issue:30990|issue:31795", + "issue:45125|issue:45375", + "issue:43502|issue:44222", + "issue:43502|issue:44961", + "issue:43010|issue:43232", + "issue:46664|issue:46726", + "issue:43827|issue:44961", + "issue:30333|issue:36296", + "issue:29870|issue:31795", + "issue:43299|issue:44556", + "issue:33453|issue:34689", + "issue:44355|issue:45071", + "issue:43576|issue:45071", + "issue:42757|issue:43576", + "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:36296|issue:43901", + "issue:43352|issue:44961", + "issue:43576|issue:44938", + "issue:43475|issue:46617", + "issue:46143|issue:46144", + "issue:41628|issue:42548", + "issue:44315|issue:46209", + "issue:44222|issue:44230" + ], + "prepared_review_unit_hash": "a2eab54e66023c374e8d4fc855f1a34af28e9c94aa7eb375401273f076b9890e", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 39, + "serialized_chars": 38292, + "estimated_input_tokens": 9573, + "estimated_eval_tokens": 19402 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: most items are unrelated Transformers regression, import, loading, docs, or model-specific bugs. None of the soft pairs look like the same underlying issue closely enough to merge, so I would reject all candidate duplicate edges.", + "confidence": 0.78, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "If forced to pick a representative, issue 43576 is the broadest v5 regression report in the set, but the cluster is too mixed for it to serve as a true canonical duplicate target.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44222", + "right": "issue:45125", + "accept": false, + "reason": "FP8 save_pretrained/serialization vs tensor-parallel plan/model-loading; different failure modes and code paths." + }, + { + "left": "issue:43824", + "right": "issue:46617", + "accept": false, + "reason": "Missing top-level class export vs cache/trust_remote_code FileNotFoundError; unrelated causes." + }, + { + "left": "issue:43502", + "right": "issue:44485", + "accept": false, + "reason": "Local-files-only network leakage vs GLM-5 RoPE implementation discussion; not the same bug." + }, + { + "left": "issue:43010", + "right": "issue:43065", + "accept": false, + "reason": "Cache update no_grad behavior vs SAM3 dummy Conv2d; completely different subsystems." + }, + { + "left": "issue:43874", + "right": "issue:46617", + "accept": false, + "reason": "Missing image-patch helper vs cache load failure; different underlying problems." + }, + { + "left": "issue:43576", + "right": "issue:46132", + "accept": false, + "reason": "Broken env CLI command vs AttentionInterface.register behavior change; not the same issue." + }, + { + "left": "issue:44387", + "right": "issue:44556", + "accept": false, + "reason": "Int4 OOM/memory regression vs checkpoint reload breakage after upgrade; distinct bugs." + }, + { + "left": "issue:30333", + "right": "issue:33453", + "accept": false, + "reason": "MLflow reporting on failed jobs vs tokenizer loading regression; unrelated." + }, + { + "left": "issue:45125", + "right": "issue:45478", + "accept": false, + "reason": "Both involve Qwen3.5 MoE, but one is missing _tp_plan and the other is a from_pretrained error; too little evidence of one concrete bug." + }, + { + "left": "issue:44355", + "right": "issue:44749", + "accept": false, + "reason": "Compiled Python file errors vs slowdown during data filtering after upgrade; different symptoms and paths." + }, + { + "left": "issue:42548", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but for different symbols and likely different export failures." + }, + { + "left": "issue:30990", + "right": "issue:31795", + "accept": false, + "reason": "Sentence Transformers loading hang vs docs confusion about model.forward arguments; unrelated." + }, + { + "left": "issue:45125", + "right": "issue:45375", + "accept": false, + "reason": "Tensor-parallel plan issue vs missing config field silently dropped by strict validation; related model family, not same bug." + }, + { + "left": "issue:43502", + "right": "issue:44222", + "accept": false, + "reason": "Local_files_only network requests vs FP8 save_pretrained; different code paths." + }, + { + "left": "issue:43502", + "right": "issue:44961", + "accept": false, + "reason": "Network leakage bug vs unrelated placeholder issue title; no shared underlying defect." + }, + { + "left": "issue:43010", + "right": "issue:43232", + "accept": false, + "reason": "No-grad decoration for cache updates vs generation kwargs handling after sync_gpus; unrelated generation internals." + }, + { + "left": "issue:46664", + "right": "issue:46726", + "accept": false, + "reason": "Video num_frames semantics vs AutoProcessor backend-missing import message; different processor bugs." + }, + { + "left": "issue:43827", + "right": "issue:44961", + "accept": false, + "reason": "Docs still referencing pipeline() vs unrelated placeholder issue; not a duplicate." + }, + { + "left": "issue:30333", + "right": "issue:36296", + "accept": false, + "reason": "MLflow reporting vs tensor-parallel training bug; unrelated." + }, + { + "left": "issue:29870", + "right": "issue:31795", + "accept": false, + "reason": "Bart generate UnboundLocalError vs documentation confusion; different artifact types and problems." + }, + { + "left": "issue:43299", + "right": "issue:44556", + "accept": false, + "reason": "Qwen3VL/Qwen3.5 MoE load regression vs checkpoint reload after version upgrade; both version-related, but not the same concrete failure." + }, + { + "left": "issue:33453", + "right": "issue:34689", + "accept": false, + "reason": "Tokenizer loading regression vs Llama 3.2 Vision model loading breakage; similar theme, different model-specific bugs." + }, + { + "left": "issue:44355", + "right": "issue:45071", + "accept": false, + "reason": "Compiled-file loading errors vs PretrainedConfig type checking regression; unrelated." + }, + { + "left": "issue:43576", + "right": "issue:45071", + "accept": false, + "reason": "Broken env command vs typing regression; no shared code path." + }, + { + "left": "issue:42757", + "right": "issue:43576", + "accept": false, + "reason": "huggingface_hub import issue vs transformers env command bug; different packages and failures." + }, + { + "left": "issue:44315", + "right": "issue:44360", + "accept": false, + "reason": "Liger Kernel/model_init application vs DSA indexer ReLU discussion; unrelated." + }, + { + "left": "issue:31795", + "right": "issue:38617", + "accept": false, + "reason": "Documentation confusion vs missing layer_type_validation import; different problem classes." + }, + { + "left": "issue:43576", + "right": "issue:43824", + "accept": false, + "reason": "CLI env regression vs missing model class export; not the same bug." + }, + { + "left": "issue:44961", + "right": "issue:45468", + "accept": false, + "reason": "Placeholder title vs Gemma4 audio positional encoding bug; no relation." + }, + { + "left": "issue:44485", + "right": "issue:44961", + "accept": false, + "reason": "GLM-5 RoPE implementation discussion vs placeholder title; unrelated." + }, + { + "left": "issue:44351", + "right": "issue:46617", + "accept": false, + "reason": "Missing HybridCache export vs cache FileNotFoundError; different root causes." + }, + { + "left": "issue:36296", + "right": "issue:43901", + "accept": false, + "reason": "Tensor-parallel training bug vs text-classification docs mismatch; unrelated." + }, + { + "left": "issue:43352", + "right": "issue:44961", + "accept": false, + "reason": "Flash Attention 2.0 support error vs placeholder title; no overlap." + }, + { + "left": "issue:43576", + "right": "issue:44938", + "accept": false, + "reason": "Env command broken in v5 vs Python 3.14 load failure; different issues." + }, + { + "left": "issue:43475", + "right": "issue:46617", + "accept": false, + "reason": "SAM3VisionEncoderOutput attribute error vs trust_remote_code cache loading failure; unrelated." + }, + { + "left": "issue:46143", + "right": "issue:46144", + "accept": false, + "reason": "kwargs passthrough bug vs attention-interface usage in RoFormer; related area, but not the same concrete defect." + }, + { + "left": "issue:41628", + "right": "issue:42548", + "accept": false, + "reason": "Different missing top-level imports; same broad symptom, but not enough to treat as one bug." + }, + { + "left": "issue:44315", + "right": "issue:46209", + "accept": false, + "reason": "Liger Kernel/model_init issue vs FP8 inference failure on B200; different code paths." + }, + { + "left": "issue:44222", + "right": "issue:44230", + "accept": false, + "reason": "FP8 save_pretrained bug vs FP8 inference support request; same buzzwords, different problems." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 26, + "review_unit_index": 59, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:31795", + "issue:33453", + "issue:41669", + "issue:42371", + "issue:42890", + "issue:43352", + "issue:43408", + "issue:43519", + "issue:43525", + "issue:43572", + "issue:43576", + "issue:43673", + "issue:43976", + "issue:44038", + "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:44623", + "issue:44704", + "issue:44829", + "issue:44861", + "issue:44908", + "issue:45003", + "issue:45071", + "issue:45092", + "issue:45125", + "issue:45230", + "issue:45278", + "issue:45341", + "issue:45362", + "issue:45542", + "issue:45753", + "issue:45850", + "issue:46095", + "issue:46218", + "issue:46328", + "issue:46655", + "issue:46664" + ], + "soft_pairs": [ + "issue:45362|issue:45850", + "issue:43673|issue:44492", + "issue:44485|issue:46664", + "issue:44273|issue:45542", + "issue:44623|issue:45230", + "issue:44273|issue:45003", + "issue:44360|issue:46664", + "issue:44038|issue:45125", + "issue:41669|issue:43673", + "issue:44393|issue:45071", + "issue:44393|issue:45753", + "issue:43576|issue:44556", + "issue:43525|issue:45850", + "issue:43572|issue:45850", + "issue:44556|issue:45092", + "issue:44492|issue:45850", + "issue:45278|issue:45850", + "issue:42371|issue:44492", + "issue:44230|issue:46095", + "issue:41669|issue:46218", + "issue:43408|issue:44315", + "issue:44246|issue:44273", + "issue:42890|issue:44908", + "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:44492|issue:46328", + "issue:44704|issue:44908", + "issue:41669|issue:44351", + "issue:41669|issue:45341", + "issue:43976|issue:45850", + "issue:43976|issue:46218", + "issue:44077|issue:44861", + "issue:31795|issue:33453", + "issue:43576|issue:44273", + "issue:44315|issue:44829", + "issue:43519|issue:46655" + ], + "prepared_review_unit_hash": "4cf768a63b492611751e7c2fdd10ee3c2061375f5f19c364c96fe103f68149d6", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 40, + "serialized_chars": 39088, + "estimated_input_tokens": 9772, + "estimated_eval_tokens": 19800 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These are mostly unrelated closed issues around Transformers v5 regressions, model-specific bugs, and import/loading problems. No soft pair is close enough to be treated as the same underlying bug or change.", + "confidence": 0.34, + "canonical_issue_reason": "issue:45850 is the broadest umbrella for the v5 regression theme, but it is still too generic to represent a true duplicate cluster.", + "canonical_pr_reason": "No pull requests are present in the provided items.", + "best_issue_reason": "issue:45850 is the best overall representative because it captures the widest version-regression complaint, even though it is not a precise duplicate target.", + "best_pr_reason": "No pull requests are present, so there is no best PR candidate.", + "soft_edge_verdicts": [ + { + "left": "issue:45362", + "right": "issue:45850", + "accept": false, + "reason": "Same general v5 frustration, but one is a specific Qwen3.5 chat crash and the other is a broad umbrella complaint." + }, + { + "left": "issue:43673", + "right": "issue:44492", + "accept": false, + "reason": "Different cache-related topics: chunked_prefill cache handling vs a typo in cache strategy wording." + }, + { + "left": "issue:44485", + "right": "issue:46664", + "accept": false, + "reason": "Unrelated model areas: GLM-5 RoPE implementation vs SmolVLM video-frame semantics." + }, + { + "left": "issue:44273", + "right": "issue:45542", + "accept": false, + "reason": "Both involve loading/import paths, but they are different bugs in different subsystems." + }, + { + "left": "issue:44623", + "right": "issue:45230", + "accept": false, + "reason": "One is a specific missing-files bug in processor.save_pretrained; the other is a generic bug report with no clear overlap." + }, + { + "left": "issue:44273", + "right": "issue:45003", + "accept": false, + "reason": "Both touch import/lazy-loading behavior, but the concrete failure modes are different." + }, + { + "left": "issue:44360", + "right": "issue:46664", + "accept": false, + "reason": "Completely different areas: DSA indexer behavior vs video processor frame handling." + }, + { + "left": "issue:44038", + "right": "issue:45125", + "accept": false, + "reason": "Both mention Qwen3/MoE, but one is a model-support bug and the other is tensor-parallel metadata." + }, + { + "left": "issue:41669", + "right": "issue:43673", + "accept": false, + "reason": "Import-time performance regression and cache-missing runtime bug are not the same code-path problem." + }, + { + "left": "issue:44393", + "right": "issue:45071", + "accept": false, + "reason": "Different layers entirely: Qwen3-VL output behavior vs PretrainedConfig type checking." + }, + { + "left": "issue:44393", + "right": "issue:45753", + "accept": false, + "reason": "Both are Qwen3-VL-family issues, but one is bbox output error and the other is an infinite loop on a specific image." + }, + { + "left": "issue:43576", + "right": "issue:44556", + "accept": false, + "reason": "Both are v5 breakages, but one is the env command and the other is checkpoint reload compatibility." + }, + { + "left": "issue:43525", + "right": "issue:45850", + "accept": false, + "reason": "A specific Llama4Config attribute error is not the same as a broad 'many models break' complaint." + }, + { + "left": "issue:43572", + "right": "issue:45850", + "accept": false, + "reason": "StableLmConfig pad token field bug is model-specific, not the same as the generic v5 regression umbrella." + }, + { + "left": "issue:44556", + "right": "issue:45092", + "accept": false, + "reason": "Related upgrade/reload symptoms, but the checkpoint compatibility context is different and not clearly the same bug." + }, + { + "left": "issue:44492", + "right": "issue:45850", + "accept": false, + "reason": "A typo in cache strategies is unrelated to the broad v5 regression report." + }, + { + "left": "issue:45278", + "right": "issue:45850", + "accept": false, + "reason": "General upgrade breakage overlaps in theme only; it is not the same concrete issue." + }, + { + "left": "issue:42371", + "right": "issue:44492", + "accept": false, + "reason": "TF32 API guidance and cache strategy typo are unrelated." + }, + { + "left": "issue:44230", + "right": "issue:46095", + "accept": false, + "reason": "Different request types: FP8 support for Qwen3-VL/Qwen3.5 MoE vs a DeepSeekV4 weight conversion question." + }, + { + "left": "issue:41669", + "right": "issue:46218", + "accept": false, + "reason": "Import-related, but one is import-time slowdown and the other is a missing module import error." + }, + { + "left": "issue:43408", + "right": "issue:44315", + "accept": false, + "reason": "Different models and different failures: model-type mismatch warning vs Liger Kernel not applied." + }, + { + "left": "issue:44246", + "right": "issue:44273", + "accept": false, + "reason": "Both concern imports/loading, but one is intermittent slow import and the other is broken lazy loading." + }, + { + "left": "issue:42890", + "right": "issue:44908", + "accept": false, + "reason": "Test flakiness from missing set_seed calls is unrelated to scheduler kwargs handling." + }, + { + "left": "issue:43576", + "right": "issue:44908", + "accept": false, + "reason": "Different bugs in different features: env CLI vs inverse_sqrt scheduler kwargs." + }, + { + "left": "issue:43352", + "right": "issue:44393", + "accept": false, + "reason": "Flash-Attention-2 support error is unrelated to Qwen3-VL bbox output hallucination." + }, + { + "left": "issue:43976", + "right": "issue:44393", + "accept": false, + "reason": "Python-version compatibility regression is unrelated to the Qwen3-VL output bug." + }, + { + "left": "issue:44908", + "right": "issue:45071", + "accept": false, + "reason": "Scheduler kwargs handling is unrelated to PretrainedConfig type checking." + }, + { + "left": "issue:44295", + "right": "issue:45230", + "accept": false, + "reason": "Specific position_ids buffer error does not match a generic bug report." + }, + { + "left": "issue:44297", + "right": "issue:45230", + "accept": false, + "reason": "Tokenizer save_pretrained class mismatch is a distinct serializer bug, not the same as a generic report." + }, + { + "left": "issue:44492", + "right": "issue:46328", + "accept": false, + "reason": "Cache strategy typo and feature-extractor padding conversion are unrelated." + }, + { + "left": "issue:44704", + "right": "issue:44908", + "accept": false, + "reason": "Both mention kwargs, but they affect different APIs and different code paths." + }, + { + "left": "issue:41669", + "right": "issue:44351", + "accept": false, + "reason": "Import performance regression is not the same as a missing HybridCache symbol." + }, + { + "left": "issue:41669", + "right": "issue:45341", + "accept": false, + "reason": "No substantive overlap between import-star performance and a testing_utils bug." + }, + { + "left": "issue:43976", + "right": "issue:45850", + "accept": false, + "reason": "A Python-version support issue is broader regression noise, not the same underlying bug." + }, + { + "left": "issue:43976", + "right": "issue:46218", + "accept": false, + "reason": "Python support and missing module import errors are different failure classes." + }, + { + "left": "issue:44077", + "right": "issue:44861", + "accept": false, + "reason": "Optional post_init validation and tied-weights list handling are unrelated." + }, + { + "left": "issue:31795", + "right": "issue:33453", + "accept": false, + "reason": "Documentation confusion and tokenizer loading regression are different issues." + }, + { + "left": "issue:43576", + "right": "issue:44273", + "accept": false, + "reason": "CLI env command breakage and lazy loading failure only overlap at a very high level." + }, + { + "left": "issue:44315", + "right": "issue:44829", + "accept": false, + "reason": "Liger Kernel application in model_init is unrelated to flash_attention_3 training degeneration." + }, + { + "left": "issue:43519", + "right": "issue:46655", + "accept": false, + "reason": "Both are processor-related, but timestamp calculation and video padding are distinct 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": 26, + "review_unit_index": 60, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:28348", + "issue:29870", + "issue:30333", + "issue:31515", + "issue:31795", + "issue:34689", + "issue:36296", + "issue:38617", + "issue:41669", + "issue:42371", + "issue:42913", + "issue:43066", + "issue:43298", + "issue:43352", + "issue:43502", + "issue:43519", + "issue:43525", + "issue:43572", + "issue:43576", + "issue:43606", + "issue:43618", + "issue:43704", + "issue:43717", + "issue:44230", + "issue:44263", + "issue:44273", + "issue:44351", + "issue:44360", + "issue:44485", + "issue:44492", + "issue:44623", + "issue:44655", + "issue:44861", + "issue:44908", + "issue:44938", + "issue:44998", + "issue:45230", + "issue:45250", + "issue:45290", + "issue:45310", + "issue:45468", + "issue:45850", + "issue:46018", + "issue:46095", + "issue:46144", + "issue:46172", + "issue:46218", + "issue:46894" + ], + "soft_pairs": [ + "issue:44360|issue:45468", + "issue:44273|issue:44908", + "issue:45230|issue:45310", + "issue:44655|issue:44861", + "issue:45468|issue:46894", + "issue:43519|issue:44485", + "issue:43576|issue:46218", + "issue:43519|issue:44908", + "issue:41669|issue:43704", + "issue:31795|issue:34689", + "issue:44485|issue:45468", + "issue:43704|issue:44485", + "issue:46018|issue:46172", + "issue:43502|issue:43519", + "issue:42371|issue:43704", + "issue:45850|issue:46218", + "issue:43519|issue:44623", + "issue:44263|issue:45468", + "issue:43606|issue:45468", + "issue:43298|issue:46095", + "issue:43519|issue:43704", + "issue:44351|issue:46218", + "issue:43352|issue:46218", + "issue:42913|issue:43066", + "issue:28348|issue:45250", + "issue:44908|issue:44998", + "issue:44492|issue:44998", + "issue:43618|issue:44998", + "issue:44998|issue:45850", + "issue:44938|issue:44998", + "issue:44861|issue:45290", + "issue:31515|issue:38617", + "issue:43572|issue:44861", + "issue:43525|issue:44861", + "issue:44230|issue:44623", + "issue:28348|issue:43352", + "issue:29870|issue:31515", + "issue:30333|issue:31515", + "issue:31515|issue:36296", + "issue:31515|issue:31795", + "issue:44230|issue:45310", + "issue:43298|issue:44230", + "issue:43717|issue:46144" + ], + "prepared_review_unit_hash": "cb5a2f096d7f4f61484a67a85256ca265bb0b959728ccd57dfc03fcba51c539c", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 43, + "serialized_chars": 39624, + "estimated_input_tokens": 9906, + "estimated_eval_tokens": 20068 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous: the soft matches share broad themes, but none look like the same concrete bug or change. I would not merge any of the candidate pairs; 42913 is the best representative only because it is the clearest broad tokenizer-v5 regression report.", + "confidence": 0.79, + "canonical_issue_reason": "42913 is the broadest and clearest issue in the set, and the only plausible near-match is 43066; it is the best anchor if this cluster must be represented by one issue.", + "canonical_pr_reason": null, + "best_issue_reason": "42913 is the most descriptive user-facing report and the best general representative among otherwise unrelated issues.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44360", + "right": "issue:45468", + "accept": false, + "reason": "Different model/component areas: DSA indexer vs Gemma-4 audio positional encoding." + }, + { + "left": "issue:44273", + "right": "issue:44908", + "accept": false, + "reason": "Lazy loading versus scheduler kwargs are unrelated failure modes." + }, + { + "left": "issue:45230", + "right": "issue:45310", + "accept": false, + "reason": "Generic bug report does not match a specific Qwen3.5 MoE from_pretrained regression." + }, + { + "left": "issue:44655", + "right": "issue:44861", + "accept": false, + "reason": "Saving Pipeline objects and tied-weight key handling are different code paths." + }, + { + "left": "issue:45468", + "right": "issue:46894", + "accept": false, + "reason": "Different model families and symptoms: Gemma-4 audio encoding vs CohereASR training loss shift." + }, + { + "left": "issue:43519", + "right": "issue:44485", + "accept": false, + "reason": "Qwen3VL timestamp math and GLM-5 RoPE implementation are distinct bugs." + }, + { + "left": "issue:43576", + "right": "issue:46218", + "accept": false, + "reason": "Transformers env CLI breakage is unrelated to a Qwen3_5 import failure." + }, + { + "left": "issue:43519", + "right": "issue:44908", + "accept": false, + "reason": "Processor timestamp calculation and inverse_sqrt scheduler kwargs do not share a code path." + }, + { + "left": "issue:41669", + "right": "issue:43704", + "accept": false, + "reason": "Import-star cleanup/perf issue is unrelated to a VRAM leak in Qwen3ForCausalLM." + }, + { + "left": "issue:31795", + "right": "issue:34689", + "accept": false, + "reason": "Documentation confusion about inputs differs from a Llama 3.2 model-loading regression." + }, + { + "left": "issue:44485", + "right": "issue:45468", + "accept": false, + "reason": "GLM-5 RoPE math and Gemma-4 audio positional encoding are separate implementation problems." + }, + { + "left": "issue:43704", + "right": "issue:44485", + "accept": false, + "reason": "VRAM leak in Qwen3ForCausalLM is not the same as GLM-5 RoPE behavior." + }, + { + "left": "issue:46018", + "right": "issue:46172", + "accept": false, + "reason": "DeepSeek shared-expert gating and a comb transpose/dtype bug are different concrete issues." + }, + { + "left": "issue:43502", + "right": "issue:43519", + "accept": false, + "reason": "Local-files-only network access and Qwen3VL timestamp calculation are unrelated." + }, + { + "left": "issue:42371", + "right": "issue:43704", + "accept": false, + "reason": "TF32 settings and a dataloader-thread VRAM leak are not the same bug." + }, + { + "left": "issue:45850", + "right": "issue:46218", + "accept": false, + "reason": "A broad v5 breakage complaint does not specifically match a Qwen3_5 import error." + }, + { + "left": "issue:43519", + "right": "issue:44623", + "accept": false, + "reason": "Timestamp computation and missing files in processor.save_pretrained are different failures." + }, + { + "left": "issue:44263", + "right": "issue:45468", + "accept": false, + "reason": "torch.split return handling in GlmMoeDsaIndexer is unrelated to Gemma-4 audio encoding." + }, + { + "left": "issue:43606", + "right": "issue:45468", + "accept": false, + "reason": "CPU offload device mismatch in bark-small is unrelated to Gemma-4 audio positional encoding." + }, + { + "left": "issue:43298", + "right": "issue:46095", + "accept": false, + "reason": "Continuous batching support for audio models and DeepSeek-V4 weight conversion are not the same change." + }, + { + "left": "issue:43519", + "right": "issue:43704", + "accept": false, + "reason": "Processor timestamp math and VRAM leakage are unrelated symptoms." + }, + { + "left": "issue:44351", + "right": "issue:46218", + "accept": false, + "reason": "Both are import errors, but they are missing different symbols from different modules." + }, + { + "left": "issue:43352", + "right": "issue:46218", + "accept": false, + "reason": "Flash Attention 2.0 support for Nemotron is unrelated to importing Qwen3_5ForCausalLM." + }, + { + "left": "issue:42913", + "right": "issue:43066", + "accept": false, + "reason": "Both concern tokenizer v5 behavior, but one is a broad regression report and the other is a specific decoder-type bug; not clearly the same concrete issue." + }, + { + "left": "issue:28348", + "right": "issue:45250", + "accept": false, + "reason": "Specific GPT2LMHeadModel Flash Attention support request is not the same as a generic Flash Attention 2.0 topic." + }, + { + "left": "issue:44908", + "right": "issue:44998", + "accept": false, + "reason": "Scheduler kwargs and an unrelated low-signal issue title share no concrete bug." + }, + { + "left": "issue:44492", + "right": "issue:44998", + "accept": false, + "reason": "A cache-strategy typo and a generic low-signal issue are not duplicates." + }, + { + "left": "issue:43618", + "right": "issue:44998", + "accept": false, + "reason": "CLIPOutput attentions assignment and a generic low-signal issue are unrelated." + }, + { + "left": "issue:44998", + "right": "issue:45850", + "accept": false, + "reason": "A generic low-signal issue is not the same as broad v5 breakage." + }, + { + "left": "issue:44938", + "right": "issue:44998", + "accept": false, + "reason": "Python 3.14 loading failure is unrelated to a generic low-signal issue." + }, + { + "left": "issue:44861", + "right": "issue:45290", + "accept": false, + "reason": "Tied-weight key handling and chat-template tool-call crashes are different code paths." + }, + { + "left": "issue:31515", + "right": "issue:38617", + "accept": false, + "reason": "Checkpoint load slowness and missing layer_type_validation import are unrelated." + }, + { + "left": "issue:43572", + "right": "issue:44861", + "accept": false, + "reason": "Missing pad_token_idx and tied-weight key crashes are different configuration/model-save problems." + }, + { + "left": "issue:43525", + "right": "issue:44861", + "accept": false, + "reason": "Llama4Config missing pad_token_id is unrelated to tied-weight key list handling." + }, + { + "left": "issue:44230", + "right": "issue:44623", + "accept": false, + "reason": "FP8 inference support and missing processor save files are different problems." + }, + { + "left": "issue:28348", + "right": "issue:43352", + "accept": false, + "reason": "Both mention Flash Attention 2.0, but they target different models and likely different fixes." + }, + { + "left": "issue:29870", + "right": "issue:31515", + "accept": false, + "reason": "BART generate() decoder_attention_mask bug is unrelated to from_pretrained checkpoint slowness." + }, + { + "left": "issue:30333", + "right": "issue:31515", + "accept": false, + "reason": "MLFlow reporting on job failure and checkpoint loading performance are different issues." + }, + { + "left": "issue:31515", + "right": "issue:36296", + "accept": false, + "reason": "from_pretrained slowness and tensor-parallel training bugs are not the same bug." + }, + { + "left": "issue:31515", + "right": "issue:31795", + "accept": false, + "reason": "Checkpoint loading performance and forward-argument documentation are unrelated." + }, + { + "left": "issue:44230", + "right": "issue:45310", + "accept": false, + "reason": "FP8 inference support and Qwen3.5 MoE from_pretrained failure are different feature/breakage reports." + }, + { + "left": "issue:43298", + "right": "issue:44230", + "accept": false, + "reason": "Audio continuous batching support and FP8 inference support are separate feature requests." + }, + { + "left": "issue:43717", + "right": "issue:46144", + "accept": false, + "reason": "Mamba init_weights usage and RoFormer attention-interface usage are unrelated implementation concerns." + } + ] + }, + "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": 26, + "review_unit_index": 61, + "review_unit_count": 61, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-506", + "nodes": [ + "issue:37934", + "issue:42936", + "issue:43240", + "issue:43668", + "issue:43813", + "issue:44031", + "issue:46018", + "issue:46144" + ], + "soft_pairs": [ + "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:43668|issue:43813", + "issue:43240|issue:43668", + "issue:37934|issue:43668", + "issue:42936|issue:43668" + ], + "prepared_review_unit_hash": "770c8c78cc8e52be1473494a44abd6df8294f402b453d710733f3c6349d9f14d", + "component_budget": { + "node_count": 506, + "item_count": 506, + "soft_pair_count": 1861, + "serialized_chars": 965513, + "estimated_input_tokens": 241379, + "estimated_eval_tokens": 483014 + }, + "budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 11, + "serialized_chars": 7775, + "estimated_input_tokens": 1944, + "estimated_eval_tokens": 4144 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is not a duplicate set; it contains several unrelated bugs and one type-hint/typo issue across different models and subsystems. No single canonical issue or PR stands out.", + "confidence": 0.97, + "canonical_issue_reason": "No canonical issue: the items span unrelated topics (norm behavior, ignore_value handling, kwargs propagation, type hints, typos, tokenizer warning, DeepSeek gating, RoFormer attention), so they are not the same underlying bug.", + "canonical_pr_reason": null, + "best_issue_reason": "None is a strong global representative because the cluster is heterogeneous; if forced, issue 42936 is a concrete self-contained bug report, but it does not unify the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42936", + "right": "issue:46018", + "accept": false, + "reason": "Different models and different bugs: Mask2Former ignore_value handling vs DeepSeek-V4 shared expert gating." + }, + { + "left": "issue:43668", + "right": "issue:44031", + "accept": false, + "reason": "Unrelated problems: a ModernBERT config type hint vs tokenizer regex warning behavior." + }, + { + "left": "issue:42936", + "right": "issue:43240", + "accept": false, + "reason": "Mask2Former ignore_value usage is unrelated to kwargs not being forwarded to a loss function." + }, + { + "left": "issue:43240", + "right": "issue:46144", + "accept": false, + "reason": "Different code paths: loss-function kwargs propagation vs RoFormer attention interface usage." + }, + { + "left": "issue:42936", + "right": "issue:44031", + "accept": false, + "reason": "Different subsystems and symptoms: segmentation loss config vs tokenizer warning regression." + }, + { + "left": "issue:42936", + "right": "issue:46144", + "accept": false, + "reason": "No shared underlying bug; these concern different models and different implementation layers." + }, + { + "left": "issue:42936", + "right": "issue:43813", + "accept": false, + "reason": "Mask2Former loss handling and a typo in a quantization attribute are unrelated." + }, + { + "left": "issue:43668", + "right": "issue:43813", + "accept": false, + "reason": "A type-hint bug in ModernBERTConfig is not the same as a typo in quantization code." + }, + { + "left": "issue:43240", + "right": "issue:43668", + "accept": false, + "reason": "Loss kwargs forwarding and an incorrect config type hint are different kinds of issues." + }, + { + "left": "issue:37934", + "right": "issue:43668", + "accept": false, + "reason": "Llama4TextL2Norm/RMS norm question is unrelated to ModernBERTConfig norm_eps typing." + }, + { + "left": "issue:42936", + "right": "issue:43668", + "accept": false, + "reason": "Different models and bug classes; no shared concrete code-path problem." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": "c2fe418a88ac6a3a9dcbacc242f797d6255c0904363002830a5e08449378fc23", + "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": "These three closed issues are unrelated: tokenization truncation behavior, a ReduceLROnPlateau-style feature request, and Wav2Vec2 word boundary timestamps. They should not be merged as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": "No canonical issue: the items describe different features/requests in different parts of the library.", + "canonical_pr_reason": null, + "best_issue_reason": "If a representative had to be chosen, issue 10519 is the most technically specific and discussion-heavy, but it still does not represent the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:10519", + "right": "issue:11307", + "accept": false, + "reason": "Different problem domains: tokenizer truncation direction vs. Wav2Vec2 time offsets." + }, + { + "left": "issue:10519", + "right": "issue:11005", + "accept": false, + "reason": "Unrelated feature requests: truncation option vs. ReduceLROnPlateau-like scheduling behavior." + }, + { + "left": "issue:11005", + "right": "issue:11307", + "accept": false, + "reason": "No shared underlying bug or change; one is about learning-rate scheduling, the other about word timestamp extraction." + } + ] + }, + "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": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-13244-3", + "nodes": [ + "issue:13244", + "issue:15980", + "issue:16618" + ], + "soft_pairs": [ + "issue:13244|issue:16618", + "issue:15980|issue:16618" + ], + "prepared_review_unit_hash": "95d5c4a94b8695c0ce6a23d989c0acc670b337bb928a3807fe58c2040b3278ee", + "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": "The cluster is loosely about model loading/tokenization, but the issues describe different problems: one tokenizer discrepancy, one private-model auth/error-message issue, and one generic model-loading failure. I do not see a true duplicate set here.", + "confidence": 0.82, + "canonical_issue_reason": "Issue #16618 is the broadest model-loading report, but the cluster is not a single duplicate bug because #15980 is specifically about private-model authentication and #13244 is about Tapas tokenization behavior.", + "canonical_pr_reason": null, + "best_issue_reason": "#16618 is the best representative of the cluster\u2019s shared theme of model loading, but it is still not a clean canonical duplicate for the other issues.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:13244", + "right": "issue:16618", + "accept": false, + "reason": "Tokenizer mismatch for Tapas is a different underlying problem from a failure to load bert-base-uncased." + }, + { + "left": "issue:15980", + "right": "issue:16618", + "accept": false, + "reason": "Private-model download/auth error is not the same bug as a generic inability to load 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": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-26350-2", + "nodes": [ + "issue:26350", + "issue:33260" + ], + "soft_pairs": [ + "issue:26350|issue:33260" + ], + "prepared_review_unit_hash": "9a3de37a3e401568baec5273f2b8c33e879ec84080137e0b9ce86cec111db10f", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1589, + "estimated_input_tokens": 398, + "estimated_eval_tokens": 1052 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1589, + "estimated_input_tokens": 398, + "estimated_eval_tokens": 1052 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are both community contribution requests, but they target different feature areas: Flash Attention 2 support vs. GGUF support. They are not duplicates and should remain separate.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:26350", + "right": "issue:33260", + "accept": false, + "reason": "Different underlying feature requests with different target architectures/format support; they do not describe the same bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 5, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-27088-9", + "nodes": [ + "issue:27088", + "issue:28936", + "issue:38435", + "issue:43834", + "issue:44365", + "issue:44370", + "issue:46222", + "issue:46226", + "issue:46931" + ], + "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:43834|issue:46931", + "issue:44365|issue:44370", + "issue:44365|issue:46222", + "issue:44365|issue:46226", + "issue:44365|issue:46931", + "issue:44370|issue:46222", + "issue:44370|issue:46226", + "issue:44370|issue:46931", + "issue:46222|issue:46226", + "issue:46222|issue:46931", + "issue:46226|issue:46931" + ], + "prepared_review_unit_hash": "248e905d114ad4cb2b28f89b24b0c4f5a08e4c1c14ad3ab0d9a12b4d0cb6f72e", + "component_budget": { + "node_count": 9, + "item_count": 9, + "soft_pair_count": 29, + "serialized_chars": 15977, + "estimated_input_tokens": 3995, + "estimated_eval_tokens": 8246 + }, + "budget": { + "node_count": 9, + "item_count": 9, + "soft_pair_count": 29, + "serialized_chars": 15977, + "estimated_input_tokens": 3995, + "estimated_eval_tokens": 8246 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Most items are i18n docs-translation requests, but they split into separate language-specific issues rather than one duplicate. The only clear duplicate sets are the three generic placeholder-title issues (43834/44365/44370) and the identical Thai pair (46222/46226).", + "confidence": 0.68, + "canonical_issue_reason": "Issue 43834 is the best representative of the largest exact-duplicate subgroup: it matches the other placeholder-title translation issues and is the most generic/central of that set.", + "canonical_pr_reason": null, + "best_issue_reason": "Among the available issues, 43834 is the cleanest representative of a repeated placeholder translation request; it sits in the densest duplicate pocket and is the most canonical-looking issue here.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:27088", + "right": "issue:44370", + "accept": false, + "reason": "Different target language/request; not the same concrete translation issue." + }, + { + "left": "issue:27088", + "right": "issue:44365", + "accept": false, + "reason": "Different target language/request; not a duplicate." + }, + { + "left": "issue:27088", + "right": "issue:43834", + "accept": false, + "reason": "Turkish vs generic placeholder translation request; not the same issue." + }, + { + "left": "issue:27088", + "right": "issue:38435", + "accept": false, + "reason": "Different language translation requests; separate issues." + }, + { + "left": "issue:27088", + "right": "issue:28936", + "accept": false, + "reason": "Turkish vs Spanish translation request; not a duplicate." + }, + { + "left": "issue:28936", + "right": "issue:43834", + "accept": false, + "reason": "Spanish request is distinct from the generic placeholder translation issue." + }, + { + "left": "issue:28936", + "right": "issue:44365", + "accept": false, + "reason": "Spanish vs generic placeholder request; not the same underlying issue." + }, + { + "left": "issue:28936", + "right": "issue:38435", + "accept": false, + "reason": "Different languages, so these are separate translation requests." + }, + { + "left": "issue:28936", + "right": "issue:44370", + "accept": false, + "reason": "Spanish vs generic placeholder request; not a duplicate." + }, + { + "left": "issue:38435", + "right": "issue:44370", + "accept": false, + "reason": "Romanian request is distinct from the generic placeholder issue." + }, + { + "left": "issue:38435", + "right": "issue:44365", + "accept": false, + "reason": "Romanian vs generic placeholder translation request; not the same issue." + }, + { + "left": "issue:38435", + "right": "issue:43834", + "accept": false, + "reason": "Romanian request is separate from the placeholder-title subgroup." + }, + { + "left": "issue:38435", + "right": "issue:46222", + "accept": false, + "reason": "Different language/request and different title; not a duplicate." + }, + { + "left": "issue:38435", + "right": "issue:46226", + "accept": false, + "reason": "Different language/request and different title; not a duplicate." + }, + { + "left": "issue:43834", + "right": "issue:44370", + "accept": true, + "reason": "Same generic placeholder translation request; looks like an exact duplicate." + }, + { + "left": "issue:43834", + "right": "issue:44365", + "accept": true, + "reason": "Same placeholder-title issue with no meaningful difference." + }, + { + "left": "issue:43834", + "right": "issue:46222", + "accept": false, + "reason": "Different title/content and not clearly the same translation request." + }, + { + "left": "issue:43834", + "right": "issue:46226", + "accept": false, + "reason": "Different title/content and not clearly the same translation request." + }, + { + "left": "issue:43834", + "right": "issue:46931", + "accept": false, + "reason": "Different title and likely different underlying request; too weak to merge." + }, + { + "left": "issue:44365", + "right": "issue:44370", + "accept": true, + "reason": "Same placeholder-title translation issue; effectively duplicate." + }, + { + "left": "issue:44365", + "right": "issue:46222", + "accept": false, + "reason": "Different title/content; not enough to treat as the same issue." + }, + { + "left": "issue:44365", + "right": "issue:46226", + "accept": false, + "reason": "Different title/content; not the same concrete issue." + }, + { + "left": "issue:44365", + "right": "issue:46931", + "accept": false, + "reason": "Different title and likely different request; reject." + }, + { + "left": "issue:44370", + "right": "issue:46222", + "accept": false, + "reason": "Different title/content; not a safe duplicate match." + }, + { + "left": "issue:44370", + "right": "issue:46226", + "accept": false, + "reason": "Different title/content; not the same issue." + }, + { + "left": "issue:44370", + "right": "issue:46931", + "accept": false, + "reason": "Different title and underlying request; reject." + }, + { + "left": "issue:46222", + "right": "issue:46226", + "accept": true, + "reason": "Identical Thai-titled issue; clear duplicate." + }, + { + "left": "issue:46222", + "right": "issue:46931", + "accept": false, + "reason": "Similar language/theme but different title and not clearly the same concrete issue." + }, + { + "left": "issue:46226", + "right": "issue:46931", + "accept": false, + "reason": "Different title and not enough evidence they are the same underlying request." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: the only clearly exact-duplicate clusters are the three identical placeholder-title issues (43834/44365/44370) and the identical Thai pair (46222/46226). The remaining pair rejections are conservative given that the packet only shows soft similarity and no shared issue links or concrete same-bug evidence. The canonical issue choice for the larger exact-duplicate subgroup is reasonable." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 6, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-36718-21", + "nodes": [ + "issue:36718", + "issue:36817", + "issue:36879", + "issue:37015", + "issue:41115", + "issue:41211", + "issue:42503", + "issue:42738", + "issue:42954", + "issue:43671", + "issue:43837", + "issue:43909", + "issue:44121", + "issue:44205", + "issue:45027", + "issue:45177", + "issue:45322", + "issue:46133", + "issue:46645", + "issue:46653", + "issue:47004" + ], + "soft_pairs": [ + "issue:36718|issue:47004", + "issue:36718|issue:41115", + "issue:36817|issue:41211", + "issue:36817|issue:43671", + "issue:36817|issue:36879", + "issue:36817|issue:42954", + "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:42954", + "issue:42738|issue:45027", + "issue:42503|issue:42738", + "issue:42738|issue:42954", + "issue:42738|issue:46653", + "issue:42954|issue:44205", + "issue:42954|issue:43671", + "issue:42954|issue:45177", + "issue:42954|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": "e917d1021d3bca9f7af3891e01eee639061d5c948cdb4d7b1fe74eb939f54b0b", + "component_budget": { + "node_count": 21, + "item_count": 21, + "soft_pair_count": 48, + "serialized_chars": 28618, + "estimated_input_tokens": 7155, + "estimated_eval_tokens": 14566 + }, + "budget": { + "node_count": 21, + "item_count": 21, + "soft_pair_count": 48, + "serialized_chars": 28618, + "estimated_input_tokens": 7155, + "estimated_eval_tokens": 14566 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This set is only loosely related by being model/architecture request issues; the titles span unrelated models and features, so there are no true duplicates here. All soft edges should be rejected.", + "confidence": 0.89, + "canonical_issue_reason": "Issue 36879 is the clearest representative of the group: a concrete, concise model-add request, and it sits near the center of the soft-similarity web. It is still only a loose representative, not a true duplicate target.", + "canonical_pr_reason": null, + "best_issue_reason": "36879 is the best issue anchor because it is specific, closed, and broadly representative of the 'add model support' pattern that links these items. However, the underlying requests are still distinct.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:36718", + "right": "issue:47004", + "accept": false, + "reason": "RoMa keypoint matcher vs LoMa keypoint detector/matcher: related naming, but different models and different requested changes." + }, + { + "left": "issue:36718", + "right": "issue:41115", + "accept": false, + "reason": "Keypoint matcher request vs MiniCPM3 architecture request; unrelated model families and code paths." + }, + { + "left": "issue:36817", + "right": "issue:41211", + "accept": false, + "reason": "EuroBert config addition and DEIMv2 are different model requests, not the same change." + }, + { + "left": "issue:36817", + "right": "issue:43671", + "accept": false, + "reason": "EuroBert config vs Qwen3-TTS support; different model types and implementation areas." + }, + { + "left": "issue:36817", + "right": "issue:36879", + "accept": false, + "reason": "EuroBert config and RF-DETR model support are unrelated model additions." + }, + { + "left": "issue:36817", + "right": "issue:42954", + "accept": false, + "reason": "EuroBert model config vs a person-name issue; no shared underlying bug or feature." + }, + { + "left": "issue:36879", + "right": "issue:41211", + "accept": false, + "reason": "RF-DETR and DEIMv2 are different model additions with different integration work." + }, + { + "left": "issue:36879", + "right": "issue:43909", + "accept": false, + "reason": "RF-DETR vision model vs LFM2.5 audio model; distinct artifacts and code paths." + }, + { + "left": "issue:36879", + "right": "issue:44121", + "accept": false, + "reason": "RF-DETR model support is unrelated to OpenAI weight-sparse transformer support." + }, + { + "left": "issue:36879", + "right": "issue:43837", + "accept": false, + "reason": "RF-DETR and Qwen3-ASR are different models and different tasks." + }, + { + "left": "issue:36879", + "right": "issue:43671", + "accept": false, + "reason": "RF-DETR vision model vs Qwen3-TTS audio support; not the same change." + }, + { + "left": "issue:37015", + "right": "issue:43837", + "accept": false, + "reason": "NeoBERT addition and Qwen3-ASR support are different model requests." + }, + { + "left": "issue:37015", + "right": "issue:44121", + "accept": false, + "reason": "NeoBERT vs weight-sparse transformer support; unrelated architectures." + }, + { + "left": "issue:37015", + "right": "issue:43909", + "accept": false, + "reason": "NeoBERT and LFM2.5 Audio address different model families." + }, + { + "left": "issue:36879", + "right": "issue:37015", + "accept": false, + "reason": "RF-DETR model support and NeoBERT support are separate model integrations." + }, + { + "left": "issue:37015", + "right": "issue:42503", + "accept": false, + "reason": "NeoBERT and ModernVBERT are different model additions, not duplicate requests." + }, + { + "left": "issue:41115", + "right": "issue:43837", + "accept": false, + "reason": "MiniCPM3 architecture request vs Qwen3-ASR support; different models and tasks." + }, + { + "left": "issue:41115", + "right": "issue:44121", + "accept": false, + "reason": "MiniCPM3 architecture work is not the same as adding weight-sparse transformer support." + }, + { + "left": "issue:41115", + "right": "issue:42503", + "accept": false, + "reason": "MiniCPM3 architecture and ModernVBERT models are separate model requests." + }, + { + "left": "issue:41115", + "right": "issue:45027", + "accept": false, + "reason": "MiniCPM3 vs Voxtral TTS support; different model families and code paths." + }, + { + "left": "issue:41211", + "right": "issue:43671", + "accept": false, + "reason": "DEIMv2 and Qwen3-TTS are unrelated model additions." + }, + { + "left": "issue:42503", + "right": "issue:43837", + "accept": false, + "reason": "ModernVBERT models and Qwen3-ASR support are different requested features." + }, + { + "left": "issue:42503", + "right": "issue:43671", + "accept": false, + "reason": "ModernVBERT and Qwen3-TTS are unrelated model integrations." + }, + { + "left": "issue:42503", + "right": "issue:42954", + "accept": false, + "reason": "ModernVBERT models and the Mirko Privitera issue are clearly unrelated." + }, + { + "left": "issue:42738", + "right": "issue:45027", + "accept": false, + "reason": "BERT-like models with RoPE and Voxtral TTS support are different model requests." + }, + { + "left": "issue:42503", + "right": "issue:42738", + "accept": false, + "reason": "ModernVBERT models vs a general BERT-like RoPE request; related theme, but not the same change." + }, + { + "left": "issue:42738", + "right": "issue:42954", + "accept": false, + "reason": "BERT-like RoPE discussion and a person-name issue have no common code-path." + }, + { + "left": "issue:42738", + "right": "issue:46653", + "accept": false, + "reason": "BERT-like RoPE support is unrelated to the Mirko Privitera issue." + }, + { + "left": "issue:42954", + "right": "issue:44205", + "accept": false, + "reason": "Person-name issue vs SAM3-LiteText; no shared underlying request." + }, + { + "left": "issue:42954", + "right": "issue:43671", + "accept": false, + "reason": "Mirko Privitera issue and Qwen3-TTS support are unrelated." + }, + { + "left": "issue:42954", + "right": "issue:45177", + "accept": false, + "reason": "Mirko Privitera issue vs DeepSeek-OCR2 request; not the same change." + }, + { + "left": "issue:42954", + "right": "issue:46653", + "accept": false, + "reason": "Mirko Privitera issue and another person-name issue are not a code change duplicate." + }, + { + "left": "issue:43671", + "right": "issue:43837", + "accept": false, + "reason": "Qwen3-TTS support and Qwen3-ASR support are different modalities and likely different implementations." + }, + { + "left": "issue:43671", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-TTS support and Voxtral-4B-TTS-2603 support are different model-specific requests." + }, + { + "left": "issue:43671", + "right": "issue:45177", + "accept": false, + "reason": "Qwen3-TTS and DeepSeek-OCR2 are unrelated model additions." + }, + { + "left": "issue:43671", + "right": "issue:44121", + "accept": false, + "reason": "Qwen3-TTS support is not the same as weight-sparse transformer support." + }, + { + "left": "issue:43837", + "right": "issue:43909", + "accept": false, + "reason": "Qwen3-ASR support and LFM2.5 Audio are different audio models with different integration work." + }, + { + "left": "issue:43837", + "right": "issue:45322", + "accept": false, + "reason": "Qwen3-ASR support and EUPE are unrelated model requests." + }, + { + "left": "issue:43837", + "right": "issue:44121", + "accept": false, + "reason": "ASR support and weight-sparse transformer support are different code paths." + }, + { + "left": "issue:43837", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-ASR and Voxtral-4B-TTS-2603 are different speech models and tasks." + }, + { + "left": "issue:37015", + "right": "issue:45027", + "accept": false, + "reason": "NeoBERT and Voxtral TTS are unrelated model requests." + }, + { + "left": "issue:43837", + "right": "issue:45177", + "accept": false, + "reason": "Qwen3-ASR support and DeepSeek-OCR2 are different model additions." + }, + { + "left": "issue:45322", + "right": "issue:46133", + "accept": false, + "reason": "EUPE and TIPSv2 are separate research models with different integration surfaces." + }, + { + "left": "issue:42503", + "right": "issue:46653", + "accept": false, + "reason": "ModernVBERT models and the Mirko Privitera issue are unrelated." + }, + { + "left": "issue:45027", + "right": "issue:46653", + "accept": false, + "reason": "Voxtral TTS support and the Mirko Privitera issue 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": 7, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-38740-2", + "nodes": [ + "issue:38740", + "issue:46443" + ], + "soft_pairs": [ + "issue:38740|issue:46443" + ], + "prepared_review_unit_hash": "75e35e3aee90439a253f2b98fd9c68800d8285105b6528cad84e0a2e15391ca1", + "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 two issues are not duplicates: one is a docs request about adding Pruna as an optimization framework, while the other is a Vision RoPE refactor in model code. They concern different subsystems and different kinds of work.", + "confidence": 0.99, + "canonical_issue_reason": "Issue #46443 is the better canonical issue only because it is the more concrete technical change request; the other is a documentation-oriented enhancement and does not appear related.", + "canonical_pr_reason": null, + "best_issue_reason": "#46443 is the best issue representative for the cluster because it describes an actual code refactor, whereas #38740 is a standalone docs item with no overlap.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:38740", + "right": "issue:46443", + "accept": false, + "reason": "The titles and context are unrelated: documentation for adding Pruna as an optimization framework versus a Vision RoPE refactor. No shared code path, filenames, or explicit target suggests the same underlying bug/change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 8, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-41084-2", + "nodes": [ + "issue:41084", + "issue:43037" + ], + "soft_pairs": [ + "issue:41084|issue:43037" + ], + "prepared_review_unit_hash": "e5835d87d37ad645790103d308718457ce31f2af65018185dfda214a61cba6d3", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1479, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1479, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues are not duplicates: one is a generic request about block decoding, while the other is a feature request for DeepSeek v3.2 support. They reference different underlying problems and likely different implementation work.", + "confidence": 0.97, + "canonical_issue_reason": "Issue #41084 is the broader, more generic software-behavior request and is the better candidate to represent the cluster if forced to pick an issue.", + "canonical_pr_reason": null, + "best_issue_reason": "#41084 is the more general issue and has clearer potential to subsume related block-decoding discussion; #43037 is model-specific and narrower.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41084", + "right": "issue:43037", + "accept": false, + "reason": "Different topics and code paths: block decoding versus adding support for a specific model version. The titles and context do not indicate the same underlying bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 9, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-41306-2", + "nodes": [ + "issue:41306", + "issue:41826" + ], + "soft_pairs": [ + "issue:41306|issue:41826" + ], + "prepared_review_unit_hash": "a944f1b5968727fb48b28f01f8504178cec4a2d207393d2caa6d2c841d9d8298", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1583, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1583, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are both about memory-efficient tiled model components, but they describe different features: one is about a `from_pretrained()` flag for FLCE/LigerFusedLinearCrossEntropy, the other about integrating TiledMLP. They do not look like duplicates or the same underlying change.", + "confidence": 0.17, + "canonical_issue_reason": "Issue #41306 is the more concrete, feature-specific request and is easier to anchor as the representative item, but it is not actually the same problem as the other issue.", + "canonical_pr_reason": null, + "best_issue_reason": "#41306 is the more narrowly scoped and actionable issue, while #41826 is a separate enhancement request for a different tiled component.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41306", + "right": "issue:41826", + "accept": false, + "reason": "Different components and different requested changes: FLCE/from_pretrained flag vs TiledMLP integration. They share a broad memory-efficiency theme only." + } + ] + }, + "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": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43598-2", + "nodes": [ + "issue:43598", + "issue:43600" + ], + "soft_pairs": [ + "issue:43598|issue:43600" + ], + "prepared_review_unit_hash": "6578ed3b63c377822c0b3b37d037b3f24dc3b3e0e072c7f0b75ce5815141eee5", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1583, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1583, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues are thematically related to Trainer custom trainers, but they describe different problems: one is about `remove_unused_column` customizability, the other about FLOPS estimation accounting. They should not be treated as duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43598 is the cleaner representative of the cluster theme because it is a concrete Trainer-behavior request, but it is not a duplicate of 43600.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43598", + "right": "issue:43600", + "accept": false, + "reason": "Related to Trainer extensibility, but they target different code paths and outcomes: column pruning customization vs FLOPS estimation for custom trainers." + } + ] + }, + "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": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43684-2", + "nodes": [ + "issue:43684", + "issue:44405" + ], + "soft_pairs": [ + "issue:43684|issue:44405" + ], + "prepared_review_unit_hash": "601b9829fb9ab71f3910c5afaf3a465f815642cc928a8aded997a9141adb270c", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1611, + "estimated_input_tokens": 403, + "estimated_eval_tokens": 1062 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1611, + "estimated_input_tokens": 403, + "estimated_eval_tokens": 1062 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are not duplicates: #43684 asks for Qwen3-Omni registration in AutoModel/AutoModelForConditionalGeneration, while #44405 requests AutoModelForSequenceClassification support for Qwen3.5 (Qwen3_5Config). They involve different model/config families and different auto-head behavior.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue for this pair; they target different model families and different feature surfaces.", + "canonical_pr_reason": null, + "best_issue_reason": "#43684 is the broader auto-registration request, but it still does not subsume #44405\u2019s sequence-classification support for Qwen3.5.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43684", + "right": "issue:44405", + "accept": false, + "reason": "Different underlying requests: Qwen3-Omni model registration vs Qwen3.5 sequence-classification support. Same broad ecosystem, but not the same bug/change and would not be merged into one issue." + } + ] + }, + "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": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": "79c3ee06538e7a2dd1fe5bcc84f0d114606746d0210fbdb4328f69f5206cb1d1", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4266, + "estimated_input_tokens": 1067, + "estimated_eval_tokens": 2390 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4266, + "estimated_input_tokens": 1067, + "estimated_eval_tokens": 2390 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Three issues share the exact same bare title \"huggingface \" and, with no other distinguishing evidence, look like duplicate reports; \"Fett\" appears unrelated. No PRs are present.", + "confidence": 0.9, + "canonical_issue_reason": "Issue 43845 is a reasonable canonical representative because it is one of the repeated identical-title reports and is the earliest of that cluster, with no evidence distinguishing it from the other two.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43845 is the best overall issue candidate since it anchors the duplicate cluster of identical \"huggingface \" reports and is the earliest among them.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43835", + "right": "issue:43845", + "accept": false, + "reason": "Different titles and no shared body, file, or target evidence; \"Fett\" does not look like the same issue as \"huggingface \"." + }, + { + "left": "issue:43835", + "right": "issue:43846", + "accept": false, + "reason": "Different titles and no shared body, file, or target evidence; \"Fett\" does not look like the same issue as \"huggingface \"." + }, + { + "left": "issue:43835", + "right": "issue:43859", + "accept": false, + "reason": "Different titles and no shared body, file, or target evidence; \"Fett\" does not look like the same issue as \"huggingface \"." + }, + { + "left": "issue:43845", + "right": "issue:43846", + "accept": true, + "reason": "Same exact title text and no distinguishing metadata or content evidence; these likely report the same issue." + }, + { + "left": "issue:43845", + "right": "issue:43859", + "accept": true, + "reason": "Same exact title text and no distinguishing metadata or content evidence; this is a plausible duplicate match." + }, + { + "left": "issue:43846", + "right": "issue:43859", + "accept": true, + "reason": "Same exact title text and no distinguishing metadata or content evidence; this is a plausible duplicate match." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The summary is mostly grounded, but two soft-edge verdict reasons overstate the evidence. For issue:43845 vs issue:43859, the packet only supports identical titles and lack of other distinguishing metadata; it does not support \u201cnear-identical metadata.\u201d For issue:43846 vs issue:43859, they are not in very close time proximity, so that wording is also too strong. The duplicate judgments may be reasonable as soft matches, but the rationale must stay more conservative." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 13, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": "287c6a7dafed58d00f051b211284112621300b6d48a75043a7f93f3b37f33fbd", + "component_budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 9, + "serialized_chars": 5785, + "estimated_input_tokens": 1447, + "estimated_eval_tokens": 3150 + }, + "budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 9, + "serialized_chars": 5785, + "estimated_input_tokens": 1447, + "estimated_eval_tokens": 3150 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "All items look like tiny closed spam/noise issues with no substantive body, links, or code context. There is no strong evidence that any pair refers to the same underlying bug, so no soft merges are justified.", + "confidence": 0.94, + "canonical_issue_reason": "No issue stands out as a true canonical duplicate target; the set is dominated by spammy or content-free issues, so there is no reliable representative bug report.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44418 is the least spam-like and has the only vaguely descriptive title, but it is still too empty and generic to serve as a strong canonical issue.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44371", + "right": "issue:44572", + "accept": false, + "reason": "Same placeholder title, but both are content-free closed issues with no evidence they describe the same underlying problem; could just be separate spam posts." + }, + { + "left": "issue:44371", + "right": "issue:44573", + "accept": false, + "reason": "Identical spam-style titles are not enough to merge; there is no body, linkage, or code context tying them to one bug." + }, + { + "left": "issue:44371", + "right": "issue:44418", + "accept": false, + "reason": "One is a spam placeholder and the other is a generic documentation suggestion; no concrete shared defect or change target." + }, + { + "left": "issue:44371", + "right": "issue:44937", + "accept": false, + "reason": "Completely different titles and no supporting details; no sign of a shared underlying issue." + }, + { + "left": "issue:44418", + "right": "issue:44572", + "accept": false, + "reason": "Generic documentation suggestion versus placeholder spam; insufficient evidence of the same report." + }, + { + "left": "issue:44418", + "right": "issue:44573", + "accept": false, + "reason": "Generic documentation suggestion versus placeholder spam; no common bug description or target." + }, + { + "left": "issue:44572", + "right": "issue:44573", + "accept": false, + "reason": "Both are spam placeholders, but that does not establish the same underlying issue; they look like separate low-effort submissions." + }, + { + "left": "issue:44572", + "right": "issue:44937", + "accept": false, + "reason": "Placeholder spam versus a specific-sounding title; no evidence they refer to the same defect." + }, + { + "left": "issue:44573", + "right": "issue:44937", + "accept": false, + "reason": "Placeholder spam versus a specific-sounding title; too little overlap to treat as duplicates." + } + ] + }, + "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": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45145-2", + "nodes": [ + "issue:45145", + "issue:45151" + ], + "soft_pairs": [ + "issue:45145|issue:45151" + ], + "prepared_review_unit_hash": "6c32a6a5efac5cbb34a8758cb5d55201428cb54bcf5b4db9758eb26b3f6d5de9", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1645, + "estimated_input_tokens": 412, + "estimated_eval_tokens": 1080 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1645, + "estimated_input_tokens": 412, + "estimated_eval_tokens": 1080 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues look like a probable duplicate pair: they have the same title, very similar body size/content, and no evidence of distinct scope. However, the evidence is limited to soft similarity only, so this should be treated as likely duplicate rather than definitive.", + "confidence": 0.8, + "canonical_issue_reason": "Issue #45145 is slightly earlier and already has an inbound reference, so it is the safest canonical representative for this likely duplicate issue pair.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #45145 is the best overall issue anchor because it appeared first and already has a reference pointing to it, making it the most stable deduplication target.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45145", + "right": "issue:45151", + "accept": true, + "reason": "Same title and highly similar long-form content, with no signs of different scope; this is a likely duplicate, though the evidence is only soft_similarity and not definitive." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The packet supports a likely duplicate pair, but the analyst overstates certainty: evidence is only soft_similarity with no diff/file/link corroboration, so calling them \"near-exact duplicates\" and assigning 0.99 confidence is too strong. Keep the conclusion conservative (probable duplicate, not definitive)." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 15, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45762-2", + "nodes": [ + "issue:45762", + "issue:45763" + ], + "soft_pairs": [ + "issue:45762|issue:45763" + ], + "prepared_review_unit_hash": "6b0913f086f89aa32485370b2e2d5e90219266124d3720f31b9eaabafed85853", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1477, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1477, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues do not appear to be duplicates; they are separate, unrelated issue reports identified only by different personal-name titles and dates.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45762", + "right": "issue:45763", + "accept": false, + "reason": "Different names and different dates; no evidence they describe the same underlying bug or request." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 17, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45764-2", + "nodes": [ + "issue:45764", + "issue:45765" + ], + "soft_pairs": [ + "issue:45764|issue:45765" + ], + "prepared_review_unit_hash": "77c92d7cf664e91d272e496099f9fc33a4120d08c81b510177a889503d4593fe", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1481, + "estimated_input_tokens": 371, + "estimated_eval_tokens": 998 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1481, + "estimated_input_tokens": 371, + "estimated_eval_tokens": 998 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues appear to be unrelated closed posts with only personal-name/date titles and no shared technical content, so this is not a duplicate cluster.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45764", + "right": "issue:45765", + "accept": false, + "reason": "Different personal-name/date titles with no body, links, or technical details; no evidence they describe the same underlying bug or request." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 18, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45810-2", + "nodes": [ + "issue:45810", + "issue:46295" + ], + "soft_pairs": [ + "issue:45810|issue:46295" + ], + "prepared_review_unit_hash": "13309a802f8475cd2acd343252158eb45ae9dd9984edcfd7327851d82f0e59a4", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1541, + "estimated_input_tokens": 386, + "estimated_eval_tokens": 1028 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1541, + "estimated_input_tokens": 386, + "estimated_eval_tokens": 1028 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are unrelated: one requests adding Qwen3_5ForTokenClassification as a value-model option, while the other requests FP8 training support. They should not be merged as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": "Issue #46295 is the broader, more general feature request and is the better representative of the cluster, though it is not a duplicate of #45810.", + "canonical_pr_reason": null, + "best_issue_reason": "#46295 is the stronger standalone issue because it describes a broader training capability request; #45810 is a model-specific API addition request.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45810", + "right": "issue:46295", + "accept": false, + "reason": "Different underlying changes: one is about adding a specific token-classification/value-model class, the other is about FP8 training support. They do not describe the same bug or feature." + } + ] + }, + "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": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": "58f7f2037af75f7ba89a00ab84c167a49456bcbe0578785778f8956a0e9a1bd6", + "component_budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 8, + "serialized_chars": 5728, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 + }, + "budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 8, + "serialized_chars": 5728, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two separate exact-duplicate issue pairs are present: 45870/45871 and 45872/45873. Issue 45869 is a different dataset offer and is not a duplicate of the example/service requests. There is no single canonical issue for the whole cluster.", + "confidence": 0.94, + "canonical_issue_reason": "No single canonical issue spans the whole cluster; the items split into two distinct duplicate pairs with different underlying requests.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45870 is the best representative of its duplicate pair because it is the earlier of the two identical \"Example: Financial time series prediction with DataForge\" reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45869", + "right": "issue:45870", + "accept": false, + "reason": "Different topics: a financial dataset bundle offer vs an example request; similar finance wording is not enough to make them the same issue." + }, + { + "left": "issue:45869", + "right": "issue:45871", + "accept": false, + "reason": "Different topics: a financial dataset bundle offer vs an example request; not the same underlying issue." + }, + { + "left": "issue:45870", + "right": "issue:45871", + "accept": true, + "reason": "Same title and effectively the same content; these are direct duplicates of the same example issue." + }, + { + "left": "issue:45870", + "right": "issue:45872", + "accept": false, + "reason": "Both mention financial time series/DataForge, but one is an example request and the other adds a scraping-service component; not the same concrete issue." + }, + { + "left": "issue:45870", + "right": "issue:45873", + "accept": false, + "reason": "Example request vs scraping-service request; related theme, but different underlying issue and likely separate work." + }, + { + "left": "issue:45871", + "right": "issue:45872", + "accept": false, + "reason": "Same reasoning as 45870/45872: related finance/DataForge theme, but different requested change." + }, + { + "left": "issue:45871", + "right": "issue:45873", + "accept": false, + "reason": "Same reasoning as 45870/45873: similar broad area, but not the same underlying issue." + }, + { + "left": "issue:45872", + "right": "issue:45873", + "accept": true, + "reason": "Same title and effectively the same content; these are direct duplicates of the same DataForge scraping-service issue." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is well grounded in the packet. The two exact-duplicate pairs are supported by identical titles/body sizes and close timestamps: 45870/45871 and 45872/45873. The decision to treat 45869 as separate is conservative and consistent with its distinct title/content. The soft-edge rejections between the example-request pair and the scraping-service pair are also appropriately cautious, and the conclusion that there is no single canonical issue for the whole cluster is supported." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 20, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46439-2", + "nodes": [ + "issue:46439", + "issue:46441" + ], + "soft_pairs": [ + "issue:46439|issue:46441" + ], + "prepared_review_unit_hash": "44d301303d7add82cb4dbe1531fa7bb1a675ed989c5d7079dd1d39f9ce24b6fa", + "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 point at static-cache initialization around linear attention, but they describe different observed failures (shape corruption vs. `KeyError: 'linear_attention'`). I would not merge them without clearer evidence they share the exact same root cause.", + "confidence": 0.78, + "canonical_issue_reason": "Issue 46439 is the more specific root-cause report: it names `early_initialization`, identifies linear-attention shape corruption, and includes the concrete `RuntimeError` behavior.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 46439 is the best single issue to keep because it is more diagnostic and likely broader than the model-specific crash report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46439", + "right": "issue:46441", + "accept": false, + "reason": "They both involve static cache and linear attention, but the failure modes differ materially (`wrong shapes`/`RuntimeError` vs `KeyError: 'linear_attention'`). That is not enough to treat them as 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": 21, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46551-2", + "nodes": [ + "issue:46551", + "issue:46552" + ], + "soft_pairs": [ + "issue:46551|issue:46552" + ], + "prepared_review_unit_hash": "7329e8f22fa8cbb2f5326ff7eb65d820a3e6dee1b5e43d0c25101eb071a0efd4", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1707, + "estimated_input_tokens": 427, + "estimated_eval_tokens": 1110 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1707, + "estimated_input_tokens": 427, + "estimated_eval_tokens": 1110 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are near-identical duplicates about the same Wav2Vec2CTCTokenizer v5 special-token bug; one is likely the cleaner canonical issue because it is open and already referenced elsewhere.", + "confidence": 0.99, + "canonical_issue_reason": "issue:46552 is the better canonical issue because it is open, has inbound references, and has the same title/body as issue:46551, making it the active tracking thread.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:46552 is the best issue to keep as the primary record for this bug since it is still open and appears to be the version others are already pointing to.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46551", + "right": "issue:46552", + "accept": true, + "reason": "Same title and same reported tokenizer bug; both describe the identical v5 word-delimiter special-token leakage into get_special_tokens_mask and convert_ids_to_tokens." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The conclusion is grounded in the packet: both issues have identical titles, the same body length, and soft-similarity evidence; 46552 is open and has inbound references while 46551 is closed, so choosing 46552 as the canonical issue is conservative. The soft-edge accept is also supported by the provided evidence. No overstatement rises to the level of requiring rejection." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 22, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46657-2", + "nodes": [ + "issue:46657", + "issue:46692" + ], + "soft_pairs": [ + "issue:46657|issue:46692" + ], + "prepared_review_unit_hash": "5106752f9772cece1b8e2381c301aa47c8828c1cd4ea66db8aaa57e071fe38c6", + "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 only loosely related by object-detection training, but they describe different failure modes: a tensor shape mismatch versus an out-of-memory condition after several epochs. They should not be merged as duplicates.", + "confidence": 0.89, + "canonical_issue_reason": "Issue #46657 is the better representative of the cluster because it is a concrete, broadly framed training failure report; #46692 appears to be a separate RT-DETR v2 memory-growth problem.", + "canonical_pr_reason": null, + "best_issue_reason": "#46657 is the most canonical issue here because it states a specific runtime error directly tied to training and is less specialized than the RT-DETR v2 OOM report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46657", + "right": "issue:46692", + "accept": false, + "reason": "The reports do not look like the same underlying bug: one is a tensor size mismatch during object detection training, while the other is an out-of-memory failure after some epochs in RT-DETR v2. Different symptoms and likely 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": 23, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46822-2", + "nodes": [ + "issue:46822", + "issue:46823" + ], + "soft_pairs": [ + "issue:46822|issue:46823" + ], + "prepared_review_unit_hash": "9c1658f34440ee95cffdc66a0c22dd074f0dfbfc6c33f5eda70a0b85337cf88d", + "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": "The two issues are not duplicates: one is about ZeRO-3 zero.Init failing to partition a composite model submodule and causing OOM during multi-GPU load, while the other is about 4-bit loading with device_map='auto' and max_memory incorrectly dispatching modules to CPU and raising ValueError.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46822", + "right": "issue:46823", + "accept": false, + "reason": "Different failure modes and code paths: ZeRO-3 partitioning/OOM during initialization vs 4-bit auto device mapping with max_memory causing CPU dispatch on a fitting model. They share only broad loading/dispatch terminology, 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": 24, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46906-7", + "nodes": [ + "issue:46906", + "issue:46940", + "issue:46948", + "issue:46969", + "issue:46985", + "issue:46998", + "issue:47037" + ], + "soft_pairs": [ + "issue:46906|issue:46940", + "issue:46906|issue:46948", + "issue:46906|issue:46969", + "issue:46906|issue:46985", + "issue:46906|issue:47037", + "issue:46940|issue:46948", + "issue:46940|issue:46969", + "issue:46940|issue:46985", + "issue:46940|issue:46998", + "issue:46948|issue:46969", + "issue:46948|issue:46985", + "issue:46948|issue:46998", + "issue:46969|issue:46985", + "issue:46969|issue:47037", + "issue:46969|issue:46998", + "issue:46985|issue:47037", + "issue:46985|issue:46998", + "issue:46998|issue:47037", + "issue:46948|issue:47037", + "issue:46940|issue:47037" + ], + "prepared_review_unit_hash": "ebc05b2bcfdbed493824eaf87a7cd44b1976f17753e4a2ced5f52aa7ff53894a", + "component_budget": { + "node_count": 7, + "item_count": 7, + "soft_pair_count": 20, + "serialized_chars": 11154, + "estimated_input_tokens": 2789, + "estimated_eval_tokens": 5834 + }, + "budget": { + "node_count": 7, + "item_count": 7, + "soft_pair_count": 20, + "serialized_chars": 11154, + "estimated_input_tokens": 2789, + "estimated_eval_tokens": 5834 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are daily \"integration failure triage\" issues for different dates, so they look like recurring triage reports rather than duplicates of one underlying bug. Use the earliest open issue as the representative thread; reject all soft-duplicate links.", + "confidence": 0.94, + "canonical_issue_reason": "issue:46906 is the earliest issue in the series and has the strongest discussion/inbound reference activity, making it the best representative for this recurring triage topic.", + "canonical_pr_reason": null, + "best_issue_reason": "As a cluster representative, issue:46906 is the clearest anchor because it starts the date-stamped sequence and appears to have the most engagement.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46906", + "right": "issue:46940", + "accept": false, + "reason": "Same template title, but different date-stamped triage entries; no evidence they report the same underlying failure." + }, + { + "left": "issue:46906", + "right": "issue:46948", + "accept": false, + "reason": "Recurring triage post on a different day; similarity is structural, not a shared bug report." + }, + { + "left": "issue:46906", + "right": "issue:46969", + "accept": false, + "reason": "Separate daily triage issue, not the same incident or code-path failure." + }, + { + "left": "issue:46906", + "right": "issue:46985", + "accept": false, + "reason": "Another date-specific triage entry; no concrete duplication signal beyond wording." + }, + { + "left": "issue:46906", + "right": "issue:47037", + "accept": false, + "reason": "Same series pattern, but a later day's triage note rather than the same issue." + }, + { + "left": "issue:46940", + "right": "issue:46948", + "accept": false, + "reason": "Consecutive daily triage reports; likely distinct snapshots of failures, not one bug thread." + }, + { + "left": "issue:46940", + "right": "issue:46969", + "accept": false, + "reason": "Different day/date indicates separate triage instances, not a duplicate issue." + }, + { + "left": "issue:46940", + "right": "issue:46985", + "accept": false, + "reason": "No concrete shared failure described; only the recurring triage wrapper matches." + }, + { + "left": "issue:46940", + "right": "issue:46998", + "accept": false, + "reason": "Distinct daily report, not a single underlying defect." + }, + { + "left": "issue:46940", + "right": "issue:47037", + "accept": false, + "reason": "Later triage entry with the same format, but not evidence of the same root cause." + }, + { + "left": "issue:46948", + "right": "issue:46969", + "accept": false, + "reason": "Separate date-stamped triage issues; similarity alone is insufficient to merge." + }, + { + "left": "issue:46948", + "right": "issue:46985", + "accept": false, + "reason": "Same broad triage series, but no shared concrete bug or fix target." + }, + { + "left": "issue:46948", + "right": "issue:46998", + "accept": false, + "reason": "Different daily triage report; no indication of a duplicate incident." + }, + { + "left": "issue:46969", + "right": "issue:46985", + "accept": false, + "reason": "Consecutive daily triage posts are recurring operational notes, not the same issue." + }, + { + "left": "issue:46969", + "right": "issue:47037", + "accept": false, + "reason": "Different day and no shared underlying failure details; reject as duplicate." + }, + { + "left": "issue:46969", + "right": "issue:46998", + "accept": false, + "reason": "Distinct triage snapshot, not one concrete bug report." + }, + { + "left": "issue:46985", + "right": "issue:47037", + "accept": false, + "reason": "Same recurring triage pattern, but no evidence they describe the same failure." + }, + { + "left": "issue:46985", + "right": "issue:46998", + "accept": false, + "reason": "Separate date-specific triage issue, not a duplicate." + }, + { + "left": "issue:46998", + "right": "issue:47037", + "accept": false, + "reason": "Later daily triage entry; title overlap is expected and not enough to merge." + }, + { + "left": "issue:46948", + "right": "issue:47037", + "accept": false, + "reason": "Recurring triage threads for different dates; no shared concrete bug evidence." + }, + { + "left": "issue:46940", + "right": "issue:47037", + "accept": false, + "reason": "Same template title, but these are distinct daily triage reports rather than duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 25, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46983-2", + "nodes": [ + "issue:46983", + "issue:46984" + ], + "soft_pairs": [ + "issue:46983|issue:46984" + ], + "prepared_review_unit_hash": "fdbfa86d003963b31aac085b7b008818d0f35f26a39a899c71eac94ab8ba749e", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1490, + "estimated_input_tokens": 373, + "estimated_eval_tokens": 1002 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1490, + "estimated_input_tokens": 373, + "estimated_eval_tokens": 1002 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both issues ask for RWKV7 model support, but the packet only provides identical titles and soft similarity. That is enough to group them as closely related, but not enough to confidently prove duplication or that #46983 is an earlier duplicate. #46984 is still the better live tracker because it is open.", + "confidence": 0.72, + "canonical_issue_reason": "#46984 is open and therefore the better active tracker for the RWKV7 support request. The evidence does not conclusively show duplicate status, so this is based on tracker suitability rather than a proven earlier/later duplicate relationship.", + "canonical_pr_reason": null, + "best_issue_reason": "#46984 is the best issue to keep as the canonical live discussion because it is open and matches the same feature request. The available evidence is too limited to justify stronger duplicate claims.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46983", + "right": "issue:46984", + "accept": false, + "reason": "The issues have the same title and look related, but the packet lacks body-level or code-level evidence to establish that they are the same underlying report. Soft similarity alone is not enough to accept a duplicate edge." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The canonical choice of the open issue is grounded, but the summary and soft-edge verdict overstate the evidence. The packet only shows same title, close timestamps, and soft similarity; it does not establish that #46983 is a duplicate or an earlier duplicate. Confidence 0.98 is also too strong for this evidence level." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 26, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-47007-2", + "nodes": [ + "issue:47007", + "issue:47015" + ], + "soft_pairs": [ + "issue:47007|issue:47015" + ], + "prepared_review_unit_hash": "bb1ea77f541ecd2c07cfafdbe808298e465cad3a247a1b4aa4f411975b6d76bc", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1626, + "estimated_input_tokens": 407, + "estimated_eval_tokens": 1070 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1626, + "estimated_input_tokens": 407, + "estimated_eval_tokens": 1070 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are only loosely similar: both discuss AI-driven vulnerability discovery, but one is about lessons from using advanced LLMs while the other is about a transformer-based cybersecurity discovery system. They read like related but distinct writeups, not duplicates.", + "confidence": 0.84, + "canonical_issue_reason": "Issue 47015 is the broader, more generic title and is the better representative of the shared AI-vulnerability-discovery theme.", + "canonical_pr_reason": null, + "best_issue_reason": "If a single representative is needed, 47015 is the closest fit because its wording is more general; 47007 is more specifically framed around lessons from advanced LLM usage.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:47007", + "right": "issue:47015", + "accept": false, + "reason": "The titles overlap on AI cybersecurity, but they describe different angles and likely different submissions: LLM lessons vs transformer-based system. Similar topic, 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": 1, + "component_count": 34, + "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-40546-42", + "nodes": [ + "pull_request:43451", + "pull_request:44517", + "pull_request:46196", + "pull_request:46266", + "pull_request:46417", + "pull_request:46658", + "pull_request:46707", + "pull_request:46836", + "pull_request:46895", + "pull_request:46903", + "pull_request:46907", + "pull_request:46908", + "pull_request:46912", + "pull_request:46920", + "pull_request:46926", + "pull_request:46927", + "pull_request:46928", + "pull_request:46933", + "pull_request:46941", + "pull_request:46942", + "pull_request:46949", + "pull_request:46950", + "pull_request:46951", + "pull_request:46954", + "pull_request:46957", + "pull_request:46958", + "pull_request:46961", + "pull_request:46965", + "pull_request:46968", + "pull_request:46970", + "pull_request:46971", + "pull_request:46974", + "pull_request:46986", + "pull_request:46995", + "pull_request:47003", + "pull_request:47006", + "pull_request:47019", + "pull_request:47025", + "pull_request:47034", + "pull_request:47035" + ], + "soft_pairs": [ + "pull_request:46895|pull_request:46958", + "pull_request:46903|pull_request:46958", + "pull_request:46950|pull_request:46951", + "pull_request:46927|pull_request:46950", + "pull_request:46950|pull_request:47006", + "pull_request:46927|pull_request:46951", + "pull_request:46951|pull_request:46971", + "pull_request:46949|pull_request:46971", + "pull_request:46927|pull_request:47006", + "pull_request:46951|pull_request:47006", + "pull_request:46971|pull_request:47006", + "pull_request:46941|pull_request:46970", + "pull_request:46908|pull_request:47006", + "pull_request:46908|pull_request:46928", + "pull_request:46907|pull_request:46926", + "pull_request:46970|pull_request:46971", + "pull_request:46926|pull_request:46928", + "pull_request:46927|pull_request:46986", + "pull_request:46941|pull_request:46986", + "pull_request:46926|pull_request:46941", + "pull_request:46933|pull_request:46942", + "pull_request:46928|pull_request:46941", + "pull_request:46907|pull_request:46928", + "pull_request:46928|pull_request:47006", + "pull_request:46941|pull_request:47006", + "pull_request:46907|pull_request:46908", + "pull_request:46908|pull_request:46951", + "pull_request:46907|pull_request:46941", + "pull_request:46926|pull_request:46927", + "pull_request:46954|pull_request:46961", + "pull_request:46954|pull_request:46968", + "pull_request:46954|pull_request:46958", + "pull_request:46958|pull_request:46961", + "pull_request:46961|pull_request:46974", + "pull_request:46958|pull_request:46974", + "pull_request:46954|pull_request:46971", + "pull_request:46920|pull_request:46968", + "pull_request:46971|pull_request:46974", + "pull_request:46958|pull_request:46968", + "pull_request:46968|pull_request:46974", + "pull_request:46920|pull_request:46961", + "pull_request:46961|pull_request:46968", + "pull_request:46961|pull_request:46971", + "pull_request:46949|pull_request:46968", + "pull_request:46266|pull_request:46957", + "pull_request:47034|pull_request:47035", + "pull_request:47019|pull_request:47025", + "pull_request:46957|pull_request:46965", + "pull_request:44517|pull_request:46417", + "pull_request:46933|pull_request:47035", + "pull_request:46266|pull_request:46707", + "pull_request:46196|pull_request:46995", + "pull_request:46707|pull_request:46961", + "pull_request:46968|pull_request:47019", + "pull_request:46707|pull_request:46968", + "pull_request:46707|pull_request:46957", + "pull_request:46912|pull_request:47025", + "pull_request:46196|pull_request:46965", + "pull_request:46961|pull_request:47019", + "pull_request:46707|pull_request:47019", + "pull_request:46266|pull_request:47034", + "pull_request:46836|pull_request:46961", + "pull_request:43451|pull_request:47019", + "pull_request:46836|pull_request:46965", + "pull_request:46957|pull_request:46961", + "pull_request:43451|pull_request:46266", + "pull_request:43451|pull_request:46995", + "pull_request:46266|pull_request:47003", + "pull_request:46961|pull_request:47035", + "pull_request:46836|pull_request:46995", + "pull_request:43451|pull_request:46933", + "pull_request:46658|pull_request:46961" + ], + "prepared_review_unit_hash": "9aa406b5b3e68b9acae4dfdc16d8edd0c081e44fc3fbf36c591c58d54a2a583a", + "component_budget": { + "node_count": 42, + "item_count": 42, + "soft_pair_count": 131, + "serialized_chars": 149662, + "estimated_input_tokens": 37416, + "estimated_eval_tokens": 75088 + }, + "budget": { + "node_count": 40, + "item_count": 40, + "soft_pair_count": 72, + "serialized_chars": 113799, + "estimated_input_tokens": 28450, + "estimated_eval_tokens": 57156 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Almost all candidate pairs are unrelated fixes across different models or docs. The only clear duplicate-style pair is the two CohereASR double-shift PRs, which address the same loss computation bug in the same code path.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46895 is the cleanest canonical choice: it directly targets issue 46894 and fixes the CohereASR training-loss double-shift in-place with a narrowly scoped code change.", + "best_issue_reason": null, + "best_pr_reason": "PR 46895 is the best representative of the underlying bug because it is explicitly issue-linked, narrowly scoped, and fixes the exact CohereASR loss path without broad unrelated edits.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46895", + "right": "pull_request:46958", + "accept": true, + "reason": "Same CohereASR training-loss double-shift bug in the same forward/loss path; one is just an alternative implementation." + }, + { + "left": "pull_request:46903", + "right": "pull_request:46958", + "accept": false, + "reason": "46903 also changes PPFormulaNet, so it is broader than the CohereASR fix and not the same concrete change." + }, + { + "left": "pull_request:46950", + "right": "pull_request:46951", + "accept": false, + "reason": "Different problems: Mamba2 OOM/runtime behavior vs PEFT loading integration." + }, + { + "left": "pull_request:46927", + "right": "pull_request:46950", + "accept": false, + "reason": "Unrelated model bugs: GLM46V input dtype handling vs Mamba2 OOM." + }, + { + "left": "pull_request:46950", + "right": "pull_request:47006", + "accept": false, + "reason": "Different code paths and models; Mamba2 OOM fix is not the same as Gemma3 generation cache logic." + }, + { + "left": "pull_request:46927", + "right": "pull_request:46951", + "accept": false, + "reason": "GLM46V dtype/input handling is unrelated to PEFT integration loading." + }, + { + "left": "pull_request:46951", + "right": "pull_request:46971", + "accept": false, + "reason": "PEFT integration load fix and Mamba2 test argument change are different issues." + }, + { + "left": "pull_request:46949", + "right": "pull_request:46971", + "accept": false, + "reason": "Whisper cache update bug and Mamba2 test loading change are unrelated." + }, + { + "left": "pull_request:46927", + "right": "pull_request:47006", + "accept": false, + "reason": "Different models and failures; no shared concrete bug." + }, + { + "left": "pull_request:46951", + "right": "pull_request:47006", + "accept": false, + "reason": "PEFT loading change does not match Gemma3 cache-generation fix." + }, + { + "left": "pull_request:46971", + "right": "pull_request:47006", + "accept": false, + "reason": "Test-side dtype/load change is unrelated to Gemma3 generation cache behavior." + }, + { + "left": "pull_request:46941", + "right": "pull_request:46970", + "accept": false, + "reason": "Both are large test-fix style PRs, but they address different models and failure modes." + }, + { + "left": "pull_request:46908", + "right": "pull_request:47006", + "accept": false, + "reason": "DeepSeekV32 load/offload fix is unrelated to Gemma3 generation caching." + }, + { + "left": "pull_request:46908", + "right": "pull_request:46928", + "accept": false, + "reason": "DeepSeekV32 loading issue and BLIP-2 tokenizer list update are unrelated." + }, + { + "left": "pull_request:46907", + "right": "pull_request:46926", + "accept": false, + "reason": "EdgeTam import/config kwarg fix and MiniMax M3 VL load-offload test fix are unrelated." + }, + { + "left": "pull_request:46970", + "right": "pull_request:46971", + "accept": false, + "reason": "Musicgen Melody output mismatch and Mamba2 OOM are different bugs." + }, + { + "left": "pull_request:46926", + "right": "pull_request:46928", + "accept": false, + "reason": "Different targets: MiniMax M3 VL load error vs BLIP-2 tokenizer auto mapping." + }, + { + "left": "pull_request:46927", + "right": "pull_request:46986", + "accept": false, + "reason": "Different model families and failures: GLM46V dtype issue vs GLM OCR output mismatch." + }, + { + "left": "pull_request:46941", + "right": "pull_request:46986", + "accept": false, + "reason": "Broader dtype/quantization fixes versus GLM OCR output mismatch are not the same change." + }, + { + "left": "pull_request:46926", + "right": "pull_request:46941", + "accept": false, + "reason": "MiniMax M3 VL load_error and DAC-related output mismatch are unrelated." + }, + { + "left": "pull_request:46928", + "right": "pull_request:46941", + "accept": false, + "reason": "Tokenizer auto-list update and DAC output mismatch are unrelated." + }, + { + "left": "pull_request:46907", + "right": "pull_request:46908", + "accept": false, + "reason": "EdgeTam import_or_config failure and DeepSeekV32 load_error are different issues." + }, + { + "left": "pull_request:46908", + "right": "pull_request:46951", + "accept": false, + "reason": "DeepSeekV32 loading fix is unrelated to PEFT integration loading." + }, + { + "left": "pull_request:46907", + "right": "pull_request:46941", + "accept": false, + "reason": "EdgeTam kwarg cleanup and DAC output mismatch are unrelated." + }, + { + "left": "pull_request:46926", + "right": "pull_request:46927", + "accept": false, + "reason": "Different model-specific test/load issues with no shared code path." + }, + { + "left": "pull_request:46954", + "right": "pull_request:46961", + "accept": false, + "reason": "Pytest-v9 compatibility fix in GlmASR is unrelated to the Serge normalization workflow." + }, + { + "left": "pull_request:46954", + "right": "pull_request:46968", + "accept": false, + "reason": "GlmASR fp32-module tweak and docs autolink cleanup are unrelated." + }, + { + "left": "pull_request:46954", + "right": "pull_request:46958", + "accept": false, + "reason": "Different models and different bugs; no shared concrete fix." + }, + { + "left": "pull_request:46958", + "right": "pull_request:46961", + "accept": false, + "reason": "CohereASR loss fix and workflow automation are unrelated." + }, + { + "left": "pull_request:46961", + "right": "pull_request:46974", + "accept": false, + "reason": "Workflow cleanup and Musicgen generate typo are unrelated." + }, + { + "left": "pull_request:46958", + "right": "pull_request:46974", + "accept": false, + "reason": "CohereASR training loss and Musicgen cache-length typo are different bugs." + }, + { + "left": "pull_request:46920", + "right": "pull_request:46968", + "accept": false, + "reason": "Voxtral realtime API adaptation and docs autolink cleanup are unrelated." + }, + { + "left": "pull_request:46961", + "right": "pull_request:46968", + "accept": false, + "reason": "Workflow automation and docs link cleanup are not the same change." + }, + { + "left": "pull_request:46961", + "right": "pull_request:46971", + "accept": false, + "reason": "Workflow cleanup is unrelated to Mamba2 test loading changes." + }, + { + "left": "pull_request:46949", + "right": "pull_request:46968", + "accept": false, + "reason": "Whisper cache bug and docs autolink cleanup are unrelated." + }, + { + "left": "pull_request:46266", + "right": "pull_request:46957", + "accept": false, + "reason": "DINO modernization and finegrained FP8 generation changes are unrelated." + }, + { + "left": "pull_request:47034", + "right": "pull_request:47035", + "accept": false, + "reason": "47034 is a broader Neuron SFT experiment; 47035 is a separate training_args backend default tweak, not the same concrete change." + }, + { + "left": "pull_request:47019", + "right": "pull_request:47025", + "accept": false, + "reason": "Masking-utils contract regression and Diffusion Gemma test skips are unrelated." + }, + { + "left": "pull_request:46957", + "right": "pull_request:46965", + "accept": false, + "reason": "Generation optimization / FP8 work and DeepSeek V4 use_cache support are different issues." + }, + { + "left": "pull_request:44517", + "right": "pull_request:46417", + "accept": false, + "reason": "Different model additions (Qwen3-TTS vs HunYuan VL); not the same artifact." + }, + { + "left": "pull_request:46933", + "right": "pull_request:47035", + "accept": false, + "reason": "Gemma4 quantized vision/audio casting fix is unrelated to Neuron backend defaults." + }, + { + "left": "pull_request:46266", + "right": "pull_request:47003", + "accept": false, + "reason": "DINO modernization and OpenVINO exporter work are unrelated." + }, + { + "left": "pull_request:46933", + "right": "pull_request:46942", + "accept": false, + "reason": "Both touch dtype casting, but on different models and code paths; too broad to treat as the same change." + }, + { + "left": "pull_request:46912", + "right": "pull_request:47025", + "accept": false, + "reason": "RoPE test coverage and Diffusion Gemma test fixes are different test-only changes." + }, + { + "left": "pull_request:46196", + "right": "pull_request:46995", + "accept": false, + "reason": "Wav2Vec2 SDPA dispatch and flash-linear-attention MLU support are unrelated backend changes." + }, + { + "left": "pull_request:46707", + "right": "pull_request:46961", + "accept": false, + "reason": "FSDP2 migration and workflow cleanup do not address the same problem." + }, + { + "left": "pull_request:46968", + "right": "pull_request:47019", + "accept": false, + "reason": "Docs autolink cleanup and masking_utils regression fix are unrelated." + }, + { + "left": "pull_request:46707", + "right": "pull_request:46968", + "accept": false, + "reason": "FSDP2 work and docs autolink cleanup are unrelated." + }, + { + "left": "pull_request:46707", + "right": "pull_request:46957", + "accept": false, + "reason": "FSDP2 migration and finegrained FP8 generation optimization are unrelated." + }, + { + "left": "pull_request:46912", + "right": "pull_request:46942", + "accept": false, + "reason": "RoPE test coverage and TimesFM 2.5 quantized cast fix are unrelated." + }, + { + "left": "pull_request:46196", + "right": "pull_request:46965", + "accept": false, + "reason": "Wav2Vec2 SDPA dispatch and DeepSeek V4 use_cache support are unrelated." + }, + { + "left": "pull_request:46836", + "right": "pull_request:46961", + "accept": false, + "reason": "Unlimited OCR model addition and workflow cleanup are unrelated." + }, + { + "left": "pull_request:43451", + "right": "pull_request:47019", + "accept": false, + "reason": "Molmo2 model addition and masking-utils regression fix are unrelated." + }, + { + "left": "pull_request:46836", + "right": "pull_request:46965", + "accept": false, + "reason": "Unlimited OCR model addition and DeepSeek V4 cache behavior are unrelated." + }, + { + "left": "pull_request:46957", + "right": "pull_request:46961", + "accept": false, + "reason": "Finegrained FP8 work and workflow normalization are unrelated." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46266", + "accept": false, + "reason": "Molmo2 model addition and DINO modernization are unrelated." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46995", + "accept": false, + "reason": "Molmo2 model addition and flash-linear-attention availability are unrelated." + }, + { + "left": "pull_request:46266", + "right": "pull_request:47003", + "accept": false, + "reason": "DINO modernization and OpenVINO exporter work are unrelated." + }, + { + "left": "pull_request:46961", + "right": "pull_request:46995", + "accept": false, + "reason": "Workflow automation and flash-linear-attention availability are unrelated." + }, + { + "left": "pull_request:46968", + "right": "pull_request:47025", + "accept": false, + "reason": "Docs autolink cleanup and Diffusion Gemma test changes are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: 46895 is explicitly issue-linked and the CohereASR pair is the only clear same-bug duplicate-style match. The other pairwise rejects are conservative and supported by the titles/diffs, with no obvious overstatement." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 1, + "component_count": 34, + "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-40546-42", + "nodes": [ + "pull_request:40546", + "pull_request:43451", + "pull_request:44517", + "pull_request:45694", + "pull_request:46196", + "pull_request:46417", + "pull_request:46658", + "pull_request:46707", + "pull_request:46836", + "pull_request:46907", + "pull_request:46908", + "pull_request:46912", + "pull_request:46926", + "pull_request:46927", + "pull_request:46949", + "pull_request:46950", + "pull_request:46951", + "pull_request:46957", + "pull_request:46961", + "pull_request:46965", + "pull_request:46968", + "pull_request:46970", + "pull_request:46971", + "pull_request:46986", + "pull_request:46995", + "pull_request:47003", + "pull_request:47006", + "pull_request:47025", + "pull_request:47034", + "pull_request:47035" + ], + "soft_pairs": [ + "pull_request:46961|pull_request:46965", + "pull_request:46658|pull_request:46965", + "pull_request:46196|pull_request:47025", + "pull_request:43451|pull_request:44517", + "pull_request:46707|pull_request:47025", + "pull_request:45694|pull_request:47035", + "pull_request:46995|pull_request:47003", + "pull_request:44517|pull_request:46836", + "pull_request:44517|pull_request:47003", + "pull_request:45694|pull_request:46707", + "pull_request:43451|pull_request:46957", + "pull_request:43451|pull_request:47025", + "pull_request:43451|pull_request:46417", + "pull_request:46196|pull_request:46961", + "pull_request:43451|pull_request:46961", + "pull_request:46961|pull_request:47003", + "pull_request:46961|pull_request:47034", + "pull_request:46912|pull_request:46961", + "pull_request:46417|pull_request:46836", + "pull_request:43451|pull_request:45694", + "pull_request:46417|pull_request:47003", + "pull_request:45694|pull_request:47003", + "pull_request:46417|pull_request:46965", + "pull_request:46196|pull_request:46836", + "pull_request:45694|pull_request:46836", + "pull_request:46912|pull_request:47034", + "pull_request:43451|pull_request:46968", + "pull_request:46968|pull_request:47003", + "pull_request:46950|pull_request:46971", + "pull_request:46949|pull_request:46950", + "pull_request:46949|pull_request:46951", + "pull_request:46927|pull_request:46949", + "pull_request:46907|pull_request:46927", + "pull_request:46927|pull_request:46971", + "pull_request:46908|pull_request:46926", + "pull_request:46986|pull_request:47006", + "pull_request:46907|pull_request:46949", + "pull_request:46970|pull_request:47006", + "pull_request:46970|pull_request:46986", + "pull_request:46907|pull_request:47006", + "pull_request:46907|pull_request:46970", + "pull_request:46907|pull_request:46986", + "pull_request:46707|pull_request:47034", + "pull_request:46658|pull_request:46836", + "pull_request:43451|pull_request:46836", + "pull_request:46836|pull_request:47003", + "pull_request:40546|pull_request:43451", + "pull_request:43451|pull_request:46658", + "pull_request:46658|pull_request:47003", + "pull_request:43451|pull_request:46196", + "pull_request:43451|pull_request:47003", + "pull_request:43451|pull_request:47034", + "pull_request:43451|pull_request:47035", + "pull_request:43451|pull_request:46912", + "pull_request:46196|pull_request:47003", + "pull_request:47003|pull_request:47034", + "pull_request:47003|pull_request:47035", + "pull_request:46912|pull_request:47003", + "pull_request:46836|pull_request:46912" + ], + "prepared_review_unit_hash": "a83a01dadfa39eae035985f1ff4394bac675ca6d24c87588d946357eed79d86c", + "component_budget": { + "node_count": 42, + "item_count": 42, + "soft_pair_count": 131, + "serialized_chars": 149662, + "estimated_input_tokens": 37416, + "estimated_eval_tokens": 75088 + }, + "budget": { + "node_count": 30, + "item_count": 30, + "soft_pair_count": 59, + "serialized_chars": 89539, + "estimated_input_tokens": 22385, + "estimated_eval_tokens": 45026 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a mix of unrelated PRs: new model integrations, isolated bug fixes, docs-only edits, and automation/workflow changes. The soft similarities are driven by shared repo scaffolding and recurring filenames, not the same underlying bug or change.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 43451 is the most central concrete change in the set: a full model addition with substantial code, docs, auto-mapping, and tests. It has the strongest structural overlap with other model-integration PRs, so it is the best anchor for this cluster.", + "best_issue_reason": null, + "best_pr_reason": "43451 is the best overall PR anchor because it is substantial, self-contained, and representative of the shared model-registration scaffolding seen across the cluster. The other PRs are narrower fixes, drafts, or unrelated subsystem changes.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46961", + "right": "pull_request:46965", + "accept": false, + "reason": "Automation/workflow cleanup vs a DeepSeek V4 cache/use_cache fix; unrelated change targets." + }, + { + "left": "pull_request:46658", + "right": "pull_request:46965", + "accept": false, + "reason": "Step 3.7 Flash model work vs DeepSeek V4 behavior fix; different models and code paths." + }, + { + "left": "pull_request:46196", + "right": "pull_request:47025", + "accept": false, + "reason": "wav2vec2 sdpa dispatch support vs diffusion gemma test adjustments; unrelated subsystems." + }, + { + "left": "pull_request:43451", + "right": "pull_request:44517", + "accept": false, + "reason": "Both add models, but Molmo2 and Qwen3-TTS are different integrations with different code paths." + }, + { + "left": "pull_request:46707", + "right": "pull_request:47025", + "accept": false, + "reason": "Native FSDP2 migration vs diffusion gemma test cleanup; no shared bug or change." + }, + { + "left": "pull_request:45694", + "right": "pull_request:47035", + "accept": false, + "reason": "Batch-size split_batches fix vs torch-compile-neuron backend default; different training arguments." + }, + { + "left": "pull_request:46995", + "right": "pull_request:47003", + "accept": false, + "reason": "MLU flash-linear-attention availability vs OpenVINO exporter work; different features entirely." + }, + { + "left": "pull_request:44517", + "right": "pull_request:46836", + "accept": false, + "reason": "Qwen3-TTS and Unlimited OCR are separate model additions." + }, + { + "left": "pull_request:44517", + "right": "pull_request:47003", + "accept": false, + "reason": "Qwen3-TTS model integration vs OpenVINO exporter platform work; unrelated." + }, + { + "left": "pull_request:45694", + "right": "pull_request:46707", + "accept": false, + "reason": "TrainingArguments batch-size semantics vs FSDP2 module migration; unrelated code paths." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46957", + "accept": false, + "reason": "Molmo2 model addition vs finegrained FP8/generation optimization work; different concerns." + }, + { + "left": "pull_request:43451", + "right": "pull_request:47025", + "accept": false, + "reason": "Molmo2 model addition vs diffusion gemma test fixes; unrelated models." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46417", + "accept": false, + "reason": "Molmo2 and HunYuan VL are distinct model integrations, not duplicates." + }, + { + "left": "pull_request:46196", + "right": "pull_request:46961", + "accept": false, + "reason": "wav2vec2 sdpa support vs branch-normalization workflow; unrelated." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46961", + "accept": false, + "reason": "Model addition vs CI workflow automation; no shared underlying issue." + }, + { + "left": "pull_request:46961", + "right": "pull_request:47003", + "accept": false, + "reason": "Workflow normalization PR vs OpenVINO exporter PR; completely different scopes." + }, + { + "left": "pull_request:46961", + "right": "pull_request:47034", + "accept": false, + "reason": "Workflow automation vs Neuron SFT experiment files; unrelated." + }, + { + "left": "pull_request:46912", + "right": "pull_request:46961", + "accept": false, + "reason": "RoPE test improvements vs workflow automation; unrelated." + }, + { + "left": "pull_request:46417", + "right": "pull_request:46836", + "accept": false, + "reason": "HunYuan VL and Unlimited OCR are separate model additions." + }, + { + "left": "pull_request:43451", + "right": "pull_request:45694", + "accept": false, + "reason": "Molmo2 model integration vs TrainingArguments batch-size fix; unrelated." + }, + { + "left": "pull_request:46417", + "right": "pull_request:47003", + "accept": false, + "reason": "HunYuan VL model integration vs OpenVINO exporter work; different domains." + }, + { + "left": "pull_request:45694", + "right": "pull_request:47003", + "accept": false, + "reason": "TrainingArguments split_batches fix vs OpenVINO exporter changes; unrelated." + }, + { + "left": "pull_request:46417", + "right": "pull_request:46965", + "accept": false, + "reason": "HunYuan VL model addition vs DeepSeek V4 cache/use_cache fix; unrelated." + }, + { + "left": "pull_request:46196", + "right": "pull_request:46836", + "accept": false, + "reason": "wav2vec2 sdpa dispatch vs Unlimited OCR model addition; different targets." + }, + { + "left": "pull_request:45694", + "right": "pull_request:46836", + "accept": false, + "reason": "TrainingArguments fix vs Unlimited OCR model addition; no duplicate underlying change." + }, + { + "left": "pull_request:46912", + "right": "pull_request:47034", + "accept": false, + "reason": "RoPE test coverage vs Neuron SFT experiment; unrelated." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46968", + "accept": false, + "reason": "Molmo2 model addition vs docs autolink cleanup; unrelated." + }, + { + "left": "pull_request:46968", + "right": "pull_request:47003", + "accept": false, + "reason": "Documentation autolink fix vs exporter implementation; unrelated." + }, + { + "left": "pull_request:46950", + "right": "pull_request:46971", + "accept": false, + "reason": "Both mention Mamba2 OOM, but one changes model code and the other only adjusts tests/load dtype; not the same concrete fix." + }, + { + "left": "pull_request:46949", + "right": "pull_request:46950", + "accept": false, + "reason": "Whisper/audioflamingo cache-state fix vs Mamba2 OOM work; unrelated models." + }, + { + "left": "pull_request:46949", + "right": "pull_request:46951", + "accept": false, + "reason": "Cache initialization fix vs PEFT loading mismatch fix; unrelated." + }, + { + "left": "pull_request:46927", + "right": "pull_request:46949", + "accept": false, + "reason": "GLM46V input handling vs Whisper/audioflamingo cache fix; unrelated models." + }, + { + "left": "pull_request:46907", + "right": "pull_request:46927", + "accept": false, + "reason": "EdgeTam processor-kwarg fix vs GLM46V input casting fix; different models and bugs." + }, + { + "left": "pull_request:46927", + "right": "pull_request:46971", + "accept": false, + "reason": "GLM46V input casting vs Mamba2 OOM test fix; unrelated." + }, + { + "left": "pull_request:46908", + "right": "pull_request:46926", + "accept": false, + "reason": "DeepSeekV3.2 load-error test fix vs MiniMax-M3-VL load-error test fix; different models." + }, + { + "left": "pull_request:46986", + "right": "pull_request:47006", + "accept": false, + "reason": "GLM OCR output mismatch fix vs Gemma3 generation-config change; unrelated." + }, + { + "left": "pull_request:46970", + "right": "pull_request:47006", + "accept": false, + "reason": "MusicGen Melody test failures vs Gemma3 generation behavior; different code paths." + }, + { + "left": "pull_request:46970", + "right": "pull_request:46986", + "accept": false, + "reason": "MusicGen Melody and GLM OCR are separate model test fixes." + }, + { + "left": "pull_request:46907", + "right": "pull_request:46949", + "accept": false, + "reason": "EdgeTam processor kwarg fix vs Whisper/audioflamingo cache fix; unrelated." + }, + { + "left": "pull_request:46907", + "right": "pull_request:47006", + "accept": false, + "reason": "EdgeTam bug fix vs Gemma3 generation-utils fix; unrelated." + }, + { + "left": "pull_request:46907", + "right": "pull_request:46970", + "accept": false, + "reason": "EdgeTam and MusicGen Melody target different model families and bugs." + }, + { + "left": "pull_request:46907", + "right": "pull_request:46986", + "accept": false, + "reason": "EdgeTam processor fix vs GLM OCR mismatch fix; unrelated." + }, + { + "left": "pull_request:46707", + "right": "pull_request:47034", + "accept": false, + "reason": "Native FSDP2 migration vs Neuron SFT experiment; unrelated." + }, + { + "left": "pull_request:46658", + "right": "pull_request:46836", + "accept": false, + "reason": "Step 3.7 Flash model implementation vs Unlimited OCR model addition; different models." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46836", + "accept": false, + "reason": "Molmo2 and Unlimited OCR are separate model integrations." + }, + { + "left": "pull_request:40546", + "right": "pull_request:43451", + "accept": false, + "reason": "VibeVoice and Molmo2 are unrelated model additions." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46658", + "accept": false, + "reason": "Molmo2 and Step 3.7 Flash are distinct model additions." + }, + { + "left": "pull_request:46658", + "right": "pull_request:47003", + "accept": false, + "reason": "Step 3.7 Flash model work vs OpenVINO exporter infrastructure; unrelated." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46196", + "accept": false, + "reason": "Molmo2 model addition vs wav2vec2 sdpa dispatch; unrelated." + }, + { + "left": "pull_request:43451", + "right": "pull_request:47003", + "accept": false, + "reason": "Molmo2 model addition vs OpenVINO exporter changes; unrelated." + }, + { + "left": "pull_request:43451", + "right": "pull_request:47034", + "accept": false, + "reason": "Molmo2 model addition vs Neuron training experiment; unrelated." + }, + { + "left": "pull_request:43451", + "right": "pull_request:47035", + "accept": false, + "reason": "Molmo2 model addition vs torch-compile backend default; unrelated." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46912", + "accept": false, + "reason": "Molmo2 model addition vs RoPE test improvements; unrelated." + }, + { + "left": "pull_request:46196", + "right": "pull_request:47003", + "accept": false, + "reason": "wav2vec2 sdpa support vs OpenVINO exporter work; unrelated." + }, + { + "left": "pull_request:47003", + "right": "pull_request:47034", + "accept": false, + "reason": "OpenVINO exporter project vs Neuron SFT experiment; unrelated." + }, + { + "left": "pull_request:47003", + "right": "pull_request:47035", + "accept": false, + "reason": "OpenVINO exporter work vs torch-compile neuron default; unrelated." + }, + { + "left": "pull_request:46912", + "right": "pull_request:47003", + "accept": false, + "reason": "RoPE tests vs OpenVINO exporter infrastructure; unrelated." + }, + { + "left": "pull_request:46836", + "right": "pull_request:46912", + "accept": false, + "reason": "Unlimited OCR model addition vs RoPE test improvements; 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": 2, + "component_count": 34, + "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": "69ba97c99602f06d1ea230cec72e2e712beb101c436ad0aa9eba0ec5633789c3", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4675, + "estimated_input_tokens": 1169, + "estimated_eval_tokens": 2594 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4675, + "estimated_input_tokens": 1169, + "estimated_eval_tokens": 2594 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two PRs are unrelated feature additions in different model families: TimesFM covariate forecasting vs. DINOv3 image classification support. They should not be clustered as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #41224 is the better representative PR in this cluster because it is a more focused, concrete model feature addition with clear code and test changes for a single capability.", + "best_issue_reason": null, + "best_pr_reason": "PR #41224 is the strongest standalone PR here: it introduces a specific new classification path for DINOv3 with supporting implementation and tests, making it the clearest representative of a distinct change.", + "soft_edge_verdicts": [ + { + "left": "pull_request:40755", + "right": "pull_request:41224", + "accept": false, + "reason": "Different models, different tasks, and different code paths; this is not the same underlying bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 3, + "component_count": 34, + "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": 34, + "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-41855-13", + "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" + ], + "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:44827", + "pull_request:43424|pull_request:43823", + "pull_request:43823|pull_request:45655", + "pull_request:43823|pull_request:43995", + "pull_request:43995|pull_request:45213", + "pull_request:43995|pull_request:44827", + "pull_request:44827|pull_request:45170", + "pull_request:44827|pull_request:46267", + "pull_request:44827|pull_request:46048", + "pull_request:44827|pull_request:45655", + "pull_request:44827|pull_request:45213", + "pull_request:43823|pull_request:45213", + "pull_request:43251|pull_request:43254" + ], + "prepared_review_unit_hash": "97afea65aff25649f61c87f1679e40de5d4f39b3f460558d6f418f9441b741f0", + "component_budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 18, + "serialized_chars": 34782, + "estimated_input_tokens": 8696, + "estimated_eval_tokens": 17648 + }, + "budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 18, + "serialized_chars": 34782, + "estimated_input_tokens": 8696, + "estimated_eval_tokens": 17648 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous. The only clear duplicate pair is the two `fixed_cross_entropy` PRs (#43251 and #43254), which both add extra kwargs to the same loss call. The other pairs are just broad topical similarity (models, auto-mapping, tests, docs) and do not look like the same concrete fix.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #43251 is the better canonical representative of the only true duplicate pair: it is the more explicit implementation of the cross-entropy kwargs fix, has an issue target, and more discussion/review context. #43254 is a narrower near-copy of the same change.", + "best_issue_reason": null, + "best_pr_reason": "PR #43251 is the best global representative because it addresses a concrete bug report, has the clearer issue linkage, and is the more fully reviewed version of the shared loss-function fix.", + "soft_edge_verdicts": [ + { + "left": "pull_request:41855", + "right": "pull_request:43823", + "accept": false, + "reason": "Unrelated PRs: Mistral tokenizer method additions vs adding a new MobileLLM model. Same general repo area, different changes." + }, + { + "left": "pull_request:41855", + "right": "pull_request:43424", + "accept": false, + "reason": "Different code paths: tokenizer API additions vs ExecuTorch dynamic-shape export test." + }, + { + "left": "pull_request:42668", + "right": "pull_request:43823", + "accept": false, + "reason": "Both are model/processor related, but one is auto/processor robustness and the other is a new MobileLLM model. Not the same fix." + }, + { + "left": "pull_request:43251", + "right": "pull_request:43823", + "accept": false, + "reason": "Cross-entropy kwargs fix vs adding a new model; no shared underlying bug or change." + }, + { + "left": "pull_request:43251", + "right": "pull_request:44827", + "accept": false, + "reason": "Loss-function kwargs change and Mistral4 test/model fixes are distinct changes with no concrete overlap." + }, + { + "left": "pull_request:43424", + "right": "pull_request:44827", + "accept": false, + "reason": "ExecuTorch export test and Mistral4 fixes are separate test concerns, not one mergeable change." + }, + { + "left": "pull_request:43424", + "right": "pull_request:43823", + "accept": false, + "reason": "Dynamic-shape export test vs new MobileLLM model; different features and failure modes." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45655", + "accept": false, + "reason": "Adding MobileLLM and fixing `cls.config` resolution are unrelated model-creation vs core metaclass behavior changes." + }, + { + "left": "pull_request:43823", + "right": "pull_request:43995", + "accept": false, + "reason": "New model addition vs Falcon output-collection refactor; same broad area but not the same concrete change." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45213", + "accept": false, + "reason": "New model addition vs an agent skill/docs/model-creation scaffolding PR; not the same underlying bug or code path." + }, + { + "left": "pull_request:43995", + "right": "pull_request:45213", + "accept": false, + "reason": "Falcon forward/refactor changes are unrelated to the model-creation skill PR." + }, + { + "left": "pull_request:43995", + "right": "pull_request:44827", + "accept": false, + "reason": "Falcon output interface refactor and Mistral4 fixes are separate model-specific changes." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45170", + "accept": false, + "reason": "Mistral4 test fixes vs CLIP-family `layrnorm` rename/model conversion fix; only broad model-family similarity." + }, + { + "left": "pull_request:44827", + "right": "pull_request:46267", + "accept": false, + "reason": "Mistral4 fixes and chat response parsing updates are unrelated subsystems." + }, + { + "left": "pull_request:44827", + "right": "pull_request:46048", + "accept": false, + "reason": "Mistral4 fixes vs adding a TDT loss kernel; different features and code paths." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45655", + "accept": false, + "reason": "Mistral4 model fixes and `cls.config` resolution are unrelated core changes." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45213", + "accept": false, + "reason": "Mistral4 tests/fixes vs model-creation skill PR; no shared underlying bug." + }, + { + "left": "pull_request:43251", + "right": "pull_request:43254", + "accept": true, + "reason": "Same underlying bug: both modify `fixed_cross_entropy` to pass supported kwargs through to `nn.functional.cross_entropy`, and both target the same issue." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The only accepted duplicate pair is correctly supported by shared file, shared issue target, and nearly identical loss-function diffs. The other pair rejections are appropriately cautious and do not overstate evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 5, + "component_count": 34, + "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": "64d11cd6698e2b37fced03084e2bdc76cd33c1783974a8d1d8fd27bcda195ee8", + "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": "Both items are broad GitHub Actions upgrade PRs, but they target different actions and failure modes: one updates checkout/cache/upload-artifact for Node 24 compatibility, while the other updates docker/build-push-action and related workflow actions. They are related maintenance work, not the same underlying change.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 42975 is the better canonical representative because it is the broader workflow-actions compatibility upgrade and its title more clearly states the maintenance goal.", + "best_issue_reason": null, + "best_pr_reason": "PR 42975 is the best representative of the cluster overall since it covers many workflow action version bumps and is the more general GitHub Actions upgrade effort.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42975", + "right": "pull_request:42976", + "accept": false, + "reason": "Different concrete upgrades: 42975 focuses on checkout/cache/upload-artifact for Node 24 compatibility, while 42976 upgrades docker/build-push-action and other workflow actions. They touch overlapping workflows but not the same code-path problem." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 6, + "component_count": 34, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-42978-4", + "nodes": [ + "pull_request:42978", + "pull_request:43448", + "pull_request:45490", + "pull_request:46562" + ], + "soft_pairs": [ + "pull_request:42978|pull_request:43448", + "pull_request:42978|pull_request:45490", + "pull_request:43448|pull_request:45490", + "pull_request:45490|pull_request:46562" + ], + "prepared_review_unit_hash": "d60010f34e0e92464645af4870a0870d68865a6c4ae7b29a40fd572e49abe4be", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 4, + "serialized_chars": 13046, + "estimated_input_tokens": 3262, + "estimated_eval_tokens": 6780 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 4, + "serialized_chars": 13046, + "estimated_input_tokens": 3262, + "estimated_eval_tokens": 6780 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "All four pull requests are distinct model additions, not duplicates. The overlap is limited to shared repository boilerplate such as docs, auto registries, and test scaffolding; the actual model targets and code paths differ.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: these PRs add different models (ViT NEPA, Molmo, CTSM, Step3p7) and only overlap on shared infra files, not the same underlying change.", + "best_issue_reason": null, + "best_pr_reason": "No single PR is a good cluster representative for duplicate triage; 46562 is the most developed by activity, but it still addresses a different model than the others.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42978", + "right": "pull_request:43448", + "accept": false, + "reason": "ViT NEPA vs Molmo are unrelated model additions; shared auto/docs files are boilerplate, not the same fix." + }, + { + "left": "pull_request:42978", + "right": "pull_request:45490", + "accept": false, + "reason": "ViT NEPA and CTSM add different model implementations with different code paths and tests." + }, + { + "left": "pull_request:43448", + "right": "pull_request:45490", + "accept": false, + "reason": "Molmo and CTSM are separate model ports; the common files reflect registry/docs updates, not duplicate functionality." + }, + { + "left": "pull_request:45490", + "right": "pull_request:46562", + "accept": false, + "reason": "CTSM and Step3p7 are different model additions; despite some overlapping auto-mapping/docs changes, they do not fix the same concrete problem." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 7, + "component_count": 34, + "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-43291-7", + "nodes": [ + "pull_request:43291", + "pull_request:43665", + "pull_request:44815", + "pull_request:46380", + "pull_request:46539", + "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: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": "7a3a96288ec154605a843968f2dbc393e9582290a6b09f0d7d73fbed5f836ddf", + "component_budget": { + "node_count": 7, + "item_count": 7, + "soft_pair_count": 15, + "serialized_chars": 19601, + "estimated_input_tokens": 4901, + "estimated_eval_tokens": 10058 + }, + "budget": { + "node_count": 7, + "item_count": 7, + "soft_pair_count": 15, + "serialized_chars": 19601, + "estimated_input_tokens": 4901, + "estimated_eval_tokens": 10058 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Most items are unrelated PRs; the only strong duplicate-looking pair is the YOLOS device-mismatch fixes, which target the same issue and file. The rest differ by model, subsystem, or bug shape and should stay separate.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": "pull_request:46922 is the best representative of the duplicate YOLOS fix: it addresses the concrete runtime device mismatch in the forward path and also includes the supporting no-split-module change. pull_request:46886 is a narrower variant of the same fix.", + "best_issue_reason": null, + "best_pr_reason": "pull_request:46922 is the strongest overall PR in the cluster because it fixes the specific YOLOS failure mode directly and subsumes the related auto device-map adjustment, unlike the other PRs which are clearly unrelated.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43291", + "right": "pull_request:43665", + "accept": false, + "reason": "Both are test-only PRs, but they target different models and different failures (Whisper decoding/tests vs CLIP/ViT config tests)." + }, + { + "left": "pull_request:43291", + "right": "pull_request:44815", + "accept": false, + "reason": "Whisper tokenizer/test changes are unrelated to FP8 dequantization and model loading logic." + }, + { + "left": "pull_request:43291", + "right": "pull_request:46380", + "accept": false, + "reason": "Different subsystems entirely: Whisper decoding tests vs generation config validation and watermarking handling." + }, + { + "left": "pull_request:43291", + "right": "pull_request:46922", + "accept": false, + "reason": "Whisper test fixes and YOLOS device-mismatch fixes are unrelated model-specific problems." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46539", + "accept": false, + "reason": "Both touch tests, but one is CLIP/ViT test stubbing and the other is Phimoe CI/integration setup; not the same bug." + }, + { + "left": "pull_request:43665", + "right": "pull_request:44815", + "accept": false, + "reason": "CLIP/ViT test edits do not match the FP8 dequant/model-loading code path." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46380", + "accept": false, + "reason": "No shared concrete failure: model config tests vs generation watermarking validation." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46922", + "accept": false, + "reason": "Different model families and fixes; the similarity is only superficial test/model code touch." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46539", + "accept": false, + "reason": "FP8 dequantization/loading changes are unrelated to Phimoe CI runner memory/device setup." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46380", + "accept": false, + "reason": "Different code paths: quantization/dequant loading vs generation configuration validation." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46922", + "accept": false, + "reason": "FP8 quantization fixes and YOLOS device placement are separate bugs." + }, + { + "left": "pull_request:46380", + "right": "pull_request:46539", + "accept": false, + "reason": "Generation watermarking validation and Phimoe integration test infrastructure are unrelated." + }, + { + "left": "pull_request:46380", + "right": "pull_request:46922", + "accept": false, + "reason": "Different models and different runtime issues; no shared concrete fix." + }, + { + "left": "pull_request:46539", + "right": "pull_request:46922", + "accept": false, + "reason": "Phimoe CI/device-map test infrastructure is not the same underlying bug as YOLOS mid-position embedding placement." + }, + { + "left": "pull_request:46886", + "right": "pull_request:46922", + "accept": true, + "reason": "Same YOLOS device-mismatch issue on the same file and same explicit target (46869); both changes address the same concrete failure path and could plausibly be merged into one PR." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative on the soft edges. The only clearly duplicate-looking pair is the YOLOS fixes (46886/46922), which share the same file and target the same issue; the other pairwise rejections are well-supported by different models, subsystems, or bug shapes. No overstatement stands out." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 8, + "component_count": 34, + "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": "917daa183669b57e36a7f5ac3c2b228b2da406b1ed7fa002ef6292511d27191d", + "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": "These are three unrelated '[don't merge]' test/demo PRs. They all have superficial similarity, but each changes a different area of the repo, so none should be merged as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "pull_request:43532 is the broadest and most representative of the cluster\u2019s test/CI demo changes.", + "best_issue_reason": null, + "best_pr_reason": "It is only a weak placeholder canonical: the three PRs do not fix the same concrete problem or implement the same change.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43488", + "right": "pull_request:46747", + "accept": false, + "reason": "Different changes: #43488 is a setup/dependency/version bump with a Vit test tweak, while #46747 only changes the PR security-gate workflow. No shared code-path or mergeable common fix." + }, + { + "left": "pull_request:43488", + "right": "pull_request:43532", + "accept": false, + "reason": "#43488 is about packaging/version/dependency table changes; #43532 is about workflows, testing utilities, and a Vit test. The shared test filename is incidental, not the same underlying bug/change." + }, + { + "left": "pull_request:43532", + "right": "pull_request:46747", + "accept": false, + "reason": "#43532 spans workflow/test utility updates and a test file, but #46747 only edits the PR security-gate workflow. They are different CI changes with different intent and code paths." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 9, + "component_count": 34, + "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": "eece3935d47f91dd761df227e50132568bf0a121ea89472c872a554fd5f40f9e", + "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 not duplicates: both add new pipeline-related docs/code, but one is for Promptable Concept Segmentation and the other for Promptable Visual Segmentation, with different pipeline classes and different model/test integrations.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "pull_request:43613 is the more substantial standalone PR, with broader model/test coverage and a clearly defined visual-segmentation code path; pull_request:43612 is a separate concept-segmentation addition, not a duplicate.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43612", + "right": "pull_request:43613", + "accept": false, + "reason": "Different underlying changes: concept segmentation vs visual segmentation. The overlap is mostly boilerplate/docs/registration; they do not target the same concrete bug or code path, so they should not be merged as duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 10, + "component_count": 34, + "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": "26f377ca772e361c2ca5d3704d14ee2e8b39ae43441d2c49de38a71efeb7cc7b", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4293, + "estimated_input_tokens": 1074, + "estimated_eval_tokens": 2404 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4293, + "estimated_input_tokens": 1074, + "estimated_eval_tokens": 2404 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two PRs both touch `trainer.py` but fix unrelated Trainer extension points: one changes gradient-accumulation loss scaling, the other extracts signature-column filtering for PEFT/custom trainers. Shared file alone is not enough to treat them as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "No single PR is a duplicate/canonical representative of the other; they target different code paths and could be merged independently.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43651", + "right": "pull_request:43663", + "accept": false, + "reason": "Different behaviors: 43651 adjusts loss normalization during training, while 43663 refactors input-signature column filtering. Same file, but not the same bug/change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 11, + "component_count": 34, + "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": "f9a0502e8d58276da58810ee2965909a1e9881a6f1a037b1bba80a6da604bd6a", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 3663, + "estimated_input_tokens": 916, + "estimated_eval_tokens": 2088 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 3663, + "estimated_input_tokens": 916, + "estimated_eval_tokens": 2088 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both pull requests target the same runtime TypeAdapter annotation failure in src/transformers/cli/serve.py; 43842 is the more focused fix, while 43656 is a broader typing cleanup around the same code path.", + "confidence": 0.84, + "canonical_issue_reason": null, + "canonical_pr_reason": "Prefer PR 43842: it directly addresses the TypeAdapter NameError when pydantic is unavailable by moving the import under TYPE_CHECKING and quoting the annotation, making it the tighter fix for the reported runtime bug.", + "best_issue_reason": null, + "best_pr_reason": "PR 43842 is the best representative because it is the most explicit, minimal correction for the concrete failure mode; PR 43656 is related but broader and less directly scoped.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43656", + "right": "pull_request:43842", + "accept": true, + "reason": "Same underlying serve.py TypeAdapter NameError in the CLI, and the two fixes are compatible enough that they could plausibly be combined into one PR." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded overall: both PRs are in src/transformers/cli/serve.py and the packet supports treating them as alternative fixes for the same TypeAdapter NameError. The soft-pair verdict is conservative enough here because the changes are compatible and could be combined in one PR, even if 43656 is broader." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 12, + "component_count": 34, + "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": 34, + "review_unit_index": 1, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43996-28", + "nodes": [ + "pull_request:43996", + "pull_request:43997", + "pull_request:44007", + "pull_request:44010", + "pull_request:44013", + "pull_request:44017", + "pull_request:44018", + "pull_request:44019", + "pull_request:44044", + "pull_request:44056", + "pull_request:44059", + "pull_request:44066", + "pull_request:44068", + "pull_request:44071", + "pull_request:44072", + "pull_request:44073", + "pull_request:44074", + "pull_request:44076", + "pull_request:44085", + "pull_request:44086", + "pull_request:44098", + "pull_request:44101", + "pull_request:44116", + "pull_request:44129", + "pull_request:44154", + "pull_request:44161", + "pull_request:44722" + ], + "soft_pairs": [ + "pull_request:43996|pull_request:44085", + "pull_request:43996|pull_request:44044", + "pull_request:44066|pull_request:44085", + "pull_request:44007|pull_request:44072", + "pull_request:44072|pull_request:44722", + "pull_request:44066|pull_request:44072", + "pull_request:44013|pull_request:44044", + "pull_request:44066|pull_request:44086", + "pull_request:44066|pull_request:44071", + "pull_request:44018|pull_request:44068", + "pull_request:44066|pull_request:44068", + "pull_request:44018|pull_request:44066", + "pull_request:44019|pull_request:44722", + "pull_request:44086|pull_request:44722", + "pull_request:44019|pull_request:44085", + "pull_request:44018|pull_request:44019", + "pull_request:44019|pull_request:44071", + "pull_request:44068|pull_request:44722", + "pull_request:44018|pull_request:44116", + "pull_request:44068|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:44722", + "pull_request:44017|pull_request:44161", + "pull_request:44017|pull_request:44073", + "pull_request:44068|pull_request:44071", + "pull_request:44071|pull_request:44086", + "pull_request:43997|pull_request:44073", + "pull_request:44017|pull_request:44071", + "pull_request:44073|pull_request:44101", + "pull_request:44073|pull_request:44161", + "pull_request:44059|pull_request:44071", + "pull_request:44018|pull_request:44059", + "pull_request:44056|pull_request:44101", + "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:43997|pull_request:44071", + "pull_request:44010|pull_request:44071", + "pull_request:44018|pull_request:44129", + "pull_request:44010|pull_request:44073", + "pull_request:43997|pull_request:44074", + "pull_request:44010|pull_request:44018", + "pull_request:44071|pull_request:44129", + "pull_request:43997|pull_request:44129", + "pull_request:43996|pull_request:44161", + "pull_request:43996|pull_request:44017", + "pull_request:44010|pull_request:44044", + "pull_request:44044|pull_request:44074", + "pull_request:44066|pull_request:44722", + "pull_request:44085|pull_request:44722", + "pull_request:44085|pull_request:44129", + "pull_request:43996|pull_request:44007", + "pull_request:43996|pull_request:44129", + "pull_request:44044|pull_request:44129", + "pull_request:44044|pull_request:44154" + ], + "prepared_review_unit_hash": "20769dffad22993f340d151e961f32da8368a1395aa22cf77e56a4f1056f0385", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 78, + "serialized_chars": 84894, + "estimated_input_tokens": 21224, + "estimated_eval_tokens": 42704 + }, + "budget": { + "node_count": 27, + "item_count": 27, + "soft_pair_count": 72, + "serialized_chars": 75093, + "estimated_input_tokens": 18774, + "estimated_eval_tokens": 37804 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a set of parallel model-by-model refactors for standardized output tracing around issue 43979, not one single bug fix. The only clear duplicate-like overlaps are within the same model code paths, especially GPT-J/GPT-Neo variants.", + "confidence": 0.8, + "canonical_issue_reason": "43979 is the umbrella tracking issue for the standardized output-tracing refactor family, and many of the PRs explicitly target it.", + "canonical_pr_reason": "44722 is the strongest representative PR in the cluster: it is the latest GPT-J/CodeGen refactor, hits the same model code path as nearby GPT-J variants, and appears to be the most complete version of that subcluster.", + "best_issue_reason": "No issue artifact is present in the items, so 43979 is best treated as the organizing issue inferred from the PR targets.", + "best_pr_reason": "44722 is the best overall PR exemplar because it is later, issue-linked, and captures the same GPT-J output-tracing refactor as the overlapping GPT-J PRs while also covering CodeGen.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43996", + "right": "pull_request:44085", + "accept": false, + "reason": "Different models and code paths; sharing the same refactor theme and issue target is not enough." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44044", + "accept": false, + "reason": "Different model families (FNet/CVT vs DeBERTaV2); only the umbrella issue is shared." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44085", + "accept": true, + "reason": "Same GPT-J output-tracing code path in modeling_gptj.py; one PR is just a broader variant of the other." + }, + { + "left": "pull_request:44007", + "right": "pull_request:44072", + "accept": false, + "reason": "Different models; both are output-tracing refactors but not the same concrete change." + }, + { + "left": "pull_request:44072", + "right": "pull_request:44722", + "accept": false, + "reason": "EfficientNet and GPT-J/CodeGen are unrelated code paths." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44072", + "accept": false, + "reason": "Different model code paths; same refactor pattern only." + }, + { + "left": "pull_request:44013", + "right": "pull_request:44044", + "accept": false, + "reason": "MobileNetV2 and DeBERTaV2 are unrelated implementations." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44086", + "accept": false, + "reason": "Different models; no shared code path." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-J/CodeGen and MPT are unrelated model implementations." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44068", + "accept": true, + "reason": "Both are GPT-Neo output-tracing refactors on the same model file and are effectively overlapping revisions." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT-J and GPT-Neo are different model code paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44066", + "accept": false, + "reason": "GPT-Neo vs GPT-J; same family of refactor, but not the same underlying change." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44722", + "accept": false, + "reason": "ResNet and GPT-J/CodeGen are different code paths." + }, + { + "left": "pull_request:44086", + "right": "pull_request:44722", + "accept": false, + "reason": "MGP-STR and GPT-J/CodeGen are unrelated models." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44085", + "accept": false, + "reason": "ResNet and GPT-J are different concrete implementations." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44019", + "accept": false, + "reason": "GPT-Neo and ResNet are unrelated model code paths." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44071", + "accept": false, + "reason": "ResNet and MPT are unrelated; only the high-level refactor theme matches." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT-Neo and GPT-J/CodeGen are different model families." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo and Flaubert are different code paths." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo and Flaubert are unrelated models." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-J and MGP-STR are unrelated code paths." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-2 and MGP-STR are different models." + }, + { + "left": "pull_request:44098", + "right": "pull_request:44101", + "accept": false, + "reason": "ViLT and Flaubert/XLM are unrelated implementations." + }, + { + "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:44722", + "accept": false, + "reason": "Segformer and GPT-J/CodeGen are unrelated models." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44161", + "accept": false, + "reason": "Segformer and LongT5 are different code paths." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44073", + "accept": false, + "reason": "Segformer and VisualBert are unrelated models." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-Neo and MPT are unrelated code paths." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44076", + "accept": false, + "reason": "MPT and ImageGPT are different model implementations." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44101", + "accept": false, + "reason": "TextNet and Flaubert/XLM are unrelated models." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44017", + "accept": false, + "reason": "RegNet and Segformer are different code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44056", + "accept": false, + "reason": "RegNet and MPNet are unrelated implementations." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44129", + "accept": false, + "reason": "MPT and SpeechT5 are different models." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44161", + "accept": false, + "reason": "FNet/CVT and LongT5 are unrelated code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44017", + "accept": false, + "reason": "FNet/CVT and Segformer are different model families." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44071", + "accept": false, + "reason": "SqueezeBert and MPT are unrelated implementations." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44073", + "accept": false, + "reason": "SqueezeBert and VisualBert are different code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44018", + "accept": false, + "reason": "RegNet and GPT-Neo are unrelated models." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44161", + "accept": false, + "reason": "RegNet and LongT5 are different code paths." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44076", + "accept": false, + "reason": "MPT and ImageGPT are unrelated model implementations." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44101", + "accept": false, + "reason": "TextNet and Flaubert/XLM are different concrete changes." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44161", + "accept": false, + "reason": "FNet/CVT and LongT5 are unrelated code paths." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44071", + "accept": false, + "reason": "ResNet and MPT are different model families." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44101", + "accept": false, + "reason": "MPNet and Flaubert/XLM are unrelated code paths." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-2 and MPT are different models." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44101", + "accept": false, + "reason": "MPNet and Flaubert/XLM are unrelated implementations." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44017", + "accept": false, + "reason": "FNet/CVT and Segformer are different model code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44044", + "accept": false, + "reason": "SqueezeBert and DeBERTaV2 are unrelated models." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44129", + "accept": false, + "reason": "DeBERTaV2 and SpeechT5 are different code paths." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44154", + "accept": false, + "reason": "DeBERTaV2 and Vits are unrelated model implementations." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44722", + "accept": true, + "reason": "Same GPT-J/CodeGen output-tracing refactor family on the same underlying code path; 44722 is the broader later variant." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44722", + "accept": true, + "reason": "Both touch GPT-J output tracing in modeling_gptj.py; 44722 is the broader variant that subsumes the same change." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-J and SpeechT5 are unrelated models." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44007", + "accept": false, + "reason": "FNet/CVT and RegNet/ResNet/RT-DETR are different code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44129", + "accept": false, + "reason": "FNet/CVT and SpeechT5 are unrelated models." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44129", + "accept": false, + "reason": "DeBERTaV2 and SpeechT5 are different code paths." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44154", + "accept": false, + "reason": "DeBERTaV2 and Vits are unrelated model implementations." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary stays grounded in the packet: 43979 is clearly the shared umbrella target across many PRs, and 44722 is a reasonable representative because it is later, issue-linked, and spans both GPT-J and CodeGen. The soft-edge decisions are mostly conservative and the accepted pairs are plausibly mergeable as overlapping revisions of the same output-tracing bugfix within the same model code paths." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 13, + "component_count": 34, + "review_unit_index": 2, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43996-28", + "nodes": [ + "pull_request:43997", + "pull_request:44010", + "pull_request:44024", + "pull_request:44056", + "pull_request:44073", + "pull_request:44074", + "pull_request:44076", + "pull_request:44129" + ], + "soft_pairs": [ + "pull_request:44024|pull_request:44076", + "pull_request:44073|pull_request:44074", + "pull_request:44056|pull_request:44073", + "pull_request:44056|pull_request:44074", + "pull_request:43997|pull_request:44010", + "pull_request:44010|pull_request:44129" + ], + "prepared_review_unit_hash": "f3bddcf8f81f7b7e4553a86d6b906b62cc5546e91429cb580876d45675123a57", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 78, + "serialized_chars": 84894, + "estimated_input_tokens": 21224, + "estimated_eval_tokens": 42704 + }, + "budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 6, + "serialized_chars": 17122, + "estimated_input_tokens": 4281, + "estimated_eval_tokens": 8818 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "A cluster of model-specific PRs refactoring output tracing to standardized capture_outputs/can_return_tuple patterns. They are thematically similar, but each touches a different model and code path, so they are not duplicates of the same concrete change.", + "confidence": 0.84, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44129 is the strongest canonical representative: it is a focused output-tracing refactor, has an explicit issue target, and includes a substantial implementation change rather than only import/decorator plumbing.", + "best_issue_reason": null, + "best_pr_reason": "PR 44129 is the best overall fit because it is concrete, self-contained, and directly tied to a tracking issue; the other PRs are similar refactors but for different model implementations.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44024", + "right": "pull_request:44076", + "accept": false, + "reason": "Both are standardized output-tracing refactors, but for different model families (FocalNet vs ImageGPT) and different classes/forward paths; not the same underlying bug or mergeable into one PR." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44074", + "accept": false, + "reason": "VisualBert and TextNet are separate model implementations with different output contracts; this is a shared refactor pattern, not one concrete code-path fix." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44073", + "accept": false, + "reason": "MPNet and VisualBert touch different encoder stacks and tracing hooks; they are similar cleanup PRs, not duplicates of the same change." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44074", + "accept": false, + "reason": "MPNet vs TextNet is only a broad output-tracing modernization theme; the code paths and model-specific hook mappings differ." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44010", + "accept": false, + "reason": "RegNet and SqueezeBert are unrelated architectures and forward implementations; no shared concrete bug or patch to merge together." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44129", + "accept": false, + "reason": "SqueezeBert and SpeechT5 both adjust output collection, but they are separate model-specific refactors with different APIs and internals, so they are not duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 14, + "component_count": 34, + "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": "f9b48bb67c0197c3ef5bb26685902bf8a6be6aebdbf38975bcebdf6f51c5e395", + "component_budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 40, + "serialized_chars": 40173, + "estimated_input_tokens": 10044, + "estimated_eval_tokens": 20344 + }, + "budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 40, + "serialized_chars": 40173, + "estimated_input_tokens": 10044, + "estimated_eval_tokens": 20344 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are separate output-tracing refactors across unrelated model families. They share a maintenance theme, but not the same concrete bug or code-path, so I do not treat any pair as duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44003 is the strongest representative of the set: it makes the broadest output-tracing plumbing change and spans two closely related Mamba-family implementations.", + "best_issue_reason": null, + "best_pr_reason": "PR 44003 is the best single representative of this cluster because it is the most substantive and central example of the refactor pattern here.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43998", + "right": "pull_request:44002", + "accept": false, + "reason": "Different model families and files (timm_backbone vs upernet); same refactor pattern only, not the same concrete change." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44000", + "accept": false, + "reason": "Different model families and files (timm_backbone vs vision_text_dual_encoder); only a shared output-tracing theme." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44001", + "accept": false, + "reason": "Different code paths in unrelated models (timm_backbone vs univnet); not the same underlying bug/change." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44003", + "accept": false, + "reason": "Different subsystems and model families (timm_backbone vs mamba/falcon_mamba); same maintenance pattern only." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44029", + "accept": false, + "reason": "Different model families and implementations (timm_backbone vs rwkv); not a duplicate fix." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44001", + "accept": false, + "reason": "MobileNetV1 and UnivNet are unrelated code paths; both refactor output handling but do not fix the same issue." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44002", + "accept": false, + "reason": "Different models and files (mobilenet_v1 vs upernet); shared refactor theme is too broad to merge as one duplicate." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44025", + "accept": false, + "reason": "MobileNetV1 vs Depth Anything are unrelated model families; not the same underlying bug or change." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44026", + "accept": false, + "reason": "Different model wrappers and code paths (mobilenet_v1 vs vision_encoder_decoder); only superficial similarity." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44027", + "accept": false, + "reason": "MobileNetV1 and Speech Encoder-Decoder are unrelated; not the same fix." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44001", + "accept": false, + "reason": "Vision-Text Dual Encoder and UnivNet are unrelated; both use can_return_tuple-style refactors, but not the same bug." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44002", + "accept": false, + "reason": "Different architectures and files (vision_text_dual_encoder vs upernet); not a duplicate change." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44003", + "accept": false, + "reason": "Vision-Text Dual Encoder and Mamba-family code paths are unrelated; same refactor motif only." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44025", + "accept": false, + "reason": "Different model families (vision_text_dual_encoder vs depth_anything); not the same underlying change." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44002", + "accept": false, + "reason": "UnivNet and UperNet are unrelated model code paths; not a duplicate bug fix." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44003", + "accept": false, + "reason": "UnivNet and Mamba-family changes address different implementations; only broad output-tracing similarity." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44025", + "accept": false, + "reason": "UnivNet vs Depth Anything are unrelated; same pattern, different code paths." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44029", + "accept": false, + "reason": "UperNet and RWKV are unrelated models; not the same concrete issue." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44003", + "accept": false, + "reason": "UperNet and Mamba-family code are different subsystems; not mergeable as one duplicate fix." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44028", + "accept": false, + "reason": "Mamba-family and SuperPoint are unrelated; same refactor theme only." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44004", + "accept": false, + "reason": "Mamba-family output tracing changes and CodeGen output tracing changes affect different model code paths; not the same change." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44004", + "accept": false, + "reason": "TimmBackbone and CodeGen are unrelated; they do not fix the same underlying issue." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44004", + "accept": false, + "reason": "UnivNet and CodeGen are different model families and implementations; no shared concrete bug." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44004", + "accept": false, + "reason": "UperNet and CodeGen are unrelated code paths; only the refactor style matches." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44004", + "accept": false, + "reason": "Vision-Text Dual Encoder and CodeGen are unrelated models; not a duplicate pair." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44026", + "accept": false, + "reason": "Depth Anything and Vision-Encoder-Decoder are different model families; same output-tracing cleanup theme only." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44027", + "accept": false, + "reason": "Depth Anything and Speech Encoder-Decoder are unrelated code paths; not the same bug/change." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44028", + "accept": false, + "reason": "Depth Anything and SuperPoint are different model implementations; only broad similarity in output handling." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44029", + "accept": false, + "reason": "Depth Anything and RWKV are unrelated models; not a duplicate fix." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44030", + "accept": false, + "reason": "Depth Anything and DPR are different model code paths; same refactor theme only." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44029", + "accept": false, + "reason": "Vision-Encoder-Decoder and RWKV are unrelated; no shared concrete bug." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44027", + "accept": false, + "reason": "Vision-Encoder-Decoder and Speech Encoder-Decoder are similar wrappers but still different concrete code paths and files." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44028", + "accept": false, + "reason": "Vision-Encoder-Decoder and SuperPoint are unrelated model families; not the same change." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44030", + "accept": false, + "reason": "Vision-Encoder-Decoder and DPR are unrelated implementations; not a duplicate pair." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44028", + "accept": false, + "reason": "Speech Encoder-Decoder and SuperPoint are unrelated models; same refactor motif only." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44029", + "accept": false, + "reason": "Speech Encoder-Decoder and RWKV are different code paths; not the same underlying bug/change." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44030", + "accept": false, + "reason": "Speech Encoder-Decoder and DPR are unrelated model families; not a duplicate fix." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44029", + "accept": false, + "reason": "SuperPoint and RWKV are unrelated implementations; no shared concrete issue." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44030", + "accept": false, + "reason": "SuperPoint and DPR are different model code paths; not mergeable as one fix." + }, + { + "left": "pull_request:44029", + "right": "pull_request:44030", + "accept": false, + "reason": "RWKV and DPR are unrelated model families; only the output-tracing refactor style matches." + } + ] + }, + "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": 34, + "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": "61401d655252433fd7eeaf00573644d740052924175f480b4c4112c0b39aff66", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6552, + "estimated_input_tokens": 1638, + "estimated_eval_tokens": 3532 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6552, + "estimated_input_tokens": 1638, + "estimated_eval_tokens": 3532 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are two unrelated model-addition PRs that only overlap on standard Transformers boilerplate (docs, auto registries, model package wiring). They add different model families and do not fix the same concrete code path.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45401 is the better representative of the cluster because it is more focused, explicitly tied to an issue, and has a narrower, self-contained Voxtral TTS implementation.", + "best_issue_reason": null, + "best_pr_reason": "PR #45401 is the strongest standalone PR: it has an explicit target issue, a coherent scope, and a clearly bounded model-support change. PR #44662 is larger and unrelated to it.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44662", + "right": "pull_request:45401", + "accept": false, + "reason": "Do not merge: they add different models (PenguinVL vs Voxtral TTS). Shared files are only generic registry/docs plumbing, not the same underlying bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 16, + "component_count": 34, + "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": "2780d2e768d80fbedcf78032f46e16b1e40bcb092746dc330e344fe8f67b4b6a", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4219, + "estimated_input_tokens": 1055, + "estimated_eval_tokens": 2366 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4219, + "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: one rewires the PR documentation workflow to convert PRs to draft, while the other tweaks CI branch routing and includes an unrelated placeholder code change in `modeling_utils.py`. They do not appear to fix the same bug or implement the same change.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "No issue is present, so there is no canonical issue target. As a PR representative, `#46845` is the better anchor because it has broader context and a more substantial CI-related change set, but it is still unrelated to `#44965`.", + "best_issue_reason": null, + "best_pr_reason": "`#46845` is the most representative PR in the cluster because it has more discussion and touches the main CI path, whereas `#44965` is a small workflow-only change. Neither should be treated as a duplicate of the other.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44965", + "right": "pull_request:46845", + "accept": false, + "reason": "Different underlying changes: `#44965` changes the PR documentation workflow to convert PRs to draft, while `#46845` changes CI branch selection and adds an unrelated code edit. They do not appear mergeable 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": 34, + "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": "0130d286cf3ab76063dc525ffafa85d74a0eebb4f4a6189e3a2df06bff309283", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4408, + "estimated_input_tokens": 1102, + "estimated_eval_tokens": 2460 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4408, + "estimated_input_tokens": 1102, + "estimated_eval_tokens": 2460 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These PRs are related to fusion, but they are not the same change: #44979 introduces a generic module-fusion API, while #45363 adds KernelConfig-based n-to-1 kernel fusion integration and an example. They share theme, not a concrete code path.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single canonical PR: #44979 is a foundational API addition, while #45363 is a narrower integration built around that idea and does not appear to be a duplicate of it.", + "best_issue_reason": null, + "best_pr_reason": "#44979 is the more central and reusable change; it defines the module fusion abstraction that a later integration PR would build on.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44979", + "right": "pull_request:45363", + "accept": false, + "reason": "Different levels of the stack: generic module-fusion API vs a specific KernelConfig-based kernel-fusion integration/example. Related, but not mergeable as one duplicate PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 18, + "component_count": 34, + "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": "eb4b8714e20c7044c6f0b93a8b944240da709720f69527a83e3448255ccb3383", + "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 target the same Mistral regex AttributeError fix in `src/transformers/tokenization_utils_tokenizers.py` and share the same issue target; `45317` is the more complete version because it adds test coverage.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "`45317` is the stronger representative PR: it implements the same code fix as `45086` and also adds a test file, making it the more complete change set.", + "best_issue_reason": null, + "best_pr_reason": "`45317` is the best PR for the cluster because it covers the same underlying bug fix with additional tests, so it is the most suitable single reference for the change.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45086", + "right": "pull_request:45317", + "accept": true, + "reason": "Same concrete bug fix in the same file, same issue target, and the second PR is just the more complete variant with tests; these are duplicates of 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": 34, + "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": "1db25c72be192eb0e642ff4b759f0467ff12fff34d64e7efb5f07458af63311a", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6065, + "estimated_input_tokens": 1517, + "estimated_eval_tokens": 3290 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6065, + "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 PRs in the NemotronH area overlap on shared model files, but they target different changes: one is a refactor to inherit GraniteMoeHybrid and add conversion mappings, while the other introduces separate NemotronH dense/sparse model variants and docs. They are related but not duplicates.", + "confidence": 0.9, + "canonical_issue_reason": null, + "canonical_pr_reason": "No issue is present. If forced to pick one PR as the primary anchor, 45569 is broader and more complete, covering the concrete productization of NemotronH into dense/sparse variants plus docs and auto-mapping updates.", + "best_issue_reason": null, + "best_pr_reason": "45569 is the better global candidate because it is the larger, more user-facing implementation with explicit new model families, documentation, and tests. 45115 is narrower and mainly a refactor/conversion-mapping change.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45115", + "right": "pull_request:45569", + "accept": false, + "reason": "Shared NemotronH files alone are not enough: 45115 focuses on inheritance/conversion-key renames, while 45569 adds distinct dense/sparse model packages and docs. They do not look like the same underlying code change 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": 20, + "component_count": 34, + "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": "01317ef4f18c5039b2b461224d9d5d8a32b96997c67eb1fb12ba1c59e801c3dd", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6091, + "estimated_input_tokens": 1523, + "estimated_eval_tokens": 3302 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6091, + "estimated_input_tokens": 1523, + "estimated_eval_tokens": 3302 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two open PRs are closely related heterogeneous-configuration work, but one is a broader model-support implementation and the other is a narrower config-only slice. They overlap on shared configuration files and docs, yet they are not the same concrete change.", + "confidence": 0.84, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45332 is the broader, more complete change: it includes heterogeneous config support plus modeling, masking, cache, skip, and model integration updates, so it subsumes the narrower config-only work.", + "best_issue_reason": null, + "best_pr_reason": "PR #45332 is the best representative because it covers the full heterogeneous-model feature set and incorporates the shared configuration/doc pieces that PR #45333 focuses on.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45332", + "right": "pull_request:45333", + "accept": false, + "reason": "They are related, but not the same underlying PR. #45333 is a config-only subset while #45332 adds broader modeling and runtime integration changes, so they should not be merged as duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 21, + "component_count": 34, + "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": "e9e0c09a0f5f0fd41782355d2c1d8713b2b75826b58432db5edbda976461f31c", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5514, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5514, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs are typing-focused and overlap in a few files, but they address different scopes: 45415 is a broad type-checking sweep across many transformers modules, while 45425 adds specific typing protocols/aliases and updates a smaller set of code paths around modeling_utils and related integrations. They are related, but not the same change set.", + "confidence": 0.9, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: 45415 is a broad repository-wide typing cleanup, whereas 45425 is a narrower typing refactor for specific interfaces and modeling_utils-related annotations.", + "best_issue_reason": null, + "best_pr_reason": "45425 is the better representative because it is more self-contained and specific; 45415 is a large umbrella typing pass that touches many unrelated files.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45415", + "right": "pull_request:45425", + "accept": false, + "reason": "Shared typing-related files are not enough to make these duplicates. 45415 is a broad mypy/type-checking enablement PR, while 45425 introduces concrete typing aliases/protocols and targeted fixes for specific modules. They would not plausibly merge as one single 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": 34, + "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": "c4ca9af17cc81199a582c4ff69b38fa66061f2244cf56eb89e24e2aa753803c1", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4508, + "estimated_input_tokens": 1127, + "estimated_eval_tokens": 2510 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4508, + "estimated_input_tokens": 1127, + "estimated_eval_tokens": 2510 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs are about expert-parallel loading, but they fix different failure modes: one targets DeepSpeed ZeRO-3 / accelerate launch state handling, the other targets FSDP2 expert parameters being overwritten during rank-0 broadcast. Related area, not the same change.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45662 is the better representative of the cluster because it addresses the more concrete runtime corruption bug in EP loading and spans the relevant loading paths, whereas #45548 is a narrower DeepSpeed/serialization fix.", + "best_issue_reason": null, + "best_pr_reason": "PR #45662 is the strongest standalone PR: it fixes a specific code-path problem with expert parameters being silently overwritten and includes the loading-path changes needed for that behavior.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45548", + "right": "pull_request:45662", + "accept": false, + "reason": "Different concrete bugs and code paths: #45548 is about preserving distributed_config / EP state for DeepSpeed ZeRO-3 loading, while #45662 is about preventing expert params from being overwritten during FSDP2-style loading. They are related but not mergeable 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": 23, + "component_count": 34, + "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": "07241d8686e2aeef0e7d8f7036fc2eb3649fcad4314f1ebdafa7a64ef5fb3b68", + "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 target the same xdist-related captured_info artifact collision in testing/debug logging. PR 45645 appears to be the broader follow-up that includes the core fix plus CI log collection, so it is the better canonical representative.", + "confidence": 0.87, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45645 is the more complete change set: it includes the xdist-safe output handling in testing_utils and also updates the workflow to collect multiple captured_info artifacts, whereas 45639 only contains the core library-side fix.", + "best_issue_reason": null, + "best_pr_reason": "PR 45645 best represents the underlying problem because it covers the same captured_info xdist collision and the CI artifact handling needed to preserve/debug those logs.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45639", + "right": "pull_request:45645", + "accept": true, + "reason": "Both PRs fix the same concrete captured_info xdist collision path; 45645 is a superset with the same core code change plus workflow handling, so they could plausibly be merged into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 24, + "component_count": 34, + "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": "fe53914da7af8a58e84175d2033625056c92bb35bb939c8d8dcfc4e179345379", + "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/modeling-rule PRs. PR #45720 changes model forward signatures/return behavior for TRF003, while PR #46259 changes weight initialization logic for TRF018. They share some broad model files, but not the same underlying fix.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45720 is the more focused, concrete code-path change: it adds `can_return_tuple` and adjusts forward-return behavior across specific models. PR #46259 is a separate, broader initialization-rule change with different semantics.", + "best_issue_reason": null, + "best_pr_reason": "PR #45720 is the best representative of the cluster because it is the more narrowly scoped and directly verifiable change, whereas #46259 is a different linter rule affecting a different part of model behavior.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45720", + "right": "pull_request:46259", + "accept": false, + "reason": "Do not merge: these fix different problems. #45720 targets tuple/return-dict handling in model forwards; #46259 targets weight initialization behavior. Shared model files are incidental and not evidence of the same underlying change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 25, + "component_count": 34, + "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": 34, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45932-2", + "nodes": [ + "pull_request:45932", + "pull_request:46100" + ], + "soft_pairs": [ + "pull_request:45932|pull_request:46100" + ], + "prepared_review_unit_hash": "dfc08ca0411ffdbfbd2f25982497b29bf0827b4e1245ecc0d3344e3b88a6a2fc", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5161, + "estimated_input_tokens": 1291, + "estimated_eval_tokens": 2838 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5161, + "estimated_input_tokens": 1291, + "estimated_eval_tokens": 2838 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two pull requests are not duplicates: PR #45932 is a broad CI/telemetry + model config sweep, while PR #46100 is a test-migration PR adding DDP config and model tests. They do not target the same concrete bug or change and would not plausibly merge into one PR.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "No clear canonical PR: the PRs address different concerns and code paths. If one must be picked, #45932 is the more distinctive infra/CI-oriented change, but it is not a duplicate of #46100.", + "best_issue_reason": null, + "best_pr_reason": "#45932 is the better representative of its own change set because it is a single infra/model-config sweep with a clear experimental intent; #46100 is a separate test migration and should not be collapsed into it.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45932", + "right": "pull_request:46100", + "accept": false, + "reason": "Different underlying work: #45932 touches CI secrets and multiple model config files, while #46100 adds DDP/test scaffolding and model tests. Shared similarity is only broad test/model maintenance, 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": 27, + "component_count": 34, + "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": "50f5cbdb62a5d95edf59db30f155eae109b15044c556c456157801b2de71bbf3", + "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": "The two PRs overlap in distributed code, but they are not the same change. PR #46126 is a focused EP/trainer integration feature, while PR #46269 is a broad distributed branch base that touches many unrelated areas. The shared filenames are not enough to treat them as duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #46126 is the better representative of a single concrete change: it targets EP routing and trainer integration with a tight scope and coherent purpose.", + "best_issue_reason": null, + "best_pr_reason": "PR #46126 is the strongest standalone candidate because it is narrowly scoped and clearly addresses one feature path, unlike #46269 which is an umbrella branch-base PR.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46126", + "right": "pull_request:46269", + "accept": false, + "reason": "They share some distributed files, but #46126 is a specific EP/trainer feature PR and #46269 is a broad branch-base PR spanning many unrelated distributed subsystems. Not the same underlying change or bug fix." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 28, + "component_count": 34, + "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:46432|pull_request:46563", + "pull_request:46504|pull_request:46716", + "pull_request:46432|pull_request:46716" + ], + "prepared_review_unit_hash": "32c2ce950f62904d13b31c8b404aae8c8d31602002356c66e5f678025c8ec1d2", + "component_budget": { + "node_count": 10, + "item_count": 10, + "soft_pair_count": 31, + "serialized_chars": 26677, + "estimated_input_tokens": 6670, + "estimated_eval_tokens": 13596 + }, + "budget": { + "node_count": 10, + "item_count": 10, + "soft_pair_count": 31, + "serialized_chars": 26677, + "estimated_input_tokens": 6670, + "estimated_eval_tokens": 13596 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is a heterogeneous set of workflow maintenance PRs. The only obvious mini-cluster is the pr-ci-security-gate.yml sequence, but the individual PRs make different changes (logic, permissions, comments, secrets), so I reject all soft edges as true duplicates.", + "confidence": 0.86, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single PR cleanly represents the whole cluster; 46554 is the best anchor because it makes the most substantive pr-ci-security-gate behavior change and sits in the only near-duplicate subgroup.", + "best_issue_reason": null, + "best_pr_reason": "46554 is the strongest representative PR: it changes actual gate logic rather than only adjacent permissions/comments/secrets, and is the most substantive item in the only related subgroup.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46138", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflows and different fixes: stdin handling for self-comment-ci vs security-gate logic changes." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46558", + "accept": false, + "reason": "Unrelated workflow changes; one fixes PR comment piping, the other adjusts security-gate permissions/approval flow." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46563", + "accept": false, + "reason": "Different workflow areas and different concerns; not the same concrete change." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46716", + "accept": false, + "reason": "Self-comment CI input handling is unrelated to pr-ci-security-gate secrets handling." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46504", + "accept": false, + "reason": "Both touch shell quoting, but in different workflows and different code paths; not a duplicate fix." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46503", + "accept": false, + "reason": "SSH runner hardening is unrelated to benchmark trigger conditions." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46501", + "accept": false, + "reason": "Different workflows and different changes; no shared bug or patch." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46502", + "accept": false, + "reason": "SSH runner changes do not overlap with assign-reviewers checkout ref handling." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflows and different security concerns; not the same fix." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46558", + "accept": false, + "reason": "SSH runner secret cleanup is unrelated to pr-ci-security-gate permissions changes." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46503", + "accept": false, + "reason": "Both are workflow chores, but they change different jobs and different triggers." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46502", + "accept": false, + "reason": "Different workflows and different behavior; no shared underlying bug." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46504", + "accept": false, + "reason": "Anti-slop permission adjustment is unrelated to build-ci-docker-images shell quoting." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46554", + "accept": false, + "reason": "Different files, different workflows, different code paths." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46503", + "accept": false, + "reason": "Assign-reviewers checkout ref change is unrelated to benchmark trigger logic." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46504", + "accept": false, + "reason": "Different workflow scripts and unrelated fixes." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46554", + "accept": false, + "reason": "No shared concrete bug; these are distinct workflow updates." + }, + { + "left": "pull_request:46503", + "right": "pull_request:46504", + "accept": false, + "reason": "Benchmark dispatch condition and docker-image commit-message quoting are different changes." + }, + { + "left": "pull_request:46503", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflows and different logic; not a duplicate." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46554", + "accept": false, + "reason": "Both are workflow tweaks, but one changes commit-message parsing and the other changes security-gate behavior." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46558", + "accept": false, + "reason": "Different workflows and different fixes; not the same underlying change." + }, + { + "left": "pull_request:46554", + "right": "pull_request:46558", + "accept": false, + "reason": "Same file, but different changes: gate logic vs permissions/approval flow." + }, + { + "left": "pull_request:46554", + "right": "pull_request:46563", + "accept": false, + "reason": "One changes security-gate behavior; the other only adds explanatory comments." + }, + { + "left": "pull_request:46554", + "right": "pull_request:46716", + "accept": false, + "reason": "Different security-gate modifications: workflow logic vs secrets inheritance." + }, + { + "left": "pull_request:46558", + "right": "pull_request:46563", + "accept": false, + "reason": "Permissions/approval changes are distinct from comment-only updates." + }, + { + "left": "pull_request:46558", + "right": "pull_request:46716", + "accept": false, + "reason": "Different adjustments to the same workflow, but not the same concrete fix." + }, + { + "left": "pull_request:46563", + "right": "pull_request:46716", + "accept": false, + "reason": "Comments-only review note is unrelated to switching secrets to empty." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46563", + "accept": false, + "reason": "Different workflows and different purposes; not a duplicate." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46563", + "accept": false, + "reason": "SSH runner hardening and security-gate comments are unrelated." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46716", + "accept": false, + "reason": "Different workflow files and different fixes." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46716", + "accept": false, + "reason": "No shared code path or bug; different workflow changes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 29, + "component_count": 34, + "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": 30, + "component_count": 34, + "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": "e8b6b3eb82a7c106d9c267c060e1724316881c9cf360ee15daf0d0cbb8808498", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4279, + "estimated_input_tokens": 1070, + "estimated_eval_tokens": 2396 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4279, + "estimated_input_tokens": 1070, + "estimated_eval_tokens": 2396 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs harden Transformers against path traversal, but they patch different code paths and different untrusted inputs: one validates shard filenames from a checkpoint index, the other validates an explicitly configured weights filename. They are related security fixes, not duplicates.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46281 is the cleaner representative of the cluster theme: it fixes traversal in the sharded-checkpoint loader via `weight_map` filenames. PR 46890 targets a separate explicit-filename path in `modeling_utils.py`, so it is not a duplicate.", + "best_issue_reason": null, + "best_pr_reason": "PR 46281 is the best representative artifact because it addresses a concrete, self-contained loader vulnerability with a focused patch and tests; PR 46890 is a distinct hardening change in another code path.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46281", + "right": "pull_request:46890", + "accept": false, + "reason": "Same broad security theme, but different inputs and different loading code paths (`get_checkpoint_shard_files` vs `_get_resolved_checkpoint_files`); 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": 31, + "component_count": 34, + "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": "8e692a8b0894ef7e10956ab9ae6475ae4f07452dab786110ed79a07e0b38528a", + "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 PRs are unrelated: one fixes CodeLlama decode stripping of real leading spaces, while the other fixes DeepSeek-Coder/DeepSeek-LLM loading the wrong tokenizer class. They should not be clustered as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: the patches address different bugs in different code paths and would not plausibly be merged into one PR.", + "best_issue_reason": null, + "best_pr_reason": "No single best PR for duplication purposes; each PR is a standalone fix for a different model family, so neither is a better representative of the other.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46574", + "right": "pull_request:46580", + "accept": false, + "reason": "Different problems and different fixes: CodeLlama decoder stripping vs DeepSeek checkpoint tokenizer-class override. No shared concrete code-path bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 32, + "component_count": 34, + "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-46781-3", + "nodes": [ + "pull_request:46781", + "pull_request:46993", + "pull_request:46994" + ], + "soft_pairs": [ + "pull_request:46781|pull_request:46993", + "pull_request:46993|pull_request:46994" + ], + "prepared_review_unit_hash": "e4b259d8016efdf659fd702780f0adaf0eb64ce8445be6822b81e848f6c89f16", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 6604, + "estimated_input_tokens": 1651, + "estimated_eval_tokens": 3558 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 6604, + "estimated_input_tokens": 1651, + "estimated_eval_tokens": 3558 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "All three PRs are generation-related but they fix different code paths: cache validation, prompt lookup decoding length capping, and logits processor input validation. They are not duplicates, so both soft edges should be rejected.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46994 is the clearest standalone bugfix: it addresses a specific validation gap in logits_process and includes a narrowly targeted change plus tests.", + "best_issue_reason": null, + "best_pr_reason": "PR 46994 is the best representative of the cluster because it has the most self-contained, precisely scoped fix and the title matches the concrete failure mode.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46781", + "right": "pull_request:46993", + "accept": false, + "reason": "Different underlying problems and code paths: unsupported QuantizedCache layer-type validation vs prompt lookup decoding overshooting max_length. Similar subsystem only, not the same fix." + }, + { + "left": "pull_request:46993", + "right": "pull_request:46994", + "accept": false, + "reason": "Both are generation fixes, but one changes candidate generation length math while the other hardens logits processor validation for empty lists. They do not fix the same bug and would not plausibly merge into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 33, + "component_count": 34, + "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-46847-2", + "nodes": [ + "pull_request:46847", + "pull_request:47009" + ], + "soft_pairs": [ + "pull_request:46847|pull_request:47009" + ], + "prepared_review_unit_hash": "5945467acf106e30b18efa60862683b7613f485dba66fc72e3272a53bfbe8bf0", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4640, + "estimated_input_tokens": 1160, + "estimated_eval_tokens": 2576 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4640, + "estimated_input_tokens": 1160, + "estimated_eval_tokens": 2576 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two near-identical Qwen3.5 PRs add the same linear_attn tensor-parallel entries to the same files for the same issue target; they look like duplicate attempts at the same change.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #46847 is the better canonical PR: it is the earlier submission, has more discussion/review activity and inbound references, and already explicitly targets the tracking issue for this change.", + "best_issue_reason": null, + "best_pr_reason": "PR #46847 is the best representative of the change because it appears to be the original, more established PR for the same fix, while #47009 is a later duplicate with no added substantive scope.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46847", + "right": "pull_request:47009", + "accept": true, + "reason": "Both PRs touch the same Qwen3.5 files and add the same linear_attn base_model_tp_plan entries for the same issue target, so they are the same underlying code 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": 34, + "component_count": 34, + "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-47032-2", + "nodes": [ + "pull_request:47032", + "pull_request:47033" + ], + "soft_pairs": [ + "pull_request:47032|pull_request:47033" + ], + "prepared_review_unit_hash": "e029b231e5ada9892fe9af845ff0a7788ab7c06233e299b7287492d50bd1614f", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4302, + "estimated_input_tokens": 1076, + "estimated_eval_tokens": 2408 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4302, + "estimated_input_tokens": 1076, + "estimated_eval_tokens": 2408 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two pull requests address unrelated fixes in different areas of Transformers: image-text-to-text stop sequence handling vs. serve tool parsing for legacy response schemas. They are not duplicates and should not be clustered as the same change.", + "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:47032", + "right": "pull_request:47033", + "accept": false, + "reason": "Different subsystems, different bug shapes, and different touched files; they do not appear to fix the same underlying code-path problem and would not plausibly merge into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + } + ] +} diff --git a/snapshots/20260703T060023Z/analysis-runs/hybrid-model-20260703t060023z/manifest.json b/snapshots/20260703T060023Z/analysis-runs/hybrid-model-20260703t060023z/manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..c571a1243e18ce782294d251fd225eb579741ac5 --- /dev/null +++ b/snapshots/20260703T060023Z/analysis-runs/hybrid-model-20260703t060023z/manifest.json @@ -0,0 +1,14 @@ +{ + "analysis_id": "hybrid-model-20260703t060023z", + "artifacts": { + "hybrid": "snapshots/20260703T060023Z/analysis-runs/hybrid-model-20260703t060023z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260703T060023Z/analysis-runs/hybrid-model-20260703t060023z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "model": null, + "published_at": "2026-07-03T06:11:41Z", + "repo": "huggingface/transformers", + "schema_version": 1, + "snapshot_id": "20260703T060023Z", + "variant": "hybrid" +} diff --git a/snapshots/20260703T060023Z/manifest.json b/snapshots/20260703T060023Z/manifest.json index 0d677eb121be789c894f76d52d5e5d8b3daeadb1..839b24b70db4c43e3c79bba67148dc297e1b4961 100644 --- a/snapshots/20260703T060023Z/manifest.json +++ b/snapshots/20260703T060023Z/manifest.json @@ -32,6 +32,24 @@ "timeline_events": 258 }, "extracted_at": "2026-07-03T06:00:23Z", + "published_analysis": { + "canonical_analysis_id": "hybrid-model-20260703t060023z", + "runs": { + "hybrid-model-20260703t060023z": { + "analysis_id": "hybrid-model-20260703t060023z", + "artifacts": { + "hybrid": "snapshots/20260703T060023Z/analysis-runs/hybrid-model-20260703t060023z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260703T060023Z/analysis-runs/hybrid-model-20260703t060023z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "manifest_path": "snapshots/20260703T060023Z/analysis-runs/hybrid-model-20260703t060023z/manifest.json", + "model": null, + "published_at": "2026-07-03T06:11:41Z", + "variant": "hybrid" + } + }, + "schema_version": 1 + }, "repo": "huggingface/transformers", "snapshot_id": "20260703T060023Z", "watermark": {