diff --git a/analysis/current/analysis-report-hybrid.json b/analysis/current/analysis-report-hybrid.json index aedfaa5999dc204cf842c2d064677df965f39ff5..e24bf39e424f0edb3003d58de7491d750f6adfc8 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": "20260617T000107Z", - "generated_at": "2026-06-17T00:10:35Z", + "snapshot_id": "20260617T060052Z", + "generated_at": "2026-06-17T06:10:51Z", "evidence_quality": "full", "llm_enrichment": true, "meta_bugs": [ @@ -1169,11 +1169,11 @@ { "left_pr_number": 41116, "right_pr_number": 45613, - "code_similarity": 0.797, - "size_similarity": 0.885, + "code_similarity": 0.808, + "size_similarity": 0.893, "file_overlap": 0.75, - "area_overlap": 0.831, - "patch_similarity": 0.693, + "area_overlap": 0.869, + "patch_similarity": 0.667, "shared_filenames": [ "docs/source/en/_toctree.yml", "docs/source/en/model_doc/minicpm3.md", @@ -1263,7 +1263,7 @@ "left_ranges": [ [ 1, - 141 + 149 ] ], "right_ranges": [ @@ -1278,7 +1278,7 @@ "left_ranges": [ [ 1, - 544 + 550 ] ], "right_ranges": [ @@ -1293,7 +1293,7 @@ "left_ranges": [ [ 1, - 444 + 394 ] ], "right_ranges": [ @@ -1308,7 +1308,7 @@ "left_ranges": [ [ 1, - 109 + 132 ] ], "right_ranges": [ @@ -1739,6 +1739,77 @@ } ] }, + { + "cluster_id": "cluster-46562-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 46562, + "issue_numbers": [], + "pr_numbers": [ + 46562, + 46658 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 46562, + "right_pr_number": 46658, + "code_similarity": 0.17, + "size_similarity": 0.397, + "file_overlap": 0.032, + "area_overlap": 0.0, + "patch_similarity": 0.537, + "shared_filenames": [ + "src/transformers/models/auto/auto_mappings.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/auto/auto_mappings.py", + "left_ranges": [ + [ + 441, + 447 + ], + [ + 591, + 598 + ], + [ + 823, + 829 + ], + [ + 883, + 889 + ], + [ + 1082, + 1088 + ], + [ + 1205, + 1211 + ] + ], + "right_ranges": [ + [ + 362, + 368 + ], + [ + 795, + 801 + ] + ] + } + ] + } + ] + }, { "cluster_id": "cluster-44018-2", "summary": "Cluster of 2 related pull requests linked by soft_similarity.", @@ -2123,6 +2194,15 @@ ], "target_issue_number": 46305, "reason": "PRs in cluster-46305-3 are treated as duplicates because they converge on issue #46305 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-46562-2", + "canonical_pr_number": 46562, + "duplicate_pr_numbers": [ + 46658 + ], + "target_issue_number": null, + "reason": "PRs in cluster-46562-2 are treated as duplicates because they share soft_similarity evidence." } ], "best_issue": { diff --git a/analysis/current/analysis-report-hybrid.llm-reviews.json b/analysis/current/analysis-report-hybrid.llm-reviews.json index 13227916f127dfab6b919843713a0c60ffd4b584..0521a665c9b0c54de0a624b95386e59db230ef94 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": "20260617T000107Z", - "generated_at": "2026-06-17T00:10:35Z", + "snapshot_id": "20260617T060052Z", + "generated_at": "2026-06-17T06:10:51Z", "model": "gpt-5.4-mini?service_tier=flex", "reviews": [ { @@ -10,13 +10,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 1, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:24643", "issue:30277", @@ -130,14 +130,14 @@ "issue:45237|issue:46015", "issue:43039|issue:44805" ], - "prepared_review_unit_hash": "4b56d51f3fb9f231e5f03a7bea86d77ee54cbe371ee62a2e56c8e657c9b8e231", + "prepared_review_unit_hash": "fc7b34c7eb574940dcda7453b066bc512ee649506bfc65572dc67fd1a8227a9c", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 47, @@ -152,390 +152,288 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous issue reports, with a few small duplicate-like subgroups: a DeepSpeed ZeRO-3 index-out-of-bounds pair, a Qwen3.5MoE rotary-embedding pair, and a torch 2.4.1 incompatibility pair. Most other soft edges are only broad similarity and should stay separate.", + "summary": "Mostly a heterogeneous issue cluster with a few true duplicate pairs. The clearest duplicate subclusters are the DeepSpeed ZeRO-3 + Bert index-out-of-bounds reports and the torch 2.4.1 compatibility reports; most other pairs only share broad subsystem similarity and should stay separate.", "confidence": 0.84, - "canonical_issue_reason": "issue:43596 is the clearest representative of the strongest duplicate-like bug here: the DeepSpeed ZeRO-3 index-out-of-bounds failure, which is mirrored very closely by issue:43638.", + "canonical_issue_reason": "Issue 43596 is the best canonical issue: it is a precise DeepSpeed ZeRO-3 + Bert reproduction of the same index-out-of-bounds failure seen in 43638, with the cleanest specific bug description.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43596 best represents the cluster's most concrete duplicate-worthy problem and has the strongest match to another issue in the set.", + "best_issue_reason": "If one issue must anchor the cluster, 43596 is the strongest representative because it sits in a real duplicate pair and has the most concrete, narrowly scoped failure mode.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44322", "right": "issue:45464", "accept": false, - "reason": "Different models and failures: missing config attribute vs streaming chat/completions API error." + "reason": "Different failures: Qwen3.5 config/model-shape error vs chat/completions streaming inference." }, { "left": "issue:45237", "right": "issue:45685", "accept": false, - "reason": "Both are device/backend failures, but on different hardware paths and different symptoms; not the same bug." + "reason": "Different backends and bugs: AMD GPU runtime issue vs MPS histogram op not implemented." }, { "left": "issue:43716", "right": "issue:45237", "accept": false, - "reason": "Different models and root causes: image-preprocessor dtype mismatch vs AMD GPU execution failure." + "reason": "Mistral image dtype mismatch is unrelated to GPT-OSS failing on AMD GPUs." }, { "left": "issue:43828", "right": "issue:45237", "accept": false, - "reason": "Both mention dtype/device issues, but they affect different models and code paths." + "reason": "Phi autocast dtype mismatch and GPT-OSS AMD GPU failure are different code paths." }, { "left": "issue:45237", "right": "issue:46468", "accept": false, - "reason": "Unrelated failures: AMD GPU incompatibility vs a broken example for Nemotron." + "reason": "AMD GPU failure for GPT-OSS is not the same as the Nemotron example breakage." }, { "left": "issue:45237", "right": "issue:45507", "accept": false, - "reason": "Different model families and different error modes; too broad to be duplicates." + "reason": "Different model-specific bugs: GPT-OSS AMD runtime issue vs GraniteMoEHybrid invalid method call." }, { "left": "issue:30277", "right": "issue:35545", "accept": false, - "reason": "DeepSpeed collective mismatch is unrelated to ONNX export errors." + "reason": "Deepspeed collective mismatch on Jamba is unrelated to ModernBERT ONNX export." }, { "left": "issue:35545", "right": "issue:42915", "accept": false, - "reason": "Both are model-specific runtime/export issues, but with different models and distinct failure mechanisms." + "reason": "ModernBERT export error and Qwen3Moe FP8 failure are different problems." }, { "left": "issue:35545", "right": "issue:41720", "accept": false, - "reason": "ONNX export error and CUDA assert during device mapping are unrelated." + "reason": "ONNX export bug and Qwen3 device-mapping CUDA assert are unrelated." }, { "left": "issue:45901", "right": "issue:46015", "accept": false, - "reason": "Different topics: task crash vs outdated RAG examples." + "reason": "Table QA crash and outdated RAG docs are different issues." }, { "left": "issue:44509", "right": "issue:44512", "accept": false, - "reason": "Both are docs cleanup items for v5, but they refer to different removed docs content." + "reason": "Both are docs cleanups for v5 removals, but they target different removed material and are not the same fix." }, { "left": "issue:35545", "right": "issue:36010", "accept": false, - "reason": "Different failure classes: export error vs import error." + "reason": "ONNX export failure is unrelated to the GenerationMixin import error." }, { "left": "issue:44918", "right": "issue:46165", "accept": false, - "reason": "TRI/SFT input-embedding unpacking issue is unrelated to resuming training with remote code." + "reason": "Qwen3.5 embedding unpacking with TRL is unrelated to resume-training with trust_remote_code." }, { "left": "issue:44805", "right": "issue:45823", "accept": false, - "reason": "Both involve tensor/device mismatches, but the reported bugs and affected paths are different." + "reason": "Mask shape mismatch during indexing is unrelated to Gemma4 device-map mismatch." }, { "left": "issue:43638", "right": "issue:44805", "accept": false, - "reason": "Zero-3 index error versus mask/tensor shape mismatch are different symptoms and likely different bugs." + "reason": "Both mention shape/index errors, but the Bert ZeRO-3 load crash and the mask/index mismatch are different bugs." }, { "left": "issue:43638", "right": "issue:46165", "accept": false, - "reason": "Training resume with remote code is not the same issue as a zero-sized tensor index error." + "reason": "Bert ZeRO-3 loading failure is unrelated to resuming local-model training with remote code." }, { "left": "issue:43596", "right": "issue:44805", "accept": false, - "reason": "Only a generic index/shape similarity; the concrete errors and code paths differ." + "reason": "Different failures and repros: Bert ZeRO-3 init index error vs generic mask/tensor shape mismatch." }, { "left": "issue:44805", "right": "issue:46294", "accept": false, - "reason": "Mask-shape IndexError is unrelated to Qwen3.5 video feature/token mismatch." + "reason": "Generic mask/index mismatch is unrelated to Qwen3.5 video-token template mismatch." }, { "left": "issue:44805", "right": "issue:45161", "accept": false, - "reason": "Different models and different failure types; not a duplicate." + "reason": "Mask/index mismatch and GPT-OSS tensor-parallel failure are not the same bug." }, { "left": "issue:46165", "right": "issue:46468", "accept": false, - "reason": "Training resume issue and broken example are unrelated." + "reason": "Remote-code resume-training issue is unrelated to the Nemotron example failure." }, { "left": "issue:44560", "right": "issue:46165", "accept": false, - "reason": "Video chat-template StopIteration and training-resume problems are unrelated." + "reason": "Qwen3-VL video StopIteration and remote-code resume-training are different problems." }, { "left": "issue:43596", "right": "issue:46165", "accept": false, - "reason": "Both mention training/loading, but the specific bug and triggering conditions are different." + "reason": "Bert ZeRO-3 load crash is unrelated to the remote-code resume-training issue." }, { "left": "issue:34634", "right": "issue:35545", "accept": false, - "reason": "Processor voice_preset bug is unrelated to ModernBERT ONNX export." + "reason": "BarkProcessor voice preset bug is unrelated to ModernBERT export." }, { "left": "issue:30277", "right": "issue:43638", "accept": false, - "reason": "Both involve DeepSpeed, but the concrete collective mismatch and zero-sized tensor index error are different." + "reason": "Both involve DeepSpeed, but the Jamba collective mismatch and Bert ZeRO-3 index error are not the same underlying bug." }, { "left": "issue:43596", "right": "issue:43638", "accept": true, - "reason": "Near-exact duplicate: both report the same ZeRO-3 index-out-of-bounds error with size-0 tensors during model loading/training." + "reason": "Same DeepSpeed ZeRO-3/Bert index-out-of-bounds failure, just described with pretrained vs non-pretrained wording." }, { "left": "issue:43039", "right": "issue:45161", "accept": false, - "reason": "Liger Kernel cross_entropy call and GPT-OSS TP failure are different issues." + "reason": "Liger Kernel cross-entropy dispatch and GPT-OSS tensor-parallel failure are unrelated." }, { "left": "issue:43975", "right": "issue:44457", "accept": false, - "reason": "Different bugs: detokenization regression vs LoRA merge/save/reload mismatch." + "reason": "DeepSeek detokenization and LoRA merge/save-load output mismatch are different issues." }, { "left": "issue:44292", "right": "issue:44457", "accept": false, - "reason": "Different models and unrelated failures." + "reason": "Qwen NVFP4 runtime failure is unrelated to the LoRA save/reload discrepancy." }, { "left": "issue:44841", "right": "issue:46661", "accept": false, - "reason": "Processor failure for Voxtral is unrelated to remote-code config_class AttributeError." + "reason": "Processor failure for Voxtral is unrelated to remote-code config_class import errors." }, { "left": "issue:34928", "right": "issue:44805", "accept": false, - "reason": "Both are training-time tensor problems, but one is activation checkpointing/FSDP and the other is a mask-shape mismatch." + "reason": "Different training-time mismatch bugs: FSDP checkpointing vs generic mask/index shape error." }, { "left": "issue:24643", "right": "issue:30277", "accept": false, - "reason": "Both mention DeepSpeed, but one is a 2-D weight training error and the other is a collective mismatch." + "reason": "DeepSpeed training weight-dimension error is unrelated to the Jamba collective mismatch." }, { "left": "issue:45859", "right": "issue:45902", - "accept": true, - "reason": "Same Qwen3.5MoE rotary-embedding area and same offload/materialization bug family; the second issue looks like the concrete manifestation of the first." + "accept": false, + "reason": "Same model family, but CPU-offload incompatibility and uninitialized-memory materialization are different bugs." }, { "left": "issue:43638", "right": "issue:45161", "accept": false, - "reason": "Zero-3 index error and GPT-OSS tensor-parallel failure are different bugs." + "reason": "Bert ZeRO-3 index error and GPT-OSS tensor-parallel failure are unrelated." }, { "left": "issue:34928", "right": "issue:36331", "accept": false, - "reason": "Activation checkpointing/FSDP recomputation mismatch is unrelated to CustomTrainer compute_loss signature breakage." + "reason": "Activation-checkpointing/FSDP mismatch is unrelated to the compute_loss signature change." }, { "left": "issue:43596", "right": "issue:45161", "accept": false, - "reason": "Different subsystems and different failure modes." + "reason": "Bert ZeRO-3 loading failure is unrelated to GPT-OSS tensor-parallel issues." }, { "left": "issue:43716", - "right": "issue:45507", + "right": "issue:44292", "accept": false, - "reason": "Different model-specific issues; one is dtype mismatch, the other is an invalid method call." + "reason": "Mistral image dtype mismatch and Qwen NVFP4 runtime error are different model-specific bugs." }, { "left": "issue:45874", "right": "issue:46294", "accept": false, - "reason": "Different Gemma/Qwen video-related bugs with different symptoms and code paths." - }, - { - "left": "issue:44292", - "right": "issue:45237", - "accept": false, - "reason": "Different model/runtime failures; not the same underlying bug." - }, - { - "left": "issue:39290", - "right": "issue:45081", - "accept": false, - "reason": "Gemma3/vLLM attribute error and Mistral tokenizer regex crash are unrelated." + "reason": "Gemma4 reverse-embedding lookup cost is unrelated to Qwen video feature/token mismatch." }, { - "left": "issue:45823", - "right": "issue:45874", + "left": "issue:43425", + "right": "issue:44292", "accept": false, - "reason": "Both concern model behavior, but one is device-map forward mismatch and the other is reverse embedding lookup cost." + "reason": "Torch version compatibility issue is unrelated to Qwen NVFP4 runtime failure." }, { "left": "issue:45800", "right": "issue:46515", "accept": true, - "reason": "These are the same compatibility complaint: transformers/torch 2.4.1 incompatibility, phrased with the same version pair." - }, - { - "left": "issue:43425", - "right": "issue:44292", - "accept": false, - "reason": "Generic torch incompatibility vs a specific Qwen-3 NVFP4 runtime failure are not the same bug." - }, - { - "left": "issue:34928", - "right": "issue:41720", - "accept": false, - "reason": "FSDP checkpointing mismatch and Qwen3 auto device-map CUDA assert are unrelated." - }, - { - "left": "issue:43828", - "right": "issue:45161", - "accept": false, - "reason": "Different models and different failure classes: autocast dtype mismatch vs TP-only failure." - }, - { - "left": "issue:43828", - "right": "issue:44292", - "accept": false, - "reason": "Both involve model/runtime errors, but they are not the same code path or symptom." - }, - { - "left": "issue:43541", - "right": "issue:43716", - "accept": false, - "reason": "Mixtral grouped_mm tracing error and Mistral image-preprocessor dtype mismatch are unrelated." - }, - { - "left": "issue:43716", - "right": "issue:46468", - "accept": false, - "reason": "Different model families and different failure modes." - }, - { - "left": "issue:43716", - "right": "issue:45685", - "accept": false, - "reason": "Different devices and symptoms; not a duplicate." - }, - { - "left": "issue:44292", - "right": "issue:44918", - "accept": false, - "reason": "Qwen-3 NVFP4 runtime issue is unrelated to Qwen3.5 embedding unpacking with TRL SFT." - }, - { - "left": "issue:45507", - "right": "issue:45685", - "accept": false, - "reason": "Invalid method call in GraniteMoEHybrid is not the same bug as the MPS histogram error." - }, - { - "left": "issue:43716", - "right": "issue:44292", - "accept": false, - "reason": "Distinct model-specific failures." - }, - { - "left": "issue:44918", - "right": "issue:45161", - "accept": false, - "reason": "TRL embedding unpacking and GPT-OSS tensor-parallel failure are unrelated." - }, - { - "left": "issue:43039", - "right": "issue:44918", - "accept": false, - "reason": "Liger Kernel cross_entropy invocation and SFT embedding unpacking are different issues." - }, - { - "left": "issue:41762", - "right": "issue:43638", - "accept": false, - "reason": "Both are ZeRO-3 index errors, but the models and triggering paths differ enough that this is not clearly the same bug." + "reason": "Same torch 2.4.1 compatibility problem; one report is just the more specific transformers 5.8.0 manifestation." }, { "left": "issue:43827", "right": "issue:44512", "accept": false, - "reason": "Both are v5 docs cleanup, but they address different removed references." + "reason": "Both are v5 docs fixes, but they cover different removed docs content and should not be merged as one bug." }, { "left": "issue:43828", - "right": "issue:45685", - "accept": false, - "reason": "Different hardware backends and different error mechanisms." - }, - { - "left": "issue:44918", - "right": "issue:45685", - "accept": false, - "reason": "Unrelated bugs: TRL SFT embedding unpacking vs MPS histogram limitation." - }, - { - "left": "issue:43039", - "right": "issue:43828", + "right": "issue:45161", "accept": false, - "reason": "Different model/backend combinations and different concrete failures." + "reason": "Phi autocast dtype mismatch is unrelated to GPT-OSS tensor-parallel failure." }, { "left": "issue:30277", "right": "issue:36331", "accept": false, - "reason": "DeepSpeed collective mismatch and trainer compute_loss signature breakage are unrelated." + "reason": "DeepSpeed collective mismatch and CustomTrainer API breakage are different layers and different bugs." }, { "left": "issue:34928", "right": "issue:36010", "accept": false, - "reason": "Activation checkpointing/FSDP issue and import error are unrelated." + "reason": "FSDP activation-checkpointing mismatch is unrelated to the GenerationMixin import error." }, { "left": "issue:45237", "right": "issue:46015", "accept": false, - "reason": "GPU runtime failure and outdated RAG examples are unrelated." + "reason": "GPT-OSS AMD GPU runtime failure is unrelated to outdated RAG examples." }, { "left": "issue:43039", "right": "issue:44805", "accept": false, - "reason": "Liger Kernel cross_entropy usage and mask/tensor shape IndexError are unrelated." + "reason": "Liger Kernel cross-entropy dispatch is unrelated to the mask shape/index error." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet and stays mostly conservative. The accepted soft edges are plausibly duplicate-like based on very similar titles/code paths (43596\u201343638, 45859\u201345902, 45800\u201346515), and the rejections are generally cautious rather than overstated. No obvious mergeability or evidence precision problems stand out." + "feedback": "The cluster summary is grounded in the packet and stays conservative. The two accepted duplicate pairs are well supported by the issue titles/body patterns, and the rejected soft edges mostly separate clearly different model/backend failures or distinct docs/API issues. I don\u2019t see evidence of overstatement in the canonical issue choice or the summary." }, "evaluator_used": true, "retried": false, @@ -548,13 +446,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 2, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:36010", "issue:39290", @@ -648,281 +546,278 @@ "issue:43296|issue:43366", "issue:44805|issue:44918" ], - "prepared_review_unit_hash": "97843236e51ae67ff4d8f830efabc5c3966d7ff8fd9447f9980149568cf0b7cd", + "prepared_review_unit_hash": "241f43f6d1ea3b7631a6ec558e9195595977355ab64cb1552eb98a4b40d78b5f", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 41, - "serialized_chars": 39819, - "estimated_input_tokens": 9955, - "estimated_eval_tokens": 20166 + "serialized_chars": 39815, + "estimated_input_tokens": 9954, + "estimated_eval_tokens": 20164 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is noisy and heterogeneous: it mixes unrelated model-loading failures, backend/version incompatibilities, tokenizer/processor regressions, and documentation issues. I don\u2019t see a safe duplicate canonicalization candidate from the titles alone, so all soft pairs should be rejected.", - "confidence": 0.91, - "canonical_issue_reason": null, + "summary": "This cluster is a loose mix of transformers/vLLM/DeepSpeed/model-loading regressions, not a single tight duplicate family. The only supported merge-worthy pairs are the ZeRO-3 index-out-of-bounds reports and the AutoTokenizer wrong-token-ID regression reports. Most other items share only broad subsystem or version-regression themes and should remain separate.", + "confidence": 0.92, + "canonical_issue_reason": "Issue 45020 is the broadest umbrella for the recurring recent-version regression reports, but the cluster remains heterogeneous and only a couple of pairs are close enough to combine.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Issue 45020 is the strongest single representative because it captures the recurring recent-version breakage theme without requiring the cluster to be much tighter than it is.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43866", "right": "issue:46402", "accept": false, - "reason": "Different models and different loading failures: corrupted Ovis2 checkpoint vs missing Sam3Model text encoder weights." + "reason": "Different models and failure modes: corrupted checkpoint vs missing text encoder weights." }, { "left": "issue:45800", "right": "issue:46304", "accept": false, - "reason": "Both are version-compatibility regressions, but they are different concrete breakages in different releases and environments." + "reason": "Both are version-compatibility complaints, but they describe different breakages and code paths." }, { "left": "issue:43296", "right": "issue:46304", "accept": false, - "reason": "PaddleOCR-VL load failure in vLLM is unrelated to the PyTorch/distributed support regression." + "reason": "PaddleOCR-VL loading in vLLM is unrelated to USE_DISTRIBUTED=0 support in PyTorch." }, { "left": "issue:41762", "right": "issue:43596", - "accept": false, - "reason": "Same error shape, but different model paths and contexts; not enough to treat Gemma3 load failure and BertModel zero3 init as the same bug." + "accept": true, + "reason": "Same ZeRO-3 load-time signature: index 0 out of bounds for dimension 0 with size 0 during model initialization." }, { "left": "issue:39290", "right": "issue:41720", "accept": false, - "reason": "Different subsystems and errors: Gemma3TextConfig attribute mismatch in vLLM vs Qwen3 auto device mapping cuda assert." + "reason": "Different symptoms and subsystems: Gemma3 config attribute error vs Qwen3 auto device-map CUDA assert." }, { "left": "issue:43901", "right": "issue:44512", "accept": false, - "reason": "Docs mismatch for TextClassificationPipeline is unrelated to the removed transformers run command docs issue." + "reason": "Both are docs issues, but they cover different documentation mistakes." }, { "left": "issue:44589", "right": "issue:46304", "accept": false, - "reason": "Storage type lookup failure is not the same as the PyTorch USE_DISTRIBUTED=0 compatibility break." + "reason": "Storage-type lookup failure is unrelated to the PyTorch distributed support regression." }, { "left": "issue:45237", "right": "issue:45901", "accept": false, - "reason": "Different runtime areas: AMD GPU support for GPT-OSS vs table-question-answering task crash." + "reason": "AMD GPU support for GPT-OSS and table-question-answering crashes are different issues." }, { "left": "issue:43039", "right": "issue:44560", "accept": false, - "reason": "Liger Kernel cross_entropy fallback issue is unrelated to Qwen3.5 video StopIteration." + "reason": "Liger Kernel cross-entropy dispatch and Qwen3.5-VL video StopIteration are different bugs." }, { "left": "issue:45823", "right": "issue:46294", "accept": false, - "reason": "Gemma4 PLE device mismatch and Qwen3.5 video chat-template mismatch are different model-specific bugs." + "reason": "Device-map mismatch in Gemma4 PLE is unrelated to Qwen3.5 video token/feature alignment." }, { "left": "issue:43975", "right": "issue:46015", "accept": false, - "reason": "DeepSeek detokenization regression is not the same as outdated RAG examples." + "reason": "Wrong detokenization and outdated RAG examples are unrelated." }, { "left": "issue:44841", "right": "issue:45823", "accept": false, - "reason": "Voxtral processor failure and Gemma4 device mismatch affect different code paths and models." + "reason": "Voxtral processor failure and Gemma4 device mismatch are different code paths." }, { "left": "issue:44863", "right": "issue:46015", "accept": false, - "reason": "NemotronH checkpoint loading bug is unrelated to documentation examples." + "reason": "NemotronH checkpoint loading is unrelated to RAG documentation drift." }, { "left": "issue:45020", "right": "issue:46304", "accept": false, - "reason": "Remote-code model breakage is broader, but it is not the same concrete bug as PyTorch USE_DISTRIBUTED=0 support." + "reason": "Both are broad regression reports, but one is about remote_code models and the other about USE_DISTRIBUTED=0 support." }, { "left": "issue:45812", "right": "issue:45920", - "accept": false, - "reason": "Both are tokenizer regressions, but they affect different model sets and likely different tokenizer logic." + "accept": true, + "reason": "Same AutoTokenizer wrong-token-ID v5 regression; 45920 is the broader follow-up covering more model families." }, { "left": "issue:43866", "right": "issue:44863", "accept": false, - "reason": "Corrupted Ovis2 checkpoint and NemotronH checkpoint loading are different model-specific failures." + "reason": "Corrupted checkpoint and NemotronH load failure are not the same bug." }, { "left": "issue:44292", "right": "issue:46015", "accept": false, - "reason": "Qwen-3-8B-NVFP4 runtime error is unrelated to RAG documentation content." + "reason": "Qwen-3-8B-NVFP4 runtime error is unrelated to outdated RAG examples." }, { "left": "issue:43872", "right": "issue:46304", "accept": false, - "reason": "bitsandbytes Int8Params incompatibility is a different dependency issue than the PyTorch distributed support regression." + "reason": "bitsandbytes Int8Params incompatibility is distinct from the PyTorch distributed build issue." }, { "left": "issue:39290", "right": "issue:43531", "accept": false, - "reason": "Both mention sliding_window, but one is a Gemma3TextConfig attribute error and the other is a Qwen3-MoE sliding_window issue." + "reason": "Both mention sliding window, but the configs and failures are different model-specific problems." }, { "left": "issue:41093", "right": "issue:43638", "accept": false, - "reason": "Same general mask/index error, but the titles point to different models and different training/loading scenarios." + "reason": "Different errors: mask/tensor shape mismatch vs ZeRO-3 index-out-of-bounds during BertModel loading." }, { "left": "issue:43854", "right": "issue:43866", "accept": false, - "reason": "GLM-4.7-Flash unit test loading failure is not the same as a corrupted Ovis2 checkpoint." + "reason": "Model load failure in tests and corrupted checkpoint are different problems." }, { "left": "issue:43792", "right": "issue:45901", "accept": false, - "reason": "Whisper load failure and table-question-answering crash are unrelated model/task failures." + "reason": "Whisper load failure and table-question-answering crash are unrelated." + }, + { + "left": "issue:43425", + "right": "issue:45070", + "accept": false, + "reason": "Torch 2.10 incompatibility and the PretrainedConfig pydantic-field regression are different issues." }, { "left": "issue:43975", "right": "issue:45901", "accept": false, - "reason": "Detokenization regression and table-question-answering crash are different bugs in different layers." + "reason": "Detokenization regression and table-question-answering crash are unrelated." }, { "left": "issue:43596", "right": "issue:45800", "accept": false, - "reason": "Both involve errors during model setup, but they are different bugs with different environments and failure modes." + "reason": "Both involve versioned infrastructure, but one is a ZeRO-3 index error and the other is torch/transformers incompatibility." }, { "left": "issue:43792", "right": "issue:45070", "accept": false, - "reason": "Whisper load failure is unrelated to the PretrainedConfig pydantic regression." + "reason": "Whisper loading and pydantic config-field breakage do not share the same bug." }, { "left": "issue:45362", "right": "issue:45464", "accept": false, - "reason": "Qwen3.5 chat crash and streaming chat/completions API failure are different request paths." - }, - { - "left": "issue:44292", - "right": "issue:45901", - "accept": false, - "reason": "Qwen-3-8B-NVFP4 runtime failure is not the same as the table-question-answering task crash." + "reason": "Different Qwen3.5 failures and different surfaces (chat vs streaming API)." }, { - "left": "issue:39290", - "right": "issue:45823", - "accept": false, - "reason": "Different models and errors: Gemma3TextConfig/vLLM attribute issue vs Gemma4 PLE device mismatch." - }, - { - "left": "issue:45464", - "right": "issue:45685", + "left": "issue:45685", + "right": "issue:45464", "accept": false, - "reason": "Streaming inference failure is unrelated to the MPS histogram_mps int tensor error." + "reason": "MPS MoE histogram failure is unrelated to Qwen3.5 streaming inference." }, { "left": "issue:43541", "right": "issue:45685", "accept": false, - "reason": "Torch dynamo grouped_mm tracing failure is a different code path from the MPS histogram implementation error." + "reason": "Different MoE-related runtime errors with different ops and failure modes." }, { "left": "issue:43541", "right": "issue:43828", "accept": false, - "reason": "Grouped_mm tracing error under dynamo is not the same as the autocast dtype mismatch on Phi-tiny-MoE." + "reason": "Grouped_mm tracing error and autocast dtype mismatch are not the same underlying bug." }, { "left": "issue:36010", "right": "issue:39290", "accept": false, - "reason": "ImportError for GenerationMixin is unrelated to Gemma3TextConfig sliding_window_pattern errors." + "reason": "ImportError for GenerationMixin is unrelated to Gemma3 sliding_window_pattern." }, { "left": "issue:43404", "right": "issue:43866", "accept": false, - "reason": "Mistral3 lm_head tying bug is unrelated to a corrupted Ovis2 checkpoint." + "reason": "Untied lm_head weights in Mistral3 are unrelated to a corrupted Ovis2 checkpoint." }, { "left": "issue:44841", "right": "issue:45084", "accept": false, - "reason": "Processor failure for Voxtral-Mini is not the same as the non-template nodes compilation error." + "reason": "Processor failure and template-node compilation error are different issues." }, { "left": "issue:41093", "right": "issue:41762", "accept": false, - "reason": "Both are IndexError-style failures, but they affect different model-loading/training flows and do not clearly share a root cause." + "reason": "Both are IndexErrors, but they are different failure signatures and code paths." }, { "left": "issue:45399", "right": "issue:46661", "accept": false, - "reason": "Flash-attn2 fallback gating is unrelated to remote-code model_class/config_class AttributeError." + "reason": "Flash-attn fallback logic and remote-code AutoModel config_class handling are unrelated." }, { "left": "issue:43572", "right": "issue:45070", "accept": false, - "reason": "StableLmConfig pad_token_idx regression is unrelated to the pydantic PretrainedConfig field breakage." + "reason": "Missing pad_token_idx and pydantic PretrainedConfig breakage are different config regressions." }, { "left": "issue:45800", "right": "issue:45823", "accept": false, - "reason": "Both are version-related, but one is a torch/transformers compatibility issue and the other is a Gemma4 device placement bug." + "reason": "Torch/transformers incompatibility and Gemma4 device mismatch are unrelated." }, { "left": "issue:43296", "right": "issue:43366", "accept": false, - "reason": "PaddleOCR-VL load failure is not the same as GGUF support for gpt-oss." + "reason": "PaddleOCR-VL loading in vLLM and GGUF gpt-oss support are different model-support requests." }, { "left": "issue:44805", "right": "issue:44918", "accept": false, - "reason": "Similar tensor-shape symptoms, but one is a mask/index mismatch and the other is Qwen3.5 embedding unpacking with TRL SFT." + "reason": "Mask shape mismatch and Qwen3.5 embedding unpacking with TRL are different bugs." } ] }, - "evaluator_result": null, - "evaluator_used": false, - "retried": false, - "accepted_nontrivial_soft_edge": false, + "evaluator_result": { + "accept": false, + "feedback": "The cluster-level takeaway is mostly grounded, but the summary overstates the evidence by claiming acceptance of \u201cthe mask/tensor shape mismatch reports\u201d even though no soft-edge verdict in the packet accepts such a pair. The supported accepted pairs are the ZeRO-3 index-out-of-bounds pair and the AutoTokenizer wrong-token-ID pair. Also, calling issue 45020 \u201ccentral\u201d is a bit stronger than the packet justifies; it is broad, but the cluster is still quite heterogeneous. Please tighten the summary to match only the explicitly supported pairings and keep the umbrella wording more conservative." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -931,13 +826,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 3, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:39186", "issue:40990", @@ -994,8 +889,8 @@ "issue:43366|issue:45464", "issue:44291|issue:45070", "issue:43531|issue:45070", - "issue:43716|issue:46294", "issue:43541|issue:44589", + "issue:43716|issue:46294", "issue:45020|issue:45800", "issue:41093|issue:43596", "issue:44479|issue:46294", @@ -1020,14 +915,14 @@ "issue:43975|issue:44292", "issue:44292|issue:45507" ], - "prepared_review_unit_hash": "99bdc13cb38c84bde33e40ae79eb8dd6c7446afcf265a21ec4c04a25d5009a87", + "prepared_review_unit_hash": "c25c597c39fc324ddb959da42755f1286f07a7c0523fbb362a6500da02fa7eaf", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 47, @@ -1042,207 +937,207 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Conservative triage: only the Qwen3.5 video-input pair looks like the same underlying bug. The custom-module-loading pair and the validation/error-message pair are related in theme but not close enough to merge as duplicates from the evidence here.", - "confidence": 0.78, - "canonical_issue_reason": "issue:44479 is the best canonical issue in the confirmed duplicate pair because it is the broader regression report, while issue:46294 reads like a specific manifestation of the same Qwen3.5 video-token mismatch.", + "summary": "This set is mostly heterogeneous; the soft pairs are largely not true duplicates. The only plausible duplicate-like pair is the T5 input-validation error messaging issue, while the rest differ by model, symptom, or code path.", + "confidence": 0.73, + "canonical_issue_reason": "issue:45005 is the broadest issue in the set and the best anchor for the tied-weights regression subgroup, with multiple nearby soft links, whereas most other items are narrower model-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44479 is the strongest issue representative overall: it captures the broader video-input regression cluster and has the clearest duplicate counterpart in 46294.", + "best_issue_reason": "If one issue must represent the cluster, issue:45005 is the best fit because it covers the widest related regression surface and sits at the center of the closest soft ties.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44387", "right": "issue:45005", "accept": false, - "reason": "Different bugs: int4 quantization OOM vs tied-weights behavior on translation models." + "reason": "Both involve v5 regressions, but one is an int4 memory/OOM problem and the other is tied-weights behavior; different failure modes and likely different fixes." }, { "left": "issue:43541", "right": "issue:45161", "accept": false, - "reason": "Different code paths: torch dynamo grouped_mm tracing failure vs tensor-parallel support failure." + "reason": "Both mention MoE models, but one is a grouped_mm torch-dynamo tracing error and the other is a tensor-parallel support issue for GPT-OSS; not the same bug." }, { "left": "issue:43828", "right": "issue:45464", "accept": false, - "reason": "Both are mismatch-style errors, but they involve different models and different failure paths." + "reason": "One is an autocast dtype mismatch in Phi-tiny-MoE; the other is a streaming chat/completions failure in Qwen3.5-0.8B. Different code paths." }, { "left": "issue:43366", "right": "issue:45464", "accept": false, - "reason": "GGUF gpt-oss support and Qwen3.5 streaming inference are unrelated issues." + "reason": "GGUF/gpt-oss support and Qwen3.5 streaming inference are unrelated issues." }, { "left": "issue:44291", "right": "issue:45070", "accept": false, - "reason": "init_empty_weights argument handling is unrelated to the pydantic PretrainedConfig regression." + "reason": "init_empty_weights/_is_hf_initialized is a constructor-signature bug; the pydantic issue is about PretrainedConfig fields. Different regressions." }, { "left": "issue:43531", "right": "issue:45070", "accept": false, - "reason": "Qwen3-MoE sliding-window behavior and the v5 pydantic config breakage are different problems." + "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to the PretrainedConfig/pydantic breakage." }, { - "left": "issue:43716", - "right": "issue:46294", + "left": "issue:43541", + "right": "issue:44589", "accept": false, - "reason": "Different multimodal bugs: image-preprocessor dtype mismatch vs video feature/token mismatch." + "reason": "Grouped_mm tracing failure and missing Float8 storage are distinct runtime errors." }, { - "left": "issue:43541", - "right": "issue:44589", + "left": "issue:43716", + "right": "issue:46294", "accept": false, - "reason": "Different runtime failures: grouped_mm tracing error vs missing Float8 storage object." + "reason": "Both are multimodal mismatches, but one is image preprocessor dtype handling for Mistral-3 and the other is Qwen3.5 video token/feature alignment; not the same underlying bug." }, { "left": "issue:45020", "right": "issue:45800", "accept": false, - "reason": "Broad remote_code breakage and a specific torch/transformers incompatibility are not the same fix." + "reason": "Remote_code breakage across versions is not the same as a torch/transformers version incompatibility report." }, { "left": "issue:41093", "right": "issue:43596", "accept": false, - "reason": "Mask-shape mismatch and zero3/BertModel index error are distinct failures." + "reason": "Mask-length mismatch in one issue and an index-out-of-bounds error during deepspeed zero3 init in the other are different failures." }, { "left": "issue:44479", "right": "issue:46294", - "accept": true, - "reason": "Same underlying Qwen3.5 video-input regression; 46294 appears to be a specific apply_chat_template manifestation of the broader bug in 44479." + "accept": false, + "reason": "Both concern video inputs, but 44479 is a broad v5.3.0 regression across several model families while 46294 is a specific apply_chat_template token/feature mismatch." }, { "left": "issue:43054", "right": "issue:43994", "accept": false, - "reason": "Same model family, but the reported symptoms and likely root causes differ." + "reason": "Both are about siglip2, but one reports poor embedding quality and the other incorrect AutoModel/pipeline outputs; the symptoms and likely fixes differ." }, { "left": "issue:45759", "right": "issue:45902", "accept": false, - "reason": "Composite-config unwrapping and uninitialized rotary embedding memory are different bugs." + "reason": "One is about unwrapping text_config in composite multimodal configs; the other is about uninitialized memory in rotary embedding after to_empty. Different code paths." }, { "left": "issue:46640", "right": "issue:46661", "accept": false, - "reason": "Flash Attention 2 issue is unrelated to the remote-code config_class AttributeError." + "reason": "Flash Attention 2 setup/troubleshooting and remote-code config_class lookup are unrelated." }, { "left": "issue:39186", "right": "issue:44918", "accept": false, - "reason": "FSDP 2-D weight error and Qwen3.5 embedding unpacking failure are unrelated." + "reason": "FSDP 2-D weight errors and TRL SFT unpacking failures are unrelated model/runtime problems." }, { "left": "issue:43716", "right": "issue:44918", "accept": false, - "reason": "Different models and different mismatch modes; not the same bug." + "reason": "Image dtype mismatch and Qwen3.5 input-embedding unpacking are distinct multimodal issues." }, { "left": "issue:40990", "right": "issue:44960", "accept": false, - "reason": "High perplexity on gpt-oss and GLM5 are unrelated model-specific reports." + "reason": "High perplexity on gpt-oss and a GLM5 issue are unrelated model-specific reports." }, { "left": "issue:45632", "right": "issue:45698", "accept": false, - "reason": "Both concern custom-module loading, but the titles describe different failure modes; the evidence is too weak to call them duplicates." + "reason": "Both involve custom-module loading, but one is a cache-path collision for local models while the other is wrong module selection after save_pretrained; too different to merge confidently." }, { "left": "issue:43645", "right": "issue:45399", "accept": false, - "reason": "Notebook custom-model initialization breakage is unrelated to the flash-attn2 fallback gating problem." + "reason": "Custom model initialization in notebooks and flash-attn2 fallback gating are unrelated." }, { "left": "issue:43404", "right": "issue:45005", "accept": false, - "reason": "Both mention tied weights, but they involve different models and different failure modes." + "reason": "Both mention tied weights, but 43404 is a specific Mistral3 multimodal lm_head tying bug and 45005 is a broader translation-model regression; not clearly the same concrete fix." }, { "left": "issue:42915", "right": "issue:45399", "accept": false, - "reason": "FP8 config failure and flash-attn2 fallback blocking are unrelated." + "reason": "Qwen3Moe FineGrainedFP8Config failure and flash-attn2 fallback checks are different issues." }, { "left": "issue:44530", "right": "issue:45005", "accept": false, - "reason": "PagedAttentionCache linear_attention crash is unrelated to translation tied-weights issues." + "reason": "PagedAttentionCache/linear_attention crash is unrelated to tied-weights behavior." }, { "left": "issue:43873", "right": "issue:45005", "accept": false, - "reason": "Quantization/offloading memory behavior is a different problem from tied-weight handling." + "reason": "Quantization/offloading behavior and tied weights are different problem areas." }, { "left": "issue:44360", "right": "issue:44512", "accept": false, - "reason": "DSA indexer discussion and a docs typo are unrelated." + "reason": "A DSA indexer implementation comment and a docs typo are unrelated." }, { "left": "issue:44863", "right": "issue:45005", "accept": false, - "reason": "NemotronH checkpoint loading failure is unrelated to translation tied-weights issues." + "reason": "NemotronH checkpoint loading and translation-model tied weights are not the same bug." }, { "left": "issue:44492", "right": "issue:44512", "accept": false, - "reason": "A typo in cache-strategy docs is not the same as stale run-command docs." + "reason": "Cache-strategy typo and removed run-command documentation are both docs-level issues, but not the same one." }, { "left": "issue:45183", "right": "issue:46463", - "accept": false, - "reason": "Both are validation/error-message bugs, but one is generic XOR messaging and the other is T5 decoder-input validation; not enough evidence they are the same bug." + "accept": true, + "reason": "Both are about input validation for input_ids/inputs_embeds and misleading/incorrect error messages; 46463 looks like a specific T5-family instance of the broader validation bug." }, { "left": "issue:46489", "right": "issue:46491", "accept": false, - "reason": "Different tokenizer regressions on different models; not the same concrete bug." + "reason": "Both are tokenizer regressions, but one is DeepSeek-Coder v1 incorrect detokenization and the other is CodeLlama leading-space round-trip loss; different model behaviors." }, { "left": "issue:44918", "right": "issue:45507", "accept": false, - "reason": "Qwen3.5 embedding unpacking and GraniteMoEHybrid invalid-method calls are unrelated." + "reason": "Qwen3.5 input-embedding unpacking in TRL and GraniteMoEHybrid calling an invalid method are unrelated failures." }, { "left": "issue:43975", "right": "issue:44292", "accept": false, - "reason": "Tokenizer detokenization regression and init_empty_weights argument handling are unrelated." + "reason": "DeepSeek-Coder tokenization and Qwen-3-8B-NVFP4 runtime errors are unrelated." }, { "left": "issue:44292", "right": "issue:45507", "accept": false, - "reason": "Unexpected _is_hf_initialized argument and invalid method calls are different failures." + "reason": "NVFP4 runtime failure and GraniteMoEHybrid invalid-method calls are different issues." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The grouping is mostly conservative, but a couple of accepted soft edges look too strong for the evidence in the packet. In particular, issue:45632 vs issue:45698 are both about custom-module loading, but the titles point to different failure modes (leaf-directory cache collision vs wrong module after save_pretrained), so treating them as the same duplicate cluster is a stretch without body-level evidence. The same applies, more mildly, to issue:45183 vs issue:46463: both are validation/error-message bugs, but one is generic input_ids/inputs_embeds XOR messaging and the other is T5 decoder-input validation. The Qwen3.5 video pair looks fine, but because at least one soft-edge accept overstates the link, I\u2019d reject this pass." + "accept": true, + "feedback": "The summary is well grounded in the packet: it stays conservative about the soft links, correctly treats most pairs as non-duplicates, and the only accepted soft pair (issue:45183 with issue:46463) matches the validation/error-message family in the titles. The choice of issue:45005 as the anchor is also supported by its multiple nearby soft ties." }, "evaluator_used": true, - "retried": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -1252,13 +1147,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 4, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:24643", "issue:28530", @@ -1323,36 +1218,36 @@ "issue:43950|issue:45902", "issue:41553|issue:43883", "issue:44910|issue:45859", - "issue:43054|issue:43493", "issue:42947|issue:44387", + "issue:43054|issue:43493", "issue:45663|issue:45902", "issue:41553|issue:43881", "issue:44403|issue:44451", "issue:43975|issue:45520", - "issue:44403|issue:44488", "issue:42947|issue:43856", + "issue:44403|issue:44488", "issue:24643|issue:28530", "issue:43643|issue:45020", "issue:41762|issue:42915", "issue:44991|issue:45399", - "issue:45685|issue:45823", "issue:43782|issue:44960", - "issue:43881|issue:43883", + "issue:45685|issue:45823", "issue:43824|issue:43872", + "issue:43881|issue:43883", "issue:45020|issue:45823", "issue:44661|issue:45020", "issue:45081|issue:45399", "issue:42491|issue:44368", "issue:43828|issue:45823" ], - "prepared_review_unit_hash": "be244d6fe8bfc91d8633eae8095110d3efc36a6923d266f2c9473860e2f97cea", + "prepared_review_unit_hash": "ca7aefed57df90a3e9363516078f6423f35ca42112e7674798e8547b37d1749e", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, @@ -1367,226 +1262,214 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly heterogeneous. The only clearly duplicate-looking subgroup is the Transformers v5 non-persistent-buffer regression around issues 43950, 44534, and the more specific 45902. Most other soft pairs share only broad subsystem/model-family overlap and should remain separate.", - "confidence": 0.81, - "canonical_issue_reason": "Issue 43950 is the best hub for the only strong duplicate subgroup: it clearly describes the v5 non-persistent-buffer corruption regression and aligns with both the shorter duplicate report in 44534 and the more specific manifestation in 45902.", + "summary": "The packet suggests a conservative thematic cluster around Transformers v5 non-persistent-buffer corruption, but the evidence is mostly title-level similarity. I would treat the closely related reports as related-but-not-proven duplicates, with issue 43950 as the best anchor only because it states the broad regression most explicitly.", + "confidence": 0.72, + "canonical_issue_reason": "issue:43950 is the broadest and most concrete description of the v5 non-persistent-buffer corruption regression, so it is the best anchor if this cluster is being tracked conservatively.", "canonical_pr_reason": null, - "best_issue_reason": "43950 is the strongest representative issue overall because it states the underlying regression most directly and is broad enough to cover the related buffer-initialization failures.", + "best_issue_reason": "issue:43950 is the strongest representative of the buffer-corruption theme, but the packet does not firmly prove the nearby reports are true duplicates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43950", "right": "issue:44534", - "accept": true, - "reason": "Same underlying v5 non-persistent-buffer corruption regression; 44534 is a shorter rephrasing of 43950." + "accept": false, + "reason": "Very similar wording, but the packet only gives title-level similarity; that is not enough to confirm a duplicate merge." }, { "left": "issue:39692", "right": "issue:43054", "accept": false, - "reason": "Both mention SigLIP2, but one is a docs/example error report and the other is a text-embedding quality regression; different bugs." + "reason": "Both mention SigLIP2, but one is a docs/example error and the other is a model-quality regression." }, { "left": "issue:43644", "right": "issue:43950", "accept": false, - "reason": "Different failures: missing fields from trust_remote_code vs buffer corruption during from_pretrained." + "reason": "Missing fields with trust_remote_code is a different failure mode from non-persistent-buffer corruption." }, { "left": "issue:36331", "right": "issue:41762", "accept": false, - "reason": "Unrelated bugs: custom compute_loss signature change vs Gemma3 loading with DeepSpeed ZeRO-3." + "reason": "Unexpected compute_loss kwarg and DeepSpeed ZeRO-3 load index errors are different bugs." }, { "left": "issue:43828", "right": "issue:44928", "accept": false, - "reason": "Different code paths and symptoms: autocast dtype mismatch vs RLHF NaN/SDPA fallback from 3D position_ids." + "reason": "Autocast dtype mismatch and Qwen3.5 position_ids/SDPA issues are not the same code-path problem." }, { "left": "issue:43872", "right": "issue:44589", "accept": false, - "reason": "Both are compatibility/runtime errors, but one is bitsandbytes Int8Params and the other is missing Float8 storage; not the same bug." + "reason": "bitsandbytes constructor incompatibility and missing Float8 storage are unrelated compatibility errors." }, { "left": "issue:45920", "right": "issue:46489", "accept": false, - "reason": "Both concern tokenizer regressions, but 45920 is a broad multi-model wrong-token-ID report while 46489 is a specific DeepSeek-Coder v1 gap tied to PR #44801." + "reason": "Both are tokenizer regressions, but 46489 is a specific DeepSeek-Coder gap in an earlier fix, not the same broad multi-model bug." }, { "left": "issue:43054", "right": "issue:45701", "accept": false, - "reason": "Different problems: SigLIP2 text embedding quality vs tokenization behavior changes." + "reason": "Text-embedding degradation and tokenization changes are only loosely related, not duplicates." }, { "left": "issue:43866", "right": "issue:44403", "accept": false, - "reason": "Corrupted checkpoint vs generic loading noise; no shared concrete bug." + "reason": "Checkpoint corruption and noisy loading messages are different symptoms with no shared concrete bug shown." }, { "left": "issue:38175", "right": "issue:43054", "accept": false, - "reason": "Related model family, but one is unexpected zero probabilities and the other is worse text embeddings; not the same failure mode." + "reason": "Unexpected zero probabilities and worse text embeddings are different SigLIP2 problems." }, { "left": "issue:43950", "right": "issue:45902", - "accept": true, - "reason": "Same underlying non-persistent-buffer initialization/materialization regression; 45902 is a specific instance of the bug described in 43950." + "accept": false, + "reason": "Highly related buffer-corruption reports, but the packet does not provide enough direct evidence to confirm a duplicate pair." }, { "left": "issue:41553", "right": "issue:43883", "accept": false, - "reason": "Different models and root causes: bad AutoTokenizer error for Voxtral vs missing all_tied_weights_keys on MolmoForCausalLM." + "reason": "Voxtral tokenizer loading and Molmo tied-weights attribute errors are different failures." }, { "left": "issue:44910", "right": "issue:45859", "accept": false, - "reason": "Different Qwen3.5 MoE issues: flash-attention illegal access vs CPU-offload incompatibility in rotary embedding." + "reason": "FlashAttention crash from 3D position_ids and CPU-offload incompatibility are distinct Qwen3.5 issues." }, { - "left": "issue:43054", - "right": "issue:43493", + "left": "issue:42947", + "right": "issue:44387", "accept": false, - "reason": "Both are SigLIP2-related, but one is text embedding quality and the other is HF vs JAX implementation discrepancy." + "reason": "Gradient checkpointing ineffectiveness and int4 memory OOM are different performance bugs." }, { - "left": "issue:42947", - "right": "issue:44387", + "left": "issue:43054", + "right": "issue:43493", "accept": false, - "reason": "Unrelated training/memory problems: LoRA gradient checkpointing vs int4 quantization reserved-memory growth." + "reason": "SigLIP2 embedding quality and HF-vs-JAX implementation discrepancy are not the same concrete defect." }, { "left": "issue:45663", "right": "issue:45902", "accept": false, - "reason": "Different models and failure mechanisms: Gemma4 FSDP2 KeyError vs Qwen3_5Moe uninitialized memory after to_empty(cuda)." + "reason": "Gemma-4 FSDP2 shared_kv_states and Qwen3_5Moe buffer initialization are unrelated model-path bugs." }, { "left": "issue:41553", "right": "issue:43881", "accept": false, - "reason": "Different loading failures for different models; no shared concrete bug." + "reason": "Bad AutoTokenizer error for Voxtral and glm-4v-9b load failure are different issues." }, { "left": "issue:44403", "right": "issue:44451", "accept": false, - "reason": "Generic loading noise vs a specific model load failure for ScandiBERT." + "reason": "Generic loading noise and a specific ScandiBERT load failure are not the same bug." }, { "left": "issue:43975", "right": "issue:45520", "accept": false, - "reason": "Different tokenizer/import problems: DeepSeek-Coder detokenization vs flash_attn import key error on Python 3.13." + "reason": "DeepSeek detokenization regression and flash_attn import-key failure are unrelated." }, { - "left": "issue:44403", - "right": "issue:44488", + "left": "issue:42947", + "right": "issue:43856", "accept": false, - "reason": "Same general loading area, but the reports are about different models and different failures." + "reason": "Both involve training/memory, but checkpointing behavior and MoE memory inefficiency are different root causes." }, { - "left": "issue:42947", - "right": "issue:43856", + "left": "issue:44403", + "right": "issue:44488", "accept": false, - "reason": "LoRA gradient checkpointing and Qwen3 MoE memory usage are distinct training regressions." + "reason": "Same broad loading area, but the reports target different models and failure modes." }, { "left": "issue:24643", "right": "issue:28530", "accept": false, - "reason": "Training with DeepSpeed weight shape error vs early-stopping metric configuration issue." + "reason": "DeepSpeed weight-shape training error and early-stopping metric mismatch are separate trainer issues." }, { "left": "issue:43643", "right": "issue:45020", "accept": false, - "reason": "One is a specific trust_remote_code config-field omission; the other is a broad report that recent versions break remote_code models." + "reason": "Generic remote_code breakage and a specific AutoConfig missing-fields bug are not duplicates." }, { "left": "issue:41762", "right": "issue:42915", "accept": false, - "reason": "Different model families and different runtime failures." + "reason": "ZeRO-3 index-out-of-bounds during loading and FineGrainedFP8Config failure are different load/config problems." }, { "left": "issue:44991", "right": "issue:45399", "accept": false, - "reason": "Tokenizer loading for EMBEDDIA/est-roberta vs flash-attn fallback gating; unrelated." - }, - { - "left": "issue:45685", - "right": "issue:45823", - "accept": false, - "reason": "MPS histogram/int issue in MoE vs Gemma4 PLE device mismatch." + "reason": "Tokenizer loading failure and flash-attn fallback gating are unrelated." }, { "left": "issue:43782", "right": "issue:44960", "accept": false, - "reason": "Qwen3VL weight_only loading failure vs GLM5-specific issue; different code paths." + "reason": "Qwen3VL weight_only load error and GLM5 are not the same bug." }, { - "left": "issue:43881", - "right": "issue:43883", + "left": "issue:45685", + "right": "issue:45823", "accept": false, - "reason": "Different model-specific failures with no shared root cause." + "reason": "MPS histogram int error and Gemma4 device mismatch are different device/runtime issues." }, { "left": "issue:43824", "right": "issue:43872", "accept": false, - "reason": "Import/export issue for Qwen2_5_VL vs bitsandbytes parameter-construction incompatibility." + "reason": "Qwen2.5-VL import failure and bitsandbytes Int8Params kwarg mismatch are unrelated." + }, + { + "left": "issue:43881", + "right": "issue:43883", + "accept": false, + "reason": "glm-4v-9b loading failure and Molmo missing all_tied_weights_keys are different code paths." }, { "left": "issue:45020", "right": "issue:45823", "accept": false, - "reason": "Broad remote_code regression vs Gemma4 device-map bug; not the same fix." + "reason": "Remote_code regression and Gemma4 device_map forward mismatch are not the same underlying defect." }, { "left": "issue:44661", "right": "issue:45020", "accept": false, - "reason": "Tokenizer mapping-name edge case vs remote_code regression; different failures." + "reason": "Tokenizer mapping edge case and remote_code loading breakage are only superficially related." }, { "left": "issue:45081", "right": "issue:45399", "accept": false, - "reason": "Tokenizer regex crash vs flash-attn fallback gating; unrelated." - }, - { - "left": "issue:42491", - "right": "issue:44368", - "accept": false, - "reason": "Qwen3_MoE LoRA incompatibility across versions vs tie_word_embeddings warning; not the same bug." - }, - { - "left": "issue:43828", - "right": "issue:45823", - "accept": false, - "reason": "Autocast dtype mismatch in Phi-tiny-MoE vs Gemma4 device mismatch." + "reason": "Tokenizer regex crash and flash-attn fallback blocking are different bugs." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The summary is well grounded in the packet and stays conservative. The only accepted duplicate subgroup is centered on 43950, which is supported by the titles of 44534 and 45902; the rest of the soft pairs are reasonably kept separate based on distinct titles/symptoms. No PR-mergeability issue is present." + "accept": false, + "feedback": "The summary is a bit too strong for the evidence in the packet. While issue:43950, issue:44534, and issue:45902 are all thematically related to non-persistent-buffer corruption / uninitialized memory, the packet\u2019s soft-pair signal marks those pairs as deterministic_accept=false, and the analyst\u2019s hard `accept: true` verdicts go beyond what is firmly grounded from the titles alone. The claim of a \"clear duplicate cluster\" is therefore overstated; a more conservative read would keep those as related-but-not-proven duplicates." }, "evaluator_used": true, - "retried": false, - "accepted_nontrivial_soft_edge": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -1595,13 +1478,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 5, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:24643", "issue:30064", @@ -1664,13 +1547,13 @@ "issue:45020|issue:46661", "issue:43792|issue:44220", "issue:42947|issue:45923", - "issue:41553|issue:44991", "issue:43257|issue:45305", + "issue:41553|issue:44991", "issue:36331|issue:41093", "issue:43940|issue:45313", "issue:43854|issue:44863", - "issue:41553|issue:44488", "issue:44928|issue:45588", + "issue:41553|issue:44488", "issue:44479|issue:44560", "issue:41093|issue:44805", "issue:44863|issue:46402", @@ -1682,208 +1565,208 @@ "issue:43366|issue:43531", "issue:45663|issue:45823" ], - "prepared_review_unit_hash": "25456581fe419cc344769a728a19f79b57a0d9b3a96d326faf9ba5b0fb0de191", + "prepared_review_unit_hash": "dae4ded0153c02b476524e6fe52f2cba865704773679e00863dd782c8a5ab2f7", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36627, - "estimated_input_tokens": 9157, - "estimated_eval_tokens": 18570 + "serialized_chars": 36623, + "estimated_input_tokens": 9156, + "estimated_eval_tokens": 18568 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous transformers bug reports, not one unified duplicate set. I would accept only two soft edges: the two docs issues about removed v5 pipeline tasks, and the two `_is_hf_initialized` loading-TypeError reports. The rest are same-subsystem but different concrete bugs/models/paths.", - "confidence": 0.91, - "canonical_issue_reason": "No single issue is a clear cluster-wide canonical because the items span many unrelated bugs; the only strong duplicate subgroup is the docs/pipeline-removal pair, where issue 43827 is the cleaner anchor.", + "summary": "This set is mostly heterogeneous Transformer regression reports, not one tight duplicate cluster. Only a few soft pairs look like true same-bug duplicates: the removed-pipeline docs issue, the `_is_hf_initialized` loading regression, the identical mask-shape IndexError, and likely the DeepSeek tokenizer regression pair.", + "confidence": 0.64, + "canonical_issue_reason": "There is no single clean canonical across the whole batch; if one must be chosen, issue 43827 is the best anchor because it is part of the clearest duplicate pair and describes the docs regression most concretely.", "canonical_pr_reason": null, - "best_issue_reason": "No global best issue stands out for the whole cluster; it is too broad and mixed. Issue 43827 is the best representative only for the documentation subgroup, not for the full set.", + "best_issue_reason": "Issue 43827 is the strongest overall issue to anchor on because it is concrete, reproducible, and has the clearest one-to-one duplicate relationship in this set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45715", "right": "issue:46315", "accept": false, - "reason": "Related tokenizer performance regressions, but one is `convert_ids_to_tokens` and the other is `convert_tokens_to_ids`; different code paths and likely different fixes." + "reason": "Related tokenizer performance regressions, but one is about `convert_ids_to_tokens` and the other about `convert_tokens_to_ids`; different code paths." }, { "left": "issue:45020", "right": "issue:45356", "accept": false, - "reason": "Both are recent loading regressions, but one is generic `remote_code` breakage and the other is a Kimi-K2.5 tokenizer codec issue; not the same bug." + "reason": "Both are broad tokenizer/model regressions, but the symptoms and likely fixes are different (`remote_code` breakage vs Kimi tokenizer codec handling)." }, { "left": "issue:45488", "right": "issue:46489", - "accept": false, - "reason": "Both involve DeepSeek tokenizers in v5, but the reported failures differ: a LlamaTokenizer pre-tokenizer override vs a DeepSeek-Coder v1 output regression. Too broad to treat as one bug." + "accept": true, + "reason": "Both point to the same DeepSeek tokenizer regression in v5: incorrect handling of tokenizer pre-tokenization/output for the DeepSeek family." }, { "left": "issue:43827", "right": "issue:44509", "accept": true, - "reason": "Same underlying docs cleanup: both report stale references to removed v5 pipeline tasks in documentation, and one docs-fix could plausibly cover both." + "reason": "Same docs bug: v5 removed pipeline tasks, but the documentation still references summarization/translation/text2text-generation pipelines." }, { "left": "issue:43404", "right": "issue:43950", "accept": false, - "reason": "One is a model-specific weight-tying bug; the other is a general `from_pretrained` buffer regression. Different mechanisms and fix scopes." + "reason": "Both involve model loading/state handling, but one is a tied-weight bug in Mistral3 and the other is a non-persistent buffer corruption regression." }, { "left": "issue:45759", "right": "issue:46661", "accept": false, - "reason": "Different AutoModel failures: composite multimodal config unwrapping vs remote-code `config_class` lookup. Same API surface, different root causes." + "reason": "Different auto-model failures: composite multimodal config unwrapping vs remote-code `config_class` lookup." }, { "left": "issue:43872", "right": "issue:44291", "accept": true, - "reason": "Same `_is_hf_initialized` incompatibility theme: both reports are the loading path passing an unexpected kwarg into object construction, causing TypeError failures." + "reason": "Same root cause: the new `_is_hf_initialized` keyword breaks model materialization/loading paths." }, { "left": "issue:42915", "right": "issue:43872", "accept": false, - "reason": "Both mention loading/training compatibility, but one is Qwen3Moe with FineGrainedFP8Config and the other is a bitsandbytes `_is_hf_initialized` constructor mismatch. Different bugs." + "reason": "Both mention loading/quantization incompatibilities, but the model/config and failure mode are different." }, { "left": "issue:45020", "right": "issue:46661", "accept": false, - "reason": "Both concern model/tokenizer loading, but one is remote-code breakage and the other is missing `config_class` on `AutoModel.from_pretrained`. Not the same defect." + "reason": "`remote_code` regressions are too broad here; these are different concrete failures and not the same bug." }, { "left": "issue:43792", "right": "issue:44220", "accept": false, - "reason": "Whisper runtime failure vs a low-level fbank feature helper issue; related area, but not clearly the same underlying bug." + "reason": "Whisper failing to run and `_torch_extract_fbank_features()` are related, but the reported problems are not clearly the same underlying bug." }, { "left": "issue:42947", "right": "issue:45923", "accept": false, - "reason": "Both mention gradient checkpointing, but one is PEFT LoRA effectiveness and the other is a missing flag on a trust-remote-code model. Different fixes." + "reason": "Gradient checkpointing issues, but one is ineffective LoRA checkpointing and the other is a missing support flag on a remote-code variant." }, { - "left": "issue:41553", - "right": "issue:44991", + "left": "issue:43257", + "right": "issue:45305", "accept": false, - "reason": "Both are tokenizer-loading problems, but they affect different models and appear to stem from different validation/parser issues." + "reason": "Both involve DeepSpeed, but one is a weight-conversion failure and the other is a gradient-averaging bug." }, { - "left": "issue:43257", - "right": "issue:45305", + "left": "issue:41553", + "right": "issue:44991", "accept": false, - "reason": "Both involve DeepSpeed, but one is weight conversion/loading and the other is gradient averaging under GAS. Different code paths and symptoms." + "reason": "Both are tokenizer-loading problems, but they affect different models and different failure causes." }, { "left": "issue:36331", "right": "issue:41093", "accept": false, - "reason": "Unrelated errors: a custom trainer signature mismatch vs a tensor/mask shape IndexError." + "reason": "Different training-time errors: custom `compute_loss` signature mismatch vs tensor/mask shape mismatch." }, { "left": "issue:43940", "right": "issue:45313", "accept": false, - "reason": "Both are ZeRO-3 weight-loading failures, but they reference different Qwen variants and different missing components (`all params` vs `language_model`). Similar class, not same concrete bug." + "reason": "Both are ZeRO-3 weight-loading complaints for Qwen models, but the model family and missing-weights pattern differ." }, { "left": "issue:43854", "right": "issue:44863", "accept": false, - "reason": "Different models and failure modes; both are checkpoint-loading issues, but not the same underlying defect." + "reason": "Different model-loading failures for unrelated architectures; no clear shared bug." }, { - "left": "issue:41553", - "right": "issue:44488", + "left": "issue:44928", + "right": "issue:45588", "accept": false, - "reason": "Both are tokenizer-loading reports, but for different models and likely different parser/pretokenizer problems." + "reason": "Different runtime paths and symptoms: RLHF gradient explosion vs flash-attention auxiliary-state crash." }, { - "left": "issue:44928", - "right": "issue:45588", + "left": "issue:41553", + "right": "issue:44488", "accept": false, - "reason": "RLHF NaN/position-id instability vs a flash_attention `s_aux=None` crash. Same broad area of attention code, different bug." + "reason": "Both are tokenizer/model loading issues, but they are different models and different errors." }, { "left": "issue:44479", "right": "issue:44560", "accept": false, - "reason": "Both are Qwen video regressions in v5.3.0, but one is a broad family regression and the other is a specific StopIteration in Qwen3-vl-embedding; not enough evidence they are the same bug." + "reason": "Different multimodal regressions: video input handling vs Qwen3-vl-embedding video StopIteration." }, { "left": "issue:41093", "right": "issue:44805", - "accept": false, - "reason": "The titles are near-identical shape-mismatch errors, but without evidence they share the same exact reproducer or root cause, I would not merge them as duplicates." + "accept": true, + "reason": "Same concrete mask/tensor shape mismatch error; the later report is essentially a rephrased duplicate." }, { "left": "issue:44863", "right": "issue:46402", "accept": false, - "reason": "Both are model-loading problems, but for different architectures (NemotronH vs Sam3Model) and different missing-weight behaviors." + "reason": "Different checkpoint-loading failures for different model families and different missing components." }, { "left": "issue:24643", "right": "issue:30064", "accept": false, - "reason": "Completely different issues: DeepSpeed training `'weight' must be 2-D` vs image processor void segmentation maps." + "reason": "Unrelated issues: DeepSpeed training weight-shape error vs image processor handling of void segmentation maps." }, { "left": "issue:44530", "right": "issue:44863", "accept": false, - "reason": "PagedAttentionCache linear_attention crash vs NemotronH checkpoint loading; different subsystems and symptoms." + "reason": "Different model-specific failures; one is a PagedAttentionCache crash, the other is a checkpoint loading problem." }, { "left": "issue:43366", "right": "issue:43828", "accept": false, - "reason": "GGUF gpt-oss architecture support vs an autocast dtype mismatch on Phi-tiny-MoE; unrelated bugs." + "reason": "Unrelated: GGUF architecture support vs autocast dtype mismatch in a different model." }, { "left": "issue:43596", "right": "issue:46515", "accept": false, - "reason": "Zero3 init/BertModel out-of-bounds index vs torch 2.4.1 incompatibility; different failure sources." + "reason": "A ZeRO-3/BERT initialization failure is not the same as a general torch compatibility report." }, { "left": "issue:45325", "right": "issue:45902", "accept": false, - "reason": "Both touch Qwen3.5/VL internals, but one is rope-index scaling for still images and the other is uninitialized `inv_freq` after materialization. Different defects." + "reason": "Both touch rotary/position handling, but they describe different bugs and different model families." }, { "left": "issue:43366", "right": "issue:43531", "accept": false, - "reason": "Qwen3-MoE GGUF support vs Qwen3-MoE sliding_window behavior; separate changes." + "reason": "Both mention Qwen/Qwen-MoE-adjacent models, but the actual bugs are unrelated." }, { "left": "issue:45663", "right": "issue:45823", "accept": false, - "reason": "Gemma4 FSDP2 shared_kv_states bug vs Gemma4 PLE device mismatch with `device_map=auto`; same model family, different runtime issue." + "reason": "Same family (Gemma4) but different failures: shared KV state construction vs device placement under `device_map='auto'`." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet and stays conservative about cluster-wide canonicals. The two accepted soft edges are plausible duplicate/mergeable bugfix pairs: the docs pipeline-removal issues and the shared `_is_hf_initialized` loading TypeError theme. The rejected pairs are also appropriately cautious and do not overclaim duplication." + "feedback": "The summary is grounded in the packet and stays appropriately cautious about the batch being heterogeneous. The accepted soft pairs are supported by the issue titles, especially the duplicate-looking docs regression, the `_is_hf_initialized` loading bug, the identical mask-shape error, and the likely DeepSeek tokenizer regression. The negative verdicts also read conservative rather than overclaimed." }, "evaluator_used": true, "retried": false, @@ -1896,13 +1779,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 6, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:36010", "issue:38175", @@ -1955,8 +1838,8 @@ ], "soft_pairs": [ "issue:43638|issue:46515", - "issue:43716|issue:44560", "issue:44155|issue:45584", + "issue:43716|issue:44560", "issue:43950|issue:45834", "issue:43366|issue:44589", "issue:43824|issue:43957", @@ -1989,20 +1872,20 @@ "issue:46459|issue:46512", "issue:44960|issue:44977" ], - "prepared_review_unit_hash": "d66605d3b49029924fcc0cb5e640e944c9a2882bdb72a1846b2481d9de7d307a", + "prepared_review_unit_hash": "d1c330c7fe78edcc382690d798508f9306e5869d8ccde5b2467032546af50630", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 37776, + "serialized_chars": 37775, "estimated_input_tokens": 9444, "estimated_eval_tokens": 19144 }, @@ -2011,90 +1894,90 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a loose collection of Transformers regression reports with several false-positive similarity links. The only clear duplicate subthreads are the Transformers v5 non-persistent-buffer corruption reports, the SigLIP2 bad-output reports, and the custom-model-init regressions.", - "confidence": 0.76, - "canonical_issue_reason": "Issue 43644 is the strongest representative: it describes a concrete Transformers v5 regression, is specific enough to search, and has an almost identical duplicate (44534) confirming the same bug.", + "summary": "The set is mostly heterogeneous, with only a few clear duplicate mini-clusters: the non-persistent-buffer regression, the Transformers 5 custom-model init regression, and the SigLIP2 inference regression. Most other pairs are related only at a broad subsystem/model-family level and should not be merged.", + "confidence": 0.82, + "canonical_issue_reason": "No single issue cleanly represents the whole cluster; the strongest duplicate-style issue is 43644 because it has an exact near-title duplicate and a precise regression description.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43644 is the best overall issue fit because it most clearly captures a duplicated, concrete code-path regression rather than a model-specific symptom or a broader compatibility complaint.", + "best_issue_reason": "43644 is the best representative issue in this set: it states the regression clearly and has the closest duplicate match (44534) with the same underlying buffer-corruption bug.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43638", "right": "issue:46515", "accept": false, - "reason": "Different failures: Deepspeed ZeRO3/BertModel index error vs a torch 2.4.1 incompatibility report." + "reason": "Both mention failures after upgrades, but one is a Deepspeed/BERT zero3 init IndexError and the other is a generic torch 2.4.1 incompatibility; no shared concrete bug." }, { - "left": "issue:43716", - "right": "issue:44560", + "left": "issue:44155", + "right": "issue:45584", "accept": false, - "reason": "Unrelated multimodal bugs: image preprocessor dtype mismatch vs Qwen3-VL video StopIteration." + "reason": "Different models and failure modes: AudioFlamingo3 track leakage vs Whisper empty-transcription handling after token alignment." }, { - "left": "issue:44155", - "right": "issue:45584", + "left": "issue:43716", + "right": "issue:44560", "accept": false, - "reason": "Different model paths and symptoms: AudioFlamingo3 track leakage vs Whisper empty transcription handling." + "reason": "Both are multimodal issues, but they affect different models and code paths (SigLIP2 preprocessing mismatch vs Qwen3-VL video StopIteration)." }, { "left": "issue:43950", "right": "issue:45834", "accept": false, - "reason": "Both are loading bugs, but the underlying failures are unrelated: buffer corruption vs long OCR index error." + "reason": "Unrelated regressions: non-persistent buffer corruption during loading vs Kosmos2.5 long-OCR index error." }, { "left": "issue:43366", "right": "issue:44589", "accept": false, - "reason": "No shared bug: GGUF/gpt-oss support request vs Float8 storage lookup failure." + "reason": "GGUF/gpt-oss support and Float8 storage lookup are unrelated loader/runtime issues." }, { "left": "issue:43824", "right": "issue:43957", "accept": false, - "reason": "Different loading problems: missing exported class vs meta-device loading breakage." + "reason": "An import/export symbol regression is not the same bug as meta-device loading failures across some models." }, { "left": "issue:44560", "right": "issue:45685", "accept": false, - "reason": "Unrelated issues: Qwen3-VL video inference error vs MPS histogram not implemented for Int." + "reason": "Different architectures and errors: Qwen3-VL video inference StopIteration vs an MPS histogram op missing for Int." }, { "left": "issue:42915", "right": "issue:43957", "accept": false, - "reason": "Different code paths: Qwen3Moe FP8 failure vs meta-device loading issue." + "reason": "Qwen3Moe FP8 config failure and meta-device loading failures are separate bugs." }, { "left": "issue:43645", "right": "issue:43646", "accept": true, - "reason": "Same underlying regression: Transformers v5 breaks custom model initialization; one is just a notebook-specific phrasing." + "reason": "Same underlying Transformers v5 custom-model initialization regression; 43645 is a notebook-specific manifestation of 43646." }, { "left": "issue:43550", "right": "issue:44910", "accept": false, - "reason": "Different attention/compile failures with different root causes and model setups." + "reason": "Different attention/compile failures in different models; no shared code-path bug is evident." }, { "left": "issue:44560", "right": "issue:44918", "accept": false, - "reason": "Both involve Qwen models, but the bugs are different: video StopIteration vs SFT embedding unpacking." + "reason": "Both involve Qwen-family multimodal behavior, but one is a video embedding StopIteration and the other is TRL SFT embedding unpacking." }, { "left": "issue:44683", "right": "issue:46515", "accept": false, - "reason": "Different incompatibilities: torch>=2.9 compiled flex_attention vs torch 2.4.1." + "reason": "Torch-version incompatibility, but on different features and releases; not the same concrete problem." }, { "left": "issue:42617", "right": "issue:43366", "accept": false, - "reason": "Unrelated: 3d_parallel.py runtime problem vs GGUF gpt-oss support." + "reason": "3D parallel script execution and GGUF architecture support are unrelated." }, { "left": "issue:45812", @@ -2106,127 +1989,121 @@ "left": "issue:36010", "right": "issue:42915", "accept": false, - "reason": "Different breakages: GenerationMixin import error vs Qwen3Moe FP8 loading failure." + "reason": "ImportError for GenerationMixin is unrelated to Qwen3Moe FP8 execution failure." }, { "left": "issue:45823", "right": "issue:46515", "accept": false, - "reason": "Different symptoms: Gemma4 device-map mismatch vs a torch version incompatibility claim." + "reason": "Gemma4 device-map mismatch and a generic torch incompatibility are not the same bug." }, { "left": "issue:44560", "right": "issue:44805", "accept": false, - "reason": "Different Qwen3-VL failures: video StopIteration vs mask/tensor shape mismatch." + "reason": "Video StopIteration and mask/tensor shape mismatch are different runtime failures." }, { "left": "issue:44661", "right": "issue:45823", "accept": false, - "reason": "Different problems: add-new-model-like/tokenizer mapping issue vs Gemma4 PLE device mismatch." + "reason": "Tokenizer-mapping validation and Gemma4 device placement are unrelated." }, { "left": "issue:38175", "right": "issue:43994", "accept": true, - "reason": "Same SigLIP2 model regression: both describe bad/degenerate inference outputs for google/siglip2-base-patch16-224." + "reason": "Same SigLIP2 base-patch16-224 inference regression: zero probabilities / nonsensical outputs on the same model indicate the same underlying bug." }, { "left": "issue:43632", "right": "issue:45834", "accept": false, - "reason": "Different regressions: _is_hf_initialized flag breakage vs long OCR index error." + "reason": "The `_is_hf_initialized` regression and Kosmos2.5 OCR index error are unrelated." }, { "left": "issue:44877", "right": "issue:45030", "accept": false, - "reason": "Both involve config validation, but the affected models and failure modes differ." + "reason": "Both are config-validation failures, but they concern different models and there is no evidence of a shared root cause." }, { "left": "issue:43638", "right": "issue:44661", "accept": false, - "reason": "Different subsystems: ZeRO3/BertModel initialization vs tokenizer mapping constraint." + "reason": "Deepspeed zero3/BERT initialization and TOKENIZER_MAPPING_NAMES handling are different issues." }, { "left": "issue:43596", "right": "issue:44661", "accept": false, - "reason": "Different failures: empty-sized tensor index error under ZeRO3 vs add-new-model-like mapping issue." + "reason": "A zero3/BERT IndexError is not the same bug as add-new-model-like failing in tokenizer mapping names." }, { "left": "issue:43317", "right": "issue:44756", "accept": false, - "reason": "No shared bug: offload/device_map loading issue vs mmap OOM workaround." - }, - { - "left": "issue:44661", - "right": "issue:44805", - "accept": false, - "reason": "Different root causes: tokenizer/model registration rule vs mask shape mismatch." + "reason": "Device-map offload loading failure and mmap/OOM on Strix Halo are different memory/loading problems." }, { "left": "issue:42491", "right": "issue:43720", "accept": false, - "reason": "Different areas: qwen3_moe LoRA compatibility vs BitNet packed-weight loading." + "reason": "Qwen3_moe LoRA compatibility across versions and BitNet packed-weight loading are unrelated." }, { "left": "issue:38175", "right": "issue:39692", "accept": false, - "reason": "Same model family, but one is an inference-output regression and the other is a documentation/example mismatch plus quantization failure." + "reason": "Same model family, but one is an inference symptom and the other is a documentation/example issue with additional quantization failure; not clearly the same bug." }, { "left": "issue:43644", "right": "issue:44534", "accept": true, - "reason": "Near-identical duplicate reports of the same Transformers v5 non-persistent-buffer corruption bug." + "reason": "Exact same Transformers v5 non-persistent-buffer regression, just reworded." }, { "left": "issue:41720", "right": "issue:42915", "accept": false, - "reason": "Different Qwen failures with different triggers: CUDA assert on device mapping vs FP8 config failure." + "reason": "Both are Qwen-related, but one is auto device mapping CUDA assert and the other is FP8 config failure." }, { "left": "issue:42491", "right": "issue:45276", "accept": false, - "reason": "Unrelated bugs in different models: qwen3_moe LoRA incompatibility vs Gemma4 embedding resizing." + "reason": "LoRA/Qwen3_moe version compatibility and Gemma4 resize_token_embeddings behavior are unrelated." }, { "left": "issue:42915", "right": "issue:44756", "accept": false, - "reason": "No overlap in root cause: Qwen3Moe FP8 vs Strix Halo mmap OOM." + "reason": "Qwen3Moe FP8 failure and Strix Halo mmap OOM are unrelated." }, { "left": "issue:44451", "right": "issue:44488", "accept": false, - "reason": "Both are model-load failures, but there is not enough evidence they share the same concrete bug or code path." + "reason": "Both report loading failures for BERT-like models, but the titles do not show the same underlying regression or code path." }, { "left": "issue:46459", "right": "issue:46512", "accept": false, - "reason": "Related dtype-handling theme, but different APIs and symptoms: AutoModelForCausalLM loading vs from_config config-state staleness." + "reason": "Both involve dtype handling, but one is ignored dtype on composite checkpoint loading and the other is stale config dtype after from_config; different bugs." }, { "left": "issue:44960", "right": "issue:44977", "accept": false, - "reason": "Different model-specific failures: GLM5 issue vs Qwen3.5 flash-attention generation bug." + "reason": "GLM5 and Qwen3.5 flash-attention generation failures are separate model-specific problems." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet, and the soft-edge verdicts stay conservative overall. The chosen canonical issue, 43644, is strongly supported by the near-identical duplicate 44534, and the other accepted pairs are plausibly same-bug pairs based on their titles. No clear overstatement or mergeability problem is evident here." + "feedback": "The cluster summary is grounded in the packet and stays conservative about soft matches. The accepted duplicates (43645/43646, 38175/43994, 43644/44534) are plausible from the titles and descriptions, and the rejected pairs mostly avoid overclaiming shared root cause where the evidence is only broad subsystem or model-family similarity." }, "evaluator_used": true, "retried": false, @@ -2239,13 +2116,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 7, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:36010", "issue:39404", @@ -2299,8 +2176,8 @@ "issue:45313|issue:45357", "issue:45507|issue:46468", "issue:44488|issue:44991", - "issue:43644|issue:43645", "issue:43856|issue:43873", + "issue:43644|issue:43645", "issue:39692|issue:43994", "issue:43708|issue:44075", "issue:43856|issue:44387", @@ -2310,238 +2187,241 @@ "issue:42491|issue:43931", "issue:44960|issue:46050", "issue:45488|issue:45812", - "issue:45362|issue:45406", "issue:45276|issue:45335", "issue:44164|issue:44291", + "issue:45362|issue:45406", "issue:44910|issue:45588", "issue:44402|issue:44488", "issue:43404|issue:43454", + "issue:42947|issue:45704", "issue:39692|issue:44960", + "issue:43873|issue:44387", "issue:41720|issue:44155", - "issue:42947|issue:45704", - "issue:43873|issue:44387", - "issue:45663|issue:45759", "issue:43425|issue:43957", + "issue:45663|issue:45759", "issue:44291|issue:44756", "issue:36010|issue:39404", "issue:45663|issue:45859", "issue:44387|issue:44756" ], - "prepared_review_unit_hash": "c609fe27b1fb0c60f073ebef8523bd3f288b572f3cf455327fda29c95ecae362", + "prepared_review_unit_hash": "fbfd6a95b5a37acdc81e5b1339b08b6ff354e4a05194820aa552eacfe4e840b2", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 36204, - "estimated_input_tokens": 9051, - "estimated_eval_tokens": 18358 + "serialized_chars": 36207, + "estimated_input_tokens": 9052, + "estimated_eval_tokens": 18360 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: mostly unrelated Transformers bugs across different models, APIs, and failure modes. None of the soft-similar pairs look like the same underlying bug/change, so I would not merge any of them as duplicates.", - "confidence": 0.95, - "canonical_issue_reason": "No canonical issue: the items describe separate bugs in different models and code paths, so there isn't one issue that subsumes the rest.", + "summary": "Most candidates are thematically related but not true duplicates. The only clear duplicate pair is the resize_token_embeddings regression across Gemma4/T5Gemma; the rest differ in model family, code path, or failure mode.", + "confidence": 0.74, + "canonical_issue_reason": "issue:45276 is the best canonical issue: it describes the broader resize_token_embeddings propagation bug and cleanly covers the t5gemma variant.", "canonical_pr_reason": null, - "best_issue_reason": "No good global representative; the cluster is too heterogeneous for a single issue to anchor triage.", + "best_issue_reason": "issue:45276 is the strongest representative issue overall because it is the most general statement of the shared embedding-resize bug.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45313", "right": "issue:45357", "accept": false, - "reason": "Both involve Qwen3.5, but one is ZeRO-3 loading of language_model and the other is save_pretrained key mapping for visual encoder; different code paths." + "reason": "Same model family, but one is a ZeRO-3 weight-loading failure and the other is a save_pretrained key regression; different code paths and symptoms." }, { "left": "issue:45507", "right": "issue:46468", "accept": false, - "reason": "Different models and failures: GraniteMoEHybrid method dispatch vs a broken Nemotron example." + "reason": "Different models and failures; too broad to be the same underlying bug." }, { "left": "issue:44488", "right": "issue:44991", "accept": false, - "reason": "Both are tokenizer/model loading failures, but they affect different models and likely different root causes." + "reason": "Both concern tokenizer/model loading in v5, but they affect different models and likely different tokenizer failure modes." }, { - "left": "issue:43644", - "right": "issue:43645", + "left": "issue:43856", + "right": "issue:43873", "accept": false, - "reason": "One is junk-filled non-persistent buffers; the other is Jupyter custom model initialization. Same release, different bugs." + "reason": "Both are memory-related, but one is Qwen3 MoE training memory usage and the other is quantization/offloading behavior; not the same bug." }, { - "left": "issue:43856", - "right": "issue:43873", + "left": "issue:43644", + "right": "issue:43645", "accept": false, - "reason": "Both mention memory/offloading, but one is Qwen3 MoE training memory usage and the other is quantization offloading behavior; not the same defect." + "reason": "Both are Transformers 5.0 regressions, but one is buffer initialization and the other is notebook custom model init; different issues." }, { "left": "issue:39692", "right": "issue:43994", "accept": false, - "reason": "Both concern SigLIP2, but one is a bad documentation example and the other is wrong outputs in AutoModel/pipeline." + "reason": "Both mention SigLIP2, but one is a documentation example with multiple errors and the other is incorrect outputs in AutoModel/pipeline usage." }, { "left": "issue:43708", "right": "issue:44075", "accept": false, - "reason": "Trainer checkpoint resumption math is unrelated to SGD optimizer args not being used." + "reason": "Trainer checkpoint math and SGD argument handling are unrelated." }, { "left": "issue:43856", "right": "issue:44387", "accept": false, - "reason": "Both are memory regressions, but one is training memory efficiency and the other is int4 CUDA reserved memory/OOM." + "reason": "Both involve memory, but one is training-time inefficiency and the other is int4 reserved-memory/OOM regression; not the same bug." }, { "left": "issue:43257", "right": "issue:43854", "accept": false, - "reason": "Qwen3 MoE weight conversion under accelerate+deepspeed is unrelated to GLM-4.7-Flash unit-test loading failures." + "reason": "Different models and load paths; no concrete shared bug is evident." }, { "left": "issue:43854", "right": "issue:43931", "accept": false, - "reason": "Different model families and different load-time failures; no shared concrete bug." + "reason": "Both are model-loading problems, but for different Qwen/GLM-VL models and different failure signatures." }, { "left": "issue:42491", "right": "issue:45292", "accept": false, - "reason": "LoRA/qwen3_moe compatibility on hf5.x is unrelated to resize_token_embeddings not updating output embeddings." + "reason": "LoRA/Qwen3 MoE compatibility is unrelated to resize_token_embeddings behavior." }, { "left": "issue:42491", "right": "issue:43931", "accept": false, - "reason": "These are different Qwen-related load problems with different symptoms and code paths." + "reason": "Different models and failures; not the same bug." }, { "left": "issue:44960", "right": "issue:46050", "accept": false, - "reason": "GLM5 is just a model-specific issue; the other is a save/load regression for sinq-quantized models." + "reason": "GLM5 loading/model issue vs sinq quantization persistence bug; unrelated." }, { "left": "issue:45488", "right": "issue:45812", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different tokenizer implementations and model families." - }, - { - "left": "issue:45362", - "right": "issue:45406", - "accept": false, - "reason": "Qwen3.5 chat crash and Gemma4 serve processor attribute error are distinct inference-stack bugs." + "reason": "Both are tokenizer regressions, but they involve different tokenizer classes, models, and failure modes." }, { "left": "issue:45276", "right": "issue:45335", - "accept": false, - "reason": "Similar symptom around resize_token_embeddings, but different model architectures and target modules; not one concrete duplicate." + "accept": true, + "reason": "Near-paraphrase of the same bug: resize_token_embeddings does not propagate to model-specific embedding/output modules in both Gemma4 and T5Gemma." }, { "left": "issue:44164", "right": "issue:44291", "accept": false, - "reason": "save/from_pretrained extra_state handling is unrelated to init_empty_weights passing _is_hf_initialized." + "reason": "save/from_pretrained extra_state handling and init_empty_weights argument compatibility are separate loading bugs." + }, + { + "left": "issue:45362", + "right": "issue:45406", + "accept": false, + "reason": "Chat crash on Qwen3.5-35B and Gemma4Processor missing _tokenizer in serve are different subsystems and errors." }, { "left": "issue:44910", "right": "issue:45588", "accept": false, - "reason": "Both hit flash_attention, but one is packed-sequence/position_ids misuse and the other is an s_aux=None attribute error." + "reason": "Both touch flash_attention, but one is a position_ids packing crash and the other is an s_aux=None AttributeError; different code paths." }, { "left": "issue:44402", "right": "issue:44488", "accept": false, - "reason": "One is a vocab-size/token placement issue; the other is a failure to load a specific model. Different root causes." + "reason": "Different models and different symptoms; not enough evidence of the same root cause." }, { "left": "issue:43404", "right": "issue:43454", "accept": false, - "reason": "Both mention lm_head tying, but they are separate model implementations with different affected behavior." - }, - { - "left": "issue:39692", - "right": "issue:44960", - "accept": false, - "reason": "SigLIP2 documentation/model loading problems are unrelated to GLM5." + "reason": "Both mention untied lm_head weights, but they target different model classes and lack evidence of a shared concrete code-path bug." }, { - "left": "issue:41720", - "right": "issue:44155", + "left": "issue:42947", + "right": "issue:45704", "accept": false, - "reason": "Qwen3 device-mapping CUDA assert and AudioFlamingo3 cross-track token leakage are unrelated." + "reason": "Gradient checkpointing/LoRA behavior and an apex RMSNorm memory leak are unrelated." }, { - "left": "issue:42947", - "right": "issue:45704", + "left": "issue:39692", + "right": "issue:44960", "accept": false, - "reason": "Gradient checkpointing with PEFT LoRA and T5 apex RMSNorm memory leak are different subsystems and bugs." + "reason": "Unrelated model families and failure modes." }, { "left": "issue:43873", "right": "issue:44387", "accept": false, - "reason": "Both are quantization/memory related, but one is offloading behavior and the other is int4 reserved-memory growth leading to OOM." + "reason": "Both are quantization/memory themed, but one is offloading not working and the other is increased reserved memory/OOM; not the same bug." }, { - "left": "issue:45663", - "right": "issue:45759", + "left": "issue:41720", + "right": "issue:44155", "accept": false, - "reason": "Gemma-4 FSDP2 shared_kv_states KeyError is unrelated to Qwen multimodal config unwrapping in from_config." + "reason": "Qwen3 auto device mapping CUDA assert and AudioFlamingo3 batched inference leakage are unrelated." }, { "left": "issue:43425", "right": "issue:43957", "accept": false, - "reason": "Torch version incompatibility is not the same as meta-device model loading breakage." + "reason": "Torch 2.10 incompatibility and meta-device loading regressions are distinct compatibility issues." + }, + { + "left": "issue:45663", + "right": "issue:45759", + "accept": false, + "reason": "Different models and different failure mechanisms; not the same underlying bug." }, { "left": "issue:44291", "right": "issue:44756", "accept": false, - "reason": "Unexpected _is_hf_initialized during init_empty_weights is unrelated to mmap OOM on Strix Halo." + "reason": "init_empty_weights loading error and Strix Halo mmap OOM are unrelated." }, { "left": "issue:36010", "right": "issue:39404", "accept": false, - "reason": "ImportError for GenerationMixin and Whisper return_language pipeline regression are different problems." + "reason": "ImportError for GenerationMixin and Whisper return_language pipeline regression are different issues." }, { "left": "issue:45663", "right": "issue:45859", "accept": false, - "reason": "Both involve Qwen/Gemma model internals, but one is FSDP2 KeyError and the other is CPU-offload rotary embedding incompatibility." + "reason": "Different Qwen/Gemma model code paths and different offload/attention issues." }, { "left": "issue:44387", "right": "issue:44756", "accept": false, - "reason": "Both mention memory pressure, but one is quantization reserved-memory growth and the other is mmap-related OOM on Strix Halo." + "reason": "Both are memory problems, but one is int4 quantization reserved-memory growth and the other is mmap on Strix Halo; not the same bug." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The main duplicate call is well grounded by the titles: both 45276 and 45335 describe resize_token_embeddings not propagating to model-specific embedding/output modules. The remaining pairwise rejections are appropriately conservative from the packet evidence, with no obvious overstatement of shared root cause. " + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -2550,13 +2430,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 8, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:39404", "issue:42796", @@ -2620,8 +2500,8 @@ "issue:43381|issue:44928", "issue:44164|issue:44756", "issue:45684|issue:45698", - "issue:45588|issue:45663", "issue:43854|issue:45216", + "issue:45588|issue:45663", "issue:42796|issue:46326", "issue:45292|issue:45335", "issue:43646|issue:43708", @@ -2636,244 +2516,238 @@ "issue:43611|issue:43645", "issue:39404|issue:46665", "issue:44530|issue:45084", - "issue:43425|issue:45823", "issue:44683|issue:45823", + "issue:43425|issue:45823", "issue:43425|issue:43716", "issue:44479|issue:44805", "issue:43906|issue:44530" ], - "prepared_review_unit_hash": "535018b8c2d61493c0b1fabb2d7ca89d0a5683c1198f4d590bb0b2642985c401", + "prepared_review_unit_hash": "c1d63970f65c4989610e2fa1b8d0abe1e1e2e779da60fa19ff839300eb23eb2d", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 33, - "serialized_chars": 37585, - "estimated_input_tokens": 9397, - "estimated_eval_tokens": 19050 + "serialized_chars": 37582, + "estimated_input_tokens": 9396, + "estimated_eval_tokens": 19048 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Nearly all candidate pairs are different bugs or different model-specific failures. The only plausible duplicate cluster is the `resize_token_embeddings` regression reports, which appear to describe the same embedding-propagation bug across model variants.", - "confidence": 0.92, - "canonical_issue_reason": "issue:45292 is the cleanest generic report of the `resize_token_embeddings` regression affecting output embeddings; the related Gemma4/T5Gemma reports read like model-specific manifestations of the same underlying bug.", + "summary": "Most items are unrelated. The only clear duplicate subcluster is the `resize_token_embeddings` regression, where a generic report and two model-specific reports point to the same embedding-update bug.", + "confidence": 0.73, + "canonical_issue_reason": "issue:45292 is the broadest formulation of the `resize_token_embeddings` bug: output embeddings are not updated. The related Gemma4 and T5Gemma reports look like concrete manifestations of the same missing propagation.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45276 is the most informative representative because it spells out the affected Gemma4 subcomponents (`embed_tokens_per_layer` and `output_embeddings`) while matching the same resize regression.", + "best_issue_reason": "issue:45292 is the best representative because it states the underlying bug at the right level of generality and can subsume the more specific model variants.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44292", "right": "issue:44912", "accept": false, - "reason": "Both are quantization/load failures, but they involve different models and different symptoms (`NVFP4` runtime error vs `MXFP4` falling back to bf16)." + "reason": "Both are quantization/load failures, but they involve different models and different quantization setups; no clear same code-path bug." }, { "left": "issue:43335", "right": "issue:45663", "accept": false, - "reason": "Different model families and different bugs: a sparse-layer construction issue versus an FSDP2 `shared_kv_states` KeyError." + "reason": "Unrelated symptoms: SwitchTransformers layer creation vs Gemma4 FSDP2 shared_kv_states handling." }, { "left": "issue:44155", "right": "issue:45823", "accept": false, - "reason": "Unrelated failures: AudioFlamingo3 track-leak during batched inference vs Gemma4 PLE device mismatch in forward." + "reason": "Different models and failures: track leakage in batched audio inference vs Gemma4 device placement mismatch." }, { "left": "issue:42947", "right": "issue:43381", "accept": false, - "reason": "Both mention gradient checkpointing, but one is about LoRA/PEFT ineffectiveness and the other is about eval-mode incompatibility." + "reason": "Both mention gradient checkpointing, but one is about ineffectiveness with LoRA and the other about eval-mode restrictions." }, { "left": "issue:43957", "right": "issue:44756", "accept": false, - "reason": "Different underlying problems: `torch.device(\"meta\")` loading regressions vs disabling mmap to avoid OOM on Strix Halo." + "reason": "Meta-device loading bug vs Strix Halo mmap/OOM issue; different root causes." }, { "left": "issue:45276", "right": "issue:45292", "accept": true, - "reason": "Same underlying `resize_token_embeddings` regression: both report that resized embeddings are not propagated to output-related modules." + "reason": "Same underlying `resize_token_embeddings` propagation bug; 45276 is a Gemma4-specific manifestation of the generic output-embeddings issue." }, { "left": "issue:44863", "right": "issue:46468", "accept": false, - "reason": "Nemotron checkpoint-loading failure and a broken official example are related by topic, but they are not the same concrete bug." + "reason": "Both are Nemotron-related, but one is checkpoint loading and the other is an example breakage; too little evidence of the same bug." }, { "left": "issue:45685", "right": "issue:46468", "accept": false, - "reason": "Completely different failures: an MPS `histogram_mps` implementation error versus a broken Nemotron example." + "reason": "No common failure mode beyond model family name." }, { "left": "issue:44479", "right": "issue:44877", "accept": false, - "reason": "Different subsystems and symptoms: video-input regression in Qwen VL models vs strict-config loading failure for `granite_speech`." + "reason": "Different subsystems: video-input regression vs strict config loading." }, { "left": "issue:43381", "right": "issue:44928", "accept": false, - "reason": "Both mention training/runtime stability, but one is eval-mode checkpointing and the other is Qwen3.5 RLHF NaN explosion from SDPA fallback." + "reason": "Gradient-checkpointing eval-mode behavior is unrelated to the RLHF NaN / SDPA fallback regression." }, { "left": "issue:44164", "right": "issue:44756", "accept": false, - "reason": "Unrelated: save/from_pretrained extra_state handling vs mmap/OOM behavior on a specific platform." + "reason": "Serialization extra_state handling and mmap/OOM are unrelated." }, { "left": "issue:45684", "right": "issue:45698", "accept": false, - "reason": "Both touch custom-model save/load, but one is a permissions propagation bug and the other is loading the wrong custom module." + "reason": "Both involve save/from_pretrained around custom modules, but one is a permissions issue and the other is loading the wrong module." }, { - "left": "issue:45588", - "right": "issue:45663", + "left": "issue:43854", + "right": "issue:45216", "accept": false, - "reason": "Different code paths and failure modes: flash-attention sinkless-model AttributeError vs Gemma4 FSDP2 KeyError." + "reason": "Different models and different load/save failures; no shared underlying bug is evident." }, { - "left": "issue:43854", - "right": "issue:45216", + "left": "issue:45588", + "right": "issue:45663", "accept": false, - "reason": "Different reports around model loading/save regressions, but they concern different models and different failure points." + "reason": "Flash-attention sink-less-model crash vs Gemma4 FSDP2 key error are different code paths." }, { "left": "issue:42796", "right": "issue:46326", "accept": false, - "reason": "Both are tokenizer bugs, but the symptoms differ: BPE marker handling vs `continue_final_message` behavior for `mistral_common` tokenizers." + "reason": "Tokenizer-related, but one is BPE marker handling and the other is `continue_final_message` behavior in mistral_common tokenizers." }, { "left": "issue:45292", "right": "issue:45335", "accept": true, - "reason": "Same `resize_token_embeddings` regression pattern: one says output embeddings are not updated, the other says decoder embeddings are not updated." + "reason": "Same core regression: resizing token embeddings does not propagate to model embedding/output layers; these are architecture-specific instances." }, { "left": "issue:43646", "right": "issue:43708", "accept": false, - "reason": "Custom-model initialization under Transformers 5.0.0 is unrelated to `resume_from_checkpoint` max_steps calculation." + "reason": "Custom model initialization in notebooks is unrelated to Trainer checkpoint resume max_steps calculation." }, { "left": "issue:43828", "right": "issue:43957", "accept": false, - "reason": "Different issues: autocast dtype mismatch for Phi-tiny-MoE vs meta-device model loading breakage." + "reason": "Autocast dtype mismatch for a specific model is not the same as meta-device loading failures." }, { "left": "issue:44402", "right": "issue:44451", "accept": false, - "reason": "Both are model-loading/tokenizer reports, but the affected models and symptoms are different." + "reason": "One is a tokenizer vocab-size inconsistency, the other is a model-loading failure for a specific checkpoint." }, { "left": "issue:43645", "right": "issue:44164", "accept": false, - "reason": "Different bugs in the save/load stack: notebook custom model initialization vs `extra_state` handling." + "reason": "Notebook-defined custom model initialization and extra_state save/load handling are different bugs." }, { "left": "issue:44589", "right": "issue:44756", "accept": false, - "reason": "Unrelated: missing Float8 storage type vs mmap-related OOM mitigation." + "reason": "Float8 storage lookup failure is unrelated to mmap/OOM." }, { "left": "issue:43425", "right": "issue:43828", "accept": false, - "reason": "Torch-version incompatibility is broader and different from the autocast dtype mismatch in one model." + "reason": "Generic Torch 2.10 incompatibility is too broad and not the same concrete dtype-mismatch bug." }, { "left": "issue:45812", "right": "issue:46315", "accept": false, - "reason": "Different problem classes: wrong token IDs for Granite models vs slower `convert_tokens_to_ids` performance." + "reason": "Both are tokenizer regressions, but one is wrong token IDs and the other is a performance regression in `convert_tokens_to_ids`." }, { "left": "issue:46315", "right": "issue:46491", "accept": false, - "reason": "Performance regression in token conversion is not the same as CodeLlama leading-space encode/decode regression." + "reason": "Performance regression vs encode/decode correctness regression; not the same underlying change." }, { "left": "issue:39404", "right": "issue:45701", "accept": false, - "reason": "Whisper pipeline `return_language` failure and a general tokenization-change report are not the same bug." + "reason": "Whisper pipeline `return_language` regression and generic tokenization changes are not the same bug." }, { "left": "issue:43611", "right": "issue:43645", "accept": false, - "reason": "Both are Transformers 5.0.0 regressions, but one is about `base_model_prefix` loading and the other about notebook-defined custom model initialization." + "reason": "Both touch custom-model loading in Transformers 5, but the failure modes are different enough that they should stay separate." }, { "left": "issue:39404", "right": "issue:46665", "accept": false, - "reason": "Different behaviors: Whisper pipeline language return vs inconsistent `from_pretrained` usage inside Python classes." + "reason": "Whisper pipeline behavior and `from_pretrained` inside Python classes do not point to the same code path." }, { "left": "issue:44530", "right": "issue:45084", "accept": false, - "reason": "No concrete overlap: PagedAttentionCache group-type crash vs a compile-time non-template-node TypeError." + "reason": "Qwen3.5 PagedAttentionCache crash and compilation/template-node type error are unrelated." }, { - "left": "issue:43425", + "left": "issue:44683", "right": "issue:45823", "accept": false, - "reason": "One is a Torch-version compatibility report; the other is a Gemma4 device-mapping forward-pass mismatch." + "reason": "Compiled flex_attention on torch>=2.9 is unrelated to Gemma4 device-map mismatch." }, { - "left": "issue:44683", + "left": "issue:43425", "right": "issue:45823", "accept": false, - "reason": "Compiled flex_attention on torch >= 2.9 is unrelated to Gemma4 PLE device mismatch." + "reason": "Broad Torch 2.10 incompatibility does not match the specific Gemma4 PLE device placement bug." }, { "left": "issue:43425", "right": "issue:43716", "accept": false, - "reason": "Torch 2.10 compatibility and Mistral-3 image-preprocessor/model dtype mismatch are different failures." - }, - { - "left": "issue:44479", - "right": "issue:44805", - "accept": false, - "reason": "Video-input regression in Qwen VL models is unrelated to the mask/tensor shape IndexError." + "reason": "Torch compatibility and Mistral-3 image preprocessor dtype mismatch are distinct issues." }, { "left": "issue:43906", "right": "issue:44530", "accept": false, - "reason": "An isolated reproduction of another issue and a PagedAttentionCache crash are different underlying reports." + "reason": "The reproduction issue does not clearly match the Qwen3.5 linear_attention group-type crash; insufficient evidence to merge." } ] }, "evaluator_result": { "accept": true, - "feedback": "The analysis is grounded in the packet and stays conservative overall. The only accepted soft-edge cluster is the resize_token_embeddings set, and those titles are close enough to support a duplicate/broader-bug grouping. The other pair rejections are generally cautious rather than overstated." + "feedback": "The clustering is conservative and grounded in the packet. The only clearly supported duplicate subcluster is the `resize_token_embeddings` regression around output/decoder embeddings, and the analyst\u2019s canonical choice of issue:45292 is reasonable. The remaining pairwise rejects avoid overclaiming shared root causes and stay within the evidence from titles/issue descriptions." }, "evaluator_used": true, "retried": false, @@ -2886,13 +2760,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 9, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:36032", "issue:39404", @@ -2948,20 +2822,20 @@ "issue:44589|issue:45084", "issue:44265|issue:44898", "issue:43994|issue:44079", - "issue:39404|issue:44912", - "issue:43425|issue:44918", "issue:44671|issue:44991", + "issue:43425|issue:44918", + "issue:39404|issue:44912", "issue:36032|issue:43452", "issue:46050|issue:46326", "issue:44164|issue:45759", - "issue:43906|issue:46326", "issue:43299|issue:43931", + "issue:43906|issue:46326", "issue:44991|issue:45081", "issue:43856|issue:44756", "issue:43596|issue:44683", "issue:44164|issue:44479", - "issue:44488|issue:45081", "issue:44462|issue:46489", + "issue:44488|issue:45081", "issue:44488|issue:44987", "issue:43611|issue:43644", "issue:42796|issue:43404", @@ -2974,20 +2848,20 @@ "issue:43643|issue:44530", "issue:43582|issue:43632" ], - "prepared_review_unit_hash": "925d468fbe0f69b386dd4e1cf4b3adf78c542505936c849290b930eae7cb04fc", + "prepared_review_unit_hash": "88d39897d6778fa7dc11d279943763a8fca88842f0f1cc43f3d2408413b10063", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36447, + "serialized_chars": 36445, "estimated_input_tokens": 9112, "estimated_eval_tokens": 18480 }, @@ -2996,186 +2870,186 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is not a true duplicate cluster. It mixes many distinct Transformers v5 regressions across tokenizers, model loading, quantization, export/compile, docs, and memory behavior, with little evidence that any soft pair is the same underlying bug.", - "confidence": 0.9, - "canonical_issue_reason": "No clear canonical duplicate: the issues are too heterogeneous to safely collapse into one source of truth.", + "summary": "This cluster is a mix of unrelated Transformers v5 regressions: tokenizer loading, model/config loading, quantization, and a few docs issues. The soft-similarity pairs mostly share broad subsystem vocabulary but not the same concrete bug or fix path.", + "confidence": 0.88, + "canonical_issue_reason": "issue:43611 is the broadest, most central report in the set: a general Transformers v5 model-loading regression. It best represents the cluster\u2019s common theme of v5 breakages, even though many other issues are narrower and model/tokenizer-specific.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43611 is the broadest representative of the cluster because it targets a generic Transformers v5 model-loading regression, which is closest to the dominant theme here.", + "best_issue_reason": "issue:43611 is the best overall representative because it captures the overarching v5 loading regression pattern that many of the other issues are variants of, whereas the others are more specialized failure modes.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43901", "right": "issue:44509", "accept": false, - "reason": "Both are docs-related, but they cover different API/task deprecations and do not describe the same defect." + "reason": "Both are docs-related, but one is about return_all_scores and the other about removed pipeline tasks; different documentation bugs." }, { "left": "issue:44589", "right": "issue:45084", "accept": false, - "reason": "Different failures in different areas: missing Float8 storage vs compile error on non-template nodes." + "reason": "Different failures: Float8 storage lookup vs template compilation error. No shared code-path bug." }, { "left": "issue:44265", "right": "issue:44898", "accept": false, - "reason": "One is a torch.export/torch_compilable_check issue; the other is a Perceiver image-size/interpolation bug." + "reason": "torch.export/export hook failure and Perceiver interpolation failure are unrelated model paths." }, { "left": "issue:43994", "right": "issue:44079", "accept": false, - "reason": "SigLIP2 bad outputs and ModelOutput key assignment are unrelated code paths." + "reason": "One is a SigLIP2 bad-output regression, the other is ModelOutput key assignment; different underlying bugs." }, { - "left": "issue:39404", - "right": "issue:44912", + "left": "issue:44671", + "right": "issue:44991", "accept": false, - "reason": "Whisper pipeline language return and MXFP4 quantization fallback are different regressions." + "reason": "Different models and symptoms: CamemBERT MLM predictions vs tokenizer loading for est-roberta." }, { "left": "issue:43425", "right": "issue:44918", "accept": false, - "reason": "Torch 2.10 compatibility and Qwen3.5 embedding unpacking in TRL are unrelated problems." + "reason": "Torch version incompatibility is unrelated to Qwen3.5 TRL embedding unpacking." }, { - "left": "issue:44671", - "right": "issue:44991", + "left": "issue:39404", + "right": "issue:44912", "accept": false, - "reason": "CamemBERT prediction regression and tokenizer loading for est-roberta are different issues." + "reason": "Whisper pipeline return_language regression and MXFP4 fallback are separate issues." }, { "left": "issue:36032", "right": "issue:43452", "accept": false, - "reason": "Both involve loading, but the T5 tokenizer method conflict and gguf_file handling are different bugs." + "reason": "T5Tokenizer method conflict and gguf_file loading failures are different tokenizer/model loading bugs." }, { "left": "issue:46050", "right": "issue:46326", "accept": false, - "reason": "Quantized model round-trip failure and mistral_common continue_final_message behavior are unrelated." + "reason": "Quantized save/load breakage and mistral_common continue_final_message handling are unrelated." }, { "left": "issue:44164", "right": "issue:45759", "accept": false, - "reason": "extra_state save/load handling and composite Qwen config unwrapping are different mechanisms." + "reason": "extra_state save/load handling is different from unwrapping text_config in composite configs." }, { - "left": "issue:43906", - "right": "issue:46326", + "left": "issue:43299", + "right": "issue:43931", "accept": false, - "reason": "The linked reproduction is not enough to show the same underlying tokenizer bug as continue_final_message." + "reason": "Both involve Qwen3-VL loading, but one is a dev0 regression and the other is a specific weight-shape mismatch; not the same concrete bug." }, { - "left": "issue:43299", - "right": "issue:43931", + "left": "issue:43906", + "right": "issue:46326", "accept": false, - "reason": "Both mention Qwen3-VL loading, but one is a v5 dev regression and the other is a specific weight-shape mismatch." + "reason": "Both mention tokenizer/chat behavior, but they are different reproductions and code paths." }, { "left": "issue:44991", "right": "issue:45081", "accept": false, - "reason": "Different tokenizer failures with different models and different failure points." + "reason": "Different tokenizer bugs: est-roberta loading vs Mistral regex patch crash." }, { "left": "issue:43856", "right": "issue:44756", "accept": false, - "reason": "Training memory inefficiency and Strix Halo mmap OOM are not the same bug." + "reason": "Memory inefficiency during Qwen3 MoE training is unrelated to disabling mmap on Strix Halo." }, { "left": "issue:43596", "right": "issue:44683", "accept": false, - "reason": "DeepSpeed ZeRO3/BertModel indexing error is unrelated to flex_attention compile failures." + "reason": "DeepSpeed ZeRO3/BertModel indexing error and compiled flex_attention on new torch are unrelated." }, { "left": "issue:44164", "right": "issue:44479", "accept": false, - "reason": "Save/load extra_state handling and video-input regression affect different paths." + "reason": "extra_state persistence and Qwen video-input regression are separate failures." }, { - "left": "issue:44488", - "right": "issue:45081", + "left": "issue:44462", + "right": "issue:46489", "accept": false, - "reason": "Loading sleng-bert and the mistral regex patch crash are different tokenizer issues." + "reason": "Tokenizer.json being ignored is not the same as DeepSeek-Coder output corruption after a prior fix." }, { - "left": "issue:44462", - "right": "issue:46489", + "left": "issue:44488", + "right": "issue:45081", "accept": false, - "reason": "Ignoring tokenizer.json and a DeepSeek-Coder tokenizer regression are not the same underlying problem." + "reason": "Both are tokenizer-loading issues, but for different models and different failure points." }, { "left": "issue:44488", "right": "issue:44987", "accept": false, - "reason": "These are different model-loading failures with no shared concrete code-path evidence." + "reason": "Different model-specific load failures with no shared root cause evident." }, { "left": "issue:43611", "right": "issue:43644", "accept": false, - "reason": "Base_model_prefix loading regression and non-persistent buffer junk are distinct model-state issues." + "reason": "Both are v5 regressions, but one is model loading/base_model_prefix and the other is non-persistent buffer initialization." }, { "left": "issue:42796", "right": "issue:43404", "accept": false, - "reason": "Mistral tokenizer BPE-marker handling and Mistral3 lm_head tying are separate bugs." + "reason": "Tokenizer BPE marker handling and Mistral lm_head tying are unrelated bugs." }, { "left": "issue:39404", "right": "issue:44671", "accept": false, - "reason": "Whisper pipeline language output and CamemBERT masked LM predictions are unrelated." + "reason": "Whisper pipeline language return and CamemBERT MLM predictions are different behaviors in different models." }, { "left": "issue:44737", "right": "issue:45859", "accept": false, - "reason": "CPU arange placement in XLNet and CPU-offload incompatibility in Qwen3_5 MoE are different device-handling bugs." + "reason": "CPU arange placement bug in XLNet and CPU-offload incompatibility in Qwen3_5Moe are separate issues." }, { "left": "issue:46285", "right": "issue:46402", "accept": false, - "reason": "CLIP structure inconsistency and Sam3 text-encoder weight loading are different model families and failures." + "reason": "CLIP structure inconsistency and Sam3 text encoder weight loading are unrelated model-specific regressions." }, { "left": "issue:36032", "right": "issue:42915", "accept": false, - "reason": "T5 tokenizer initialization conflict and Qwen3Moe FineGrainedFP8 failure are unrelated." + "reason": "T5 tokenizer method conflict and Qwen3 MoE FineGrainedFP8 failure are different code paths." }, { "left": "issue:43716", "right": "issue:45084", "accept": false, - "reason": "Mistral image/text dtype mismatch and template-node compilation error do not share a code path." + "reason": "Mistral image-preprocessor dtype mismatch is unrelated to template-node compilation failures." }, { "left": "issue:44661", "right": "issue:44683", "accept": false, - "reason": "Tokenizer-mapping add-new-model-like behavior and flex_attention compilation are unrelated." + "reason": "Tokenizer mapping-name insertion bug and compiled flex_attention regression are distinct problems." }, { "left": "issue:43643", "right": "issue:44530", "accept": false, - "reason": "trust_remote_code config field loss and Qwen3.5 PagedAttentionCache crash are different issues." + "reason": "AutoConfig trust_remote_code field loss and Qwen3.5 paged attention group-type crash are unrelated." }, { "left": "issue:43582", "right": "issue:43632", "accept": false, - "reason": "Apple Silicon warmup TypeError and _is_hf_initialized flag breakage are separate regressions." + "reason": "AppleSilicon warmup TypeError and _is_hf_initialized breakage are different regressions." } ] }, @@ -3191,13 +3065,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 10, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:28530", "issue:34567", @@ -3250,8 +3124,8 @@ ], "soft_pairs": [ "issue:40990|issue:43421", - "issue:44451|issue:44987", "issue:42915|issue:44530", + "issue:44451|issue:44987", "issue:43541|issue:45084", "issue:34567|issue:40444", "issue:43644|issue:43646", @@ -3285,136 +3159,136 @@ "issue:44265|issue:44610", "issue:36032|issue:43742" ], - "prepared_review_unit_hash": "8745539349679b90b2aa163b3f3eb89ce9952ef957b67a7ff5810a7f9e284bd0", + "prepared_review_unit_hash": "f541d63ac5e6efa2b0ab2e75169a6a49ce75eb46442fc8cbe11a5c9fd6b25f6e", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 35, - "serialized_chars": 38399, - "estimated_input_tokens": 9600, - "estimated_eval_tokens": 19456 + "serialized_chars": 38403, + "estimated_input_tokens": 9601, + "estimated_eval_tokens": 19458 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a loose mix of mostly unrelated Transformers issues. I found no soft-edge pair with enough evidence to treat as a duplicate; the audio and GGUF pairs look like plausible same-area similarities, but the packet does not show the same concrete bug or shared target/file path. Issue 43452 is the broadest representative issue, but only as a loose anchor, not a confirmed canonical duplicate.", - "confidence": 0.63, - "canonical_issue_reason": "No single canonical issue is strongly supported by the packet. If one representative issue is needed, 43452 is the broadest GGUF-loading report, but the cluster does not provide enough evidence to elevate it to a true canonical duplicate bucket.", + "summary": "This cluster is mostly a collection of unrelated closed issues that only look superficially similar because many involve loading or model-regression symptoms. The only soft pair that looks like the same underlying bug is the GGUF loading regression pair.", + "confidence": 0.82, + "canonical_issue_reason": "issue:43452 is the best canonical issue because it describes the broadest, most concrete failure mode in the cluster: GGUF loading breaks across both tokenizer and causal LM from_pretrained paths. It is more general than the specific Qwen2.5-GGUF report and best represents that duplicate subgroup.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43452 is the best representative issue because it spans the broadest loading failure description in the set: GGUF breaks for both AutoTokenizer and AutoModelForCausalLM from_pretrained. It is still only a representative, not a confirmed duplicate hub.", + "best_issue_reason": "issue:43452 is the strongest issue candidate overall: it is the most general report among the only clearly related pair, and its wording matches a reusable code-path regression rather than a model-specific symptom.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:40990", "right": "issue:43421", "accept": false, - "reason": "Perplexity/quality regression versus runtime post-processor update after special-token changes; same NLP area, different bug." + "reason": "Different symptoms and code paths: perplexity/eval behavior versus runtime special-token post-processing updates." }, { - "left": "issue:44451", - "right": "issue:44987", + "left": "issue:42915", + "right": "issue:44530", "accept": false, - "reason": "Both are loading failures, but they involve different models and different failure modes; no shared concrete code-path evidence." + "reason": "Both mention Qwen models, but one is an FP8 loading/config failure and the other is a paged-attention cache/group-type crash; not the same bug." }, { - "left": "issue:42915", - "right": "issue:44530", + "left": "issue:44451", + "right": "issue:44987", "accept": false, - "reason": "FP8 config failure and PagedAttentionCache linear_attention crash are different bugs in the same broad inference area." + "reason": "Both are model-loading failures, but they concern different models and likely different regressions." }, { "left": "issue:43541", "right": "issue:45084", "accept": false, - "reason": "Both mention tracing/compilation, but the reported errors point to different code paths and different failure mechanisms." + "reason": "A torch dynamo grouped_mm tracing error is unrelated to a template compilation TypeError." }, { "left": "issue:34567", "right": "issue:40444", "accept": false, - "reason": "Trainer token accounting and multimodal iterable-dataset behavior are unrelated." + "reason": "Trainer token-count tracking and multimodal iterable-dataset finetuning are unrelated." }, { "left": "issue:43644", "right": "issue:43646", "accept": false, - "reason": "Both are v5 regressions, but one is non-persistent buffer initialization and the other is custom model init in notebooks; not the same bug." + "reason": "Both are v5 regressions, but one is about non-persistent buffers being corrupted while the other is about custom model initialization; too broad and not the same concrete bug." }, { "left": "issue:42175", "right": "issue:46178", "accept": false, - "reason": "Dependency/backend composition and BACKENDS_MAPPING lookup failure are different issues." + "reason": "Both concern backend/package handling, but one is missing tensorflow in a torch extra and the other is a missing tensorflow_text mapping entry." }, { "left": "issue:44811", "right": "issue:45584", "accept": false, - "reason": "Whisper batch_decode skip_special_tokens handling is distinct from empty-transcription generation after align_special_tokens." + "reason": "Both involve Whisper, but one is batch_decode skip_special_tokens handling and the other is empty-transcription generation after align_special_tokens." }, { "left": "issue:44206", "right": "issue:44220", "accept": false, - "reason": "Both touch audio feature extraction, but the packet only supports a broad same-area similarity; the specific crash conditions differ and there is no shared target/file evidence." + "reason": "Both are audio-feature-related, but the titles point to different failures and there is not enough evidence they share the same exact code-path bug." }, { "left": "issue:43883", "right": "issue:44488", "accept": false, - "reason": "Different model-loading failures with different models and different reported errors." + "reason": "One is a missing attribute on MolmoForCausalLM and the other is a specific tokenizer/model loading failure; not the same issue." }, { "left": "issue:40444", "right": "issue:46326", "accept": false, - "reason": "Multimodal iterable-dataset failures and mistral_common continue_final_message handling are unrelated." + "reason": "Different domains: multimodal finetuning with iterable datasets versus mistral_common continue_final_message behavior." }, { "left": "issue:36032", "right": "issue:43645", "accept": false, - "reason": "Tokenizer constructor conflict and custom model initialization breakage are separate regressions." + "reason": "Tokenizer method-name conflict and Jupyter custom-model initialization are unrelated." }, { "left": "issue:43452", "right": "issue:43482", - "accept": false, - "reason": "Both are GGUF-loading related, but 43452 is a broad from_pretrained breakage while 43482 is a model-specific v5 loading failure; the packet gives no stronger linkage than title similarity." + "accept": true, + "reason": "Both describe the same GGUF loading regression in transformers v5, with 43482 appearing to be a model-specific instance of the broader from_pretrained GGUF failure in 43452." }, { "left": "issue:43317", "right": "issue:43482", "accept": false, - "reason": "Dequantized model offload failure and GGUF loading failure are different loading paths." + "reason": "Device-map/offload loading failure and GGUF loading failure are different code paths." }, { "left": "issue:45084", "right": "issue:45507", "accept": false, - "reason": "Different invalid-method failures with different model contexts; same broad theme is not enough." + "reason": "Template-node compilation error and GraniteMoEHybrid invalid-method call are different failures." }, { "left": "issue:44987", "right": "issue:44991", "accept": false, - "reason": "Same loading theme, but different models and no evidence of the same root cause." + "reason": "Both are loading regressions, but they affect different models/tokenizers and do not look like one concrete bug." }, { "left": "issue:43262", "right": "issue:43377", "accept": false, - "reason": "Audio processor sampling-rate defaulting and MIMI padding-mask behavior are unrelated." + "reason": "Audio chat-template sampling-rate defaults and MIMI padding-mask batching are distinct issues." }, { "left": "issue:43746", @@ -3426,77 +3300,113 @@ "left": "issue:43883", "right": "issue:44991", "accept": false, - "reason": "Different model-loading regressions with different symptoms; no shared code-path evidence." + "reason": "Missing all_tied_weights_keys and tokenizer loading for a different model are not the same problem." }, { "left": "issue:44464", "right": "issue:45987", "accept": false, - "reason": "Chunked generation inconsistency under compilation is distinct from StaticCache.get_seq_length returning the wrong type." + "reason": "Chunked generation inconsistency under compiled forward and StaticCache return-type contract are related only at a very high level." }, { "left": "issue:43646", "right": "issue:44534", "accept": false, - "reason": "Both are v5 regressions, but custom model init and non-persistent buffer junk are different concrete problems." + "reason": "Both mention v5/custom model behavior, but the specific symptoms differ enough that they should not be merged." }, { "left": "issue:44403", "right": "issue:45081", "accept": false, - "reason": "Generic loading noise and a Mistral regex AttributeError are not the same underlying bug." + "reason": "Noise on load and a Mistral regex patch crash are different loading-related issues." }, { "left": "issue:42915", - "right": "issue:44291", - "accept": false, - "reason": "FP8 quantization/config failure and init_empty_weights unexpected-argument failure are separate regressions." - }, - { - "left": "issue:43089", - "right": "issue:45987", + "right": "issue:45084", "accept": false, - "reason": "Generation performance overhead and a wrong return type from StaticCache.get_seq_length are different problems." + "reason": "Qwen3Moe FP8 failure and template compilation TypeError are unrelated." }, { - "left": "issue:44265", - "right": "issue:44610", + "left": "issue:43645", + "right": "issue:44534", "accept": false, - "reason": "torch.export compatibility with torch_compilable_check is unrelated to the OmDet-Turbo input-size mismatch." + "reason": "Custom model initialization in notebooks and non-persistent-buffer corruption are not the same bug." }, { - "left": "issue:36032", - "right": "issue:43742", + "left": "issue:43404", + "right": "issue:44466", "accept": false, - "reason": "Tokenizer method conflict and MobileLLM key-error loading failure are different bugs." + "reason": "Both mention lm_head/tied weights, but one is a model-specific untied-weight bug and the other is a device-dependent serialization issue." }, { "left": "issue:44849", "right": "issue:45325", "accept": false, - "reason": "Qwen3.5 hidden-state handling and Qwen2.5-VL rope-index scaling are model-specific regressions with different symptoms." + "reason": "Different Qwen model families and different failure modes." }, { "left": "issue:43377", "right": "issue:43819", "accept": false, - "reason": "MIMI batched-vs-single mismatch from padding-mask support is not the same as DAC.from_latents versus forward mismatch from missing STE." + "reason": "Batched-vs-single MIMI encoder outputs and DAC.from_latents/forward mismatch are unrelated approximation bugs." }, { "left": "issue:43994", "right": "issue:44610", "accept": false, - "reason": "SigLIP2 nonsensical outputs and OmDet-Turbo processor/model size mismatch are unrelated model issues." + "reason": "Nonsensical model outputs and wrong processor input size are different problems." + }, + { + "left": "issue:28530", + "right": "issue:34567", + "accept": false, + "reason": "Early-stopping metric selection and token-count tracking are unrelated Trainer issues." + }, + { + "left": "issue:44451", + "right": "issue:45081", + "accept": false, + "reason": "Different models and different tokenizer/loading failures." + }, + { + "left": "issue:44987", + "right": "issue:45081", + "accept": false, + "reason": "Model-loading regression for one model versus a Mistral tokenizer regex crash for another." + }, + { + "left": "issue:42915", + "right": "issue:44291", + "accept": false, + "reason": "FineGrainedFP8Config failure and init_empty_weights unexpected-argument failure are different loading paths." + }, + { + "left": "issue:43089", + "right": "issue:45987", + "accept": false, + "reason": "Generation overhead/per-token syncs is a performance issue, not a cache-length contract bug." + }, + { + "left": "issue:44265", + "right": "issue:44610", + "accept": false, + "reason": "torch.export/torch_compilable_check and a processor size mismatch are unrelated." + }, + { + "left": "issue:36032", + "right": "issue:43742", + "accept": false, + "reason": "Tokenizer method conflict and MobileLLM key error are not the same failure." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The summary is a bit too confident for the evidence in the packet. In particular, both accepted soft edges (44206\u219444220 and 43452\u219443482) are only supported by title-level similarity; the packet gives no stronger linkage such as shared targets, filenames, or explicit issue links, and both are marked deterministic_accept=false in the packet. The GGUF pair especially is overstated as the \u2018same regression\u2019 rather than a plausible same-area similarity. Please keep soft-edge verdicts more conservative and avoid calling them duplicates unless the packet clearly supports it." + "accept": true, + "feedback": "The summary is grounded in the packet and stays appropriately conservative. The only accepted soft pair, issue:43452/issue:43482, is a plausible same-bug relation based on the shared GGUF loading regression, and the rejections avoid overclaiming beyond what the titles support." }, "evaluator_used": true, - "retried": true, - "accepted_nontrivial_soft_edge": false, + "retried": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -3505,13 +3415,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 11, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:33666", "issue:35707", @@ -3573,8 +3483,8 @@ "issue:43611|issue:44534", "issue:43425|issue:44403", "issue:43296|issue:43531", - "issue:40444|issue:45325", "issue:44220|issue:44683", + "issue:40444|issue:45325", "issue:45812|issue:46491", "issue:43632|issue:44756", "issue:42796|issue:43746", @@ -3587,25 +3497,25 @@ "issue:44291|issue:44589", "issue:43421|issue:44568", "issue:43632|issue:44534", - "issue:35707|issue:40444", "issue:41553|issue:43582", + "issue:35707|issue:40444", "issue:43482|issue:43525", "issue:43482|issue:45538" ], - "prepared_review_unit_hash": "f4b0d4fab61f39cfb4ffa8fb12afaef1d8f250aae93b6a5f633c6ddf6e473f29", + "prepared_review_unit_hash": "39d0bedffdd40dd28aae295250ed72830d46afd68254a030171be55fd99ff64f", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 29, - "serialized_chars": 35622, + "serialized_chars": 35621, "estimated_input_tokens": 8906, "estimated_eval_tokens": 18068 }, @@ -3614,198 +3524,192 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is mostly a heterogeneous grab-bag of distinct Transformers v5 regressions and model-specific loading/runtime bugs, especially around tokenizers and multimodal models. The only clear duplicate-like pair is the Qwen3.5 save_pretrained regression reports.", - "confidence": 0.71, - "canonical_issue_reason": "Issue 45812 is the strongest anchor: it describes a broad, clearly framed v4\u2192v5 tokenizer regression affecting all Granite models and already has inbound references, so it best represents the tokenizer-regression subgroup in this mixed cluster.", + "summary": "Most candidate edges are false positives across unrelated bugs, models, or failure modes. The only clear duplicate is the Qwen3.5 `save_pretrained` checkpoint-regression pair.", + "confidence": 0.77, + "canonical_issue_reason": "Issue 45216 is the broader, earlier report of the Qwen3.5 checkpoint serialization regression; 45357 reads like a later, narrower manifestation focused on the visual encoder keys.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45812 is the best representative issue overall because it is broad, regression-focused, and high-signal compared with the many narrower model-specific reports.", + "best_issue_reason": "Issue 45216 is the best representative because it captures the full `save_pretrained` regression at the model level, while still matching the later visual-key corruption report.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43322", "right": "issue:43540", "accept": false, - "reason": "Both are Qwen-related model failures, but one is a segmentation fault on Llava Next loading and the other is a video-processing ValueError in Qwen3OmniMoe; different code paths and symptoms." + "reason": "Different multimodal models and different failure types; both are video/image-related but not the same code path." }, { "left": "issue:44062", "right": "issue:44987", "accept": false, - "reason": "Both mention loading failures, but one is a tokenizer AddedToken argument conflict and the other is a model load failure for physical-intelligence/fast; no evidence they are the same defect." + "reason": "One is an AddedToken argument conflict, the other is a model-specific tokenizer load failure; no clear shared bug." }, { "left": "issue:42915", "right": "issue:44589", "accept": false, - "reason": "These are different quantization/runtime failures: FineGrainedFP8Config for Qwen3Moe versus missing Float8 storage object." + "reason": "Both involve FP8-related errors, but they point to different underlying failures and code paths." }, { "left": "issue:44479", "right": "issue:45325", "accept": false, - "reason": "Related multimodal/Qwen regression area, but one is a broad video-input regression across several models and the other is a specific still-image temporal position_id scaling bug; not clearly the same underlying issue." + "reason": "`44479` is a broad video-input regression across several models; `45325` is a specific still-image RoPE scaling bug in Qwen2.5-VL." }, { "left": "issue:45216", "right": "issue:45357", "accept": true, - "reason": "Both describe the same Qwen3.5 save_pretrained regression: incorrect checkpoint output, with 45357 narrowing it to incorrect visual encoder keys in the same save path." + "reason": "Same model family and same `save_pretrained` checkpoint corruption regression; `45357` is a narrower follow-up about the visual encoder keys." }, { "left": "issue:44991", "right": "issue:46315", "accept": false, - "reason": "Both are tokenizer-related regressions, but one is loading a specific tokenizer and the other is a performance slowdown in convert_tokens_to_ids; different bugs." + "reason": "Both are tokenizer issues in v5, but one is a load failure and the other is a performance regression in `convert_tokens_to_ids`." }, { "left": "issue:36032", "right": "issue:41720", "accept": false, - "reason": "T5 tokenizer class-name conflict and Qwen3 auto device-map CUDA assert are unrelated failures." + "reason": "Unrelated failures: a T5 tokenizer attribute conflict versus a Qwen3 auto device-mapping CUDA assert." }, { "left": "issue:43531", "right": "issue:44877", "accept": false, - "reason": "One is a Qwen3-MoE sliding_window bug, the other is a strict-config loading issue for granite_speech; different problems." + "reason": "Different models and unrelated bugs; one is sliding-window handling, the other is config strictness." }, { "left": "issue:43611", "right": "issue:44534", "accept": false, - "reason": "Base_model_prefix loading breakage and non-persistent buffer junk initialization are unrelated transformer v5 regressions." + "reason": "Both are v5 regressions, but they affect different loading/init paths and are not the same concrete bug." }, { "left": "issue:43425", "right": "issue:44403", "accept": false, - "reason": "Torch 2.10 incompatibility and noisy loading logs are not the same bug or change." + "reason": "Torch-version incompatibility and noisy load logs are unrelated." }, { "left": "issue:43296", "right": "issue:43531", "accept": false, - "reason": "PaddleOCR-VL load failure and Qwen3-MoE sliding_window behavior are unrelated model-specific issues." + "reason": "Different model families and different failure modes; no shared underlying bug." }, { - "left": "issue:40444", - "right": "issue:45325", + "left": "issue:44220", + "right": "issue:44683", "accept": false, - "reason": "Both involve Qwen multimodal models, but one is an iterable-dataset multi-image finetuning failure and the other is a rope-index temporal-position scaling bug; not the same defect." + "reason": "Audio feature extraction issue versus compiled flex_attention failure; unrelated." }, { - "left": "issue:44220", - "right": "issue:44683", + "left": "issue:40444", + "right": "issue:45325", "accept": false, - "reason": "Audio feature extraction and compiled flex_attention on torch>=2.9 are separate subsystems and failure modes." + "reason": "Same model family but different symptoms and paths: multi-image iterable training failure versus still-image temporal-position scaling." }, { "left": "issue:45812", "right": "issue:46491", "accept": false, - "reason": "Both are tokenizer regressions, but Granite token IDs being wrong and CodeLlama round-trip space stripping are distinct tokenizer bugs." + "reason": "Both are tokenizer regressions, but they affect different tokenizers and different behaviors." }, { "left": "issue:43632", "right": "issue:44756", "accept": false, - "reason": "_is_hf_initialized handling and mmap/OOM on Strix Halo are unrelated issues." + "reason": "Unrelated: HF initialization flag regression versus mmap/OOM on Strix Halo." }, { "left": "issue:42796", "right": "issue:43746", "accept": false, - "reason": "Tokenizer BPE marker handling and GraniteSpeech PEFT adapter loading are different code paths and symptoms." + "reason": "Tokenizer BPE-marker handling and PEFT checkpoint loading are unrelated." }, { "left": "issue:44403", "right": "issue:44991", "accept": false, - "reason": "Loading-log noise and a tokenizer load failure for EMBEDDIA/est-roberta are not the same bug." + "reason": "Generic load noise is not the same bug as a tokenizer load failure." }, { "left": "issue:43381", "right": "issue:45923", "accept": false, - "reason": "Gradient checkpointing being unusable in eval mode is different from a missing supports_gradient_checkpointing flag on a trust_remote_code model." + "reason": "One is about eval-mode gradient checkpointing misuse, the other is a missing capability flag on a trust_remote_code model." }, { "left": "issue:33666", "right": "issue:40444", "accept": false, - "reason": "Both are Qwen VL training issues, but multi-GPU training and multi-image iterable-dataset finetuning failures are different problems." + "reason": "Both involve Qwen-VL training, but they target different models and different failure scenarios." }, { "left": "issue:43792", "right": "issue:44589", "accept": false, - "reason": "Whisper runtime failure and missing Float8 storage object are unrelated." + "reason": "Whisper load failure and missing Float8 storage type are unrelated." }, { "left": "issue:40990", "right": "issue:44811", "accept": false, - "reason": "High perplexity on a language-model benchmark and Whisper batch_decode skip_special_tokens behavior are not the same issue." + "reason": "Perplexity regression and Whisper batch_decode special-token handling are unrelated." }, { "left": "issue:44079", "right": "issue:45663", "accept": false, - "reason": "ModelOutput key assignment and Gemma4 FSDP2 shared_kv_states rebuilding are different defects." + "reason": "ModelOutput key assignment and Gemma4 FSDP2 attention state rebuilding are different bugs." }, { "left": "issue:44291", "right": "issue:44589", "accept": false, - "reason": "init_empty_weights/_is_hf_initialized argument handling and Float8 storage lookup failure are unrelated." + "reason": "Both are load-time TypeErrors, but they stem from different causes and code paths." }, { "left": "issue:43421", "right": "issue:44568", "accept": false, - "reason": "Runtime special-token post-processor updates and missing BOS/EOS insertion for a specific tokenizer are different tokenizer bugs." + "reason": "Both mention special tokens, but one is runtime post-processor updates and the other is a tokenizer-specific BOS/EOS addition regression." }, { "left": "issue:43632", "right": "issue:44534", "accept": false, - "reason": "_is_hf_initialized regression and non-persistent buffer junk initialization are separate v5 problems." + "reason": "Different v5 regressions affecting different internal mechanisms; not the same issue." }, { - "left": "issue:35707", - "right": "issue:40444", + "left": "issue:41553", + "right": "issue:43582", "accept": false, - "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to multi-image IterableDataset finetuning failures." + "reason": "Bad tokenizer error messaging and Apple Silicon allocator TypeError are unrelated." }, { - "left": "issue:41553", - "right": "issue:43582", + "left": "issue:35707", + "right": "issue:40444", "accept": false, - "reason": "AutoTokenizer Voxtral error messaging and AppleSilicon caching_allocator_warmup TypeError are unrelated." + "reason": "Progressive generation with `inputs_embeds`/`past_key_values` is unrelated to Qwen2.5-VL multi-image finetuning." }, { "left": "issue:43482", "right": "issue:43525", "accept": false, - "reason": "Qwen2.5-GGUF loading failure and Llama4Config missing pad_token_id are different model-loading bugs." + "reason": "Different model load failures: GGUF/v5 compatibility versus missing `pad_token_id` on Llama4Config." }, { "left": "issue:43482", "right": "issue:45538", "accept": false, - "reason": "Qwen2.5-GGUF loading failure and CLIPTokenizer model_max_length behavior are unrelated." - }, - { - "left": "issue:43421", - "right": "issue:44568", - "accept": false, - "reason": "Runtime special-token post-processor updates and missing BOS/EOS insertion for a specific tokenizer are different tokenizer bugs." + "reason": "Unrelated tokenizer/load bugs; no shared underlying defect." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet: it correctly treats the cluster as mostly heterogeneous, identifies the Qwen3.5 save_pretrained pair as the only clear duplicate-like match, and uses issue 45812 as a plausible broad canonical anchor because it is regression-focused and has inbound references. The soft-edge rejections stay conservative and do not overstate equivalence." + "feedback": "The summary is well grounded in the packet, and the lone accepted edge (45216/45357) is conservative given both titles reference the same Qwen3.5 `save_pretrained` regression. The rejected pairs are also framed cautiously, mostly distinguishing by model, symptom, or code path rather than overstating duplication." }, "evaluator_used": true, "retried": false, @@ -3818,13 +3722,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 12, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:40444", "issue:41720", @@ -3892,8 +3796,8 @@ "issue:44265|issue:45072", "issue:42175|issue:43296", "issue:45325|issue:45381", - "issue:40444|issue:45381", "issue:43493|issue:43994", + "issue:40444|issue:45381", "issue:43454|issue:44466", "issue:45325|issue:45834", "issue:43792|issue:43957", @@ -3907,196 +3811,220 @@ "issue:43606|issue:44265", "issue:44479|issue:45823" ], - "prepared_review_unit_hash": "aebea4de0432910f8f32ba179da72cd10b0841d51eeb19e6944e7b482d07d961", + "prepared_review_unit_hash": "516d94f624fd8758911b145575a806e2ff0db83e5f30d47b0ce3fbf7752807d5", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36831, - "estimated_input_tokens": 9208, - "estimated_eval_tokens": 18672 + "serialized_chars": 36835, + "estimated_input_tokens": 9209, + "estimated_eval_tokens": 18674 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly unrelated issues, with two plausible duplicate pairs in Qwen2.5-VL video position handling and SigLIP2 correctness. No PRs were present.", - "confidence": 0.62, - "canonical_issue_reason": "Issue 43493 is the best root-cause-oriented anchor: it describes the SigLIP2 HF-vs-JAX implementation discrepancy, which cleanly subsumes the downstream 'nonsensical results' symptom reported in 43994.", + "summary": "This cluster is mostly heterogeneous: many unrelated Transformers bugs were grouped by loose title similarity. I only see two likely duplicate pairs: the Qwen2.5-VL position-id regressions and the SigLIP2 correctness regressions. No PRs are present.", + "confidence": 0.71, + "canonical_issue_reason": "Issue 43646 is the broadest representative of the set: it describes a general Transformers v5 model-initialization regression, whereas most other items are either unrelated or narrower model-specific bugs.", "canonical_pr_reason": null, - "best_issue_reason": "43493 is the strongest triage target because it is specific to a concrete model/code path, frames the underlying correctness problem rather than just a symptom, and is the clearest candidate for consolidation.", + "best_issue_reason": "Issue 43646 is the best available issue representative because it is the most generic loading/initialization regression among the related v5-model issues; it is still not a true umbrella for the whole cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43317", "right": "issue:43856", "accept": false, - "reason": "Both mention bad offloading/memory behavior, but one is a device-map/quantization loading failure and the other is a Qwen3 MoE training memory-efficiency issue; different bugs." + "reason": "Both mention offload/memory, but one is a device-map dequantized loading failure and the other is a Qwen3 MoE training memory issue." }, { "left": "issue:42175", "right": "issue:45542", "accept": false, - "reason": "Both touch TensorFlow backend packaging, but the concrete failures differ: missing TF in the torch extra vs tensorboard-only installs triggering an undefined backend error." + "reason": "Both touch TensorFlow packaging, but one is a missing backend dependency in the torch extra and the other is a runtime undefined-tf-backend error with only tensorboard installed." }, { "left": "issue:43531", "right": "issue:44291", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and init_empty_weights/_is_hf_initialized argument handling are different code paths and symptoms." + "reason": "Qwen3-MoE sliding-window handling and init_empty_weights/_is_hf_initialized are different code paths and different failure modes." }, { "left": "issue:44829", "right": "issue:45479", "accept": false, - "reason": "FlashAttention-3 degenerate training and single-label classification zero-loss are distinct training bugs with different causes." + "reason": "Both are training regressions, but one is flash_attention_3 degenerate training and the other is single-label classification with num_labels=1 causing zero loss." }, { "left": "issue:44079", "right": "issue:44534", "accept": false, - "reason": "ModelOutput key assignment and non-persistent buffer initialization are separate internal behaviors; same broad area, but not the same bug." + "reason": "ModelOutput key assignment and non-persistent buffer initialization are unrelated bugs." }, { "left": "issue:44462", "right": "issue:45812", "accept": false, - "reason": "Tokenizer.json being ignored and Granite token IDs being wrong are related to tokenization, but they point to different failure modes and likely different fixes." + "reason": "Both are AutoTokenizer regressions, but one is ignoring tokenizer.json and the other is wrong Granite token IDs; not enough to treat as the same bug." }, { "left": "issue:43818", "right": "issue:46402", "accept": false, - "reason": "Video-LLaVA temporal-attention/weight-sharing and Sam3Model text-encoder loading are unrelated model-specific regressions." + "reason": "Video-LLaVA vision-tower/temporal-attention issues are unrelated to Sam3Model text-encoder weight loading." }, { "left": "issue:41720", "right": "issue:45081", "accept": false, - "reason": "A800 cudaErrorAssert with auto device mapping is unrelated to the Mistral tokenizer regex/backend_tokenizer crash." + "reason": "CUDA assert in Qwen3 auto device mapping is unrelated to a tokenizer regex patch AttributeError." }, { "left": "issue:43818", "right": "issue:45357", "accept": false, - "reason": "Video-LLaVA encoder issues and Qwen3.5 save_pretrained visual-key serialization are different bugs." + "reason": "Video-LLaVA architecture problems and Qwen3.5 save_pretrained key serialization are different bugs." }, { "left": "issue:43377", "right": "issue:45491", "accept": false, - "reason": "Both are batching/padding-related at a high level, but one is MIMI encoder padding-mask support and the other is Gemma3 sliding-window all-padding NaNs; not the same underlying change." + "reason": "Both involve batching/masks, but one is MIMI encoder padding-mask mismatch and the other is Gemma3 sliding-window NaNs on all-padding windows." }, { "left": "issue:43611", "right": "issue:43646", "accept": false, - "reason": "Both are Transformers v5 loading regressions, but one centers on base_model_prefix handling and the other on custom model initialization; too broad to treat as one bug." + "reason": "Both are Transformers v5 loading/init regressions, but they are distinct concrete failures: base_model_prefix loading versus custom model initialization." }, { "left": "issue:43644", "right": "issue:44079", "accept": false, - "reason": "Non-persistent buffer junk initialization and ModelOutput key assignment are different internal regressions." + "reason": "Non-persistent buffer junk initialization is unrelated to ModelOutput key bookkeeping." }, { "left": "issue:43749", "right": "issue:44079", "accept": false, - "reason": "FSDP CPU RAM efficient loading and ModelOutput key assignment are unrelated issues." + "reason": "FSDP CPU RAM efficient loading and ModelOutput key assignment are different subsystems and bugs." }, { "left": "issue:44265", "right": "issue:45072", "accept": false, - "reason": "torch.export failures with torch_compilable_check and dtype mismatches in bfloat16 inference are different problems." + "reason": "torch.export with torch_compilable_check is unrelated to bfloat16 dtype mismatches in inference wrappers." }, { "left": "issue:42175", "right": "issue:43296", "accept": false, - "reason": "Packaging/backend installation and PaddleOCR-VL loading in vLLM are unrelated." + "reason": "TensorFlow backend packaging and PaddleOCR-VL load failures in vLLM are different problems." }, { "left": "issue:45325", "right": "issue:45381", "accept": true, - "reason": "Both describe the same Qwen2.5-VL video-position/rope-index regression: wrong temporal/vision position_ids for video inputs in transformers 5.3.0." + "reason": "Both are Qwen2.5-VL 5.3.0 position-id regressions in the same RoPE/vision-position-id code path, one for still images and one for video input." + }, + { + "left": "issue:43493", + "right": "issue:43994", + "accept": true, + "reason": "Both report SigLIP2 producing incorrect outputs; one compares HF to JAX and the other shows nonsensical AutoModel/pipeline results, pointing to the same model implementation bug." }, { "left": "issue:40444", "right": "issue:45381", "accept": false, - "reason": "Both involve Qwen2.5-VL multimodal inputs, but one is multiple images per prompt in finetuning and the other is video position_id scaling; not the same bug." + "reason": "IterableDataset with multiple images per prompt is a finetuning/input-packing bug, not the same as Qwen2.5-VL video position-id handling." }, { - "left": "issue:43493", - "right": "issue:43994", - "accept": true, - "reason": "Same SigLIP2 correctness regression: one reports HF vs JAX implementation divergence, the other reports nonsensical outputs from the same model family, plausibly the same fix." + "left": "issue:43454", + "right": "issue:44466", + "accept": false, + "reason": "Tied lm_head weight initialization and save_pretrained serialization inconsistency are related only at a broad theme level, not the same concrete bug." + }, + { + "left": "issue:45325", + "right": "issue:45834", + "accept": false, + "reason": "Qwen2.5-VL position-id scaling and Kosmos2.5 OCR index errors are unrelated." + }, + { + "left": "issue:43792", + "right": "issue:43957", + "accept": false, + "reason": "Whisper inference failure and meta-device model loading regressions are different model-loading issues." + }, + { + "left": "issue:42915", + "right": "issue:43482", + "accept": false, + "reason": "Qwen3MoE FP8 config failure and Qwen2.5 GGUF loading failure are unrelated." }, { "left": "issue:43632", "right": "issue:45072", "accept": false, - "reason": "_is_hf_initialized flag handling and dtype mismatch in bfloat16 inference are unrelated regressions." + "reason": "_is_hf_initialized flag handling and bfloat16 dtype mismatches are separate regressions." }, { "left": "issue:42796", "right": "issue:45381", "accept": false, - "reason": "Tokenizer BPE-marker handling and Qwen2.5-VL video position_ids are unrelated." + "reason": "Tokenizer BPE-marker handling and Qwen2.5-VL video position ids are unrelated." }, { "left": "issue:43701", "right": "issue:45325", "accept": false, - "reason": "Checkpoint key mismatch and Qwen2.5-VL rope/position-id scaling are different issues." + "reason": "Checkpoint key mismatch is unrelated to Qwen2.5-VL position-id scaling." }, { "left": "issue:44291", "right": "issue:44877", "accept": false, - "reason": "Unexpected _is_hf_initialized arg during empty-weight init is unrelated to strict config rejection for granite_speech." + "reason": "init_empty_weights argument handling and strict Granite speech config loading are different bugs." }, { "left": "issue:43873", "right": "issue:45701", "accept": false, - "reason": "Offloading with quantization and tokenizer changes across versions are not the same bug." + "reason": "Quantization/offloading behavior and tokenization changing across versions are not the same issue." }, { "left": "issue:43296", "right": "issue:45800", "accept": false, - "reason": "PaddleOCR-VL load failure in vLLM and torch/transformers compatibility are separate environment/model-loading issues." + "reason": "PaddleOCR-VL load failure in vLLM and torch/transformers version incompatibility are unrelated." }, { "left": "issue:43606", "right": "issue:44265", "accept": false, - "reason": "CPU-offload device mismatch and torch.export/exportability failures are different code paths." + "reason": "CPU offload device mismatch and torch.export/torch_compilable_check are different failure modes." }, { "left": "issue:44479", "right": "issue:45823", "accept": false, - "reason": "Qwen video-input regression and Gemma4 PLE device mismatch are unrelated model-specific bugs." + "reason": "Qwen video-input regression and Gemma4 device-map forward mismatch are unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet: it stays conservative, correctly notes that all items are issues (no PRs), and the two accepted pairs are the only ones that look plausibly duplicate from the titles alone. The reject verdicts are also appropriately cautious and do not overclaim sameness where the titles suggest different bugs." + "feedback": "The summary is grounded in the packet and stays conservative: it identifies the two strongest duplicate-like pairs supported by the issue titles/body themes, and the other soft-edge rejections are appropriately cautious. The note that issue 43646 is only a rough representative, not a true umbrella, is also consistent with the heterogeneous cluster. No PR pair analysis was needed because the packet contains no PRs." }, "evaluator_used": true, "retried": false, @@ -4109,13 +4037,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 13, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:34634", "issue:35707", @@ -4170,8 +4098,8 @@ "issue:43262|issue:43746", "issue:40990|issue:45584", "issue:43881|issue:44488", - "issue:35707|issue:41720", "issue:34634|issue:35707", + "issue:35707|issue:41720", "issue:43454|issue:43550", "issue:43606|issue:43819", "issue:39692|issue:40444", @@ -4186,8 +4114,8 @@ "issue:43317|issue:43873", "issue:43377|issue:44898", "issue:43454|issue:43720", - "issue:42796|issue:44977", "issue:40990|issue:42947", + "issue:42796|issue:44977", "issue:43749|issue:44534", "issue:42617|issue:42915", "issue:43572|issue:45759", @@ -4197,214 +4125,214 @@ "issue:43881|issue:44991", "issue:44387|issue:45704" ], - "prepared_review_unit_hash": "ad01bf96174247725518f433a5c417059e30c898373ce1741f5f58c667da9f6e", + "prepared_review_unit_hash": "1a4f9c0d0607a85e8d8d9eb9786662653541fcc999353f4c622fee8ead7fc973", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36490, - "estimated_input_tokens": 9123, - "estimated_eval_tokens": 18502 + "serialized_chars": 36488, + "estimated_input_tokens": 9122, + "estimated_eval_tokens": 18500 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly false-positive similarity matches: the items span many unrelated bugs across tokenizers, multimodal processors, quantization, loading, offloading, and training. Only the quantization/offloading pair looks plausibly like the same underlying defect.", - "confidence": 0.34, - "canonical_issue_reason": "issue:43873 is the only plausible umbrella duplicate target because it matches issue:43317's quantized/offloaded model-loading failure on the same code path; the rest of the set is too heterogeneous to share one canonical bug.", + "summary": "This cluster is mostly heterogeneous: nearly all pairs are superficially similar topic-wise but do not look like the same concrete bug. The only plausible duplicate is the quantized/offloading loading failure pair (#43317 vs #43873).", + "confidence": 0.33, + "canonical_issue_reason": "#43873 is the broadest umbrella issue in the set and is the only one that plausibly subsumes a nearby duplicate around offloading + quantization loading failures.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43873 is the best issue anchor because it is open, broad, and aligns with the one credible duplicate edge in the set (43317).", + "best_issue_reason": "#43873 is the most representative issue for the small offloading/quantization subgroup; the rest of the cluster is too mixed to treat as one duplicate set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43262", "right": "issue:43746", "accept": false, - "reason": "Different audio-model problems: sampling-rate defaulting in chat templates vs PEFT checkpoint loading." + "reason": "Audio chat-template sampling-rate default vs PEFT/local-checkpoint loading failure; different code paths and bugs." }, { "left": "issue:40990", "right": "issue:45584", "accept": false, - "reason": "Unrelated failures: high perplexity on a language model vs Whisper empty-transcription generation after token alignment." + "reason": "Perplexity evaluation issue on GPT-OSS is unrelated to Whisper empty-transcription generation failure." }, { "left": "issue:43881", "right": "issue:44488", "accept": false, - "reason": "Both are loading failures, but for different models and likely different causes; not the same bug." + "reason": "Both are model-loading failures, but they involve different models and likely different root causes." }, { - "left": "issue:35707", - "right": "issue:41720", + "left": "issue:34634", + "right": "issue:35707", "accept": false, - "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to a device-mapping CUDA assert." + "reason": "Bark voice_preset handling vs progressive generation with inputs_embeds/past_key_values are unrelated bugs." }, { - "left": "issue:34634", - "right": "issue:35707", + "left": "issue:35707", + "right": "issue:41720", "accept": false, - "reason": "Bark voice_preset behavior is a different processor issue from generation with inputs_embeds and caches." + "reason": "Generation-cache behavior and CUDA assert under auto device mapping are different failures." }, { "left": "issue:43454", "right": "issue:43550", "accept": false, - "reason": "Weight tying failure in a multimodal config is unrelated to a torch.compile/SDPA model failure." + "reason": "lm_head weight tying bug vs torch.compile/SDPA failure are not the same underlying issue." }, { "left": "issue:43606", "right": "issue:43819", "accept": false, - "reason": "Both touch audio models, but one is CPU-offload device mismatch and the other is DAC latent/STE behavior; different bugs." + "reason": "CPU offload device mismatch in Bark is unrelated to DAC latent reconstruction/STE mismatch." }, { "left": "issue:39692", "right": "issue:40444", "accept": false, - "reason": "Documentation/example errors for SigLIP2 do not match a finetuning failure on Qwen2.5-VL iterable multi-image batches." + "reason": "Different multimodal examples and failures: doc example mismatch/quantization vs iterable-dataset multi-image finetuning." }, { "left": "issue:44186", "right": "issue:44265", "accept": false, - "reason": "Tokenizer crashes on NER/padding are unrelated to torch.export failures around torch_compilable_check." + "reason": "Tokenizer crash on NER/padding is unrelated to torch.export failures in torch_compilable_check models." }, { "left": "issue:41720", "right": "issue:45823", "accept": false, - "reason": "Both involve device_map=auto, but the models and failure modes differ; not one concrete bug." + "reason": "Both mention device mapping, but the models and failure modes differ (CUDA assert vs forward-time device mismatch)." }, { "left": "issue:46305", "right": "issue:46315", "accept": false, - "reason": "Streaming mask off-by-one and tokenizer conversion slowdown are unrelated changes." + "reason": "Streaming mask off-by-one bug is unrelated to tokenizer conversion performance regression." }, { "left": "issue:43377", "right": "issue:44610", "accept": false, - "reason": "MIMI padding-mask batching bug is unrelated to OmDet-Turbo processor/model input-size mismatch." + "reason": "MIMI padding-mask correctness and OmDet-Turbo processor image-size mismatch are different subsystems and bugs." }, { "left": "issue:44568", "right": "issue:45920", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different models and distinct token-ID/special-token behavior." + "reason": "Both are tokenizer regressions, but one is missing BOS/EOS insertion and the other is wrong token IDs across many models." }, { "left": "issue:43257", "right": "issue:45923", "accept": false, - "reason": "Qwen3 MoE loading with accelerate+deepspeed is unrelated to a missing gradient-checkpointing flag on a different model." + "reason": "MOE weight conversion during accelerate+deepspeed loading is unrelated to a missing gradient-checkpointing flag." }, { "left": "issue:42796", "right": "issue:46209", "accept": false, - "reason": "Tokenizer BPE-marker handling and FP8 inference on B200 are completely different issues." + "reason": "Tokenizer BPE-marker handling and FP8 inference on B200 are unrelated." }, { "left": "issue:43454", "right": "issue:45072", "accept": false, - "reason": "lm_head weight tying and dtype mismatches in inference do not describe the same bug." + "reason": "Weight-tying regression is unrelated to bfloat16 dtype mismatches in inference." }, { "left": "issue:43317", "right": "issue:43873", "accept": true, - "reason": "Both describe quantization/offloading failures during model loading with device_map/offload; they plausibly share the same root cause." + "reason": "Both describe offloading failing for quantized/dequantized models under device_map/accelerate-style loading, plausibly the same underlying bug." }, { "left": "issue:43377", "right": "issue:44898", "accept": false, - "reason": "A batching/padding-mask discrepancy in MIMI is unrelated to a Perceiver non-default-resolution failure." + "reason": "Audio encoder padding-mask correctness is unrelated to Perceiver image-resolution handling." }, { "left": "issue:43454", "right": "issue:43720", "accept": false, - "reason": "Weight-tying problems and packed-weight unpacking during accelerate loading are different code paths and failures." + "reason": "lm_head tying bug and packed-weight unpacking during accelerate loading are distinct issues." }, { - "left": "issue:42796", - "right": "issue:44977", + "left": "issue:40990", + "right": "issue:42947", "accept": false, - "reason": "Tokenizer BPE-marker handling is unrelated to Qwen3.5 flash-attention generation behavior." + "reason": "High perplexity evaluation on a model is unrelated to gradient checkpointing behavior with PEFT LoRA." }, { - "left": "issue:40990", - "right": "issue:42947", + "left": "issue:42796", + "right": "issue:44977", "accept": false, - "reason": "Language-model perplexity on WikiText-2 is unrelated to gradient checkpointing with PEFT LoRA." + "reason": "Different tokenizer/regression bugs: BPE-marker handling vs flash-attention generation failure." }, { "left": "issue:43749", "right": "issue:44534", "accept": false, - "reason": "FSDP RAM-efficient loading and non-persistent buffer junk are different regression classes." + "reason": "FSDP CPU RAM efficient loading and non-persistent buffer junk are different v5 regressions." }, { "left": "issue:42617", "right": "issue:42915", "accept": false, - "reason": "A 3d_parallel.py runtime problem is unrelated to a Qwen3 MoE FP8 config failure." + "reason": "3d_parallel.py runtime failure and FineGrainedFP8Config failure are unrelated." }, { "left": "issue:43572", "right": "issue:45759", "accept": false, - "reason": "Missing pad_token_idx in StableLmConfig and text_config unwrapping for Qwen multimodal configs are distinct config bugs." + "reason": "Missing pad_token_idx in a config is unrelated to not unwrapping text_config for composite multimodal configs." }, { "left": "issue:43322", "right": "issue:46050", "accept": false, - "reason": "Llava Next loading segfaults and sinq save/load breakage are unrelated model-loading problems." + "reason": "Llava Next segmentation fault and sinq save/load breakage are different failure modes and code paths." }, { "left": "issue:43526", "right": "issue:43883", "accept": false, - "reason": "BEiT label reduction and a missing all_tied_weights_keys attribute are unrelated." + "reason": "BEiT label reduction bug is unrelated to Molmo weight-tying attribute errors." }, { "left": "issue:41720", "right": "issue:43322", "accept": false, - "reason": "Both are model-loading/runtime failures, but they involve different models and failure mechanisms." + "reason": "Both are load-time crashes, but the models and root causes are different." }, { "left": "issue:43881", "right": "issue:44991", "accept": false, - "reason": "A model loading failure for glm-4v-9b is unrelated to tokenizer loading regressions for EMBEDDIA/est-roberta." + "reason": "Different tokenizer/model-loading regressions for different models; no same concrete bug is evident." }, { "left": "issue:44387", "right": "issue:45704", "accept": false, - "reason": "CUDA reserved-memory OOM under int4 quantization is not the same as a T5 apex RMSNorm memory leak." + "reason": "Int4 CUDA reserved-memory OOM and T5 apex RMSNorm memory leak are separate quantization/runtime problems." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet: most pairings are clearly heterogeneous by title, and the lone accepted edge (issue:43317 \u2194 issue:43873) is a conservative match between quantization/offloading loading failures. No soft-edge verdict appears overstated, and there are no PR-pair mergeability concerns here." + "feedback": "The summary is well grounded in the packet and stays conservative. It correctly treats the cluster as heterogeneous and only elevates the 43317 \u2194 43873 pair, which is the one soft pair whose titles and descriptions both point to offloading/quantization loading behavior. The canonical/best-issue choice is subjective but not overstated, and there are no unsupported duplicate claims." }, "evaluator_used": true, "retried": false, @@ -4417,13 +4345,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 14, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:26413", "issue:28530", @@ -4439,6 +4367,7 @@ "issue:43317", "issue:43404", "issue:43450", + "issue:43526", "issue:43531", "issue:43550", "issue:43572", @@ -4448,9 +4377,9 @@ "issue:43708", "issue:43720", "issue:43873", + "issue:43881", "issue:44008", "issue:44186", - "issue:44361", "issue:44423", "issue:44466", "issue:44610", @@ -4475,18 +4404,18 @@ ], "soft_pairs": [ "issue:44466|issue:45357", - "issue:41720|issue:44977", "issue:36331|issue:42617", + "issue:41720|issue:44977", "issue:36331|issue:42915", "issue:43450|issue:45701", - "issue:44877|issue:44991", "issue:26413|issue:28530", - "issue:40990|issue:45701", + "issue:44877|issue:44991", "issue:43065|issue:46402", + "issue:40990|issue:45701", "issue:44186|issue:45072", "issue:43873|issue:44756", - "issue:43550|issue:43720", "issue:41720|issue:42617", + "issue:43550|issue:43720", "issue:43577|issue:43708", "issue:43404|issue:45357", "issue:43550|issue:43606", @@ -4504,222 +4433,229 @@ "issue:44671|issue:44743", "issue:43531|issue:43572", "issue:42915|issue:46291", - "issue:44186|issue:44361" + "issue:43526|issue:43881", + "issue:36331|issue:41720" ], - "prepared_review_unit_hash": "39b33881b4556ce825ba876af083a01a129797481cc3d25e89fb69e94fb2c273", + "prepared_review_unit_hash": "bad94399beb68e6f091b1af9e503609cff03e1431501106e9f9ee075fc8dc939", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 31, - "serialized_chars": 36498, - "estimated_input_tokens": 9125, - "estimated_eval_tokens": 18506 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37363, + "estimated_input_tokens": 9341, + "estimated_eval_tokens": 18938 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a heterogeneous set of Transformers issues; several are superficially similar around tokenization, loading, offloading, or model-specific regressions, but none look like true duplicates.", - "confidence": 0.74, - "canonical_issue_reason": "issue:45701 is the broadest and most reusable regression report in the set: it captures version-driven tokenization behavior changes, which is the closest common theme among the tokenizer/loading-related items.", + "summary": "This cluster is heterogeneous: the items span unrelated trainer, tokenizer, multimodal, quantization, loading, and model-specific bugs. None of the soft pairs look like true duplicates, so there is no safe canonical duplicate hub.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue is safe to choose because the set is a broad semantic bucket rather than one underlying bug; even the closest trainer-related items have different failure modes.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45701 is the best anchor because it is broad, clearly framed, and has inbound references, making it the most useful representative for the small tokenization-related subset.", + "best_issue_reason": "Issue 43708 is the strongest standalone report: it is concrete, reproducible, and narrowly scoped around a specific resume_from_checkpoint/max_steps bug.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44466", "right": "issue:45357", "accept": false, - "reason": "Both are serialization/save_pretrained regressions, but one is about tied lm_head weights and the other about visual encoder keys; different concrete bugs." + "reason": "Both involve save/tied-weight behavior, but one is about lm_head serialization and the other about incorrect visual encoder keys; different code paths and symptoms." }, { - "left": "issue:41720", - "right": "issue:44977", + "left": "issue:36331", + "right": "issue:42617", "accept": false, - "reason": "Both involve Qwen3-family models, but the failures are different code paths: auto device mapping/assertion vs flash-attention generation." + "reason": "Unexpected compute_loss kwargs and failure to run 3d_parallel.py are unrelated trainer/runtime problems." }, { - "left": "issue:36331", - "right": "issue:42617", + "left": "issue:41720", + "right": "issue:44977", "accept": false, - "reason": "Unexpected compute_loss kwargs and a 3d_parallel launch failure are unrelated issues." + "reason": "Both mention Qwen models, but one is a device-mapping crash and the other is flash-attention generation behavior; not the same bug." }, { "left": "issue:36331", "right": "issue:42915", "accept": false, - "reason": "Trainer API kwarg mismatch and Qwen3Moe FineGrainedFP8Config failure are different bugs." + "reason": "A compute_loss signature mismatch is unrelated to a Qwen3-MoE fine-grained FP8 failure." }, { "left": "issue:43450", "right": "issue:45701", "accept": false, - "reason": "Video processor batch-shape handling is unrelated to tokenization changes across versions." - }, - { - "left": "issue:44877", - "right": "issue:44991", - "accept": false, - "reason": "Both are loading failures, but one is strict config loading and the other is tokenizer loading for a specific model; not the same defect." + "reason": "Batched video-shape handling and version-dependent tokenization are different subsystems and failure modes." }, { "left": "issue:26413", "right": "issue:28530", "accept": false, - "reason": "Resume-from-checkpoint epoch_iterator failure and early-stopping metric lookup are distinct trainer behaviors." + "reason": "Both are trainer-related, but one is checkpoint resume failing on epoch iteration and the other is early stopping being disabled due to a missing metric." }, { - "left": "issue:40990", - "right": "issue:45701", + "left": "issue:44877", + "right": "issue:44991", "accept": false, - "reason": "High perplexity on gpt-oss-20b is too different from a generic tokenization regression to treat as the same bug." + "reason": "Strict config loading for granite_speech and tokenizer loading for est-roberta are unrelated bugs." }, { "left": "issue:43065", "right": "issue:46402", "accept": false, - "reason": "Both concern Sam3, but one is a dummy Conv2d in the pixel decoder and the other is missing text encoder weights." + "reason": "A dummy Conv2d in Sam3PixelDecoder is not the same as missing text-encoder weights in Sam3Model." + }, + { + "left": "issue:40990", + "right": "issue:45701", + "accept": false, + "reason": "High perplexity on gpt-oss-20b and tokenization changes across versions are not the same underlying issue." }, { "left": "issue:44186", "right": "issue:45072", "accept": false, - "reason": "Tokenizer crash on NER/batching and bfloat16 dtype mismatches are unrelated." + "reason": "LayoutLMv2 tokenizer crashes and dtype mismatches in bfloat16 inference are unrelated." }, { "left": "issue:43873", "right": "issue:44756", "accept": false, - "reason": "Quantized offloading behavior and mmap OOM on Strix Halo are different memory-management problems." + "reason": "Quantization/offload malfunction and mmap OOM on Strix Halo are distinct problems." }, { - "left": "issue:43550", - "right": "issue:43720", + "left": "issue:41720", + "right": "issue:42617", "accept": false, - "reason": "torch.compile+SDPA failure in Bamba and packed-weight unpacking in BitNet are unrelated model-loading/compile bugs." + "reason": "Multi-GPU training for Qwen2-VL and a 3d_parallel.py runtime failure are not the same bug." }, { - "left": "issue:41720", - "right": "issue:42617", + "left": "issue:43550", + "right": "issue:43720", "accept": false, - "reason": "Auto device mapping assertion and inability to run 3d_parallel.py are different issues." + "reason": "torch.compile with SDPA on Bamba-9B-v2 and packed-weight handling in BitNet loading are different code paths." }, { "left": "issue:43577", "right": "issue:43708", "accept": false, - "reason": "dtype propagation in Blip2 is unrelated to checkpoint resume max_steps calculation." + "reason": "Blip2 dtype loading and Trainer resume/max_steps calculation are unrelated." }, { "left": "issue:43404", "right": "issue:45357", "accept": false, - "reason": "Weight tying in Mistral3 and incorrect visual encoder keys in Qwen3.5 are different serialization/model-wiring bugs." + "reason": "Tied lm_head weights in Mistral3ForConditionalGeneration and incorrect visual encoder keys on save_pretrained are different issues." }, { "left": "issue:43550", "right": "issue:43606", "accept": false, - "reason": "Both mention device/compile behavior, but the concrete failures and code paths differ." + "reason": "A compile/SDPA problem and a CPU-offload device mismatch are not the same defect." }, { "left": "issue:43295", "right": "issue:45701", "accept": false, - "reason": "Processor.tokenizer access regression is not the same as a broad tokenization-behavior change; they overlap only loosely." + "reason": "A regression in processor.tokenizer access is not the same as version-dependent tokenization changes." }, { "left": "issue:43646", "right": "issue:46665", "accept": false, - "reason": "Both involve custom model loading, but one is a v5 initialization breakage and the other is inconsistent from_pretrained behavior inside Python classes." + "reason": "Both mention custom loading behavior, but one is broad Transformers 5.0 init breakage and the other is from_pretrained behavior inside Python classes." }, { "left": "issue:44186", "right": "issue:44610", "accept": false, - "reason": "Tokenizer crash on NER inputs and an image processor/model resolution mismatch are different bugs." + "reason": "Tokenizer crashes on NER/padding and an image processor output-size mismatch are unrelated." }, { "left": "issue:44008", "right": "issue:45823", "accept": false, - "reason": "Variable-name collision in Gemma3n forward and Gemma4 PLE device mismatch are unrelated forward-pass failures." + "reason": "Gemma3n variable-name collision in forward and Gemma4 PLE device mismatch are different bugs." }, { "left": "issue:44610", "right": "issue:44898", "accept": false, - "reason": "Both are image-resolution issues, but they involve different models and different failure mechanisms." + "reason": "Both involve image resolution, but one is wrong processor output size and the other is a model failure despite interpolate_pos_encoding." }, { "left": "issue:42915", "right": "issue:43317", "accept": false, - "reason": "FineGrainedFP8Config failure and dequantized model offload loading failure are not the same underlying bug." + "reason": "FineGrainedFP8Config failure and dequantized model offload loading failure are different quantization/device-map problems." }, { "left": "issue:44423", "right": "issue:45406", "accept": false, - "reason": "Both crash in serve, but one is a continuous-batching multimodal input handling bug and the other is a missing processor attribute." + "reason": "Both are serve multimodal crashes, but the failing attributes and processor stacks differ, so they are not clearly the same bug." }, { "left": "issue:36010", "right": "issue:42617", "accept": false, - "reason": "GenerationMixin import error and 3d_parallel execution failure are unrelated." + "reason": "GenerationMixin import failure and 3d_parallel.py execution failure are unrelated." }, { "left": "issue:45632", "right": "issue:45684", "accept": false, - "reason": "Cache-path collisions and read-only permission propagation are both filesystem-related but distinct bugs." + "reason": "Cache-path collisions for trust_remote_code and read-only permission propagation in save_pretrained are unrelated." }, { "left": "issue:41720", "right": "issue:45161", "accept": false, - "reason": "Different model families and different parallelism/device-placement failures; not a duplicate." + "reason": "Both involve distributed/parallel use, but they target different models and different failures." }, { "left": "issue:28530", "right": "issue:33666", "accept": false, - "reason": "Early stopping metric detection and Qwen2-VL multi-GPU training are unrelated." + "reason": "Early stopping metric handling and Qwen2-VL multi-GPU training are not the same issue." }, { "left": "issue:44671", "right": "issue:44743", "accept": false, - "reason": "CamemBERT masked-LM prediction regression and Qwen3.5 recurrent-state reset are different model-specific bugs." + "reason": "CamemBERT masked-LM regression and Qwen3.5 recurrent-state reset are unrelated model-specific bugs." }, { "left": "issue:43531", "right": "issue:43572", "accept": false, - "reason": "Sliding-window behavior in Qwen3-MoE and missing pad_token_idx in StableLmConfig are unrelated." + "reason": "Sliding-window behavior in Qwen3-MoE and a missing pad_token_idx in StableLmConfig are different regressions." }, { "left": "issue:42915", "right": "issue:46291", "accept": false, - "reason": "Qwen3Moe FP8 failure and kernels import/revision-version requirement are unrelated." + "reason": "A Qwen3-MoE FP8 failure and a kernels import requirement change are unrelated." }, { - "left": "issue:44186", - "right": "issue:44361", + "left": "issue:43526", + "right": "issue:43881", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and glm-4v-9b loading failure are different problems." + }, + { + "left": "issue:36331", + "right": "issue:41720", "accept": false, - "reason": "Both are tokenizer crashes, but on different tokenizers and different task paths; not the same concrete defect." + "reason": "A trainer compute_loss signature issue is unrelated to Qwen2-VL multi-GPU training." } ] }, @@ -4735,13 +4671,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 15, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:28530", "issue:29159", @@ -4751,7 +4687,6 @@ "issue:34634", "issue:35707", "issue:36032", - "issue:36331", "issue:41720", "issue:42490", "issue:42915", @@ -4759,18 +4694,19 @@ "issue:43322", "issue:43329", "issue:43388", - "issue:43526", "issue:43531", "issue:43540", "issue:43550", "issue:43606", + "issue:43644", "issue:43653", "issue:43720", "issue:43746", "issue:43782", "issue:43873", - "issue:43881", "issue:44077", + "issue:44186", + "issue:44361", "issue:44387", "issue:44451", "issue:44492", @@ -4793,8 +4729,7 @@ "issue:46665" ], "soft_pairs": [ - "issue:43526|issue:43881", - "issue:36331|issue:41720", + "issue:44186|issue:44361", "issue:43316|issue:44913", "issue:42915|issue:43322", "issue:28530|issue:30064", @@ -4814,217 +4749,218 @@ "issue:42490|issue:43329", "issue:43606|issue:43653", "issue:45406|issue:45464", - "issue:34567|issue:35707", "issue:45357|issue:46402", + "issue:34567|issue:35707", "issue:34634|issue:41720", "issue:45759|issue:45859", "issue:44451|issue:44991", "issue:43531|issue:44589", "issue:44625|issue:45479", - "issue:44077|issue:44625" + "issue:44077|issue:44625", + "issue:43644|issue:46665" ], - "prepared_review_unit_hash": "b2995d1b83e384ff7937d38d4db05a6b6148b2b927f3fec4387311e2a2ea9c7c", + "prepared_review_unit_hash": "68a43e621647e571cde589a1110ec32d26fbae6efcd9f77c4403be8bee592cbf", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36416, - "estimated_input_tokens": 9104, - "estimated_eval_tokens": 18464 + "serialized_chars": 36387, + "estimated_input_tokens": 9097, + "estimated_eval_tokens": 18450 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These items are a heterogeneous set of Transformers issues spanning tokenizers, configs, multimodal loading, training, quantization, offloading, and docs. None of the soft-edge pairs look like the same underlying bug/change, so all candidate duplicate links should be rejected.", - "confidence": 0.95, - "canonical_issue_reason": "No single canonical issue is a good fit here: the cluster is too broad and the soft pairs are mostly adjacent-sounding but materially different defects.", + "summary": "The set is mostly a heterogeneous mix of unrelated bug reports. I found no strong duplicate pair among the listed soft edges; the closest thematic overlap is the non-persistent-buffer regression, but the provided pairings do not line up tightly enough to merge.", + "confidence": 0.81, + "canonical_issue_reason": "No clear duplicate cluster emerges from the listed pairs. Issue 43644 is the best representative because it is a concrete, well-scoped regression report with a crisp failure mode and version context.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43873 is the best representative issue in this set because it has the broadest cross-cutting loading/offloading/quantization theme and relatively high discussion/inbound activity.", + "best_issue_reason": "Issue 43644 is the strongest standalone issue in the set: specific, reproducible, and easy to action. It is the clearest representative even though the other items are not true duplicates of it.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43526", - "right": "issue:43881", - "accept": false, - "reason": "BEiT Fast label reduction bug vs GLM-4v model loading failure; different models and failure modes." - }, - { - "left": "issue:36331", - "right": "issue:41720", + "left": "issue:44186", + "right": "issue:44361", "accept": false, - "reason": "Trainer custom loss signature regression vs Qwen3 device-mapping CUDA assert; unrelated code paths." + "reason": "Both are tokenizer/model-input crashes, but they affect different models and different failure modes (LayoutLMv2 NER padding/truncation vs MLuke AttributeError)." }, { "left": "issue:43316", "right": "issue:44913", "accept": false, - "reason": "Gemma3TextConfig API inconsistency vs GPTNeoX rotary_pct reload bug; not the same defect." + "reason": "Both mention config inconsistencies, but one is a Gemma3TextConfig API mismatch and the other is GPTNeoX rotary_pct not persisting on reload; different fields and code paths." }, { "left": "issue:42915", "right": "issue:43322", "accept": false, - "reason": "Qwen3 MoE FP8 config failure vs Llava Next segfault; different models and causes." + "reason": "Different subsystems and failures: Qwen3Moe + FineGrainedFP8Config versus Llava Next segmentation fault on model load." }, { "left": "issue:28530", "right": "issue:30064", "accept": false, - "reason": "Early stopping metric selection issue vs void segmentation-map processing failure; unrelated training vs image-processing bugs." + "reason": "Early-stopping metric resolution and image processor segmentation-map handling are unrelated bugs." }, { "left": "issue:33357", "right": "issue:35707", "accept": false, - "reason": "MacOS CLIP bus error vs generation with inputs_embeds/past_key_values; no shared code-path bug." + "reason": "A MacOS bus error in CLIP loading and a generation bug with inputs_embeds/past_key_values are different runtime paths." }, { "left": "issue:43540", "right": "issue:43782", "accept": false, - "reason": "Qwen3OmniMoe video-input validation bug vs Qwen3VL weight_only load error; different multimodal loading problems." + "reason": "Both are Qwen multimodal issues, but one is video-input processing in Qwen3OmniMoe and the other is a weight_only from_pretrained failure in Qwen3VL; not the same bug." }, { "left": "issue:29159", "right": "issue:36032", "accept": false, - "reason": "Slow/fast tokenizer inconsistency vs T5 tokenizer special-method conflict; both tokenizer-related but not the same bug." + "reason": "Slow/fast tokenizer inconsistency is a broad tokenizer behavior issue, while T5Tokenizer load fails due to an add_special_tokens method conflict." }, { "left": "issue:43550", "right": "issue:43653", "accept": false, - "reason": "Bamba torch.compile/SDPA failure vs BigBirdTokenizer special-token registration bug; unrelated subsystems." + "reason": "Bamba SDPA/torch.compile failure and BigBirdTokenizer special-token registration are unrelated." }, { "left": "issue:43873", "right": "issue:45704", "accept": false, - "reason": "Quantization/offloading memory behavior vs T5 apex RMSNorm memory leak; different root causes and code paths." + "reason": "Quantization/offloading behavior versus an Apex RMSNorm memory leak are different underlying problems." }, { "left": "issue:43606", "right": "issue:43720", "accept": false, - "reason": "Bark CPU offload device mismatch vs BitNet packed-weights loading issue; both around loading, but distinct failures." + "reason": "Both involve loading/offload machinery, but one is a Bark CPU-offload device mismatch and the other is BitNet packed-weight unpacking during accelerate loading." }, { "left": "issue:44387", "right": "issue:45701", "accept": false, - "reason": "int4 quantization OOM regression vs tokenization changes across versions; not the same underlying issue." + "reason": "CUDA memory growth under int4 quantization is unrelated to tokenizer changes across versions." }, { "left": "issue:44492", "right": "issue:44509", "accept": false, - "reason": "Cache-strategy typo vs docs mentioning removed pipeline tasks; documentation-only issues with different scopes." + "reason": "A typo in cache strategies is not the same as stale docs mentioning removed pipeline tasks." }, { "left": "issue:30064", "right": "issue:34634", "accept": false, - "reason": "Void segmentation-map processing vs Bark voice_preset handling; different processors and bugs." + "reason": "Void segmentation-map processing and BarkProcessor voice_preset handling are separate model/component bugs." }, { "left": "issue:44534", "right": "issue:46665", "accept": false, - "reason": "Non-persistent buffer initialization bug vs from_pretrained behavior inside Python classes; not a duplicate." + "reason": "Both touch loading behavior in v5, but one is non-persistent buffer initialization and the other is from_pretrained used inside Python classes; different code paths." }, { "left": "issue:41720", "right": "issue:43746", "accept": false, - "reason": "Qwen3 auto device mapping CUDA assert vs GraniteSpeech PEFT local checkpoint loading; different model/load paths." + "reason": "Qwen3 auto device mapping cudaErrorAssert and GraniteSpeech PEFT local checkpoint loading are unrelated failures." }, { "left": "issue:43388", "right": "issue:45059", "accept": false, - "reason": "gather_for_metrics last-batch label truncation vs SAM3 text+box behavior; unrelated training utility vs model interaction." + "reason": "A metrics-gathering batch truncation bug is unrelated to SAM3 text/bounding-box processing behavior." }, { "left": "issue:43531", "right": "issue:45759", "accept": false, - "reason": "Qwen3-MoE sliding_window issue vs from_config not unwrapping multimodal text_config; both Qwen-related, but different bugs." + "reason": "Sliding-window handling in Qwen3-MoE and from_config text_config unwrapping for composite Qwen multimodal configs are different bugs." }, { "left": "issue:42490", "right": "issue:43329", "accept": false, - "reason": "Trailing-slash import warning vs undefined video-branch variables in multimodal token counting; unrelated." + "reason": "Import warning for trailing slashes and an undefined video branch in multimodal token counting are unrelated." }, { "left": "issue:43606", "right": "issue:43653", "accept": false, - "reason": "Bark CPU offload mismatch vs BigBirdTokenizer mask-token registration; different components and symptoms." + "reason": "CPU offload device mismatch in Bark and BigBirdTokenizer special-token decode behavior are different issues." }, { "left": "issue:45406", "right": "issue:45464", "accept": false, - "reason": "Gemma4Processor missing _tokenizer in serve vs Qwen3.5 streaming inference failure; different APIs and failure modes." + "reason": "Gemma4Processor missing _tokenizer in serve and Qwen3.5 streaming chat/completions failure are separate serving bugs." }, { - "left": "issue:34567", - "right": "issue:35707", + "left": "issue:45357", + "right": "issue:46402", "accept": false, - "reason": "TrainerState token counter not updating vs generation with inputs_embeds/pkv; no shared underlying defect." + "reason": "Both concern multimodal weight loading, but one is incorrect visual encoder keys on save_pretrained and the other is Sam3Model text encoder weights not loading." }, { - "left": "issue:45357", - "right": "issue:46402", + "left": "issue:34567", + "right": "issue:35707", "accept": false, - "reason": "Qwen3.5 save_pretrained key mismatch vs Sam3Model text encoder weights loading; both model I/O, but distinct bugs." + "reason": "TrainerState token-count tracking and progressive generation with inputs_embeds/past_key_values are unrelated." }, { "left": "issue:34634", "right": "issue:41720", "accept": false, - "reason": "BarkProcessor voice_preset bug vs Qwen3 device mapping CUDA assert; unrelated." + "reason": "Bark voice_preset handling and Qwen3 auto device mapping are not the same failure mode." }, { "left": "issue:45759", "right": "issue:45859", "accept": false, - "reason": "Composite config unwrapping bug vs CPU-offload incompatibility in Qwen3_5MoeTextRotaryEmbedding; different code paths." + "reason": "Composite config unwrapping and CPU-offload incompatibility in Qwen3_5MoeTextRotaryEmbedding are different problems." }, { "left": "issue:44451", "right": "issue:44991", "accept": false, - "reason": "ScandiBERT loading failure vs tokenizer loading regression for est-roberta; separate model-specific issues." + "reason": "Both are tokenizer-loading regressions, but they affect different models and are not clearly the same root cause." }, { "left": "issue:43531", "right": "issue:44589", "accept": false, - "reason": "Qwen3-MoE sliding-window bug vs Float8 storage lookup error; different failure surfaces." + "reason": "Sliding-window logic and missing Float8 storage object errors are unrelated." }, { "left": "issue:44625", "right": "issue:45479", "accept": false, - "reason": "num_labels propagation bug vs zero-loss classification bug; related to labels, but not the same defect." + "reason": "A config propagation bug for Qwen3.5 and a sequence-classification loss degeneration bug are different issues." }, { "left": "issue:44077", "right": "issue:44625", "accept": false, - "reason": "PatchTSMixer post_init restriction vs Qwen3.5 num_labels propagation; unrelated configuration issues." + "reason": "Optional post_init on patchtsmixer and num_labels propagation in Qwen3.5 are unrelated." + }, + { + "left": "issue:43644", + "right": "issue:46665", + "accept": false, + "reason": "Non-persistent buffer junk on v5 and from_pretrained behavior inside Python classes do not describe the same defect." } ] }, @@ -5040,13 +4976,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 16, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:30064", "issue:36010", @@ -5056,7 +4992,7 @@ "issue:43381", "issue:43421", "issue:43531", - "issue:43644", + "issue:43550", "issue:43645", "issue:43653", "issue:43720", @@ -5094,11 +5030,9 @@ "issue:46050", "issue:46463", "issue:46655", - "issue:46665", "issue:46693" ], "soft_pairs": [ - "issue:43644|issue:46665", "issue:44898|issue:45072", "issue:44062|issue:44451", "issue:44368|issue:45357", @@ -5110,16 +5044,16 @@ "issue:45704|issue:45923", "issue:44918|issue:44936", "issue:41720|issue:45704", - "issue:43746|issue:46050", "issue:43381|issue:45704", + "issue:43746|issue:46050", "issue:44514|issue:44545", "issue:43531|issue:43645", "issue:42886|issue:44843", "issue:46032|issue:46463", - "issue:43421|issue:43653", "issue:45759|issue:45923", - "issue:43994|issue:44671", + "issue:43421|issue:43653", "issue:39692|issue:43873", + "issue:43994|issue:44671", "issue:43746|issue:43873", "issue:43782|issue:45381", "issue:43421|issue:44186", @@ -5127,222 +5061,223 @@ "issue:30064|issue:36010", "issue:45356|issue:45907", "issue:43720|issue:43819", - "issue:43653|issue:43720" + "issue:43653|issue:43720", + "issue:43550|issue:45072" ], - "prepared_review_unit_hash": "bbf4a3036b5cc109d3867edd4c8654804364492c3ec55726658b8317d05c0590", + "prepared_review_unit_hash": "9737738d72c5f31cc657b3eed3e28be62b279c2974782106192134c9c3d74595", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 37055, - "estimated_input_tokens": 9264, - "estimated_eval_tokens": 18784 + "serialized_chars": 36489, + "estimated_input_tokens": 9123, + "estimated_eval_tokens": 18502 }, "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. Only two soft edges look like true duplicates: the Qwen2_5_VLProcessor padding=False crash, and the HF_HUB_OFFLINE tokenizer/cache regression. The rest are different bugs in different code paths.", - "confidence": 0.78, - "canonical_issue_reason": "issue:44843 is the strongest canonical issue here: it names a concrete tokenizer offline-mode failure, identifies the specific code path (_patch_mistral_regex / model_info), and is more diagnostic than the broader cache-failure report.", + "summary": "Most items are unrelated bug reports. Only two soft edges look like true duplicates: the Qwen2.5-VL batched chat-template crash, and the HF_HUB_OFFLINE tokenizer regression around _patch_mistral_regex/model_info.", + "confidence": 0.88, + "canonical_issue_reason": "Issue 44843 is the clearest root-cause report for the offline tokenizer/Mistral-regex regression; issue 42886 reads as the broader symptom report of the same failure.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44843 is the best standalone issue in this set because it is specific, actionable, and tied to a clear root cause and reproduction surface.", + "best_issue_reason": "44843 is the strongest canonical issue in this set: it names a specific code path, a reproducible offline-mode regression, and cleanly subsumes the related cache-loading complaint.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43644", - "right": "issue:46665", - "accept": false, - "reason": "Different problems: one is non-persistent buffer initialization, the other is from_pretrained behavior inside Python classes." - }, { "left": "issue:44898", "right": "issue:45072", "accept": false, - "reason": "Different models and failure modes: image resolution handling vs dtype mismatch in inference." + "reason": "Different models and failures: image-resolution/interpolation bug vs bfloat16 dtype mismatch in another model family." }, { "left": "issue:44062", "right": "issue:44451", "accept": false, - "reason": "Tokenizer AddedToken argument error is unrelated to loading a specific model from the hub." + "reason": "AddedToken special-argument error is unrelated to the ScandiBERT loading failure." }, { "left": "issue:44368", "right": "issue:45357", "accept": false, - "reason": "A warning about tie_word_embeddings is unrelated to incorrect visual encoder keys on save_pretrained." + "reason": "Different problems: tie_word_embeddings warning during LoRA fine-tuning vs incorrect visual-encoder keys on save_pretrained." }, { "left": "issue:43381", "right": "issue:43844", "accept": false, - "reason": "Eval-mode gradient checkpointing and ZeRO-3 gradient explosion are unrelated training issues." + "reason": "Eval-mode gradient checkpointing restriction is not the same as the ZeRO-3 gradient explosion regression." }, { "left": "issue:44987", "right": "issue:45042", "accept": false, - "reason": "Model loading from the hub is unrelated to PIL image processors incorrectly requiring torchvision." + "reason": "Different subsystems and symptoms; one is a model-loading regression, the other an image-processor torchvision dependency issue." }, { "left": "issue:46655", "right": "issue:46693", "accept": false, - "reason": "Video padding calculation bug is unrelated to FA2 long-context performance regression." + "reason": "Video padding bug in SmolVLM is unrelated to the FA2 long-context performance regression." }, { "left": "issue:39692", "right": "issue:44610", "accept": false, - "reason": "Doc example/model mismatch and quantization errors are unrelated to OmDet-Turbo processor size mismatch." + "reason": "Both mention processor/model mismatch, but they are different models and different concrete bugs." }, { "left": "issue:43421", "right": "issue:43927", "accept": false, - "reason": "Runtime special-token post-processing and save/load token ID loss are different tokenizer configuration bugs." + "reason": "Runtime special-token/post-processor update bug is distinct from DiaConfig losing custom token IDs on save/load." }, { "left": "issue:45704", "right": "issue:45923", "accept": false, - "reason": "A T5 RMSNorm memory leak is unrelated to a missing supports_gradient_checkpointing flag." + "reason": "T5 RMSNorm memory leak and missing gradient-checkpointing flag are unrelated issues." }, { "left": "issue:44918", "right": "issue:44936", "accept": false, - "reason": "TRL SFT embedding unpacking failure is unrelated to trainer.evaluate() failing after trainer.train()." + "reason": "TRL input-embedding unpacking failure is not the same as a generic evaluate-after-train failure." }, { "left": "issue:41720", "right": "issue:45704", "accept": false, - "reason": "Qwen3 device-mapping CUDA assert is unrelated to the T5 Apex RMSNorm leak." + "reason": "Qwen3 cudaErrorAssert and T5 apex memory leak are unrelated." }, { - "left": "issue:43746", - "right": "issue:46050", + "left": "issue:43381", + "right": "issue:45704", "accept": false, - "reason": "PEFT adapter checkpoint loading and sinq quantized save/load breakage are different persistence paths." + "reason": "Gradient checkpointing eval-mode restriction and apex memory leak do not share a code path." }, { - "left": "issue:43381", - "right": "issue:45704", + "left": "issue:43746", + "right": "issue:46050", "accept": false, - "reason": "Gradient checkpointing eval-mode behavior is unrelated to a fused RMSNorm memory leak." + "reason": "PEFT adapter local-checkpoint loading differs from sinq quantized save/load corruption." }, { "left": "issue:44514", "right": "issue:44545", "accept": true, - "reason": "These are the same Qwen2_5_VLProcessor.apply_chat_template batched-input/padding=False crash, just reported twice." + "reason": "Same Qwen2_5_VLProcessor.apply_chat_template batched-input crash with padding=False; these are duplicate reports of the same bug." }, { "left": "issue:43531", "right": "issue:43645", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior is unrelated to custom model initialization in Jupyter notebooks." + "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to the Jupyter custom-model initialization regression." }, { "left": "issue:42886", "right": "issue:44843", "accept": true, - "reason": "Both describe the same offline tokenizer-loading regression; 44843 gives the concrete cause in _patch_mistral_regex/model_info, while 42886 is the broader symptom report." + "reason": "Both report tokenizer loading breaking in HF_HUB_OFFLINE mode from the same _patch_mistral_regex/model_info code path." }, { "left": "issue:46032", "right": "issue:46463", "accept": false, - "reason": "Mamba2Mixer use_cache correctness and T5 decoder-input validation error are unrelated." - }, - { - "left": "issue:43421", - "right": "issue:43653", - "accept": false, - "reason": "Special-token post-processor updates and BigBirdTokenizer mask-token registration are different tokenizer bugs." + "reason": "Incorrect use_cache results in Mamba2Mixer is unrelated to a T5 error-message validation bug." }, { "left": "issue:45759", "right": "issue:45923", "accept": false, - "reason": "Composite Qwen config handling is unrelated to a trust_remote_code capability flag." + "reason": "Composite-config unwrapping for Qwen multimodal models is unrelated to a missing gradient-checkpointing flag." }, { - "left": "issue:43994", - "right": "issue:44671", + "left": "issue:43421", + "right": "issue:43653", "accept": false, - "reason": "SigLIP2 AutoModel/pipeline behavior is unrelated to CamemBERT masked-LM predictions." + "reason": "Both involve special tokens, but 43653 is a missing registration bug while 43421 is about post-processor refresh after runtime edits." }, { "left": "issue:39692", "right": "issue:43873", "accept": false, - "reason": "Documentation example/model mismatch is unrelated to quantization offloading behavior." + "reason": "SigLIP2 docs/model mismatch is unrelated to quantization offloading behavior." + }, + { + "left": "issue:43994", + "right": "issue:44671", + "accept": false, + "reason": "Different models and failure modes: SigLIP2 nonsensical output vs CamemBERT masked-LM regression." }, { "left": "issue:43746", "right": "issue:43873", "accept": false, - "reason": "Local PEFT checkpoint loading is unrelated to offloading with quantization." + "reason": "PEFT adapter loading and quantization offloading are separate bugs." }, { "left": "issue:43782", "right": "issue:45381", "accept": false, - "reason": "Qwen3VL weight_only loading error is unrelated to Qwen2.5-VL vision_position_ids." + "reason": "Qwen3VL weight_only loading error is not the same as Qwen2.5-VL video position-id corruption." }, { "left": "issue:43421", "right": "issue:44186", "accept": false, - "reason": "Runtime special-token handling is unrelated to LayoutLMv2Tokenizer padding/truncation crashes." + "reason": "LayoutLMv2 tokenizer crashes on NER/padding are not the same bug as special-token post-processor refresh." }, { "left": "issue:45081", "right": "issue:45356", "accept": false, - "reason": "Both involve tokenizer regex/codec topics, but they describe different failures and are not clearly the same concrete bug." + "reason": "Both touch fix_mistral_regex, but one is an AttributeError crash and the other is a codec-handling/warning regression; not clearly the same underlying bug." }, { "left": "issue:30064", "right": "issue:36010", "accept": false, - "reason": "Void segmentation map processing is unrelated to the GenerationMixin import error." + "reason": "Void segmentation-map processing and a GenerationMixin import error are unrelated." }, { "left": "issue:45356", "right": "issue:45907", "accept": false, - "reason": "Tokenizer codec/regex regression is unrelated to list_repo_files revision forwarding." + "reason": "Tokenizer codec/regression issues are unrelated to list_repo_files missing the revision argument." }, { "left": "issue:43720", "right": "issue:43819", "accept": false, - "reason": "Packed-weight loading in BitNet is unrelated to DAC.from_latents/STE mismatch." + "reason": "BitNet packed-weight loading and DAC from_latents/STE mismatch are different code paths." }, { "left": "issue:43653", "right": "issue:43720", "accept": false, - "reason": "BigBirdTokenizer special-token registration is unrelated to BitNet packed-weight loading." + "reason": "BigBird special-token decode issue is unrelated to BitNet packed-weight loading." + }, + { + "left": "issue:43550", + "right": "issue:45072", + "accept": false, + "reason": "Torch.compile+SDPA failure in Bamba-9B-v2 is unrelated to bfloat16 dtype mismatches in other models." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative on the soft edges. The two accepted pairs are well-supported by their titles and issue descriptions, and the remaining rejections do not overclaim duplicate status. The canonical/standalone issue choice for 44843 is also reasonable based on specificity and diagnostic detail." + "feedback": "Grounded overall and appropriately conservative. The two accepted duplicate pairs are well supported by the titles, and the other soft-edge rejections avoid overclaiming despite some superficial similarity." }, "evaluator_used": true, "retried": false, @@ -5355,13 +5290,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 17, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:30064", "issue:36331", @@ -5399,7 +5334,6 @@ "issue:44877", "issue:44898", "issue:44991", - "issue:45072", "issue:45081", "issue:45084", "issue:45183", @@ -5413,231 +5347,224 @@ "issue:46693" ], "soft_pairs": [ - "issue:43550|issue:45072", - "issue:43540|issue:46294", "issue:43295|issue:45834", + "issue:43540|issue:46294", "issue:43550|issue:43927", "issue:44062|issue:44488", - "issue:45381|issue:46693", "issue:43596|issue:44220", + "issue:45381|issue:46693", "issue:43618|issue:43697", "issue:43611|issue:46665", "issue:43653|issue:43819", - "issue:43531|issue:43643", "issue:44186|issue:44898", + "issue:43531|issue:43643", "issue:44220|issue:44661", "issue:43335|issue:43756", "issue:43550|issue:43819", "issue:44843|issue:45081", "issue:41720|issue:44991", - "issue:43540|issue:44008", "issue:43643|issue:44877", "issue:30064|issue:36331", + "issue:43540|issue:44008", "issue:44821|issue:45084", "issue:39692|issue:45834", "issue:44186|issue:44442", + "issue:44485|issue:44509", "issue:44448|issue:44671", "issue:45183|issue:45479", - "issue:44485|issue:44509", - "issue:43322|issue:45685", - "issue:43388|issue:46032" + "issue:43388|issue:46032", + "issue:43322|issue:45685" ], - "prepared_review_unit_hash": "9e8faafaa7d3c55e2690566576454c24cf28fcfb817c2cd1acc980cfc9eb0427", + "prepared_review_unit_hash": "ff40d8f9a3fde6f94cac4e4da52b927ac19bb1045aa21b822c58a1e4e1b3eb84", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 28, - "serialized_chars": 36410, - "estimated_input_tokens": 9103, - "estimated_eval_tokens": 18462 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35521, + "estimated_input_tokens": 8881, + "estimated_eval_tokens": 18018 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is highly heterogeneous: the items cover unrelated bugs across tokenizers, vision/multimodal models, configs, docs, and generation/runtime regressions. None of the soft pairs look like the same underlying issue, so all soft edges should be rejected.", - "confidence": 0.97, - "canonical_issue_reason": "No single canonical issue stands out; the artifacts do not describe one shared bug and span multiple unrelated subsystems and model families.", + "summary": "The cluster is heterogeneous: it contains many unrelated model, tokenizer, config, and runtime bugs across different code paths. The soft-similarity pairs are mostly broad thematic overlaps rather than the same underlying defect, so none should be merged as duplicates. No single issue is a clear canonical representative.", + "confidence": 0.93, + "canonical_issue_reason": "No clear canonical issue: the items span unrelated subsystems and failure modes, with only loose topical similarity.", "canonical_pr_reason": null, - "best_issue_reason": "No issue is a good global representative because the cluster is not a coherent duplicate set.", + "best_issue_reason": "No strong global representative issue stands out because the cluster is too mixed to summarize with one artifact.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43550", - "right": "issue:45072", + "left": "issue:43295", + "right": "issue:45834", "accept": false, - "reason": "Different failures in different models: torch.compile/SDPA on Bamba vs bfloat16 dtype mismatch in SwitchTransformers/TimmWrapperModel." + "reason": "Different bugs: processor/tokenizer regression vs long OCR index error in a different model." }, { "left": "issue:43540", "right": "issue:46294", "accept": false, - "reason": "Both are video-related but the bugs differ: Qwen3OmniMoe input processing ValueError vs Qwen3.5-9B chat-template token/feature count mismatch." - }, - { - "left": "issue:43295", - "right": "issue:45834", - "accept": false, - "reason": "Unrelated regression types: processor.tokenizer/images API breakage vs a Kosmos2.5 long-OCR index error." + "reason": "Both involve video inputs, but they fail in different Qwen models and different code paths (processing vs chat template token/feature mismatch)." }, { "left": "issue:43550", "right": "issue:43927", "accept": false, - "reason": "Different code paths and symptoms: SDPA/torch.compile crash vs config save/load losing custom token IDs and causing generation IndexError." + "reason": "Unrelated: torch.compile/SDPA failure in Bamba vs DiaConfig token-id persistence bug." }, { "left": "issue:44062", "right": "issue:44488", "accept": false, - "reason": "Tokenizer constructor argument collision vs a model loading failure for cjvt/sleng-bert; not the same underlying bug." + "reason": "Both tokenizer-related, but the errors and affected models are different; not the same defect." }, { - "left": "issue:45381", - "right": "issue:46693", + "left": "issue:43596", + "right": "issue:44220", "accept": false, - "reason": "Different regressions: Qwen2.5-VL video position ids bug vs a performance regression with FA2 and long-context generation." + "reason": "Completely different areas: Deepspeed/BertModel init error vs audio fbank feature extraction." }, { - "left": "issue:43596", - "right": "issue:44220", + "left": "issue:45381", + "right": "issue:46693", "accept": false, - "reason": "Different failures: DeepSpeed ZeRO-3/BertModel initialization IndexError vs an audio fbank feature extraction issue." + "reason": "Different regressions: Qwen2.5-VL video position IDs vs FA2 long-context performance regression." }, { "left": "issue:43618", "right": "issue:43697", "accept": false, - "reason": "CLIPOutput attentions field regression is unrelated to RTDetrV2 producing different outputs in v5." + "reason": "Both are v5 vision-model regressions, but they affect different outputs and code paths; not one bug." }, { "left": "issue:43611", "right": "issue:46665", "accept": false, - "reason": "Different loading problems: base_model_prefix handling in Transformers 5.0 vs inconsistent from_pretrained behavior inside Python classes." + "reason": "Both concern loading, but one is base_model_prefix behavior and the other is from_pretrained inside Python classes." }, { "left": "issue:43653", "right": "issue:43819", "accept": false, - "reason": "Different tokenizer/storage issues: BigBirdTokenizer special-token registration vs DAC.from_latents not matching forward without STE." + "reason": "Tokenizer special-token registration bug vs audio codec forward/from_latents mismatch; unrelated." }, { - "left": "issue:43531", - "right": "issue:43643", + "left": "issue:44186", + "right": "issue:44898", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior is unrelated to AutoConfig.from_pretrained with trust_remote_code missing fields." + "reason": "Different models and failures: LayoutLMv2 tokenizer padding/truncation vs Perceiver image classification resolution handling." }, { - "left": "issue:44186", - "right": "issue:44898", + "left": "issue:43531", + "right": "issue:43643", "accept": false, - "reason": "Both are image-related but the bugs differ: LayoutLMv2Tokenizer crashes on NER/batched padding vs Perceiver non-default resolution classification failure." + "reason": "Sliding-window MoE behavior is unrelated to trust_remote_code config field loss." }, { "left": "issue:44220", "right": "issue:44661", "accept": false, - "reason": "Audio feature extraction issue vs add-new-model-like failing for models already in TOKENIZER_MAPPING_NAMES." + "reason": "Audio feature extraction bug vs tooling failure in add-new-model-like; different problem domains." }, { "left": "issue:43335", "right": "issue:43756", "accept": false, - "reason": "SwitchTransformers sparse-layer creation bug is unrelated to Smollm3 RoPE layer-dropping behavior." + "reason": "Both are architecture/config issues, but they concern different models and different layer-selection logic." }, { "left": "issue:43550", "right": "issue:43819", "accept": false, - "reason": "Different models and mechanisms: torch.compile with SDPA on Bamba vs DAC latent decoding mismatch from missing STE." + "reason": "Different bug types and code paths: compile/SDPA vs DAC latent decoding mismatch." }, { "left": "issue:44843", "right": "issue:45081", "accept": false, - "reason": "Both touch tokenizer regex logic, but one is offline model_info() access in _patch_mistral_regex and the other is a backend_tokenizer attribute crash; not the same bug." + "reason": "Same helper function name, but distinct failures: offline hub call vs attribute error on tokenizer object." }, { "left": "issue:41720", "right": "issue:44991", "accept": false, - "reason": "Unrelated model/runtime failures: Qwen3 auto device mapping cudaErrorAssert vs tokenizer loading failure for EMBEDDIA/est-roberta." - }, - { - "left": "issue:43540", - "right": "issue:44008", - "accept": false, - "reason": "Different multimodal failures: video input ValueError in Qwen3OmniMoe vs Gemma 3n forward path tensor/name collision causing AttributeError." + "reason": "Unrelated: Qwen3 device-mapping assert vs tokenizer loading failure for a specific RoBERTa model." }, { "left": "issue:43643", "right": "issue:44877", "accept": false, - "reason": "Both are config-loading issues, but one concerns trust_remote_code missing fields and the other strict config blocking granite_speech; not the same bug." + "reason": "Different config-loading problems with different triggers; not the same underlying defect." }, { "left": "issue:30064", "right": "issue:36331", "accept": false, - "reason": "Image Processor void segmentation map failure is unrelated to CustomTrainer.compute_loss() receiving an unexpected num_items_in_batch argument." + "reason": "Image processor segmentation-map failure vs CustomTrainer compute_loss signature regression are unrelated." + }, + { + "left": "issue:43540", + "right": "issue:44008", + "accept": false, + "reason": "Both mention multimodal inputs, but one is a video-processing ValueError and the other is a Gemma3n variable collision." }, { "left": "issue:44821", "right": "issue:45084", "accept": false, - "reason": "Different loading/template failures: AutoImageProcessor from URL vs a template compilation TypeError." + "reason": "Different issues: AutoImageProcessor URL loading vs template-node compilation error." }, { "left": "issue:39692", "right": "issue:45834", "accept": false, - "reason": "SigLIP2 docs example errors and Kosmos2.5 long OCR index error are separate problems, not duplicates." + "reason": "Docs/example mistakes in SigLIP2 vs runtime OCR index error in Kosmos2.5; not the same bug." }, { "left": "issue:44186", "right": "issue:44442", "accept": false, - "reason": "LayoutLMv2Tokenizer NER/padding crash is unrelated to AutoTokenizer failing to load FastSpeech2ConformerTokenizer." + "reason": "Tokenizer padding/truncation crash vs AutoTokenizer loading failure for another tokenizer; unrelated." + }, + { + "left": "issue:44485", + "right": "issue:44509", + "accept": false, + "reason": "RoPE implementation discussion vs stale docs mentioning removed pipeline tasks; different changes." }, { "left": "issue:44448", "right": "issue:44671", "accept": false, - "reason": "Different output regressions in different models: Pegasus v4/v5 output mismatch vs CamemBERT masked LM prediction errors." + "reason": "Both are v5 output regressions, but they affect different models and behaviors, so not mergeable as one fix." }, { "left": "issue:45183", "right": "issue:45479", "accept": false, - "reason": "These are different validation bugs: wrong XOR error messaging for input_ids/inputs_embeds vs zero-loss classification when num_labels=1." + "reason": "Validation XOR error across models vs degenerate single-label classification loss; different bugs." }, { - "left": "issue:44485", - "right": "issue:44509", + "left": "issue:43388", + "right": "issue:46032", "accept": false, - "reason": "GLM-5 RoPE implementation discussion is unrelated to docs still listing removed pipeline tasks." + "reason": "Metric-gathering last-batch truncation vs Mamba2Mixer use_cache correctness issue; unrelated." }, { "left": "issue:43322", "right": "issue:45685", "accept": false, - "reason": "Different runtime issues: Llava Next loading segfault vs MPS histogram not implemented for Int in moe path." - }, - { - "left": "issue:43388", - "right": "issue:46032", - "accept": false, - "reason": "Different training/inference bugs: gather_for_metrics dropping tuple labels vs Mamba2Mixer use_cache producing incorrect results for seq_len > 1." + "reason": "Segfault loading Llava Next models vs MPS histogram implementation error in moe; no shared underlying bug." } ] }, @@ -5653,13 +5580,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 18, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:28218", "issue:29159", @@ -5710,9 +5637,9 @@ "issue:46665" ], "soft_pairs": [ - "issue:43716|issue:44008", "issue:43644|issue:43749", "issue:45584|issue:45834", + "issue:43716|issue:44008", "issue:43606|issue:45072", "issue:34634|issue:36010", "issue:43950|issue:45987", @@ -5720,8 +5647,8 @@ "issue:43582|issue:46512", "issue:43825|issue:43994", "issue:43577|issue:43818", - "issue:46294|issue:46463", "issue:43421|issue:44811", + "issue:46294|issue:46463", "issue:43782|issue:44977", "issue:44991|issue:46291", "issue:44561|issue:45020", @@ -5739,209 +5666,212 @@ "issue:42175|issue:45081", "issue:34634|issue:36331" ], - "prepared_review_unit_hash": "ed41ef1380c9793e725f2af008e774e5b172713532613ec96e942e169ed08615", + "prepared_review_unit_hash": "7607169b9b999cf28a11fdd562af472cbadbd37474a8021c37c6538b11a4f1f2", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 28, - "serialized_chars": 35834, - "estimated_input_tokens": 8959, - "estimated_eval_tokens": 18174 + "serialized_chars": 35837, + "estimated_input_tokens": 8960, + "estimated_eval_tokens": 18176 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is mostly heterogeneous; the soft pairs share broad themes (tokenizers, loading, video inputs, dtype issues) but not the same concrete bug or mergeable code change, so I would not consolidate any of them.", - "confidence": 0.88, - "canonical_issue_reason": "No single canonical issue stands out: the items span many unrelated regressions and the soft-linked pairs are mostly broad-topic similarities rather than true duplicates.", - "canonical_pr_reason": "No pull requests are present in the provided set.", - "best_issue_reason": "No single issue is a strong global representative because the cluster is not actually one duplicate family; any choice would be arbitrary.", - "best_pr_reason": "No pull requests are present in the provided set.", + "summary": "This cluster is mostly a set of unrelated regression reports, but one tokenizer pair looks like the same underlying bug: added-token spacing / slow-vs-fast tokenizer inconsistency. All other soft edges appear to be different failures in different code paths.", + "confidence": 0.63, + "canonical_issue_reason": "issue:28218 is the most concrete tokenizer regression in the set: it describes a specific, reproducible added-token spacing bug and is a better anchor than the broader inconsistency report.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:28218 is the best representative issue for the only likely duplicate subgroup, because it states the core tokenizer symptom clearly and narrowly.", + "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43716", - "right": "issue:44008", - "accept": false, - "reason": "Different bugs: one is a dtype mismatch between image preprocessor and model, the other is a Gemma3n forward variable-name collision causing an AttributeError." - }, { "left": "issue:43644", "right": "issue:43749", "accept": false, - "reason": "Both involve loading-related regressions, but one is non-persistent buffer corruption and the other is FSDP CPU RAM efficient loading failure; not the same code-path bug." + "reason": "Both are loading-related regressions, but one is about non-persistent buffers being corrupted and the other is FSDP CPU RAM efficient loading; too different to be the same bug." }, { "left": "issue:45584", "right": "issue:45834", "accept": false, - "reason": "Whisper empty-transcription failure after special-token alignment is unrelated to Kosmos2.5 long OCR input index errors." + "reason": "Whisper empty-transcription generation and Kosmos2.5 long OCR index errors are unrelated model-specific failures." + }, + { + "left": "issue:43716", + "right": "issue:44008", + "accept": false, + "reason": "Different models and different failure modes: dtype mismatch in preprocessing vs an AttributeError from variable collision." }, { "left": "issue:43606", "right": "issue:45072", "accept": false, - "reason": "Device-mismatch with CPU offload is a different failure mode from bfloat16 dtype mismatches in other models." + "reason": "CPU offload device mismatch for Bark is unrelated to bfloat16 dtype mismatches in other models." }, { "left": "issue:34634", "right": "issue:36010", "accept": false, - "reason": "BarkProcessor voice_preset breakage is unrelated to the GenerationMixin import error." + "reason": "BarkProcessor voice_preset behavior and a GenerationMixin import error are unrelated issues." }, { "left": "issue:43950", "right": "issue:45987", "accept": false, - "reason": "Non-persistent buffer corruption on from_pretrained is a loading-state bug; StaticCache.get_seq_length returning a Tensor is an API contract issue." + "reason": "Non-persistent buffer corruption during loading and StaticCache sequence-length return-type mismatch are different code paths." }, { "left": "issue:44479", "right": "issue:45834", "accept": false, - "reason": "These are different video/OCR regressions affecting different models and failure points." + "reason": "Both mention vision/video or indexing failures, but they are different model families and distinct regressions." }, { "left": "issue:43582", "right": "issue:46512", "accept": false, - "reason": "Apple Silicon caching_allocator_warmup TypeError is unrelated to from_config leaving config.dtype stale." + "reason": "Apple Silicon allocator TypeError is unrelated to from_config leaving config.dtype stale." }, { "left": "issue:43825", "right": "issue:43994", "accept": false, - "reason": "One is an incorrect error message about translation tasks, the other is a SigLIP2 inference/surrounding pipeline failure." + "reason": "An error-message/doc mismatch for pipeline support is not the same as SigLIP2 producing nonsensical outputs." }, { "left": "issue:43577", "right": "issue:43818", "accept": false, - "reason": "A dtype mismatch in BLIP2 loading is not the same as Video-LLaVA missing temporal attention and shared weights." + "reason": "Blip2 dtype loading and Video-LLaVA missing temporal attention are different model-specific bugs." }, { - "left": "issue:46294", - "right": "issue:46463", + "left": "issue:43421", + "right": "issue:44811", "accept": false, - "reason": "Qwen3.5-9B video/chat-template mismatch is unrelated to T5 decoder-input validation messaging." + "reason": "One is runtime special-token post-processor syncing; the other is Whisper batch_decode ignoring skip_special_tokens. Similar theme, different bug." }, { - "left": "issue:43421", - "right": "issue:44811", + "left": "issue:46294", + "right": "issue:46463", "accept": false, - "reason": "Both mention token handling, but one is runtime special-token post-processor updates and the other is Whisper batch_decode ignoring skip_special_tokens; different components and fixes." + "reason": "Qwen video/chat-template feature-token mismatch and a T5 decoder-input error message are unrelated." }, { "left": "issue:43782", "right": "issue:44977", "accept": false, - "reason": "A weight_only loading error for Qwen3VL is not the same bug as flash-attention generation issues in Qwen3.5." + "reason": "Qwen3VL weight_only loading and Qwen3.5 flash-attention generation failures are different problems." }, { "left": "issue:44991", "right": "issue:46291", "accept": false, - "reason": "Tokenizer load failure for a specific model is unrelated to kernels v0.15 breaking transformers.pipelines import." + "reason": "Tokenizer loading failure and kernels package import failure are unrelated dependency/runtime issues." }, { "left": "issue:44561", "right": "issue:45020", "accept": false, - "reason": "44561 is a specific missing-helper regression; 45020 is a broad umbrella over remote_code breakages, so they are not clearly the same concrete bug." + "reason": "Both involve remote_code breakage, but the specific removed API and the broader recent-version incompatibility are not clearly the same bug." }, { "left": "issue:44805", "right": "issue:44936", "accept": false, - "reason": "A tensor/mask shape mismatch is a different failure from trainer.evaluate breaking after trainer.train()." + "reason": "A mask/index shape mismatch and trainer.evaluate() failing after train are not the same underlying defect." }, { "left": "issue:43540", "right": "issue:45381", "accept": false, - "reason": "Both are video-input problems, but they affect different model families and different mechanics (processing failure vs vision_position_ids regression)." + "reason": "Both are video-input regressions, but they affect different Qwen variants and have different reported symptoms; not enough to merge as one bug." }, { "left": "issue:44610", "right": "issue:45072", "accept": false, - "reason": "Processor image-size mismatch is unrelated to bfloat16 dtype mismatch issues." + "reason": "Processor image-size mismatch and dtype mismatch in inference are unrelated." }, { "left": "issue:43720", "right": "issue:45072", "accept": false, - "reason": "Packed-weight unpacking during accelerate loading is a different bug from dtype mismatches." + "reason": "BitNet packed-weight unpacking during accelerate loading is unrelated to generic dtype mismatch issues." }, { "left": "issue:28218", "right": "issue:29159", - "accept": false, - "reason": "They are both tokenizer-related, but one specifically reports extra space after an added token while the other is a broad slow-vs-fast inconsistency claim without the same concrete symptom." + "accept": true, + "reason": "These both point to tokenizer behavior differences, and the specific added-token spacing bug plausibly explains the slow-vs-fast inconsistency." }, { "left": "issue:43825", "right": "issue:43827", "accept": false, - "reason": "These are related to v5 pipeline removals/docs, but one is an error message and the other is stale documentation; not the same underlying issue." + "reason": "One is a misleading error message about translation tasks; the other is outdated docs still referencing pipeline. Same area, different artifacts." }, { "left": "issue:43749", "right": "issue:44291", "accept": false, - "reason": "FSDP CPU RAM efficient loading and init_empty_weights/_is_hf_initialized are different loading pathways with different failures." + "reason": "FSDP CPU RAM efficient loading and init_empty_weights/_is_hf_initialized are different loading-path failures." }, { "left": "issue:45987", "right": "issue:46192", "accept": false, - "reason": "StaticCache return-type contract and return_tensors being ignored are unrelated API bugs." + "reason": "StaticCache return type and return_tensors being ignored are unrelated API bugs." }, { "left": "issue:43645", "right": "issue:46665", "accept": false, - "reason": "Both involve model initialization, but Jupyter custom-model initialization and from_pretrained inside Python classes are different contexts and failure modes." + "reason": "Notebook custom-model initialization and from_pretrained inside Python classes are similar in spirit but not clearly the same code-path bug." }, { "left": "issue:44671", "right": "issue:45701", "accept": false, - "reason": "Incorrect CamemBERT masked-LM predictions are not the same as a generic tokenization-change report." + "reason": "Masked LM prediction regression and a broad tokenization-change report are not the same specific failure." }, { "left": "issue:43577", "right": "issue:43582", "accept": false, - "reason": "BLIP2 dtype-loading problems are unrelated to the Apple Silicon caching_allocator_warmup TypeError." + "reason": "Blip2 dtype mismatch and Apple Silicon allocator TypeError are unrelated." }, { "left": "issue:42175", "right": "issue:45081", "accept": false, - "reason": "Missing TensorFlow in a torch-only install is unrelated to the Mistral regex patch crash during tokenizer loading." + "reason": "Backend install extras and Mistral tokenizer regex patching are unrelated." }, { "left": "issue:34634", "right": "issue:36331", "accept": false, - "reason": "BarkProcessor voice_preset failure and CustomTrainer.compute_loss signature mismatch are entirely separate issues." + "reason": "BarkProcessor voice_preset and CustomTrainer.compute_loss signature mismatch are unrelated bugs." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays appropriately cautious. The only positive soft edge (28218/29159) is plausibly supported by the titles, and the rest of the verdicts are conservative given the limited evidence. If anything, keep the tokenizer linkage phrased as 'likely related' rather than asserting the same underlying bug too strongly." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -5950,13 +5880,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 19, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:36010", "issue:36331", @@ -6008,8 +5938,8 @@ ], "soft_pairs": [ "issue:45003|issue:46285", - "issue:43653|issue:44912", "issue:43582|issue:44857", + "issue:43653|issue:44912", "issue:43262|issue:43819", "issue:43782|issue:45325", "issue:45081|issue:45406", @@ -6033,205 +5963,205 @@ "issue:43720|issue:44610", "issue:42175|issue:46291", "issue:39692|issue:44912", - "issue:38175|issue:44610", - "issue:42490|issue:43257" + "issue:42490|issue:43257", + "issue:38175|issue:44610" ], - "prepared_review_unit_hash": "831fce02edb8acbf1627be97f6cf344302a36b2de01db941e4698320fb24ad68", + "prepared_review_unit_hash": "677404128b274debad9ebf88332c8e08a2e9b6ed835aadc2827c4945097e0061", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 28, - "serialized_chars": 35934, - "estimated_input_tokens": 8984, - "estimated_eval_tokens": 18224 + "serialized_chars": 35939, + "estimated_input_tokens": 8985, + "estimated_eval_tokens": 18226 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The soft pairs are mostly false positives: they mention similar APIs or subsystems, but the concrete failures and code paths are different. I found no clear duplicate cluster, so there is no strong canonical issue/PR here.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "This cluster is mostly heterogeneous: the soft-similarity pairs share only broad subsystem/model-family overlap, not the same underlying bug or change. I would not merge any of the candidate pairs as duplicates.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 45020 is the broadest representative of the set\u2019s recurring theme of recent version-related loading regressions, but the cluster is too diverse to support a true duplicate canonical.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Issue 45020 is the closest thing to a cluster anchor because it covers a wide class of loading/compatibility breakages, but it still does not subsume the other reports well enough to treat them as duplicates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45003", "right": "issue:46285", "accept": false, - "reason": "Different bugs: unsafe sys.modules access vs CLIP text-model structure inconsistency." + "reason": "Different bug classes: unsafe sys.modules access vs a CLIP text-model structure inconsistency." }, { - "left": "issue:43653", - "right": "issue:44912", + "left": "issue:43582", + "right": "issue:44857", "accept": false, - "reason": "Unrelated tokenizer special-token registration bug vs MXFP4 quantization loading fallback." + "reason": "Unrelated failure modes on different platforms: Apple Silicon TypeError vs CUDA AMP crash." }, { - "left": "issue:43582", - "right": "issue:44857", + "left": "issue:43653", + "right": "issue:44912", "accept": false, - "reason": "Apple Silicon allocator TypeError is unrelated to CUDA AMP loss crashing." + "reason": "Tokenizer special-token registration bug vs MXFP4 quantization loading/fallback issue." }, { "left": "issue:43262", "right": "issue:43819", "accept": false, - "reason": "Audio chat-template sampling-rate default is unrelated to DAC latent/STE mismatch." + "reason": "Audio chat-template sampling-rate default vs DAC latent/STE mismatch are separate code paths." }, { "left": "issue:43782", "right": "issue:45325", "accept": false, - "reason": "Qwen3VL loading error and Qwen2.5-VL rope-index scaling are different problems." + "reason": "Both involve VL models, but one is a from_pretrained weight-only loading error and the other is a rope-index scaling bug." }, { "left": "issue:45081", "right": "issue:45406", "accept": false, - "reason": "Mistral regex patch crash and Gemma4 processor _tokenizer error are different code paths." + "reason": "Different tokenizer/processor crashes in different model integrations and different attributes." }, { "left": "issue:45372", "right": "issue:46326", "accept": false, - "reason": "Both involve mistral_common, but one is an import error for Gemma 4 and the other is continue_final_message behavior." + "reason": "One is a missing import breaking Gemma 4 processor loading; the other is a continue_final_message bug for mistral_common tokenizers." }, { "left": "issue:44060", "right": "issue:44368", "accept": false, - "reason": "Both mention tied-weights warnings, but one is a mistaken internal weight tie and the other is a config warning during LoRA fine-tuning." + "reason": "Different Qwen model variants and different warning/content: tied-weights miswire vs tie_word_embeddings warning." }, { "left": "issue:43329", "right": "issue:45600", "accept": false, - "reason": "Video multimodal token counting bug vs stale auto_mappings config references." + "reason": "Multimodal token counting bug vs stale auto_mappings references are unrelated." }, { "left": "issue:43742", "right": "issue:45020", "accept": false, - "reason": "A model load KeyError is not the same as broad remote_code breakage." + "reason": "Both are loading regressions, but they are not the same concrete failure or code path." }, { "left": "issue:44589", "right": "issue:45042", "accept": false, - "reason": "Float8 storage lookup failure is unrelated to the torchvision dependency regression." + "reason": "Storage-object lookup failure is unrelated to PIL backend torchvision dependency handling." }, { "left": "issue:43526", "right": "issue:43697", "accept": false, - "reason": "BEiT label reduction bug and RTDetrV2 output drift are different model/processor issues." + "reason": "BEiT label reduction bug vs RTDetr output-drift regression are different model behaviors." }, { "left": "issue:43550", "right": "issue:44912", "accept": false, - "reason": "torch.compile with SDPA on Bamba is unrelated to quantized loading fallback on git-oss-20b." + "reason": "torch.compile/SDPA failure is unrelated to MXFP4 quantization fallback during loading." }, { "left": "issue:45084", "right": "issue:45538", "accept": false, - "reason": "Non-template-node compile error is unrelated to CLIPTokenizer max_length behavior." + "reason": "Template compilation error vs tokenizer model_max_length constant are unrelated." }, { "left": "issue:43526", "right": "issue:44554", "accept": false, - "reason": "BEiT processor label bug is unrelated to the MPS attention correctness issue." + "reason": "Different functionality entirely: image label reduction vs MPS attention correctness." }, { "left": "issue:43452", "right": "issue:44991", "accept": false, - "reason": "Both involve loading, but gguf_file regressions and a specific tokenizer load failure are not the same concrete bug." + "reason": "GGUF file handling breakage is not the same as a specific tokenizer-loading regression for one model." }, { "left": "issue:43906", "right": "issue:46209", "accept": false, - "reason": "Isolated reproduction of one earlier issue is unrelated to FP8 inference failure on B200." + "reason": "An isolated reproduction of an old issue is not the same bug as FP8 inference failure on B200." }, { "left": "issue:43257", "right": "issue:43329", "accept": false, - "reason": "Qwen3 MoE loading/conversion issue is unrelated to multimodal token counting in the video branch." + "reason": "Qwen3 MoE loading/accelerate conversion issue vs multimodal token counting undefined names are unrelated." }, { "left": "issue:43819", "right": "issue:43927", "accept": false, - "reason": "DAC from_latents mismatch and DiaConfig save/load token-id loss are different failures." + "reason": "DAC latent reconstruction bug and DiaConfig custom-token-ID persistence bug are different code paths." }, { "left": "issue:43638", "right": "issue:44936", "accept": false, - "reason": "DeepSpeed ZeRO-3 index error is unrelated to evaluate-after-train behavior." + "reason": "DeepSpeed zero3/Bert index error vs trainer.evaluate() after train are different training/runtime issues." }, { "left": "issue:36010", "right": "issue:36331", "accept": false, - "reason": "Importing GenerationMixin and CustomTrainer.compute_loss signature mismatch are unrelated." + "reason": "Import path breakage for GenerationMixin is unrelated to CustomTrainer.compute_loss signature mismatch." }, { "left": "issue:36010", "right": "issue:43645", "accept": false, - "reason": "ImportError for GenerationMixin is not the same as Jupyter custom model initialization breakage." + "reason": "Importing GenerationMixin and Jupyter custom-model initialization are different failures." }, { "left": "issue:43531", "right": "issue:45042", "accept": false, - "reason": "Qwen3-MoE sliding_window issue is unrelated to PIL image processors requiring torchvision." + "reason": "Qwen3-MoE sliding_window behavior is unrelated to PIL backend torchvision dependency enforcement." }, { "left": "issue:43720", "right": "issue:44610", "accept": false, - "reason": "BitNet packed-weight loading and OmDet-Turbo input-size mismatch are different bugs." + "reason": "BitNet packed-weight loading bug and OmDet-Turbo input-size mismatch are not the same bug." }, { "left": "issue:42175", "right": "issue:46291", "accept": false, - "reason": "Packaging extras missing TensorFlow backend is unrelated to kernels version/revision requirements." + "reason": "Package extra composition and kernels import requirements are unrelated dependency issues." }, { "left": "issue:39692", "right": "issue:44912", "accept": false, - "reason": "SigLIP2 doc-example errors are unrelated to MXFP4 quantization loading fallback." + "reason": "SigLIP2 documentation mistakes do not match a quantization-loading fallback bug." }, { - "left": "issue:38175", - "right": "issue:44610", + "left": "issue:42490", + "right": "issue:43257", "accept": false, - "reason": "Unexpected zero probabilities in SigLIP2 is unrelated to OmDet-Turbo processor/model size mismatch." + "reason": "Trailing-slash local model import warning is unrelated to Qwen3 MoE weight-conversion loading." }, { - "left": "issue:42490", - "right": "issue:43257", + "left": "issue:38175", + "right": "issue:44610", "accept": false, - "reason": "Trailing-slash local-model warning and Qwen3 MoE loading/conversion are different issues." + "reason": "Unexpected zero probabilities in SigLIP2 is not the same as OmDet-Turbo preprocessing size mismatch." } ] }, @@ -6247,13 +6177,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 20, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:29159", "issue:34567", @@ -6266,12 +6196,12 @@ "issue:43450", "issue:43454", "issue:43525", + "issue:43540", "issue:43577", "issue:43606", "issue:43643", "issue:43650", "issue:43696", - "issue:43697", "issue:43742", "issue:43746", "issue:43749", @@ -6293,7 +6223,6 @@ "issue:45042", "issue:45127", "issue:45356", - "issue:45357", "issue:45459", "issue:45538", "issue:45759", @@ -6301,7 +6230,8 @@ "issue:45859", "issue:46291", "issue:46421", - "issue:46424" + "issue:46424", + "issue:46701" ], "soft_pairs": [ "issue:43746|issue:44811", @@ -6314,207 +6244,207 @@ "issue:46421|issue:46424", "issue:43329|issue:44479", "issue:43257|issue:43931", - "issue:43650|issue:43906", "issue:43329|issue:43450", - "issue:42915|issue:43742", "issue:43606|issue:45859", + "issue:42915|issue:43742", "issue:43749|issue:44756", + "issue:43650|issue:43906", "issue:34567|issue:40990", "issue:44291|issue:45042", + "issue:43540|issue:46701", "issue:42915|issue:44821", - "issue:29159|issue:34567", "issue:43454|issue:43577", - "issue:44568|issue:45356", + "issue:29159|issue:34567", "issue:43927|issue:44912", + "issue:44568|issue:45356", "issue:43525|issue:46291", - "issue:44987|issue:45538", "issue:43404|issue:45127", - "issue:43697|issue:45357" + "issue:44987|issue:45538" ], - "prepared_review_unit_hash": "cf5f09cd249adcd9f4297c8aba03c6fd1db9ee52c307744cb381cee53e2bb33e", + "prepared_review_unit_hash": "5b1d95fb4c4ec5b7f48134dbf5b6b16f6405a6183e0db74facb958c6810b4103", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 26, - "serialized_chars": 35134, - "estimated_input_tokens": 8784, - "estimated_eval_tokens": 17824 + "serialized_chars": 35072, + "estimated_input_tokens": 8768, + "estimated_eval_tokens": 17792 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a loose collection of mostly unrelated closed issues spanning loading, cache, tokenizer, multimodal, and model-specific regressions. None of the soft-pair candidates look like the same underlying bug/change, so I would not merge any of them as duplicates.", - "confidence": 0.64, - "canonical_issue_reason": "Issue 43749 has the strongest triage signal in this set (highest discussion/inbound reference activity) and is a broad infrastructure/loading regression, so it is the best single anchor even though the cluster is heterogeneous.", + "summary": "This set is a heterogeneous mix of unrelated issue reports. None of the soft pairs look like true duplicates; most share only a broad subsystem label (tokenizers, video, loading, offload, cache, etc.) rather than the same concrete bug.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 46421 is the best weak anchor for the cluster only because it is a clearly scoped, active cache-related bug with discussion and references; however, the overall set is too diverse to have a true canonical duplicate issue.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43749 is the most representative high-signal issue in the group for cluster-level triage, but it should not be treated as a duplicate of the others.", + "best_issue_reason": "Issue 46421 has the strongest triage value among the listed items: specific reproduction context, open status, and clear user impact. It is not a true representative of the whole cluster, just the most actionable single issue here.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43746", "right": "issue:44811", "accept": false, - "reason": "Different code paths and symptoms: GraniteSpeech PEFT local-checkpoint loading vs Whisper processor batch_decode skip_special_tokens handling." + "reason": "Different bugs and code paths: GraniteSpeech PEFT/local checkpoint loading vs Whisper batch_decode ignoring skip_special_tokens." }, { "left": "issue:44387", "right": "issue:44617", "accept": false, - "reason": "Both mention OOM, but the affected models and likely root causes differ: int4 quantization reserved-memory growth vs Sam3Video memory exhaustion." + "reason": "One is a broad int4 quantization memory regression; the other is a model-specific Sam3Video OOM. Not the same underlying issue." }, { "left": "issue:43854", "right": "issue:44977", "accept": false, - "reason": "Different models and regressions: GLM-4.7-Flash loading test failure vs Qwen3.5 flash-attention generation problem." + "reason": "Different model families and failures: GLM-4.7-Flash test loading vs Qwen3.5 flash-attention generation behavior." }, { "left": "issue:43643", "right": "issue:45759", "accept": false, - "reason": "Both are config-related, but one is missing fields from trust_remote_code AutoConfig and the other is unwrapping text_config for composite multimodal configs." + "reason": "Config-loading with trust_remote_code missing fields is distinct from from_config failing to unwrap text_config in multimodal configs." }, { "left": "issue:43257", "right": "issue:43696", "accept": false, - "reason": "Qwen3 MOE weight conversion under accelerate+deepspeed is unrelated to GPT-oss-20b CUDA OOM." + "reason": "Qwen3 MOE weight conversion with accelerate+deepspeed is unrelated to GPT-oss-20b CUDA OOM." }, { "left": "issue:43262", "right": "issue:44811", "accept": false, - "reason": "Both involve audio processors, but apply_chat_template sampling-rate defaults and batch_decode skip_special_tokens are separate functions and bugs." + "reason": "Both are processor-related, but one is chat template sampling-rate defaulting and the other is Whisper decoding token filtering." }, { "left": "issue:45459", "right": "issue:45803", "accept": false, - "reason": "Both concern exception handling, but they hit different components: tokenizer protobuf decode import vs FuyuBatchFeature tensor conversion." + "reason": "Both involve exception handling, but they are different functions and different failure modes: protobuf decode import handling vs bare except in tensor conversion." }, { "left": "issue:46421", "right": "issue:46424", "accept": false, - "reason": "Cache-related, but one is spurious recompilation from lazy init and the other is max_cache_length being ignored outside quantized cache." + "reason": "Related cache area, but one is lazy init causing recompilation and the other is max_cache_length being ignored outside Quantized Cache. Same subsystem, different bug." }, { "left": "issue:43329", "right": "issue:44479", "accept": false, - "reason": "The first is an undefined-variable bug in multimodal token counting; the second is a broader video-input regression across Qwen VL models." + "reason": "Undefined video-token helper variables in multimodal token counting is not the same as the broader v5.3.0 video input regression across several Qwen variants." }, { "left": "issue:43257", "right": "issue:43931", "accept": false, - "reason": "Different load failures: Qwen3 MOE conversion with accelerate/deepspeed vs Qwen3-VL weight-shape mismatch." + "reason": "Different loading problems: Qwen3 MOE conversion failure vs Qwen3-VL weight shape mismatch." }, { - "left": "issue:43650", - "right": "issue:43906", + "left": "issue:43329", + "right": "issue:43450", "accept": false, - "reason": "No relation: placeholder issue text versus an isolated reproduction of an older issue." + "reason": "Token-counting helper bug with undefined video branch variables is different from batched video processor shape output." }, { - "left": "issue:43329", - "right": "issue:43450", + "left": "issue:43606", + "right": "issue:45859", "accept": false, - "reason": "Video-token counting helper bug is not the same as incorrect batched video processor output shape." + "reason": "Both mention CPU offload, but they affect different models and code paths: Bark device mismatch vs Qwen3_5MoeTextRotaryEmbedding compatibility." }, { "left": "issue:42915", "right": "issue:43742", "accept": false, - "reason": "Different failures on different models: FineGrainedFP8Config on Qwen3Moe vs key error loading MobileLLM-125M." + "reason": "FineGrainedFP8Config failure for Qwen3Moe is unrelated to a key error while loading MobileLLM-125M." }, { - "left": "issue:43606", - "right": "issue:45859", + "left": "issue:43749", + "right": "issue:44756", "accept": false, - "reason": "Both involve CPU offload, but one is a bark-small device mismatch and the other is a Qwen3_5Moe rotary embedding compatibility issue." + "reason": "Different memory/loading problems: FSDP CPU-RAM-efficient loading vs disabling mmap on Strix Halo to avoid OOM." }, { - "left": "issue:43749", - "right": "issue:44756", + "left": "issue:43650", + "right": "issue:43906", "accept": false, - "reason": "One is FSDP CPU RAM efficient loading, the other is mmap-related OOM on Strix Halo; related theme, but not the same bug." + "reason": "No substantive overlap: one is a nonspecific placeholder title, the other is a reproduction of an external issue." }, { "left": "issue:34567", "right": "issue:40990", "accept": false, - "reason": "TrainerState token counting is unrelated to a perplexity regression on gpt-oss-20b." + "reason": "TrainerState token-count tracking and high perplexity on GPT-oss-20b are unrelated bugs." }, { "left": "issue:44291", "right": "issue:45042", "accept": false, - "reason": "Unrelated: init_empty_weights constructor argument breakage vs PIL image processors wrongly requiring torchvision." + "reason": "Different areas: init_empty_weights / _is_hf_initialized argument handling vs PIL image processors incorrectly requiring torchvision." }, { - "left": "issue:42915", - "right": "issue:44821", + "left": "issue:43540", + "right": "issue:46701", "accept": false, - "reason": "Different areas entirely: Qwen3Moe FP8 loading vs AutoImageProcessor URL loading." + "reason": "Different model/pipeline failures: Qwen3OmniMoe video input ValueError vs DiffusionGemmaGenerationOutput pipeline not working." }, { - "left": "issue:29159", - "right": "issue:34567", + "left": "issue:42915", + "right": "issue:44821", "accept": false, - "reason": "Tokenizer slow/fast inconsistency is unrelated to TrainerState num_input_tokens_seen not updating." + "reason": "Unrelated issues: FineGrainedFP8Config failure for Qwen3Moe vs AutoImageProcessor loading from URL." }, { "left": "issue:43454", "right": "issue:43577", "accept": false, - "reason": "Different model families and failures: tied lm_head bug in AyaVision vs dtype propagation issue in Blip2." + "reason": "Weight-tying failure causing generation corruption is distinct from dtype propagation remaining float32 in BLIP2." }, { - "left": "issue:44568", - "right": "issue:45356", + "left": "issue:29159", + "right": "issue:34567", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different models and behaviors: special-token insertion vs codec/warning handling." + "reason": "Tokenizer slow/fast inconsistency and TrainerState token accounting are entirely different components." }, { "left": "issue:43927", "right": "issue:44912", "accept": false, - "reason": "Different symptoms and subsystems: DiaConfig save/load token IDs vs MXFP4 quantization fallback to bf16." + "reason": "DiaConfig losing custom token IDs on save/load is unrelated to MXFP4 quantization falling back to bf16." }, { - "left": "issue:43525", - "right": "issue:46291", + "left": "issue:44568", + "right": "issue:45356", "accept": false, - "reason": "Missing pad_token_id in Llama4Config is unrelated to kernels v0.15 import-time requirement changes." + "reason": "Both are tokenizer regressions, but they concern different models and different mechanics (special-token insertion vs codec handling/warnings)." }, { - "left": "issue:44987", - "right": "issue:45538", + "left": "issue:43525", + "right": "issue:46291", "accept": false, - "reason": "No shared underlying bug: loading failure for physical-intelligence/fast vs CLIPTokenizer model_max_length behavior." + "reason": "Missing pad_token_id on Llama4Config is unrelated to a kernels package version/revision requirement blocking transformers.pipelines import." }, { "left": "issue:43404", "right": "issue:45127", "accept": false, - "reason": "Both mention tying embeddings, but one is a specific Mistral3 lm_head tying bug and the other is LoRA merge collapse with extended vocabulary." + "reason": "Missing lm_head tying in Mistral3 is not the same as LoRA merge collapse with extended vocabulary." }, { - "left": "issue:43697", - "right": "issue:45357", + "left": "issue:44987", + "right": "issue:45538", "accept": false, - "reason": "Different regression types: RTDetrV2 output changes in v5 vs Qwen3.5 save_pretrained writing wrong visual encoder keys." + "reason": "Model loading failure for physical-intelligence/fast is unrelated to CLIPTokenizer's model_max_length constant behavior." } ] }, @@ -6530,13 +6460,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 21, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:24643", "issue:27957", @@ -6558,6 +6488,7 @@ "issue:43540", "issue:43577", "issue:43582", + "issue:43697", "issue:43742", "issue:43749", "issue:43756", @@ -6568,15 +6499,13 @@ "issue:44488", "issue:44560", "issue:44568", - "issue:44792", "issue:44821", "issue:44849", "issue:44898", - "issue:44964", "issue:44987", "issue:44991", - "issue:45081", "issue:45137", + "issue:45357", "issue:45381", "issue:45405", "issue:45488", @@ -6588,16 +6517,17 @@ "issue:46693" ], "soft_pairs": [ + "issue:43697|issue:45357", "issue:28282|issue:34689", - "issue:42915|issue:45538", "issue:45137|issue:45405", + "issue:42915|issue:45538", "issue:33357|issue:34567", "issue:43425|issue:46291", "issue:44479|issue:46693", "issue:42175|issue:43824", "issue:28282|issue:29127", - "issue:44062|issue:44991", "issue:43742|issue:44291", + "issue:44062|issue:44991", "issue:43452|issue:43582", "issue:43577|issue:46512", "issue:43317|issue:45538", @@ -6608,231 +6538,217 @@ "issue:44568|issue:45488", "issue:24643|issue:29159", "issue:44987|issue:45907", - "issue:44479|issue:45381", "issue:27957|issue:28282", + "issue:44479|issue:45381", "issue:43540|issue:44479", "issue:42175|issue:44488", "issue:38175|issue:44898", "issue:43257|issue:43756", "issue:34567|issue:36331", - "issue:44291|issue:44821", - "issue:44792|issue:44964", - "issue:44062|issue:45081" + "issue:44291|issue:44821" ], - "prepared_review_unit_hash": "ddf7fad77b5503672dbe045243901d6c7d5ed2b89ab59b5e678cc97a8b50d0e8", + "prepared_review_unit_hash": "77db16a41d8d4ecf55ef2c77165a2b0854e348120430f9701325be0f4dadc02f", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36702, - "estimated_input_tokens": 9176, - "estimated_eval_tokens": 18608 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35770, + "estimated_input_tokens": 8943, + "estimated_eval_tokens": 18142 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of unrelated regression/load-error reports. The only strong duplicate-like overlap is the Qwen video-input regression pair, where one issue is a broad umbrella and the other is a specific manifestation.", - "confidence": 0.84, - "canonical_issue_reason": "issue:44479 is the best umbrella issue here: it describes the broader Qwen video-input regression in v5.3.0, and issue:45381 reads like a concrete instance of the same bug (wrong vision_position_ids for qwen2.5-vl video input).", + "summary": "Mostly heterogeneous issues; only the Qwen video-input regression pair looks like the same underlying bug. The rest are distinct failures across unrelated subsystems.", + "confidence": 0.8, + "canonical_issue_reason": "issue:44479 is the broader, earlier regression report and best captures the underlying Qwen video-input bug covered by the accepted duplicate.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44479 is the most suitable cluster anchor because it is concrete, broader than the narrow follow-up report, and directly captures the shared video-input regression path.", + "best_issue_reason": "issue:44479 is the strongest single issue to keep: it states the versioned regression, names multiple affected model families, and reads like the umbrella report for the duplicate pair.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:28282", - "right": "issue:34689", + "left": "issue:43697", + "right": "issue:45357", "accept": false, - "reason": "Different failures: missing PyTorch vs a Llama 3.2 Vision model loading regression." + "reason": "Different models and failures: RTDetrV2 output drift vs Qwen3.5 save_pretrained key mapping." }, { - "left": "issue:42915", - "right": "issue:45538", + "left": "issue:28282", + "right": "issue:34689", "accept": false, - "reason": "Unrelated: Qwen3Moe FP8 loading vs CLIPTokenizer max-length behavior." + "reason": "Both are loading errors, but one is missing-PyTorch ImportError and the other is a Llama vision model load regression." }, { "left": "issue:45137", "right": "issue:45405", "accept": false, - "reason": "Different areas: DeepSpeed ZeRO3 deque error vs PEFT version bump issue." + "reason": "DeepSpeed ZeRO3 deque error is unrelated to a PEFT version bump complaint." + }, + { + "left": "issue:42915", + "right": "issue:45538", + "accept": false, + "reason": "Qwen3Moe FP8 failure and CLIPTokenizer max_length are unrelated bugs." }, { "left": "issue:33357", "right": "issue:34567", "accept": false, - "reason": "MacOS bus error with CLIP is not the same as TrainerState token-count tracking." + "reason": "MacOS bus error in CLIP loading is unrelated to TrainerState token-count bookkeeping." }, { "left": "issue:43425", "right": "issue:46291", "accept": false, - "reason": "Torch 2.10 compatibility and kernels revision handling are separate problems." + "reason": "Torch 2.10 incompatibility is not the same as kernels revision/version handling in pipelines import." }, { "left": "issue:44479", "right": "issue:46693", "accept": false, - "reason": "Both mention regressions, but one is Qwen video-input handling and the other is FA2 long-context performance." + "reason": "Both mention Qwen/video or generation regressions, but one is a video-input correctness bug and the other is a long-context performance regression." }, { "left": "issue:42175", "right": "issue:43824", "accept": false, - "reason": "Different failures: backend dependency packaging vs missing Qwen2.5-VL class import." + "reason": "Torch backend packaging issue is unrelated to a missing Qwen2_5_VL class export." }, { "left": "issue:28282", "right": "issue:29127", "accept": false, - "reason": "Missing PyTorch and LayoutLMv3 box-validation messaging are unrelated." + "reason": "Missing-PyTorch ImportError and LayoutLMv3 box-validation messaging are different problems." }, { - "left": "issue:44062", - "right": "issue:44991", + "left": "issue:43742", + "right": "issue:44291", "accept": false, - "reason": "Both tokenizer-related, but the concrete bugs differ: AddedToken duplication vs tokenizer loading failure for a specific model." + "reason": "MobileLLM key loading and init_empty_weights unexpected argument are distinct load-path failures." }, { - "left": "issue:43742", - "right": "issue:44291", + "left": "issue:44062", + "right": "issue:44991", "accept": false, - "reason": "KeyError on a specific model load vs unexpected init_empty_weights argument handling." + "reason": "Tokenizer load failure may be related, but the titles do not show the same concrete AddedToken duplication bug." }, { "left": "issue:43452", "right": "issue:43582", "accept": false, - "reason": "GGUF loading breakage and Apple Silicon allocator TypeError are different code paths." + "reason": "gguf_file loading breakage and Apple Silicon caching_allocator_warmup TypeError are unrelated." }, { "left": "issue:43577", "right": "issue:46512", "accept": false, - "reason": "Both involve dtype state, but one is BLIP2 loading and the other is from_config/config staleness." + "reason": "BLIP2 dtype mismatch during loading is different from stale config.dtype after from_config(dtype=...)." }, { "left": "issue:43317", "right": "issue:45538", "accept": false, - "reason": "Dequantized device_map offload failure is unrelated to CLIPTokenizer model_max_length." + "reason": "Device-map/offload load failure and tokenizer max_length behavior are unrelated." }, { "left": "issue:44849", "right": "issue:45701", "accept": false, - "reason": "Qwen3.5 output_hidden_states bug is not the same as a broad tokenization-change report." + "reason": "Qwen3.5 hidden-states bug is unrelated to version-dependent tokenization changes." }, { "left": "issue:42915", "right": "issue:43749", "accept": false, - "reason": "Qwen3Moe FP8 loading and FSDP CPU RAM efficient loading are different issues." + "reason": "Qwen3Moe FP8 failure and FSDP CPU RAM efficient loading are different distributed-loading bugs." }, { "left": "issue:41720", "right": "issue:43540", "accept": false, - "reason": "Auto device-mapping CUDA assert and Qwen3OmniMoe video-input error are not the same bug." + "reason": "Qwen3 auto-device-map cuda assert and Qwen3OmniMoe video ValueError are different code paths." }, { "left": "issue:43540", "right": "issue:44560", "accept": false, - "reason": "Both are video-input bugs, but in different Qwen model families and different failure modes." + "reason": "Both concern video inputs, but they target different Qwen model families and different failures." }, { "left": "issue:44568", "right": "issue:45488", "accept": false, - "reason": "Tokenizer special-token handling and LlamaTokenizer pre-tokenizer override are distinct regressions." + "reason": "Tokenizer special-token regression and LlamaTokenizer pre-tokenizer override are separate tokenizer bugs." }, { "left": "issue:24643", "right": "issue:29159", "accept": false, - "reason": "DeepSpeed training weight-shape error is unrelated to slow/fast tokenizer inconsistency." + "reason": "DeepSpeed training weight-dimension error is unrelated to slow/fast tokenizer inconsistency." }, { "left": "issue:44987", "right": "issue:45907", "accept": false, - "reason": "Model loading regression and list_repo_files revision forwarding are separate bugs." - }, - { - "left": "issue:44479", - "right": "issue:45381", - "accept": true, - "reason": "Strong overlap: both describe the same Qwen2.5-VL/Qwen video-input regression in v5.3.0, with 45381 appearing to be a specific manifestation of the broader bug." + "reason": "Model-loading failure and list_repo_files revision forwarding are different issues." }, { "left": "issue:27957", "right": "issue:28282", "accept": false, - "reason": "Flash Attention 2 behavior for XLM-Roberta is unrelated to missing PyTorch import errors." + "reason": "Flash Attention 2 XLMRoberta issue is unrelated to missing PyTorch ImportError." }, { - "left": "issue:43540", - "right": "issue:44479", - "accept": false, - "reason": "Related theme, but different models and different video-processing failures." + "left": "issue:44479", + "right": "issue:45381", + "accept": true, + "reason": "Both describe the same Qwen2.5-VL/Qwen video-input regression; the narrower vision_position_ids report fits the broader regression issue." }, { "left": "issue:42175", "right": "issue:44488", "accept": false, - "reason": "Packaging/backend dependency issue and a specific model-load failure are not duplicates." + "reason": "Backend packaging with pip extras is unrelated to loading a specific community model." }, { "left": "issue:38175", "right": "issue:44898", "accept": false, - "reason": "Different vision-model bugs: zero probabilities vs non-default-resolution Perceiver failure." + "reason": "Siglip2 zero-probability output and Perceiver interpolation failure are different vision-model bugs." }, { "left": "issue:43257", "right": "issue:43756", "accept": false, - "reason": "Qwen3 MoE weight-conversion and Smollm3 RoPE-layer issues are unrelated." + "reason": "Qwen3 MOE weight conversion and Smollm3 RoPE layer count are unrelated model-specific issues." }, { "left": "issue:34567", "right": "issue:36331", "accept": false, - "reason": "TrainerState metric updating and custom compute_loss signature mismatch are separate trainer API issues." + "reason": "TrainerState token-count tracking and CustomTrainer compute_loss signature mismatch are different trainer API issues." }, { "left": "issue:44291", "right": "issue:44821", "accept": false, - "reason": "init_empty_weights loading TypeError and AutoImageProcessor URL loading are unrelated." - }, - { - "left": "issue:44792", - "right": "issue:44964", - "accept": false, - "reason": "Janus image-generation test failure and Phi-4 multimodal loading breakage are different regressions." - }, - { - "left": "issue:44062", - "right": "issue:45081", - "accept": false, - "reason": "AddedToken special-argument duplication is a different tokenizer bug than the Mistral regex patch crash." + "reason": "init_empty_weights unexpected argument is unrelated to loading AutoImageProcessor from URL." } ] }, "evaluator_result": { "accept": true, - "feedback": "The clustering is grounded in the packet and stays conservative overall. The only accepted duplicate-like pair, issue:44479 and issue:45381, is supported by the titles as the same Qwen2.5-VL/video-input regression family in v5.3.0, with 44479 plausibly serving as the broader umbrella. The rejected soft pairs are explained with appropriately cautious distinctions, and the summary does not overstate the level of duplication beyond the evidence provided." + "feedback": "Grounded and conservative overall. The selected canonical issue (44479) is supported by the title as a broader Qwen video-input regression, and the one accepted soft pair (44479 vs 45381) is a plausible same-bug match based on the packet. The remaining rejects read cautiously and are not overstated." }, "evaluator_used": true, "retried": false, @@ -6845,48 +6761,48 @@ "component_index": 1, "component_count": 22, "review_unit_index": 22, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:24643", "issue:28282", "issue:29127", "issue:32090", - "issue:33357", "issue:43404", "issue:43452", "issue:43526", "issue:43531", "issue:43540", "issue:43577", - "issue:43643", "issue:43697", + "issue:43716", "issue:43720", "issue:43742", "issue:43756", "issue:43906", + "issue:44062", "issue:44488", "issue:44568", "issue:44589", "issue:44625", "issue:44737", + "issue:44792", "issue:44821", "issue:44877", "issue:44898", "issue:44913", + "issue:44964", "issue:45042", "issue:45072", "issue:45081", - "issue:45200", "issue:45216", "issue:45325", "issue:45357", - "issue:45375", "issue:45405", "issue:45479", "issue:45588", @@ -6899,23 +6815,27 @@ "issue:46463", "issue:46468", "issue:46491", - "issue:46693" + "issue:46693", + "issue:46701" ], "soft_pairs": [ + "issue:44792|issue:44964", + "issue:44062|issue:45081", "issue:44737|issue:46463", "issue:43756|issue:46399", + "issue:43716|issue:46701", "issue:44568|issue:46491", - "issue:43742|issue:44589", "issue:43756|issue:45325", + "issue:43742|issue:44589", "issue:45042|issue:45081", "issue:43577|issue:45072", - "issue:43906|issue:46468", "issue:45823|issue:46463", + "issue:43906|issue:46468", "issue:45405|issue:45600", "issue:43531|issue:44821", - "issue:43906|issue:46693", "issue:44737|issue:45588", "issue:43404|issue:45987", + "issue:43906|issue:46693", "issue:43452|issue:44488", "issue:43526|issue:45479", "issue:44625|issue:44877", @@ -6924,202 +6844,199 @@ "issue:44913|issue:46121", "issue:43720|issue:44898", "issue:43697|issue:45216", - "issue:24643|issue:29127", "issue:28282|issue:32090", - "issue:43643|issue:45375", - "issue:45200|issue:46463", - "issue:28282|issue:33357" + "issue:24643|issue:29127" ], - "prepared_review_unit_hash": "d7990b5e4854a601c53cd28a92855ad3b11cc1f461bba43f8be6ee5d1f3bf893", + "prepared_review_unit_hash": "641f90ac03fab3f4d0169e7b2215fcc82db2cc1342505f9e1c36d8aebda7b116", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 27, - "serialized_chars": 35716, - "estimated_input_tokens": 8929, - "estimated_eval_tokens": 18114 + "serialized_chars": 36222, + "estimated_input_tokens": 9056, + "estimated_eval_tokens": 18368 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These issues are mostly distinct bugs with only superficial overlap in areas like tokenizers, rope/config handling, loading, and dtype/device behavior. None of the soft pairs look like true duplicates or a single mergeable fix.", - "confidence": 0.92, - "canonical_issue_reason": "issue:45920 is the broadest and most representative regression report in the set: a v5 AutoTokenizer bug affecting multiple model families with a clear, concrete symptom.", + "summary": "Heterogeneous cluster: the soft pairs share only broad vocabulary (tokenizer/config/multimodal/dtype/loading) but not the same concrete bug or code path, so none should be merged.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue is safe here; the items are a grab-bag of unrelated bugs, so collapsing them would over-merge distinct problems.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45920 has the best combination of scope, clarity, and practical triage value as a cluster anchor.", + "best_issue_reason": "If one issue must be kept as the representative, issue:45920 is the most useful stand-alone anchor because it is a broad active regression report with inbound references; still, it is not a duplicate sink for the rest of the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44792", + "right": "issue:44964", + "accept": false, + "reason": "Both are multimodal-loading failures, but one is a janus test failure and the other is a Phi-4 model load error; different models and different failure paths." + }, + { + "left": "issue:44062", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer-related, but one is an AddedToken keyword collision and the other is a Mistral regex patch crash; not the same bug." + }, { "left": "issue:44737", "right": "issue:46463", "accept": false, - "reason": "XLNet CPU device-placement bug vs T5 decoder-input validation message; unrelated code paths." + "reason": "XLNet CPU device placement versus T5 decoder-input validation; unrelated code paths and symptoms." }, { "left": "issue:43756", "right": "issue:46399", "accept": false, - "reason": "Both touch RoPE, but one is missing layers in Smollm3 and the other is a Qwen3.5 kernel registration crash." + "reason": "Both mention rotary-related internals, but one is Smollm3 RoPE layer dropping and the other is a Qwen3.5 kernel registration error; distinct fixes." }, { - "left": "issue:44568", - "right": "issue:46491", + "left": "issue:43716", + "right": "issue:46701", "accept": false, - "reason": "Different tokenizer regressions: BOS/EOS insertion vs CodeLlama leading-space round-trip." + "reason": "Both are multimodal, but one is a dtype mismatch between image preprocessor and model and the other is an image-text pipeline failure for DiffusionGemmaGenerationOutput." }, { - "left": "issue:43742", - "right": "issue:44589", + "left": "issue:44568", + "right": "issue:46491", "accept": false, - "reason": "MobileLLM key error on load vs missing Float8 storage type; unrelated failures." + "reason": "Tokenizer regressions, but one is BOS/EOS insertion and the other is CodeLlama leading-space round-trip behavior; different underlying bugs." }, { "left": "issue:43756", "right": "issue:45325", "accept": false, - "reason": "Both involve RoPE-related behavior, but the bugs are model-specific and the failure mechanisms differ." + "reason": "Both involve RoPE/position handling, but one is missing layers in Smollm3 and the other is Qwen2.5-VL still-image temporal position scaling; not mergeable into one fix." + }, + { + "left": "issue:43742", + "right": "issue:44589", + "accept": false, + "reason": "Loading problems, but one is a key error for MobileLLM and the other is a missing Float8 storage type; different failure mechanisms." }, { "left": "issue:45042", "right": "issue:45081", "accept": false, - "reason": "Image processor torchvision dependency issue vs Mistral tokenizer regex patch crash; different subsystems." + "reason": "Image-processor torchvision dependency versus tokenizer regex patch crash; broad subsystem overlap only, no shared bug." }, { "left": "issue:43577", "right": "issue:45072", "accept": false, - "reason": "Both are dtype-related, but they affect different models and different mismatch scenarios." + "reason": "Both are dtype mismatches, but they affect different model families and different forward paths; too broad to treat as the same bug." }, { - "left": "issue:43906", - "right": "issue:46468", + "left": "issue:45823", + "right": "issue:46463", "accept": false, - "reason": "One is a reproduction of an earlier issue; the other is a broken official Nemotron example with no clear match." + "reason": "Gemma4 device-map forward mismatch and T5 decoder-input validation are unrelated issues." }, { - "left": "issue:45823", - "right": "issue:46463", + "left": "issue:43906", + "right": "issue:46468", "accept": false, - "reason": "Gemma4 device-map mismatch and T5 decoder validation are unrelated." + "reason": "An isolated reproduction of one issue versus a broken official Nemotron example; no evidence of the same underlying bug." }, { "left": "issue:45405", "right": "issue:45600", "accept": false, - "reason": "Released-version dependency bump vs stale auto_mappings config references; different maintenance problems." + "reason": "PEFT version pinning versus stale auto_mappings config references; one is dependency metadata, the other is config cleanup." }, { "left": "issue:43531", "right": "issue:44821", "accept": false, - "reason": "Qwen3-MoE sliding-window behavior vs AutoImageProcessor URL loading; no shared underlying bug." - }, - { - "left": "issue:43906", - "right": "issue:46693", - "accept": false, - "reason": "Isolated reproduction issue vs FA2 long-context performance regression; different symptoms and paths." + "reason": "Qwen3-MoE sliding-window behavior and AutoImageProcessor URL loading are unrelated." }, { "left": "issue:44737", "right": "issue:45588", "accept": false, - "reason": "XLNet rotary CPU allocation bug vs flash_attention None handling bug; unrelated." + "reason": "XLNet CPU arange bug versus flash_attention s_aux=None crash; different modules and triggers." }, { "left": "issue:43404", "right": "issue:45987", "accept": false, - "reason": "Mistral3 weight tying bug vs StaticCache return-type contract bug; separate code paths." + "reason": "Mistral3 lm_head tying and StaticCache.get_seq_length return type are unrelated implementation bugs." + }, + { + "left": "issue:43906", + "right": "issue:46693", + "accept": false, + "reason": "Reproduction of an old issue versus a FA2 long-context performance regression; no shared concrete code-path defect." }, { "left": "issue:43452", "right": "issue:44488", "accept": false, - "reason": "Both are loading-related, but one is a gguf_file regression and the other is a specific model not loading; not the same bug." + "reason": "Both are model-loading failures, but one is gguf_file handling and the other is a specific model not loading; too different to merge." }, { "left": "issue:43526", "right": "issue:45479", "accept": false, - "reason": "BeitImageProcessorFast label reduction bug vs sequence-classification zero-loss bug; different layers and behavior." + "reason": "Both mention labels, but one is an image processor returning the wrong array shape and the other is a sequence-classification zero-loss regression." }, { "left": "issue:44625", "right": "issue:44877", "accept": false, - "reason": "Qwen3.5 num_labels propagation bug vs strict config rejection for granite_speech; both config-related but distinct." + "reason": "Configuration-related, but one is num_labels propagation in Qwen3.5 and the other is strict config blocking granite_speech loading; not the same bug." }, { "left": "issue:45357", "right": "issue:45920", "accept": false, - "reason": "Qwen3.5 save_pretrained key bug is unrelated to the AutoTokenizer token-ID regression." + "reason": "One is save_pretrained key serialization for Qwen3.5, the other is wrong token IDs in AutoTokenizer across several models; different layers of the stack." }, { "left": "issue:43540", "right": "issue:46693", "accept": false, - "reason": "Video-input ValueError in Qwen3OmniMoe vs long-context generation performance regression; not the same issue." + "reason": "Video-input processing crash versus FA2 long-context performance regression; unrelated symptoms and code paths." }, { "left": "issue:44913", "right": "issue:46121", "accept": false, - "reason": "GPTNeoX rotary_pct reload bug vs JSON list type error in rope validation; related domain, different defects." + "reason": "Both concern RoPE config handling, but one is silent reloading of rotary_pct and the other is a TypeError from JSON-loaded ignore_keys; not the same defect." }, { "left": "issue:43720", "right": "issue:44898", "accept": false, - "reason": "BitNet packed-weight loading failure vs Perceiver interpolation failure; different models and bugs." + "reason": "Accelerate loading of packed BitNet weights versus Perceiver non-default-resolution interpolation failure; different models and mechanisms." }, { "left": "issue:43697", "right": "issue:45216", "accept": false, - "reason": "RTDetrV2 output drift in v5 vs Qwen3.5 save_pretrained regression; unrelated changes." - }, - { - "left": "issue:24643", - "right": "issue:29127", - "accept": false, - "reason": "DeepSpeed training weight-shape error vs LayoutLMv3 error-message clarity issue; not duplicates." + "reason": "RTDetrV2 output drift in v5 and Qwen3.5 checkpoint save regression are separate regression classes." }, { "left": "issue:28282", "right": "issue:32090", "accept": false, - "reason": "PyTorch-missing ImportError vs Trainer broadcast TypeError; different failure modes." - }, - { - "left": "issue:43643", - "right": "issue:45375", - "accept": false, - "reason": "Both involve missing config fields, but one is generic remote-code config loading and the other is a specific Qwen3_5MoeVisionConfig field being dropped." + "reason": "Environment/import error for PyTorch versus Trainer broadcast handling of NoneType; unrelated." }, { - "left": "issue:45200", - "right": "issue:46463", - "accept": false, - "reason": "Gemma4 mm_token_type_ids defaulting issue vs T5 decoder validation error; unrelated model families and code paths." - }, - { - "left": "issue:28282", - "right": "issue:33357", + "left": "issue:24643", + "right": "issue:29127", "accept": false, - "reason": "AutoModel missing PyTorch vs MacOS bus error on CLIP; completely different problems." + "reason": "DeepSpeed weight-shape runtime error versus LayoutLMv3 boxes-validation messaging; different bugs and code paths." } ] }, @@ -7135,14 +7052,15 @@ "component_index": 1, "component_count": 22, "review_unit_index": 23, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ + "issue:28282", "issue:29159", "issue:33357", "issue:38175", @@ -7150,8 +7068,10 @@ "issue:42898", "issue:43295", "issue:43452", + "issue:43525", "issue:43531", "issue:43577", + "issue:43643", "issue:43646", "issue:43653", "issue:43688", @@ -7160,18 +7080,13 @@ "issue:43749", "issue:43756", "issue:43819", - "issue:43927", "issue:43994", "issue:44361", "issue:44442", "issue:44464", "issue:44466", - "issue:44479", - "issue:44484", "issue:44493", "issue:44568", - "issue:44625", - "issue:44704", "issue:44743", "issue:44811", "issue:44843", @@ -7179,259 +7094,254 @@ "issue:44936", "issue:44991", "issue:45005", + "issue:45200", "issue:45356", "issue:45372", - "issue:45440", + "issue:45375", "issue:45542", - "issue:45701", "issue:45704", "issue:45823", "issue:46050", "issue:46170", "issue:46178", "issue:46402", - "issue:46491" + "issue:46463", + "issue:46491", + "issue:46701" ], "soft_pairs": [ + "issue:43643|issue:45375", + "issue:45200|issue:46463", + "issue:28282|issue:33357", "issue:44466|issue:45005", "issue:41762|issue:44936", - "issue:43452|issue:44843", "issue:43688|issue:43749", + "issue:43452|issue:44843", "issue:44361|issue:44442", "issue:44493|issue:46050", - "issue:43653|issue:44361", + "issue:43525|issue:46701", "issue:43819|issue:44811", - "issue:43994|issue:44898", + "issue:43653|issue:44361", "issue:45372|issue:46170", + "issue:43994|issue:44898", "issue:38175|issue:43295", "issue:44936|issue:45823", "issue:42898|issue:46491", "issue:45542|issue:46178", "issue:43295|issue:45356", - "issue:43653|issue:44568", "issue:43531|issue:43742", + "issue:43653|issue:44568", "issue:43452|issue:43577", "issue:43577|issue:43697", - "issue:44843|issue:44991", "issue:43646|issue:43749", - "issue:29159|issue:33357", "issue:43756|issue:46402", "issue:43688|issue:45704", - "issue:43653|issue:43927", - "issue:44464|issue:44743", - "issue:43688|issue:45701", - "issue:43688|issue:45440", - "issue:44484|issue:44704", - "issue:44479|issue:44625" + "issue:44843|issue:44991", + "issue:29159|issue:33357", + "issue:44464|issue:44743" ], - "prepared_review_unit_hash": "7bb5456b658eee46344e9872b8b9eaf704821973cf42a4156d4f81ab8beb208e", + "prepared_review_unit_hash": "85886fe8abea617f86fe74f8680524f2d02656ba2070b6828eff34dc44e43701", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36906, - "estimated_input_tokens": 9227, - "estimated_eval_tokens": 18710 + "soft_pair_count": 29, + "serialized_chars": 36688, + "estimated_input_tokens": 9172, + "estimated_eval_tokens": 18600 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is a loose collection of mostly unrelated issue reports; none of the suggested soft pairs look like true duplicates or the same underlying change. The most concrete tokenizer-specific report is the best anchor, but the cluster overall is not cohesive.", - "confidence": 0.74, - "canonical_issue_reason": "issue:43653 is the most self-contained and actionable report here: it describes a specific tokenizer special-token registration bug with a clear user-visible failure and CI signal.", + "summary": "This cluster is mostly a grab bag of unrelated Transformers bug reports across tokenizers, model loading, generation, and backend-specific regressions. I would not merge any of the soft pairs as duplicates; they do not appear to share the same concrete code-path failure.", + "confidence": 0.95, + "canonical_issue_reason": "There is no strong single canonical duplicate here because the issues span different models, symptoms, and subsystems. If a representative anchor is needed, issue 44843 is the most concrete and broadly actionable loading-regression report in the set.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43653 is a strong representative issue because it is narrow, reproducible, and likely to map to a single fix path, unlike several broader regression or compatibility reports in the set.", + "best_issue_reason": "Issue 44843 is the best representative issue: it has a specific reproducible regression, a clear root area (_patch_mistral_regex / model_info), and a well-defined failure mode. The other issues are mostly model-specific or symptom-level reports that are not duplicate-equivalent.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43643", + "right": "issue:45375", + "accept": false, + "reason": "Both mention missing config fields, but they are different models and likely different code paths (trust_remote_code config loading vs @strict dropping a Qwen3 field)." + }, + { + "left": "issue:45200", + "right": "issue:46463", + "accept": false, + "reason": "Unrelated model-family issues: Gemma 4 mm_token_type_ids defaults vs T5 decoder-input validation/error messaging." + }, + { + "left": "issue:28282", + "right": "issue:33357", + "accept": false, + "reason": "Different failures entirely: missing PyTorch ImportError vs MacOS bus error loading a CLIP community model." + }, { "left": "issue:44466", "right": "issue:45005", "accept": false, - "reason": "Both concern tied-weights behavior, but one is about inconsistent serialization by device and the other about translation-model issues in v5; different symptoms and likely different code paths." + "reason": "Both involve tied weights, but one is a device-dependent serialization bug and the other is a broader translation-model issue; not clearly the same concrete defect." }, { "left": "issue:41762", "right": "issue:44936", "accept": false, - "reason": "Both mention IndexError-like failures, but one is a DeepSpeed ZeRO-3 loading crash and the other is trainer state/evaluate behavior after training; not the same bug." + "reason": "Different surfaces and code paths: ZeRO-3 loading IndexError vs trainer.evaluate() failing after training." }, { - "left": "issue:43452", - "right": "issue:44843", + "left": "issue:43688", + "right": "issue:43749", "accept": false, - "reason": "One is about gguf_file breaking AutoTokenizer/AutoModel loading, the other about AutoTokenizer calling model_info offline; both touch loading, but they are distinct failure modes." + "reason": "Different subsystems: auxiliary-loss normalization in MoE models vs FSDP CPU RAM efficient loading." }, { - "left": "issue:43688", - "right": "issue:43749", + "left": "issue:43452", + "right": "issue:44843", "accept": false, - "reason": "Both involve model loading/training regressions, but one is auxiliary-loss normalization and the other is FSDP CPU RAM efficient loading; unrelated code paths." + "reason": "Both touch AutoTokenizer.from_pretrained, but one is about gguf_file handling and the other about an unconditional model_info() call breaking offline mode." }, { "left": "issue:44361", "right": "issue:44442", "accept": false, - "reason": "One is an MLukeTokenizer AttributeError on tasks, the other is AutoTokenizer failing to load a different tokenizer class; same area, different tokenizer and failure." + "reason": "Different tokenizer classes and failures; no evidence they share the same underlying bug." }, { "left": "issue:44493", "right": "issue:46050", "accept": false, - "reason": "Both concern model loading/saving state, but one is unexpected position-id keys and the other is sinq quantized models breaking after save/load; not the same underlying bug." + "reason": "One is a position-id key regression, the other is quantization save/load corruption." }, { - "left": "issue:43653", - "right": "issue:44361", + "left": "issue:43525", + "right": "issue:46701", "accept": false, - "reason": "Both are tokenizer-related, but one is a BigBird special-token decode issue and the other is an MLukeTokenizer task AttributeError; different models and symptoms." + "reason": "AttributeError in Llama4Config pad_token_id vs a diffusion/image-text pipeline failure; unrelated." }, { "left": "issue:43819", "right": "issue:44811", "accept": false, - "reason": "One is DAC.from_latents vs forward mismatch, the other is Whisper batch_decode skip_special_tokens handling; both are decoding-related but not the same defect." + "reason": "Different behaviors: DAC from_latents vs Whisper batch_decode skip_special_tokens handling." }, { - "left": "issue:43994", - "right": "issue:44898", + "left": "issue:43653", + "right": "issue:44361", "accept": false, - "reason": "Both involve vision model inference regressions, but one is SigLIP2 nonsensical outputs and the other is Perceiver resize/interpolation failure; different architectures and bugs." + "reason": "Tokenizer bugs in different models and different failure modes; not the same underlying issue." }, { "left": "issue:45372", "right": "issue:46170", "accept": false, - "reason": "One is a Gemma 4 processor import problem from mistral_common, the other is a Windows/ROCm torch.distributed dependency issue; unrelated failures." + "reason": "Different dependency problems: mistral_common ReasoningEffort import vs torch.distributed dependency on Windows/ROCm." + }, + { + "left": "issue:43994", + "right": "issue:44898", + "accept": false, + "reason": "Different vision-model failures with different symptoms and model families." }, { "left": "issue:38175", "right": "issue:43295", "accept": false, - "reason": "Both are regressions involving model/processor behavior, but one is zero probabilities for SigLIP2 and the other is processor.tokenizer/image handling breakage; not the same bug." + "reason": "Siglip2 zero-probabilities bug vs processor.tokenizer/image-passing regression; unrelated." }, { "left": "issue:44936", "right": "issue:45823", "accept": false, - "reason": "Trainer evaluate/train state issues and Gemma4 PLE device-mismatch during forward are different problems in different subsystems." + "reason": "Training/evaluation lifecycle failure vs Gemma4 device_map auto device mismatch." }, { "left": "issue:42898", "right": "issue:46491", "accept": false, - "reason": "One is a future v5 tokenization-space cleanup behavior change, the other is a CodeLlama encode/decode leading-space regression; both tokenizer-related but not the same issue." + "reason": "Both are tokenizer-behavior regressions, but one is a v5 global cleanup behavior change and the other is a CodeLlama-specific round-trip space loss." }, { "left": "issue:45542", "right": "issue:46178", "accept": false, - "reason": "Both are backend/dependency errors, but one is about tensorboard without TensorFlow and the other is tensorflow_text not in BACKENDS_MAPPING; distinct package/backend plumbing." + "reason": "Both are backend availability issues, but for different backends and different error surfaces; not a duplicate bug." }, { "left": "issue:43295", "right": "issue:45356", "accept": false, - "reason": "Both involve processor/tokenizer regressions, but one is custom model code using processor.tokenizer/images and the other is Kimi-K2.5 codec handling plus warning text; different defects." + "reason": "Different regressions affecting different tokenizer/processor paths and different model families." }, { - "left": "issue:43653", - "right": "issue:44568", + "left": "issue:43531", + "right": "issue:43742", "accept": false, - "reason": "Both are tokenizer special-token issues, but BigBird mask-token registration and mdeberta-v3 BOS/EOS addition are separate tokenizer implementations and likely separate fixes." + "reason": "Different model-loading/generation issues with no shared concrete failure path." }, { - "left": "issue:43531", - "right": "issue:43742", + "left": "issue:43653", + "right": "issue:44568", "accept": false, - "reason": "Sliding-window behavior in Qwen3-MoE and a key error when loading MobileLLM-125M are unrelated model/configuration bugs." + "reason": "Different tokenizer bugs: special-token registration vs add_special_tokens not emitting BOS/EOS." }, { "left": "issue:43452", "right": "issue:43577", "accept": false, - "reason": "One is gguf_file breaking loading APIs, the other is Blip2 dtype propagation; no shared underlying code path beyond general loading." + "reason": "Tokenizer/model loading with gguf_file vs BLIP2 dtype propagation; unrelated." }, { "left": "issue:43577", "right": "issue:43697", "accept": false, - "reason": "Blip2 dtype placement and RTDetrV2 output divergence are unrelated model-specific regressions." - }, - { - "left": "issue:44843", - "right": "issue:44991", - "accept": false, - "reason": "One is an offline-mode failure in AutoTokenizer's Mistral regex patching, the other is tokenizer loading failure for a specific model; same surface area, different root causes." + "reason": "Different problems: BLIP2 dtype staying float32 vs RTDetrV2 output differences." }, { "left": "issue:43646", "right": "issue:43749", "accept": false, - "reason": "Custom model initialization breakage in v5 and FSDP RAM-efficient loading are different breakages with different lifecycles and code paths." - }, - { - "left": "issue:29159", - "right": "issue:33357", - "accept": false, - "reason": "Slow/fast tokenizer inconsistency and a MacOS bus error with a CLIP community model are not the same bug; one is behavior mismatch, the other is a crash." + "reason": "AutoConfig trust_remote_code field loss vs FSDP CPU RAM efficient loading; unrelated." }, { "left": "issue:43756", "right": "issue:46402", "accept": false, - "reason": "Smollm3 RoPE-layer mismatch and Sam3Model text-encoder weights not loading are different model-implementation issues." + "reason": "Smollm3 RoPE layer issue vs Sam3Model text encoder weight loading; different model-specific bugs." }, { "left": "issue:43688", "right": "issue:45704", "accept": false, - "reason": "Auxiliary-loss normalization and T5 apex.FusedRMSNorm memory leak are unrelated; one is math/scale, the other is an external dependency leak." - }, - { - "left": "issue:43653", - "right": "issue:43927", - "accept": false, - "reason": "BigBird mask-token special-token registration and DiaConfig custom token ID persistence are both token-related, but one is tokenizer metadata and the other is config save/load behavior." - }, - { - "left": "issue:44464", - "right": "issue:44743", - "accept": false, - "reason": "Chunked generation with compiled forward and Qwen3_5 recurrent-state reset are both generation/cache issues, but they target different mechanisms and model code." + "reason": "Different issues in different models: MoE auxiliary-loss normalization vs T5 apex FusedRMSNorm memory leak." }, { - "left": "issue:43688", - "right": "issue:45701", - "accept": false, - "reason": "Auxiliary-loss normalization and version-dependent tokenization changes are different classes of bug; no clear shared code path." - }, - { - "left": "issue:43688", - "right": "issue:45440", + "left": "issue:44843", + "right": "issue:44991", "accept": false, - "reason": "Auxiliary-loss normalization and divergence from remote DeepSeekV3 implementation are unrelated model behavior issues." + "reason": "Both concern tokenizer loading, but one is an offline-mode regression from a specific regex patch and the other is a model-specific tokenizer load failure." }, { - "left": "issue:44484", - "right": "issue:44704", + "left": "issue:29159", + "right": "issue:33357", "accept": false, - "reason": "A question about default max_shard_size and an AutoProcessor kwargs-passing bug are unrelated API issues." + "reason": "Tokenizer inconsistency vs MacOS bus error; no shared code-path." }, { - "left": "issue:44479", - "right": "issue:44625", + "left": "issue:44464", + "right": "issue:44743", "accept": false, - "reason": "A video-input regression in multiple Qwen VL variants and num_labels propagation in Qwen3.5 are different configuration/inference problems." + "reason": "Both are generation/caching consistency reports, but they describe different mechanisms and are not obviously mergeable into one fix." } ] }, @@ -7447,13 +7357,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 24, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:32090", "issue:33666", @@ -7464,15 +7374,13 @@ "issue:43262", "issue:43278", "issue:43295", - "issue:43452", "issue:43493", "issue:43504", - "issue:43525", "issue:43550", - "issue:43572", "issue:43577", "issue:43606", "issue:43643", + "issue:43653", "issue:43688", "issue:43720", "issue:43742", @@ -7480,23 +7388,25 @@ "issue:43825", "issue:43927", "issue:44361", - "issue:44451", "issue:44464", "issue:44466", + "issue:44479", + "issue:44484", "issue:44514", "issue:44554", "issue:44561", "issue:44610", "issue:44625", + "issue:44704", "issue:44734", "issue:44821", "issue:44964", "issue:44977", "issue:45290", "issue:45405", + "issue:45440", "issue:45491", - "issue:45584", - "issue:45823", + "issue:45701", "issue:46002", "issue:46050", "issue:46209", @@ -7505,17 +7415,22 @@ "issue:46693" ], "soft_pairs": [ + "issue:43653|issue:43927", + "issue:43688|issue:45701", + "issue:43688|issue:45440", + "issue:44484|issue:44704", + "issue:44479|issue:44625", "issue:45491|issue:46002", "issue:43550|issue:44361", "issue:33666|issue:36010", "issue:32090|issue:33666", "issue:44734|issue:46612", - "issue:44464|issue:44977", "issue:41720|issue:44464", "issue:43742|issue:44821", "issue:44466|issue:46050", - "issue:43504|issue:43720", + "issue:44464|issue:44977", "issue:43504|issue:43550", + "issue:43504|issue:43720", "issue:44977|issue:46693", "issue:43643|issue:44561", "issue:43504|issue:44610", @@ -7523,241 +7438,237 @@ "issue:43720|issue:43927", "issue:43782|issue:46693", "issue:43606|issue:43927", - "issue:44514|issue:45290", "issue:33666|issue:36331", "issue:43278|issue:43688", + "issue:44514|issue:45290", "issue:44554|issue:46365", "issue:43262|issue:43295", "issue:43720|issue:44361", "issue:39692|issue:43493", - "issue:44977|issue:46209", "issue:44964|issue:45405", "issue:43493|issue:43825", - "issue:43452|issue:44451", - "issue:43525|issue:43572", - "issue:44625|issue:45584", - "issue:45823|issue:46050" + "issue:44977|issue:46209" ], - "prepared_review_unit_hash": "9d1ad7904d27e5277531a1ddac4524250287d6fe4b3ce49eb707a067dc3a777c", + "prepared_review_unit_hash": "ed5cc5b7e81635e6ce655abbbd0fdfef09b3f68e6dc3ce27907c226ce68880b9", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 37376, - "estimated_input_tokens": 9344, - "estimated_eval_tokens": 18944 + "soft_pair_count": 33, + "serialized_chars": 37716, + "estimated_input_tokens": 9429, + "estimated_eval_tokens": 19114 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is heterogeneous: most items are distinct bugs/regressions across different models, loaders, and runtime paths. No soft pair looks like the same underlying fix, so all soft edges are rejected.", - "confidence": 0.34, - "canonical_issue_reason": "Issue 33666 is the broadest, most discussion-heavy item here, but it is not truly duplicate with the rest of the cluster; it is only the least-specific representative.", + "summary": "This cluster is a heterogeneous mix of unrelated Transformers bugs (trainer, tokenizers, multimodal processors, generation, loading, and model-specific regressions). The soft-similarity pairs mostly share broad keywords but not the same concrete failure mode or code path, so none look like true duplicates. No PRs are present.", + "confidence": 0.21, + "canonical_issue_reason": "Issue 43688 is the closest thing to a representative bug report here: it is a concrete model correctness regression with enough discussion to serve as a cluster anchor, even though the surrounding items are not true duplicates of it.", "canonical_pr_reason": null, - "best_issue_reason": "33666 is the best issue representative because it captures a general multi-GPU training problem and has the strongest evidence of being a recurring user-facing topic, even though it does not subsume the other reports.", + "best_issue_reason": "Issue 43688 is the best single issue to represent the set because it is a substantive, non-doc bug with a clear technical symptom; however, this is only a representative choice, not evidence of duplication.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43653", + "right": "issue:43927", + "accept": false, + "reason": "Both involve token/config handling, but one is a tokenizer special-token registration bug and the other is a config save/load field-loss bug; different root causes and code paths." + }, + { + "left": "issue:43688", + "right": "issue:45701", + "accept": false, + "reason": "Aux-loss normalization and tokenization-version changes are unrelated problems despite both being model behavior issues." + }, + { + "left": "issue:43688", + "right": "issue:45440", + "accept": false, + "reason": "Different models and different failures: auxiliary-loss normalization vs. DeepSeek implementation divergence." + }, + { + "left": "issue:44484", + "right": "issue:44704", + "accept": false, + "reason": "Save-pretrained shard size behavior is unrelated to AutoProcessor kwargs propagation." + }, + { + "left": "issue:44479", + "right": "issue:44625", + "accept": false, + "reason": "A multimodal video-input regression and a Qwen3.5 num_labels propagation bug are not the same defect." + }, { "left": "issue:45491", "right": "issue:46002", "accept": false, - "reason": "Both mention NaNs, but one is a GPU mixed-length batching issue and the other is a classification-head checkout/init problem; different failure modes and code paths." + "reason": "Both mention NaNs, but one is Gemma3 sliding-window attention on GPU and the other is classification-head init/checkout; different failure mechanisms." }, { "left": "issue:43550", "right": "issue:44361", "accept": false, - "reason": "Completely different subsystems: torch.compile/SDPA model forward bug vs tokenizer/task AttributeError." + "reason": "torch.compile/SDPA on Bamba is unrelated to an MLukeTokenizer AttributeError." }, { "left": "issue:33666", "right": "issue:36010", "accept": false, - "reason": "Multi-GPU training and a GenerationMixin import error are unrelated issues." + "reason": "Multi-GPU Qwen2-VL training and a GenerationMixin import error are unrelated." }, { "left": "issue:32090", "right": "issue:33666", "accept": false, - "reason": "Trainer NoneType broadcast error is a distributed-training bug, but not the same as Qwen2-VL multi-GPU training support." + "reason": "Both touch Trainer/multi-GPU contexts, but one is a NoneType broadcast error and the other is a Qwen2-VL training request; not the same bug." }, { "left": "issue:44734", "right": "issue:46612", "accept": false, - "reason": "Both involve generation internals, but one is KV-cache continuation indexing and the other is beam-search cache reorder skipping for certain model classes." - }, - { - "left": "issue:44464", - "right": "issue:44977", - "accept": false, - "reason": "Compiled-forward inconsistency and flash-attention generation breakage are different bugs with different triggers." + "reason": "Both involve generation caches, but one is KV-cache continuation tensor indexing in serve and the other is beam-search cache reorder being skipped for specific model classes." }, { "left": "issue:41720", "right": "issue:44464", "accept": false, - "reason": "Device-map/cuda assert on A800 is unrelated to chunked generation output inconsistency under compiled forward." + "reason": "Auto device-map cudaErrorAssert on Qwen3 is unrelated to compiled-forward chunked generation inconsistency." }, { "left": "issue:43742", "right": "issue:44821", "accept": false, - "reason": "Loading a model checkpoint key error is not the same as failing to load an image processor from URL." + "reason": "Model loading KeyError and AutoImageProcessor-from-URL failure are different loading paths and symptoms." }, { "left": "issue:44466", "right": "issue:46050", "accept": false, - "reason": "Both are save/load related, but one is tied-weight serialization variance by device and the other is quantized-sinq persistence failure." + "reason": "Both concern save/load behavior, but one is tied-weight serialization and the other is quantized sinq round-tripping; different concrete bugs." }, { - "left": "issue:43504", - "right": "issue:43720", + "left": "issue:44464", + "right": "issue:44977", "accept": false, - "reason": "Legacy preset loading for BEiT semantic segmentation is unrelated to BitNet packed-weight unpacking during accelerate loading." + "reason": "Compiled-forward chunked generation inconsistency is not the same as a flash-attention-specific Qwen3.5 generation failure." }, { "left": "issue:43504", "right": "issue:43550", "accept": false, - "reason": "Different model families and different bugs: BEiT preset loading vs Bamba torch.compile SDPA failure." + "reason": "A legacy-field model-load failure and a Bamba torch.compile/SDPA failure are unrelated." + }, + { + "left": "issue:43504", + "right": "issue:43720", + "accept": false, + "reason": "Both are loading-related, but one is a Beit legacy config field issue and the other is packed-weight unpacking during accelerate loading." }, { "left": "issue:44977", "right": "issue:46693", "accept": false, - "reason": "Flash-attention generation correctness/performance issue and long-context FA2 performance regression are not the same defect." + "reason": "One is a correctness bug in Qwen3.5 generation; the other is a performance regression with FA2 long-context generation." }, { "left": "issue:43643", "right": "issue:44561", "accept": false, - "reason": "Both touch trust_remote_code ecosystems, but one is missing fields in AutoConfig and the other is removal of a helper breaking model code." + "reason": "Both mention trust_remote_code, but one is missing fields in AutoConfig and the other is removal of is_torch_fx_available breaking remote-code models." }, { "left": "issue:43504", "right": "issue:44610", "accept": false, - "reason": "Legacy field loading and processor output size mismatch are unrelated." + "reason": "Legacy config loading and processor/model input-size mismatch are unrelated." }, { "left": "issue:43577", "right": "issue:44554", "accept": false, - "reason": "Blip2 dtype propagation and MPS attention correctness are separate issues in different model/runtime paths." + "reason": "A dtype-loading bug in BLIP2 is unrelated to an MPS attention correctness issue." }, { "left": "issue:43720", "right": "issue:43927", "accept": false, - "reason": "Packed-weight loading bug and DiaConfig token-ID persistence bug do not share the same underlying fix." + "reason": "BitNet packed-weight loading and DiaConfig custom token ID persistence are different model/config issues." }, { "left": "issue:43782", "right": "issue:46693", "accept": false, - "reason": "Qwen3VL weight_only loading error is unrelated to FA2 long-context performance regression." + "reason": "Qwen3VL weight_only loading error is unrelated to the FA2 long-context performance regression." }, { "left": "issue:43606", "right": "issue:43927", "accept": false, - "reason": "CPU offload device mismatch and config token-ID loss are different classes of bugs." - }, - { - "left": "issue:44514", - "right": "issue:45290", - "accept": false, - "reason": "Both are apply_chat_template crashes, but one is batched padding=False handling and the other is assistant tool-call/no-content handling." + "reason": "CPU offload device mismatch and config custom-token-ID loss are unrelated." }, { "left": "issue:33666", "right": "issue:36331", "accept": false, - "reason": "Qwen2-VL multi-GPU training is not the same as CustomTrainer.compute_loss signature breakage." + "reason": "Qwen2-VL multi-GPU training and CustomTrainer.compute_loss signature mismatch are different Trainer issues." }, { "left": "issue:43278", "right": "issue:43688", "accept": false, - "reason": "Embedding dtype mismatch at eval and auxiliary-loss normalization are unrelated model issues." + "reason": "Embedding dtype drift across train/eval is not the same as auxiliary-loss normalization." + }, + { + "left": "issue:44514", + "right": "issue:45290", + "accept": false, + "reason": "Both are apply_chat_template crashes, but the triggers differ materially: batched padding=False vs. assistant messages with tool calls and no content." }, { "left": "issue:44554", "right": "issue:46365", "accept": false, - "reason": "An MPS attention correctness issue is not the same as failing to import Trainer because of datasets." + "reason": "An MPS attention correctness issue is unrelated to an import conflict preventing Trainer from loading." }, { "left": "issue:43262", "right": "issue:43295", "accept": false, - "reason": "Audio sampling-rate defaulting and processor.tokenizer/image-tokenization regression are different processor bugs." - }, - { - "left": "issue:43720", - "right": "issue:44361", - "accept": false, - "reason": "BitNet packed-weight loading and MLukeTokenizer task AttributeError are unrelated." - }, - { - "left": "issue:39692", - "right": "issue:43493", - "accept": false, - "reason": "SigLIP2 documentation example errors are not the same as a deeper HF-vs-JAX implementation discrepancy." - }, - { - "left": "issue:44977", - "right": "issue:46209", - "accept": false, - "reason": "Qwen3.5 flash-attention generation issues and FP8 B200 inference failure are distinct hardware/runtime problems." + "reason": "Audio chat-template sampling-rate defaulting and processor.tokenizer/image-tokenization regression are different processor bugs." }, { "left": "issue:44964", "right": "issue:45405", "accept": false, - "reason": "Loading Phi-4 multimodal and a PEFT minimum-version bump are not duplicate bugs." + "reason": "A Phi-4 multimodal load failure and a bumped MIN_PEFT_VERSION are unrelated." }, { "left": "issue:43493", "right": "issue:43825", "accept": false, - "reason": "SigLIP2 implementation discrepancy and a pipeline error-message regression are unrelated." - }, - { - "left": "issue:43452", - "right": "issue:44451", - "accept": false, - "reason": "gguf_file breaks generic AutoTokenizer/AutoModel loading, while ScandiBERT load failure is a model-specific regression." - }, - { - "left": "issue:43525", - "right": "issue:43572", - "accept": false, - "reason": "Both are pad-token config regressions, but they affect different fields and different config classes, so they are not clearly the same bug." + "reason": "SigLIP2 implementation divergence and a pipeline error-message regression are unrelated." }, { - "left": "issue:44625", - "right": "issue:45584", + "left": "issue:39692", + "right": "issue:43493", "accept": false, - "reason": "Qwen3.5 num_labels propagation and Whisper empty-transcription generation failure are unrelated." + "reason": "One is a documentation example bug; the other is a core implementation discrepancy for SigLIP2. Different artifacts and fixes." }, { - "left": "issue:45823", - "right": "issue:46050", + "left": "issue:44977", + "right": "issue:46209", "accept": false, - "reason": "Gemma4 device-map forward mismatch and sinq quantized save/load breakage are different issues." + "reason": "Qwen3.5 flash-attention generation failure and FP8 inference failure on B200 are different model/runtime problems." } ] }, @@ -7773,13 +7684,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 25, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:28218", "issue:33357", @@ -7790,8 +7701,10 @@ "issue:43295", "issue:43299", "issue:43334", + "issue:43452", "issue:43493", "issue:43504", + "issue:43525", "issue:43572", "issue:43575", "issue:43606", @@ -7803,20 +7716,17 @@ "issue:43927", "issue:43994", "issue:44112", - "issue:44291", "issue:44361", - "issue:44464", + "issue:44451", "issue:44466", "issue:44488", "issue:44493", "issue:44568", "issue:44589", + "issue:44625", "issue:44792", "issue:44843", "issue:44849", - "issue:44964", - "issue:45200", - "issue:45229", "issue:45357", "issue:45405", "issue:45520", @@ -7824,13 +7734,17 @@ "issue:45759", "issue:45812", "issue:45823", + "issue:46050", "issue:46192", "issue:46209", "issue:46285", - "issue:46424", "issue:46489" ], "soft_pairs": [ + "issue:43452|issue:44451", + "issue:43525|issue:43572", + "issue:44625|issue:45584", + "issue:45823|issue:46050", "issue:28218|issue:33357", "issue:43994|issue:46192", "issue:43278|issue:43844", @@ -7838,252 +7752,227 @@ "issue:39692|issue:43295", "issue:44849|issue:46209", "issue:44568|issue:46489", - "issue:44488|issue:44843", "issue:33357|issue:33666", "issue:43646|issue:44466", + "issue:44488|issue:44843", "issue:44568|issue:45812", - "issue:43927|issue:44361", "issue:43278|issue:43701", + "issue:43927|issue:44361", "issue:44849|issue:45357", "issue:44792|issue:45405", - "issue:43504|issue:43606", "issue:43575|issue:45405", "issue:43257|issue:43299", + "issue:43504|issue:43606", "issue:44589|issue:45520", - "issue:43334|issue:43572", "issue:43606|issue:44112", - "issue:43334|issue:45759", "issue:43824|issue:44589", + "issue:43334|issue:43572", + "issue:43334|issue:45759", "issue:44493|issue:46285", - "issue:44464|issue:46424", - "issue:43504|issue:43653", - "issue:45200|issue:45229", - "issue:43572|issue:44291", - "issue:45229|issue:45823", - "issue:44464|issue:45584", - "issue:43844|issue:44964", - "issue:43994|issue:44493" + "issue:43504|issue:43653" ], - "prepared_review_unit_hash": "2b7d12b0e9a3e8ee99d96b34c8ef30df2f02125f822474c5d483dbd992f4ce26", + "prepared_review_unit_hash": "90a8e403a386e74e9e2dfb50c6f95cd59c129c66d38c1d7b4ec868e8155a91db", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 37453, - "estimated_input_tokens": 9364, - "estimated_eval_tokens": 18984 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35919, + "estimated_input_tokens": 8980, + "estimated_eval_tokens": 18216 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a loose similarity set, not a duplicate set. Most pairs are only thematically related (tokenizers, configs, multimodal loading, CI, or performance regressions) but describe different code paths and failure modes, so I would reject every soft edge.", - "confidence": 0.8, - "canonical_issue_reason": "No single issue is a safe canonical duplicate anchor here; the items span unrelated bugs across tokenizer behavior, config migration, multimodal loading, CI failures, and memory/device issues.", + "summary": "This cluster is a loose set of unrelated tokenizer, config, loading, model, and CI regressions. None of the soft edges look like true duplicates, so there is no single canonical issue here.", + "confidence": 0.98, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "If a representative issue must be picked, issue:45812 is the broadest and most central tokenizer regression, with multiple inbound references; however, it still should not be treated as a duplicate of the other items.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43452", + "right": "issue:44451", + "accept": false, + "reason": "Different models and failure modes: gguf/from_pretrained breakage versus a specific ScandiBERT load failure." + }, + { + "left": "issue:43525", + "right": "issue:43572", + "accept": false, + "reason": "Both mention missing token-id fields, but they affect different configs/models and appear to be separate regressions." + }, + { + "left": "issue:44625", + "right": "issue:45584", + "accept": false, + "reason": "Unrelated bugs: Qwen3.5 label propagation versus Whisper empty-transcription generation after special-token alignment." + }, + { + "left": "issue:45823", + "right": "issue:46050", + "accept": false, + "reason": "Different subsystems: device-map forward mismatch versus quantized model save/load corruption." + }, { "left": "issue:28218", "right": "issue:33357", "accept": false, - "reason": "Tokenizer spacing bug vs MacOS bus error in a CLIP model; completely different failure modes." + "reason": "Tokenizer spacing bug versus MacOS bus error on a CLIP model; no shared code-path." }, { "left": "issue:43994", "right": "issue:46192", "accept": false, - "reason": "SigLIP2 inference quality issue vs return_tensors handling in multimodal kwargs; not the same bug." + "reason": "SigLIP2 output quality issue versus return_tensors being ignored when text_kwargs is passed." }, { "left": "issue:43278", "right": "issue:43844", "accept": false, - "reason": "Dtype change in evaluation vs ZeRO-3 gradient growth; different symptoms and likely different causes." + "reason": "Both involve training behavior, but one is a dtype flip at eval and the other is a ZeRO-3 gradient anomaly." }, { "left": "issue:43493", "right": "issue:43844", "accept": false, - "reason": "SigLIP2 implementation discrepancy vs DeepSpeed training instability; unrelated subsystems." + "reason": "SigLIP2 implementation discrepancy is unrelated to the ZeRO-3 gradient increase bug." }, { "left": "issue:39692", "right": "issue:43295", "accept": false, - "reason": "Broken documentation example vs regression in processor.tokenizer/custom image passing; distinct issues." + "reason": "Doc/example errors are not the same bug as the processor.tokenizer regression in custom model code." }, { "left": "issue:44849", "right": "issue:46209", "accept": false, - "reason": "Qwen3.5 hidden-states bug vs FP8 inference failure on B200; different model classes and code paths." + "reason": "Different domains: Qwen3.5 hidden-state behavior versus FP8 inference failure on B200 hardware." }, { "left": "issue:44568", "right": "issue:46489", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different models and do not look like the same underlying defect." + "reason": "Both are tokenizer regressions, but they affect different tokenizers and different concrete behaviors." }, { - "left": "issue:44488", - "right": "issue:44843", + "left": "issue:33357", + "right": "issue:33666", "accept": false, - "reason": "Model load failure vs offline-mode hub call in mistral regex patch; not the same root cause." + "reason": "MacOS bus error on CLIP loading is unrelated to Qwen2-VL multi-GPU training." }, { - "left": "issue:33357", - "right": "issue:33666", + "left": "issue:43646", + "right": "issue:44466", "accept": false, - "reason": "MacOS CLIP bus error vs Qwen2-VL multi-GPU training request; unrelated." + "reason": "Custom model initialization breakage is not the same bug as device-dependent lm_head serialization." }, { - "left": "issue:43646", - "right": "issue:44466", + "left": "issue:44488", + "right": "issue:44843", "accept": false, - "reason": "Custom model initialization regression vs tied-weight serialization inconsistency; different code paths." + "reason": "Loading a specific model family is different from an offline-mode failure caused by an unconditional hub call." }, { "left": "issue:44568", "right": "issue:45812", "accept": false, - "reason": "Both concern tokenization regressions, but one is mDeBERTa special-token behavior and the other is Granite token IDs; not the same bug." + "reason": "Both are tokenizer-related, but the affected models and symptoms differ: missing BOS/EOS versus wrong token IDs for Granite." }, { - "left": "issue:43927", - "right": "issue:44361", + "left": "issue:43278", + "right": "issue:43701", "accept": false, - "reason": "DiaConfig save/load token ID loss vs MLukeTokenizer AttributeError; different models and failures." + "reason": "Embedding dtype drift in eval is unrelated to resume_from_checkpoint key mismatch." }, { - "left": "issue:43278", - "right": "issue:43701", + "left": "issue:43927", + "right": "issue:44361", "accept": false, - "reason": "Embedding dtype drift vs resume_from_checkpoint key mismatch; unrelated." + "reason": "Different root causes: lost custom token IDs in DiaConfig versus an AttributeError in MLukeTokenizer task handling." }, { "left": "issue:44849", "right": "issue:45357", "accept": false, - "reason": "Hidden-states output bug vs incorrect visual encoder keys on save_pretrained; not the same issue." + "reason": "Qwen3.5 hidden-state bug and incorrect visual-encoder keys in save_pretrained are separate regressions." }, { "left": "issue:44792", "right": "issue:45405", "accept": false, - "reason": "Failed janus image generation test vs an unreleased PEFT version bump; unrelated." - }, - { - "left": "issue:43504", - "right": "issue:43606", - "accept": false, - "reason": "Legacy-field load failure vs CPU offload device mismatch; different bugs." + "reason": "A failing janus test is not the same issue as the PEFT minimum-version bump." }, { "left": "issue:43575", "right": "issue:45405", "accept": false, - "reason": "Qwen2-57B TP OOM vs PEFT version pin problem; unrelated." + "reason": "TP OOM on a large Qwen2 model is unrelated to the PEFT dependency version change." }, { "left": "issue:43257", "right": "issue:43299", "accept": false, - "reason": "Accelerate+DeepSpeed conversion issue vs dev regression loading Qwen3VL MoE; similar family, but different failure paths." + "reason": "Both involve Qwen MOE models, but one is weight conversion during accelerate/deepspeed load and the other is model loading breakage in dev." }, { - "left": "issue:44589", - "right": "issue:45520", + "left": "issue:43504", + "right": "issue:43606", "accept": false, - "reason": "Missing storage object in FP8 path vs flash_attn import KeyError on Python 3.13; not the same defect." + "reason": "BEiT legacy-field loading and Bark CPU-offload device mismatch are unrelated bugs." }, { - "left": "issue:43334", - "right": "issue:43572", + "left": "issue:44589", + "right": "issue:45520", "accept": false, - "reason": "Missing pad_token_id in Qwen3-VL vs missing pad_token_idx in StableLmConfig; both config regressions, but different models and fields." + "reason": "Float8 storage lookup failure is a different issue from the flash_attn import KeyError on Python 3.13." }, { "left": "issue:43606", "right": "issue:44112", "accept": false, - "reason": "CPU offload mismatch in bark-small vs stale GraniteSpeech device override test; not the same bug." - }, - { - "left": "issue:43334", - "right": "issue:45759", - "accept": false, - "reason": "pad_token_id missing in Qwen3-VL vs from_config not unwrapping text_config for composite Qwen 3.5/3.6; related area but different root cause." + "reason": "A real CPU-offload device mismatch is not the same as a stale CI test in GraniteSpeech." }, { "left": "issue:43824", "right": "issue:44589", "accept": false, - "reason": "ImportError for Qwen2.5-VL class vs FP8 storage lookup failure; unrelated." - }, - { - "left": "issue:44493", - "right": "issue:46285", - "accept": false, - "reason": "Unexpected position-id key warnings vs CLIP text-model structure change; different models and APIs." - }, - { - "left": "issue:44464", - "right": "issue:46424", - "accept": false, - "reason": "Chunked generation inconsistency with compiled forward vs cache max length behavior; not the same change." - }, - { - "left": "issue:43504", - "right": "issue:43653", - "accept": false, - "reason": "Legacy field load failure vs BigBirdTokenizer mask-token special-token registration; different tokenizer/model bugs." - }, - { - "left": "issue:45200", - "right": "issue:45229", - "accept": false, - "reason": "Training-time mm_token_type_ids defaulting issue vs multi-GPU inference OOM; distinct problems." - }, - { - "left": "issue:43572", - "right": "issue:44291", - "accept": false, - "reason": "Missing pad_token_idx in StableLmConfig vs unexpected _is_hf_initialized argument with init_empty_weights; different migration issues." + "reason": "Importing a missing Qwen2.5-VL class is unrelated to the Float8 storage-object TypeError." }, { - "left": "issue:45229", - "right": "issue:45823", + "left": "issue:43334", + "right": "issue:43572", "accept": false, - "reason": "OOM during multi-GPU inference vs device mismatch during forward; not the same failure." + "reason": "Similar missing token-id theme, but different models and different attribute names imply different bugs." }, { - "left": "issue:44464", - "right": "issue:45584", + "left": "issue:43334", + "right": "issue:45759", "accept": false, - "reason": "Compiled-forward chunked generation inconsistency vs Whisper empty-transcription failure after align_special_tokens; unrelated." + "reason": "Missing pad_token_id in Qwen3-VL is not the same as failing to unwrap text_config in composite Qwen configs." }, { - "left": "issue:43844", - "right": "issue:44964", + "left": "issue:44493", + "right": "issue:46285", "accept": false, - "reason": "DeepSpeed gradient anomaly vs inability to load Phi-4 multimodal model; different code paths." + "reason": "Unexpected position-id keys across many models is a different structural regression than CLIP text-model class inconsistency." }, { - "left": "issue:43994", - "right": "issue:44493", + "left": "issue:43504", + "right": "issue:43653", "accept": false, - "reason": "SigLIP2 wrong outputs vs unexpected position-id keys across many models; not the same root cause." + "reason": "BEiT legacy-field loading and BigBirdTokenizer special-token registration are unrelated." } ] }, @@ -8099,13 +7988,13 @@ "component_index": 1, "component_count": 22, "review_unit_index": 26, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:15354", "issue:28218", @@ -8118,56 +8007,65 @@ "issue:38175", "issue:42490", "issue:43065", - "issue:43232", "issue:43295", "issue:43450", "issue:43493", "issue:43540", - "issue:43550", "issue:43572", - "issue:43575", "issue:43696", "issue:43701", - "issue:43824", "issue:43844", "issue:43873", + "issue:43937", "issue:43957", - "issue:44186", + "issue:43994", + "issue:44291", "issue:44423", + "issue:44464", + "issue:44493", "issue:44554", - "issue:44610", "issue:44734", - "issue:44849", - "issue:44987", + "issue:44964", + "issue:44977", "issue:45072", "issue:45084", "issue:45198", "issue:45200", - "issue:45356", + "issue:45229", "issue:45406", "issue:45446", "issue:45520", + "issue:45584", "issue:45588", "issue:45759", "issue:45823", "issue:46050", "issue:46121", - "issue:46178", - "issue:46192", - "issue:46369" + "issue:46369", + "issue:46424", + "issue:46701" ], "soft_pairs": [ + "issue:45200|issue:45229", + "issue:43572|issue:44291", + "issue:44464|issue:46424", + "issue:43937|issue:46701", + "issue:45229|issue:45823", + "issue:44464|issue:45584", + "issue:43844|issue:44964", "issue:43873|issue:46050", + "issue:43994|issue:44493", "issue:45588|issue:46050", "issue:29942|issue:30333", "issue:42490|issue:43696", "issue:29159|issue:32090", + "issue:44977|issue:46701", "issue:45084|issue:45520", "issue:43065|issue:46369", "issue:28218|issue:34567", "issue:44423|issue:44734", - "issue:15354|issue:38175", "issue:44554|issue:45446", + "issue:15354|issue:38175", "issue:45406|issue:46121", "issue:43493|issue:43701", "issue:45072|issue:45198", @@ -8175,234 +8073,204 @@ "issue:37428|issue:43957", "issue:43540|issue:45200", "issue:43295|issue:43450", - "issue:45200|issue:45823", - "issue:44186|issue:46192", - "issue:45823|issue:46192", - "issue:43295|issue:43873", - "issue:43295|issue:44610", - "issue:43550|issue:43844", - "issue:45520|issue:46178", - "issue:43450|issue:43575", - "issue:43232|issue:44849", - "issue:43572|issue:44987", - "issue:44554|issue:45356", - "issue:43572|issue:43957", - "issue:37428|issue:43824" + "issue:45200|issue:45823" ], - "prepared_review_unit_hash": "788ef56a0c035bf5361f26818196d3328529a13301b5f23c6b47d9987afb5c1b", + "prepared_review_unit_hash": "0a496ec75b7255cd03892a2ea15c91a164d851f9a02612ebd06f6938c9a1fbab", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 37063, - "estimated_input_tokens": 9266, - "estimated_eval_tokens": 18788 + "soft_pair_count": 28, + "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": "This cluster is mostly a grab-bag of unrelated runtime bugs across tokenization, loading, serving, attention, and model-specific regressions. None of the soft pairs look like the same underlying bug/change, so I would not merge any of them as duplicates.", - "confidence": 0.91, - "canonical_issue_reason": "issue:43295 is the strongest representative because it describes a concrete regression in a public multimodal API with a clear user-facing failure mode, which is more general than the many highly model-specific reports in the set.", + "summary": "The cluster is heterogeneous; the soft pairs are mostly broad topical similarities rather than the same underlying bug. I would not merge any of the suggested edges, and there is no clear canonical issue/PR for the whole cluster.", + "confidence": 0.95, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:43295 is the best standalone issue for the cluster: it is a clear regression, easy to reproduce, and broadly relevant to custom model code using processor/tokenizer APIs.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:45200", + "right": "issue:45229", + "accept": false, + "reason": "Gemma 4 token-type defaults vs a multi-GPU OOM; different failure modes and code paths." + }, + { + "left": "issue:43572", + "right": "issue:44291", + "accept": false, + "reason": "Both are loading-related regressions, but one is a missing config field and the other is an unexpected init argument." + }, + { + "left": "issue:44464", + "right": "issue:46424", + "accept": false, + "reason": "Both mention caching, but one is compiled-forward generation mismatch and the other is a cache-length config limitation." + }, + { + "left": "issue:43937", + "right": "issue:46701", + "accept": false, + "reason": "GenerationConfig validation for GLM-5 is unrelated to a diffusion/image-text pipeline failure." + }, + { + "left": "issue:45229", + "right": "issue:45823", + "accept": false, + "reason": "Gemma 4 OOM during inference is not the same as the device mismatch bug in forward." + }, + { + "left": "issue:44464", + "right": "issue:45584", + "accept": false, + "reason": "Compiled-forward chunked generation mismatch is unrelated to Whisper empty-transcription handling." + }, + { + "left": "issue:43844", + "right": "issue:44964", + "accept": false, + "reason": "ZeRO-3 gradient growth and Phi-4 multimodal loading are unrelated bugs." + }, { "left": "issue:43873", "right": "issue:46050", "accept": false, - "reason": "Both involve quantization-related behavior, but one is about offloading with quantization and the other is save/load breakage for sinq; different bugs." + "reason": "Quantization offloading behavior and sinq save/load breakage are different defects." + }, + { + "left": "issue:43994", + "right": "issue:44493", + "accept": false, + "reason": "SigLIP2 wrong outputs and unexpected position-id keys are not the same code-path issue." }, { "left": "issue:45588", "right": "issue:46050", "accept": false, - "reason": "Different failure modes and code paths: flash-attention sink handling versus quantized model serialization/loading." + "reason": "Flash-attention s_aux=None crash is unrelated to sinq quantization persistence." }, { "left": "issue:29942", "right": "issue:30333", "accept": false, - "reason": "Flash Attention test failures and MLflow job status reporting are unrelated." + "reason": "Failing attention tests and MLflow reporting on job failure are unrelated." }, { "left": "issue:42490", "right": "issue:43696", "accept": false, - "reason": "A trailing-slash import warning and an OOM on GPT-oss are unrelated." + "reason": "Import warning on trailing slash vs CUDA out-of-memory; no shared bug." }, { "left": "issue:29159", "right": "issue:32090", "accept": false, - "reason": "Tokenizer consistency bug versus Trainer broadcast NoneType error; no shared underlying defect." + "reason": "Tokenizer slow/fast inconsistency is unrelated to a NoneType broadcast error in Trainer." + }, + { + "left": "issue:44977", + "right": "issue:46701", + "accept": false, + "reason": "Qwen3.5 flash-attention generation and DiffusionGemma pipeline failure are different models and paths." }, { "left": "issue:45084", "right": "issue:45520", "accept": false, - "reason": "Generic compile error versus a Python 3.13 import-utils KeyError; unrelated." + "reason": "Template-node compilation error and flash_attn import lookup error are distinct issues." }, { "left": "issue:43065", "right": "issue:46369", "accept": false, - "reason": "Same model family, but a dummy Conv2d implementation issue is not the same as a memory overflow report." + "reason": "Dummy Conv2d in Sam3PixelDecoder is not the same as memory overflow in sam3-litetext." }, { "left": "issue:28218", "right": "issue:34567", "accept": false, - "reason": "Tokenizer spacing behavior and TrainerState token counting are different subsystems and bugs." + "reason": "Added-token spacing behavior is unrelated to TrainerState token-count tracking." }, { "left": "issue:44423", "right": "issue:44734", "accept": false, - "reason": "Both are serve crashes, but they hit different endpoints and different root causes." + "reason": "Continuous-batching multimodal crash and KV-cache continuation indexing bug are different serve failures." }, { - "left": "issue:15354", - "right": "issue:38175", + "left": "issue:44554", + "right": "issue:45446", "accept": false, - "reason": "TorchScript export support for GeneratorExp is unrelated to SigLIP2 zero probabilities." + "reason": "MPS attention correctness and flex_attention version-check import are unrelated." }, { - "left": "issue:44554", - "right": "issue:45446", + "left": "issue:15354", + "right": "issue:38175", "accept": false, - "reason": "MPS attention correctness and a PyTorch version check in flex_attention are not the same bug." + "reason": "TorchScript GeneratorExp export failure is unrelated to SigLIP2 zero-probability outputs." }, { "left": "issue:45406", "right": "issue:46121", "accept": false, - "reason": "Gemma4Processor _tokenizer crash and rope-validation TypeError are unrelated." + "reason": "Gemma4Processor missing _tokenizer and rope-params list handling are different bugs." }, { "left": "issue:43493", "right": "issue:43701", "accept": false, - "reason": "SigLIP2 implementation discrepancy and resume_from_checkpoint key mismatch are unrelated." + "reason": "SigLIP2 HF-vs-JAX discrepancy and checkpoint resume key mismatch are unrelated." }, { "left": "issue:45072", "right": "issue:45198", "accept": false, - "reason": "Bfloat16 dtype mismatches and Wav2Vec2 save/tokenization failure are different issues." + "reason": "bfloat16 dtype mismatch in inference and Wav2Vec2 save/tokenization failure are separate problems." }, { "left": "issue:45588", "right": "issue:45759", "accept": false, - "reason": "Flash-attention sink handling and composite Qwen config unwrapping are unrelated." + "reason": "Flash-attention sink handling and Qwen multimodal config unwrapping are not the same underlying bug." }, { "left": "issue:37428", "right": "issue:43957", "accept": false, - "reason": "An import name regression and meta-device model loading breakage are different problems." + "reason": "Flash-support import error and meta-device model loading breakage are different loading regressions." }, { "left": "issue:43540", "right": "issue:45200", "accept": false, - "reason": "Qwen3OmniMoe video input processing and Gemma 4 mm_token_type_ids defaults are unrelated." + "reason": "Video input processing error in Qwen3OmniMoe is unrelated to Gemma 4 token-type defaults." }, { "left": "issue:43295", "right": "issue:43450", "accept": false, - "reason": "Processor.tokenizer regression versus batched video shape bug; not the same underlying change." + "reason": "Processor.tokenizer regression and batched video shape bug do not share the same code path." }, { "left": "issue:45200", "right": "issue:45823", "accept": false, - "reason": "Text-only fine-tuning defaults and Gemma4 PLE device mismatch are separate issues." - }, - { - "left": "issue:44186", - "right": "issue:46192", - "accept": false, - "reason": "LayoutLMv2Tokenizer crashes on NER/batched padding versus return_tensors being ignored are different tokenizer bugs." - }, - { - "left": "issue:45823", - "right": "issue:46192", - "accept": false, - "reason": "Gemma4 device placement and return_tensors argument handling do not share a code-path bug." - }, - { - "left": "issue:43295", - "right": "issue:43873", - "accept": false, - "reason": "Processor.tokenizer regression and quantization/offloading behavior are unrelated." - }, - { - "left": "issue:43295", - "right": "issue:44610", - "accept": false, - "reason": "Processor/tokenizer regression and OmDet-Turbo processor image-size mismatch are different issues." - }, - { - "left": "issue:43550", - "right": "issue:43844", - "accept": false, - "reason": "torch.compile SDPA failure in a specific model and gradient inflation under ZeRO-3 are unrelated." - }, - { - "left": "issue:45520", - "right": "issue:46178", - "accept": false, - "reason": "Both are backend/import related, but they concern different backend registries and different errors." - }, - { - "left": "issue:43450", - "right": "issue:43575", - "accept": false, - "reason": "Video processor batch-shape bug versus TP/OOM on Qwen2-57B-A14B-Instruct are unrelated." - }, - { - "left": "issue:43232", - "right": "issue:44849", - "accept": false, - "reason": "Generation kwargs update after sync_gpus and Qwen3.5 hidden-state output bug are different code paths." - }, - { - "left": "issue:43572", - "right": "issue:44987", - "accept": false, - "reason": "StableLmConfig pad token regression and loading failure for physical-intelligence/fast are not the same bug." - }, - { - "left": "issue:44554", - "right": "issue:45356", - "accept": false, - "reason": "MPS attention correctness issue and Kimi-K2.5 tokenizer regression are unrelated." - }, - { - "left": "issue:43572", - "right": "issue:43957", - "accept": false, - "reason": "Config default missing pad_token_idx and meta-device loading failures are different issues." - }, - { - "left": "issue:37428", - "right": "issue:43824", - "accept": false, - "reason": "Both are import-related, but they concern different symbols and different modules." + "reason": "Token-type-id defaults and PLE device mismatch are separate Gemma 4 issues." } ] }, @@ -8418,68 +8286,81 @@ "component_index": 1, "component_count": 22, "review_unit_index": 27, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:28218", "issue:35141", + "issue:37428", "issue:38175", "issue:39692", - "issue:41720", "issue:42490", "issue:42915", + "issue:43232", "issue:43262", "issue:43278", + "issue:43295", "issue:43334", - "issue:43388", "issue:43450", "issue:43452", "issue:43493", "issue:43525", - "issue:43526", + "issue:43550", + "issue:43572", + "issue:43575", "issue:43653", - "issue:43819", "issue:43824", - "issue:43827", + "issue:43844", "issue:43856", - "issue:43901", + "issue:43873", "issue:43950", + "issue:43957", "issue:43975", "issue:44112", - "issue:44442", + "issue:44186", "issue:44466", - "issue:44479", "issue:44521", - "issue:44561", + "issue:44554", "issue:44610", - "issue:44792", - "issue:44857", + "issue:44849", "issue:44898", + "issue:44987", "issue:45127", - "issue:45198", - "issue:45200", "issue:45290", - "issue:45325", + "issue:45356", "issue:45381", "issue:45412", - "issue:45563", + "issue:45520", + "issue:45823", + "issue:46178", "issue:46192", "issue:46209", - "issue:46234", "issue:46285", "issue:46512" ], "soft_pairs": [ - "issue:44521|issue:45290", + "issue:44186|issue:46192", + "issue:45823|issue:46192", + "issue:43295|issue:43873", + "issue:43295|issue:44610", + "issue:43550|issue:43844", + "issue:45520|issue:46178", + "issue:43450|issue:43575", + "issue:43232|issue:44849", + "issue:43572|issue:44987", + "issue:44554|issue:45356", + "issue:43572|issue:43957", "issue:44466|issue:45127", + "issue:37428|issue:43824", "issue:42490|issue:43493", - "issue:45381|issue:46209", "issue:39692|issue:46192", + "issue:44521|issue:45290", + "issue:45381|issue:46209", "issue:43653|issue:44112", "issue:38175|issue:43493", "issue:43278|issue:43856", @@ -8490,202 +8371,204 @@ "issue:43334|issue:43525", "issue:43824|issue:43975", "issue:44898|issue:45412", - "issue:43262|issue:43450", - "issue:44442|issue:45198", - "issue:44792|issue:45563", - "issue:45200|issue:45325", - "issue:43819|issue:44112", - "issue:44479|issue:46192", - "issue:43388|issue:43526", - "issue:43827|issue:43901", - "issue:41720|issue:46209", - "issue:44561|issue:46234", - "issue:44610|issue:44857" + "issue:43262|issue:43450" ], - "prepared_review_unit_hash": "e7e3c9a4884c4ef2326401db760bd726fb9fc1330b91374daadd0a35113a2cf1", + "prepared_review_unit_hash": "78001f71b6c7f784cb9e13ed4a7fd5b9df2504d48a0a4bbc0dd79652afacfb66", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 26, - "serialized_chars": 35333, - "estimated_input_tokens": 8834, - "estimated_eval_tokens": 17924 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "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 largely heterogeneous; the soft-paired issues mostly share only broad subsystem keywords (tokenizers, chat templates, vision/video, model loading) rather than the same underlying bug. I would not merge any of the proposed soft edges. As a representative issue, #44521 is the strongest active, user-facing bug report in the set, but it is not a duplicate of the others.", - "confidence": 0.33, - "canonical_issue_reason": "#44521 is open, well-discussed, has multiple inbound references, and describes a concrete user-facing failure with a clear repro path. It is a good representative of the cluster's general multimodal regression theme, even though the cluster itself is not a true duplicate set.", + "summary": "This cluster is mostly heterogeneous; the soft pairs share broad subsystems at most, but none look like the same concrete bug/change. I would use issue 44521 as the representative issue because it is the most central/high-signal open bug in the closest recurring theme here (`apply_chat_template` / multimodal processor behavior).", + "confidence": 0.81, + "canonical_issue_reason": "Issue 44521 is the best representative because it has the strongest discussion/inbound-reference signal and describes a broad, user-facing multimodal `apply_chat_template` failure, which is the closest repeated theme in the cluster.", "canonical_pr_reason": null, - "best_issue_reason": "#44521 is the best issue to keep as the cluster representative because it is active, specific, and broadly impactful. The other issues are mostly unrelated model-specific regressions or docs/loading quirks.", + "best_issue_reason": "Issue 44521 is the most suitable anchor for this cluster: it is active, broadly applicable, and more central than the other candidates, which are mostly narrow, unrelated regressions.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44521", - "right": "issue:45290", + "left": "issue:44186", + "right": "issue:46192", "accept": false, - "reason": "Both involve apply_chat_template, but one is all-zero assistant masks for multimodal inputs and the other is a crash with tool-call assistant messages. Different failure modes and likely different fixes." + "reason": "Both are tokenizer-related, but one is a LayoutLMv2 batched NER/padding crash and the other is `return_tensors` being ignored when `text_kwargs` is passed; different code paths." }, { - "left": "issue:44466", - "right": "issue:45127", + "left": "issue:45823", + "right": "issue:46192", "accept": false, - "reason": "Both mention tied weights/lm_head behavior, but #44466 is a serialization/device-dependent save issue while #45127 is LoRA merge collapse after extending vocab. Not the same bug." + "reason": "Gemma4 device-map forward mismatch vs processor kwargs handling; unrelated failures." }, { - "left": "issue:42490", - "right": "issue:43493", + "left": "issue:43295", + "right": "issue:43873", "accept": false, - "reason": "Unrelated problems: local-path import warning vs SigLIP2 implementation discrepancy." + "reason": "Processor/tokenizer API regression and quantization offloading are different subsystems and bugs." }, { - "left": "issue:45381", - "right": "issue:46209", + "left": "issue:43295", + "right": "issue:44610", "accept": false, - "reason": "Different domains and failures: qwen2.5-vl vision_position_ids bug vs FP8 inference failure on B200." + "reason": "Both involve processors, but one is `processor.tokenizer`/image passthrough regression and the other is an OmDet-Turbo input-size mismatch." }, { - "left": "issue:39692", - "right": "issue:46192", + "left": "issue:43550", + "right": "issue:43844", "accept": false, - "reason": "Docs example/model-processor mismatch and quantization failure is unrelated to return_tensors being ignored when text_kwargs is passed." + "reason": "torch.compile+SDPA correctness bug vs DeepSpeed ZeRO-3 gradient growth; no shared code path." }, { - "left": "issue:43653", - "right": "issue:44112", + "left": "issue:45520", + "right": "issue:46178", "accept": false, - "reason": "BigBirdTokenizer special-token registration bug and GraniteSpeech CI device-override test failure are unrelated." + "reason": "Both are import/backend errors, but for different symbols and different backend registry problems." }, { - "left": "issue:38175", - "right": "issue:43493", + "left": "issue:43450", + "right": "issue:43575", "accept": false, - "reason": "Both are about SigLIP2, but one is unexpected zero probabilities on a specific checkpoint and the other is a broader HF-vs-JAX discrepancy. Similar area, not clearly the same concrete bug." + "reason": "Video processor batch-shape bug vs tensor-parallel OOM when loading a model; unrelated." }, { - "left": "issue:43278", - "right": "issue:43856", + "left": "issue:43232", + "right": "issue:44849", "accept": false, - "reason": "dtype drift in evaluation vs memory inefficiency during Qwen3 MoE training are different issues." + "reason": "Generation kwargs update after `sync_gpus` vs Qwen3.5 hidden-states bug; different behaviors." }, { - "left": "issue:28218", - "right": "issue:35141", + "left": "issue:43572", + "right": "issue:44987", "accept": false, - "reason": "Tokenizer adds extra space after added token is unrelated to embedding reinitialization after resize when tie_word_embedding is false." + "reason": "A missing `pad_token_idx` config issue and a model-loading failure are too different to treat as the same bug." }, { - "left": "issue:43950", - "right": "issue:46285", + "left": "issue:44554", + "right": "issue:45356", "accept": false, - "reason": "Non-persistent buffer corruption on from_pretrained is unrelated to CLIP model structure inconsistency." + "reason": "MPS attention correctness issue vs tokenizer codec/regex regression; unrelated." }, { - "left": "issue:42915", + "left": "issue:43572", + "right": "issue:43957", + "accept": false, + "reason": "StableLm pad-token config bug vs meta-device loading regression; different failures." + }, + { + "left": "issue:44466", + "right": "issue:45127", + "accept": false, + "reason": "Both touch tied weights, but one is device-dependent serialization and the other is LoRA merge collapse after vocab extension; not the same bug." + }, + { + "left": "issue:37428", "right": "issue:43824", "accept": false, - "reason": "FineGrainedFP8Config failure for Qwen3Moe and a missing Qwen2_5_VL import are separate model/loading problems." + "reason": "Different missing-import symbols from different modules; not duplicates." }, { - "left": "issue:43452", - "right": "issue:46512", + "left": "issue:42490", + "right": "issue:43493", "accept": false, - "reason": "gguf_file loading breakage for AutoTokenizer/AutoModel is unrelated to from_config(dtype=...) leaving config.dtype stale." + "reason": "Trailing-slash local-path warning vs SigLIP2 implementation discrepancy; unrelated." }, { - "left": "issue:43334", - "right": "issue:43525", + "left": "issue:39692", + "right": "issue:46192", "accept": false, - "reason": "Both are missing pad_token_id on different configs, but they affect different model families and likely different code paths." + "reason": "SigLIP2 doc-example errors vs processor `return_tensors` handling; different issues." }, { - "left": "issue:43824", - "right": "issue:43975", + "left": "issue:44521", + "right": "issue:45290", "accept": false, - "reason": "ImportError for Qwen2_5_VLForConditionalGeneration is unrelated to DeepSeek detokenization regression." + "reason": "Both mention `apply_chat_template`, but one is all-zero assistant masks for multimodal inputs and the other is a crash on tool-call assistant messages with no content." }, { - "left": "issue:44898", - "right": "issue:45412", + "left": "issue:45381", + "right": "issue:46209", "accept": false, - "reason": "Perceiver non-default resolution failure and RT-DETR memory release issues are unrelated." + "reason": "Qwen2.5-VL video position IDs vs FP8 inference failure on B200; unrelated." }, { - "left": "issue:43262", - "right": "issue:43450", + "left": "issue:43653", + "right": "issue:44112", "accept": false, - "reason": "Audio chat-template sampling-rate default and batched video shape regression are different processor bugs." + "reason": "BigBirdTokenizer special-token bug vs a stale GraniteSpeech CI test; no shared underlying defect." }, { - "left": "issue:44442", - "right": "issue:45198", + "left": "issue:38175", + "right": "issue:43493", "accept": false, - "reason": "Different tokenizer/model-loading failures for different speech models; no evidence they share the same underlying bug." + "reason": "Both are SigLIP2-related, but one reports zero probabilities and the other a HF/JAX implementation discrepancy; not enough to call them the same concrete bug." }, { - "left": "issue:44792", - "right": "issue:45563", + "left": "issue:43278", + "right": "issue:43856", "accept": false, - "reason": "Janus image-generation test failure is unrelated to a stale generate warning about num_return_sequences." + "reason": "Embedding dtype regression vs Qwen3 MoE memory inefficiency; different code paths." }, { - "left": "issue:45200", - "right": "issue:45325", + "left": "issue:28218", + "right": "issue:35141", "accept": false, - "reason": "Gemma 4 mm_token_type_ids defaulting issue and qwen2.5-vl rope/position_id scaling issue are different multimodal bugs." + "reason": "Tokenizer extra-space bug vs embedding resize/post-init reinitialization bug; unrelated." }, { - "left": "issue:43819", - "right": "issue:44112", + "left": "issue:43950", + "right": "issue:46285", "accept": false, - "reason": "DAC latent/STE mismatch and GraniteSpeech CI device-override failure are unrelated." + "reason": "Non-persistent buffer corruption during `from_pretrained` vs CLIP text model structure change; different regressions." }, { - "left": "issue:44479", - "right": "issue:46192", + "left": "issue:42915", + "right": "issue:43824", "accept": false, - "reason": "A video-input regression for Qwen multimodal models is unrelated to return_tensors being ignored with explicit text_kwargs." + "reason": "Qwen3Moe FP8 config failure vs VL class import error; unrelated." }, { - "left": "issue:43388", - "right": "issue:43526", + "left": "issue:43452", + "right": "issue:46512", "accept": false, - "reason": "Different label-handling bugs in different components: gather_for_metrics tuple truncation vs BeitImageProcessorFast reduce_labels returning one label." + "reason": "`gguf_file` loading failure vs stale `config.dtype` after `from_config`; different loading semantics." }, { - "left": "issue:43827", - "right": "issue:43901", + "left": "issue:43334", + "right": "issue:43525", "accept": false, - "reason": "Both are docs cleanup items about old pipeline arguments, but they concern different APIs and are not the same concrete change." + "reason": "Both are missing-`pad_token_id` AttributeErrors, but on different model families and loaders; not clearly the same concrete bug." }, { - "left": "issue:41720", - "right": "issue:46209", + "left": "issue:43824", + "right": "issue:43975", "accept": false, - "reason": "Different hardware/model failures: auto device mapping cudaErrorAssert on A800 vs FP8 inference failure on B200." + "reason": "ImportError for a VL class vs DeepSeek tokenizer detokenization regression; unrelated." }, { - "left": "issue:44561", - "right": "issue:46234", + "left": "issue:44898", + "right": "issue:45412", "accept": false, - "reason": "Removal of is_torch_fx_available breaking trust_remote_code models is unrelated to a missing import_utils helper for Falcon Mamba." + "reason": "Perceiver resolution failure vs RT-DETR memory leak after deletion; different bugs." }, { - "left": "issue:44610", - "right": "issue:44857", + "left": "issue:43262", + "right": "issue:43450", "accept": false, - "reason": "OmDet-Turbo processor size mismatch and LwDetrImageLoss float16/CUDA crash are different bugs in different models/components." + "reason": "Audio `apply_chat_template` sampling-rate default vs video processor batch-shape bug; different modalities and code paths." } ] }, @@ -8701,69 +8584,78 @@ "component_index": 1, "component_count": 22, "review_unit_index": 28, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ - "issue:29127", "issue:29942", "issue:32090", - "issue:34689", - "issue:35141", "issue:37428", "issue:38175", + "issue:41720", "issue:43278", "issue:43335", + "issue:43388", "issue:43425", "issue:43504", + "issue:43526", "issue:43550", "issue:43606", "issue:43611", "issue:43653", - "issue:43720", "issue:43819", "issue:43825", + "issue:43827", + "issue:43901", "issue:43931", "issue:44112", "issue:44206", - "issue:44351", + "issue:44442", "issue:44448", - "issue:44514", - "issue:44521", + "issue:44479", + "issue:44561", + "issue:44610", "issue:44743", + "issue:44792", "issue:44855", "issue:44857", "issue:45083", - "issue:45127", + "issue:45198", "issue:45200", - "issue:45245", - "issue:45290", + "issue:45325", "issue:45412", - "issue:45459", "issue:45520", + "issue:45563", "issue:45584", - "issue:45588", "issue:45593", "issue:45759", - "issue:45859", "issue:46170", + "issue:46192", + "issue:46209", "issue:46234", "issue:46285", - "issue:46326", - "issue:46431", - "issue:46512", - "issue:46566" + "issue:46512" ], "soft_pairs": [ + "issue:44442|issue:45198", + "issue:45200|issue:45325", + "issue:44792|issue:45563", + "issue:43819|issue:44112", + "issue:44479|issue:46192", + "issue:43388|issue:43526", + "issue:43827|issue:43901", + "issue:44561|issue:46234", + "issue:44610|issue:44857", + "issue:41720|issue:46209", "issue:43335|issue:43550", "issue:43819|issue:45083", "issue:45759|issue:46285", - "issue:44206|issue:45584", "issue:44857|issue:45593", + "issue:44206|issue:45584", "issue:38175|issue:45412", "issue:45520|issue:46170", "issue:43606|issue:45083", @@ -8775,273 +8667,205 @@ "issue:44448|issue:44743", "issue:45759|issue:46512", "issue:43335|issue:43504", - "issue:44514|issue:44521", "issue:44855|issue:45083", - "issue:43278|issue:43825", - "issue:43720|issue:44112", - "issue:29127|issue:32090", - "issue:45290|issue:46326", - "issue:43335|issue:45083", - "issue:46431|issue:46566", - "issue:43335|issue:44112", - "issue:45127|issue:45245", - "issue:43335|issue:45593", - "issue:37428|issue:44351", - "issue:45588|issue:45859", - "issue:45200|issue:45588", - "issue:34689|issue:35141", - "issue:45459|issue:46234" + "issue:43278|issue:43825" ], - "prepared_review_unit_hash": "0b5205ead8bdf393691cf500bceccdc22969fe0d3df43c4ae79cdbc6628cfc96", + "prepared_review_unit_hash": "8153038d670726ead7b2a6ff7aa4a491d5ccb18e2c8ec1b16275b94b252da3e7", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 37645, - "estimated_input_tokens": 9412, - "estimated_eval_tokens": 19080 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35795, + "estimated_input_tokens": 8949, + "estimated_eval_tokens": 18154 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The items span many unrelated Transformers bugs and regressions. I found no true duplicate pairs here, so this cluster should not be merged aggressively; one chat-template issue is the closest representative, but the set is heterogeneous overall.", - "confidence": 0.22, - "canonical_issue_reason": "Issue 44521 is the closest cluster representative: it is user-facing, has multiple inbound references, and covers a broad chat-template failure mode. That said, the overall set is too diverse to have a strong canonical duplicate.", + "summary": "These are mostly unrelated closed issues that only share broad Transformers subsystems or similar keywords; none of the soft edges look like the same underlying bug/change, so I would not merge any pair as duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue 44521 is the best single issue to anchor the cluster because it has the most visible impact and the strongest cross-link signal among the listed issues.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43335", - "right": "issue:43550", + "left": "issue:44442", + "right": "issue:45198", "accept": false, - "reason": "Sparse-layer config creation vs Bamba torch.compile/SDPA failure; different model and different bug." + "reason": "Different tokenizer failures in different models: AutoTokenizer loading vs Wav2Vec2 save_pretrained/tokenization." }, { - "left": "issue:43819", - "right": "issue:45083", + "left": "issue:45200", + "right": "issue:45325", "accept": false, - "reason": "DAC forward/from_latents mismatch vs qwen3_omni_moe length helper behavior; unrelated code paths." + "reason": "Different Qwen2.5-VL bugs: missing mm_token_type_ids default vs RoPE position scaling." }, { - "left": "issue:45759", - "right": "issue:46285", + "left": "issue:44792", + "right": "issue:45563", "accept": false, - "reason": "Both involve model construction, but one is composite Qwen text_config unwrapping and the other is CLIP class structure drift; not the same bug." + "reason": "One is a Janus image-generation test failure; the other is a stale generate() warning." }, { - "left": "issue:44206", - "right": "issue:45584", + "left": "issue:43819", + "right": "issue:44112", "accept": false, - "reason": "LASR feature extractor crash on unsupported arg vs Whisper empty-transcription generation bug; different components and symptoms." + "reason": "Different failures: DAC latent/forward mismatch vs a GraniteSpeech CI device-override test." }, { - "left": "issue:44857", - "right": "issue:45593", + "left": "issue:44479", + "right": "issue:46192", "accept": false, - "reason": "LwDetr AMP float16 crash vs D-FINE auxiliary-loss omission; distinct issues in detection models." + "reason": "Different regressions: multimodal video input handling vs return_tensors being ignored with text_kwargs." }, { - "left": "issue:38175", - "right": "issue:45412", + "left": "issue:43388", + "right": "issue:43526", "accept": false, - "reason": "Siglip2 zero probabilities vs RT-DETR memory leak on deletion; unrelated failure modes." + "reason": "Both involve labels, but one is a distributed metric gathering bug and the other is a Beit image processor label-reduction bug." }, { - "left": "issue:45520", - "right": "issue:46170", + "left": "issue:43827", + "right": "issue:43901", "accept": false, - "reason": "flash_attn import key error on Python 3.13 vs Windows/ROCm torch.distributed import problem; different dependency issues." + "reason": "Both are docs issues, but they concern different APIs and different v5 behavior changes." }, { - "left": "issue:43606", - "right": "issue:45083", + "left": "issue:44561", + "right": "issue:46234", "accept": false, - "reason": "CPU offload device mismatch in bark-small vs qwen3_omni_moe feature-length helper; unrelated subsystems." + "reason": "Different missing utilities: removal of is_torch_fx_available vs missing is_falcon_mamba_ssm_available." }, { - "left": "issue:43335", - "right": "issue:43653", - "accept": false, - "reason": "SwitchTransformers sparse-layer bug vs BigBirdTokenizer special-token registration; no shared underlying defect." - }, - { - "left": "issue:29942", - "right": "issue:32090", - "accept": false, - "reason": "Flash Attention 2 test failures vs Trainer _gpu_broadcast_one NoneType TypeError; different runtime paths." - }, - { - "left": "issue:43550", - "right": "issue:44112", - "accept": false, - "reason": "Bamba SDPA compile failure vs GraniteSpeech stale device override CI test; not the same code-path problem." - }, - { - "left": "issue:37428", - "right": "issue:43425", - "accept": false, - "reason": "Missing internal flash-attention symbol vs Torch 2.10 compatibility; related to versioning, but not the same concrete bug." - }, - { - "left": "issue:43611", - "right": "issue:43931", - "accept": false, - "reason": "base_model_prefix loading regression vs Qwen3-VL weight shape mismatch; different loading failures." - }, - { - "left": "issue:44448", - "right": "issue:44743", + "left": "issue:44610", + "right": "issue:44857", "accept": false, - "reason": "Pegasus v4/v5 output drift vs recurrent-state reset with cache; unrelated model behaviors." + "reason": "Different code paths: OmDet-Turbo processor size mismatch vs LwDetrImageLoss AMP/CUDA crash." }, { - "left": "issue:45759", - "right": "issue:46512", + "left": "issue:41720", + "right": "issue:46209", "accept": false, - "reason": "Both are from_config-related, but one unwraps text_config for multimodal configs while the other leaves config.dtype stale; separate bugs." + "reason": "Different hardware/runtime failures: Qwen3 auto device mapping cudaErrorAssert vs FP8 inference on B200." }, { "left": "issue:43335", - "right": "issue:43504", - "accept": false, - "reason": "SwitchTransformers config creates sparse layer vs Beit pretrained preset legacy-field load failure; different models and causes." - }, - { - "left": "issue:44514", - "right": "issue:44521", + "right": "issue:43550", "accept": false, - "reason": "Same API family, but one is a batched padding=False crash and the other is zero assistant_masks for multimodal input; not clearly the same defect." + "reason": "Different bugs in different models: SwitchTransformers sparse-layer creation vs Bamba torch.compile+SDPA failure." }, { - "left": "issue:44855", + "left": "issue:43819", "right": "issue:45083", "accept": false, - "reason": "Python 3.13 IndentationError in DebertaV2Model vs qwen3_omni_moe feature-length helper bug; unrelated." + "reason": "Different helper/model bugs; DAC latent mismatch is unrelated to qwen3_omni_moe feature-length computation." }, { - "left": "issue:43278", - "right": "issue:43825", + "left": "issue:45759", + "right": "issue:46285", "accept": false, - "reason": "Training/evaluate dtype mismatch vs pipeline translation-task error message regression; different concerns." + "reason": "Different structural issues: Qwen multimodal config unwrapping vs CLIP text-model attribute inconsistency." }, { - "left": "issue:43720", - "right": "issue:44112", + "left": "issue:44857", + "right": "issue:45593", "accept": false, - "reason": "Packed-weight unpacking during accelerate loading vs stale device override CI test; not the same underlying issue." + "reason": "Different training-time failures: AMP loss crash vs auxiliary-loss logic when denoising is off." }, { - "left": "issue:29127", - "right": "issue:32090", + "left": "issue:44206", + "right": "issue:45584", "accept": false, - "reason": "LayoutLMV3 box validation message clarity vs Trainer broadcast TypeError; completely different bugs." + "reason": "Different behaviors: feature extractor unsupported arg vs Whisper empty-transcription generation after token alignment." }, { - "left": "issue:45290", - "right": "issue:46326", + "left": "issue:38175", + "right": "issue:45412", "accept": false, - "reason": "Tool-call assistant message crash vs continue_final_message tokenizer behavior; same area, but different failure conditions and not one fix." + "reason": "Different model problems: unexpected zero probabilities vs RT-DETR memory not being released." }, { - "left": "issue:43335", - "right": "issue:45083", + "left": "issue:45520", + "right": "issue:46170", "accept": false, - "reason": "SwitchTransformers sparse-layer creation vs qwen3_omni_moe helper output lengths; unrelated." + "reason": "Different platform/import issues: flash_attn lookup on Python 3.13 vs Windows ROCm torch.distributed dependency failure." }, { - "left": "issue:46431", - "right": "issue:46566", + "left": "issue:43606", + "right": "issue:45083", "accept": false, - "reason": "Missing chat template guard vs DiffusionGemma meta-device loading problem; different components." + "reason": "Different bugs: CPU offload device mismatch vs qwen3_omni_moe output-length helper behavior." }, { "left": "issue:43335", - "right": "issue:44112", + "right": "issue:43653", "accept": false, - "reason": "Configuration bug in SwitchTransformers vs GraniteSpeech CI test instability; no shared code-path." + "reason": "Different components and symptoms: SwitchTransformers config bug vs BigBirdTokenizer special-token registration/empty decode bug." }, { - "left": "issue:45127", - "right": "issue:45245", + "left": "issue:29942", + "right": "issue:32090", "accept": false, - "reason": "LoRA merge collapse with extended vocab vs category-cardinality runtime limit; unrelated model logic." + "reason": "One is Flash Attention 2 test failures; the other is a Trainer broadcast TypeError with NoneType." }, { - "left": "issue:43335", - "right": "issue:45593", + "left": "issue:43550", + "right": "issue:44112", "accept": false, - "reason": "SwitchTransformers sparse-layer config vs D-FINE denoising/aux-loss behavior; different subsystems." + "reason": "Different issues: Bamba torch.compile with SDPA vs a stale GraniteSpeech CI device override test." }, { "left": "issue:37428", - "right": "issue:44351", - "accept": false, - "reason": "flash_supports_window_size import regression vs missing HybridCache export; both import-related, but not the same symbol or fix." - }, - { - "left": "issue:45588", - "right": "issue:45859", - "accept": false, - "reason": "Flash-attention s_aux=None crash vs Qwen3_5Moe rotary embedding CPU offload incompatibility; distinct MoE/attention issues." - }, - { - "left": "issue:45200", - "right": "issue:45588", - "accept": false, - "reason": "Gemma mm_token_type_ids defaulting bug vs flash_attention sink-less model crash; different APIs and causes." - }, - { - "left": "issue:34689", - "right": "issue:35141", + "right": "issue:43425", "accept": false, - "reason": "Llama 3.2 vision model loading regression vs post_init reinitializing output embeddings after resize; separate loading/training bugs." + "reason": "Different scope: missing flash_attention utility import vs general Torch 2.10 incompatibility." }, { - "left": "issue:45459", - "right": "issue:46234", + "left": "issue:43611", + "right": "issue:43931", "accept": false, - "reason": "Tokenizer error masking without protobuf vs missing import_utils availability helper; unrelated import/util issues." + "reason": "Different loading bugs: base_model_prefix compatibility vs Qwen3-VL weight-shape mismatch." }, { - "left": "issue:44514", - "right": "issue:44521", + "left": "issue:44448", + "right": "issue:44743", "accept": false, - "reason": "Both hit apply_chat_template, but one is a batched padding=False crash and the other produces empty assistant masks; too different to merge." + "reason": "Different regressions in different models: Pegasus output drift across v4/v5 vs Qwen recurrent-state reset with cache." }, { - "left": "issue:43825", - "right": "issue:44521", + "left": "issue:45759", + "right": "issue:46512", "accept": false, - "reason": "Translation-task error message regression vs all-zero assistant masks in multimodal chat templates; different user-visible failures." + "reason": "Different config/dtype problems: multimodal text_config unwrapping vs stale config.dtype after from_config(dtype=...)." }, { - "left": "issue:46326", - "right": "issue:46431", + "left": "issue:43335", + "right": "issue:43504", "accept": false, - "reason": "continue_final_message bug in mistral_common tokenizers vs missing chat template on processor; separate validation/templating issues." + "reason": "Different SwitchTransformers configuration vs Beit pretrained-model legacy-field loading bug." }, { - "left": "issue:44206", - "right": "issue:45290", + "left": "issue:44855", + "right": "issue:45083", "accept": false, - "reason": "Unsupported center arg in feature extractor vs assistant message tool-call crash; unrelated code paths." + "reason": "Different failures: Python 3.13 JIT parsing/IndentationError vs qwen3_omni_moe feature-length helper behavior." }, { - "left": "issue:44743", - "right": "issue:45759", + "left": "issue:43278", + "right": "issue:43825", "accept": false, - "reason": "Recurrent-state reset in modular_qwen3_5 cache handling vs from_config text_config unwrapping; different aspects of model wiring." + "reason": "Different issues: train/eval embedding dtype mismatch vs pipeline error messaging about translation support." } ] }, @@ -9057,301 +8881,302 @@ "component_index": 1, "component_count": 22, "review_unit_index": 29, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ - "issue:10000", - "issue:15354", "issue:26413", - "issue:27957", "issue:28218", - "issue:29942", + "issue:29127", "issue:31501", "issue:32090", "issue:33357", "issue:34689", "issue:35141", - "issue:38175", + "issue:37428", "issue:39692", - "issue:40444", "issue:42490", - "issue:43278", - "issue:43329", + "issue:43335", "issue:43526", - "issue:43550", - "issue:43575", "issue:43606", "issue:43701", + "issue:43720", "issue:43819", "issue:43854", + "issue:44112", "issue:44206", + "issue:44351", "issue:44360", "issue:44448", "issue:44493", + "issue:44514", + "issue:44521", "issue:44561", "issue:44617", "issue:44779", - "issue:44855", "issue:44857", - "issue:44977", "issue:45072", "issue:45083", - "issue:45229", + "issue:45127", + "issue:45200", + "issue:45245", "issue:45290", "issue:45310", - "issue:45381", + "issue:45459", "issue:45478", + "issue:45588", + "issue:45593", "issue:45823", - "issue:45920", - "issue:46192", - "issue:46399", - "issue:46655", + "issue:45859", + "issue:46234", + "issue:46326", + "issue:46431", + "issue:46566", "issue:46697" ], "soft_pairs": [ + "issue:44514|issue:44521", + "issue:43720|issue:44112", + "issue:29127|issue:32090", + "issue:43335|issue:45083", + "issue:43335|issue:44112", + "issue:43335|issue:45593", + "issue:45127|issue:45245", + "issue:46431|issue:46566", + "issue:37428|issue:44351", + "issue:45290|issue:46326", + "issue:45588|issue:45859", + "issue:45200|issue:45588", + "issue:34689|issue:35141", + "issue:45459|issue:46234", "issue:42490|issue:43854", "issue:39692|issue:43526", - "issue:43819|issue:44493", "issue:26413|issue:31501", + "issue:43819|issue:44493", "issue:45310|issue:45478", - "issue:44617|issue:46697", "issue:43606|issue:44857", "issue:44448|issue:44779", + "issue:44617|issue:46697", "issue:39692|issue:44493", - "issue:44206|issue:45823", "issue:28218|issue:32090", + "issue:44206|issue:45823", "issue:44360|issue:44561", "issue:44857|issue:45072", "issue:43701|issue:45083", - "issue:32090|issue:33357", - "issue:38175|issue:44493", - "issue:42490|issue:43575", - "issue:43550|issue:45083", - "issue:29942|issue:35141", - "issue:45290|issue:45381", - "issue:45920|issue:46192", - "issue:44448|issue:44977", - "issue:27957|issue:34689", - "issue:40444|issue:43526", - "issue:43278|issue:43329", - "issue:44977|issue:45229", - "issue:45290|issue:46655", - "issue:44855|issue:46399", - "issue:10000|issue:15354" + "issue:32090|issue:33357" ], - "prepared_review_unit_hash": "2dfc36d2b7bbc86e54a0a02457a096d09bfbe30ea58cc8872d0e90c937fd8d60", + "prepared_review_unit_hash": "bc3736562e708c5342864b60bdc30718b66e4b08cf3c88a1b761a3537aefb261", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 35798, - "estimated_input_tokens": 8950, - "estimated_eval_tokens": 18156 + "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": "Mostly heterogeneous issues with one clear duplicate pair: the two Qwen3.5 MoE `from_pretrained` reports. The rest describe different models, subsystems, or failure modes and should not be merged.", - "confidence": 0.34, - "canonical_issue_reason": "issue:45920 is the broadest, most central regression report in the set, affecting many model families and attracting inbound references.", + "summary": "This cluster is mostly heterogeneous and does not form a coherent duplicate set. The only clear duplicate pair is the repeated Qwen3.5 Moe `from_pretrained` regression reports (45310/45478).", + "confidence": 0.95, + "canonical_issue_reason": "No single global canonical exists for the whole cluster; if anchoring to the only true duplicate pair, `issue:45310` is the best canonical issue because it is earlier and `issue:45478` is the same reported Qwen3.5 Moe `from_pretrained` error.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45920 is the best representative issue to keep because it captures a broad, user-visible tokenizer regression rather than a narrow one-off symptom.", + "best_issue_reason": "`issue:45310` is the best representative issue because it cleanly matches `issue:45478` and has the earliest report for that exact bug.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:42490", - "right": "issue:43854", + "left": "issue:44514", + "right": "issue:44521", "accept": false, - "reason": "Different loading bugs: local-path trailing-slash import warning vs a specific unit-test model loading failure." + "reason": "Both are `apply_chat_template` problems, but one is a batching crash and the other is zero assistant masks for multimodal inputs; different failure modes." }, { - "left": "issue:39692", - "right": "issue:43526", + "left": "issue:43720", + "right": "issue:44112", "accept": false, - "reason": "Docs/example errors in SigLIP2 are unrelated to BeitImageProcessorFast label reduction behavior." + "reason": "Different models and different bugs: packed-weight loading in BitNet vs a stale GraniteSpeech CI test." }, { - "left": "issue:43819", - "right": "issue:44493", + "left": "issue:29127", + "right": "issue:32090", "accept": false, - "reason": "Different code paths and symptoms: DAC latents mismatch vs position-id key issues during model loading." + "reason": "Unrelated subsystems: LayoutLMv3 error messaging vs Trainer `_gpu_broadcast_one` NoneType handling." }, { - "left": "issue:26413", - "right": "issue:31501", + "left": "issue:43335", + "right": "issue:45083", "accept": false, - "reason": "Both involve training startup, but the concrete failures are different and not the same underlying bug." + "reason": "Both involve model internals, but one is SwitchTransformers sparse-layer config creation and the other is a qwen3_omni_moe feature-length helper issue." }, { - "left": "issue:45310", - "right": "issue:45478", - "accept": true, - "reason": "Same model family, same version gate, and identical `from_pretrained` error report; these look like duplicate reports of the same bug." + "left": "issue:43335", + "right": "issue:44112", + "accept": false, + "reason": "Different code paths and symptoms; no shared underlying defect." }, { - "left": "issue:44617", - "right": "issue:46697", + "left": "issue:43335", + "right": "issue:45593", "accept": false, - "reason": "CUDA OOM for Sam3Video is unrelated to Swin2SR padding overrun in the image processor." + "reason": "SwitchTransformers config bug is unrelated to D-FINE auxiliary-loss behavior." }, { - "left": "issue:43606", - "right": "issue:44857", + "left": "issue:45127", + "right": "issue:45245", "accept": false, - "reason": "CPU offload device mismatch and float16 AMP loss crash are different failures in different code paths." + "reason": "LoRA/embedding collapse and the 2^24 categories limit are unrelated." }, { - "left": "issue:44448", - "right": "issue:44779", + "left": "issue:46431", + "right": "issue:46566", "accept": false, - "reason": "Different models and regression types: Pegasus generation changes vs DeepSeek tokenizer output changes." + "reason": "Chat-template absence is unrelated to DiffusionGemma weights staying on meta with `device_map=\"auto\"`." }, { - "left": "issue:39692", - "right": "issue:44493", + "left": "issue:37428", + "right": "issue:44351", "accept": false, - "reason": "A SigLIP2 docs example issue is not the same as a broad position-id key warning across many models." + "reason": "Both are import errors, but they concern different missing symbols and likely different fixes." }, { - "left": "issue:44206", - "right": "issue:45823", + "left": "issue:45290", + "right": "issue:46326", "accept": false, - "reason": "Unsupported feature-extractor argument and Gemma4 device-map mismatch are unrelated bugs." + "reason": "Both touch chat templates, but one is tool-call tokenization with empty assistant content and the other is `continue_final_message` on mistral_common tokenizers." }, { - "left": "issue:28218", - "right": "issue:32090", + "left": "issue:45588", + "right": "issue:45859", "accept": false, - "reason": "Tokenizer spacing behavior and distributed broadcast NoneType errors are unrelated." + "reason": "Different failures: flash-attention `s_aux=None` crash vs CPU-offload incompatibility in Qwen rotary embedding." }, { - "left": "issue:44360", - "right": "issue:44561", + "left": "issue:45200", + "right": "issue:45588", "accept": false, - "reason": "A missing ReLU in DSA indexer is not the same as removal of `is_torch_fx_available` breaking remote-code models." + "reason": "Gemma4 `mm_token_type_ids` defaulting and flash_attention `s_aux=None` are unrelated issues." }, { - "left": "issue:44857", - "right": "issue:45072", + "left": "issue:34689", + "right": "issue:35141", "accept": false, - "reason": "Both mention dtype issues, but they are different models and distinct failure mechanisms." + "reason": "Different model-loading regression vs post-init reinitialization of output embeddings." }, { - "left": "issue:43701", - "right": "issue:45083", + "left": "issue:45459", + "right": "issue:46234", "accept": false, - "reason": "Checkpoint key mismatches are unrelated to the qwen3_omni_moe feature-length helper bug." + "reason": "Tokenizer protobuf error masking and a missing import-utils availability helper are unrelated." }, { - "left": "issue:32090", - "right": "issue:33357", + "left": "issue:42490", + "right": "issue:43854", "accept": false, - "reason": "NoneType GPU broadcast and MacOS bus-error CLIP loading are different runtime failures." + "reason": "A trailing-slash local path warning is not the same bug as GLM-4.7-Flash unit-test load failure." }, { - "left": "issue:38175", - "right": "issue:44493", + "left": "issue:39692", + "right": "issue:43526", "accept": false, - "reason": "Unexpected zero probabilities in SigLIP2 is not the same bug as unexpected position-id keys in loading." + "reason": "SigLIP2 docs/example errors and BeitImageProcessorFast `reduce_labels` are different issues." }, { - "left": "issue:42490", - "right": "issue:43575", + "left": "issue:26413", + "right": "issue:31501", "accept": false, - "reason": "Trailing-slash local model warning is unrelated to Qwen2-57B-A14B-Instruct OOM during tensor-parallel loading." + "reason": "The resume-from-checkpoint iterator error and FineWeb training not starting are not the same concrete bug." }, { - "left": "issue:43550", - "right": "issue:45083", + "left": "issue:43819", + "right": "issue:44493", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba is unrelated to the qwen3_omni_moe feature-length helper issue." + "reason": "DAC `from_latents`/STE mismatch is unrelated to unexpected position-id keys." }, { - "left": "issue:29942", - "right": "issue:35141", - "accept": false, - "reason": "Flash Attention 2 test failures and embedding reinitialization after resize are different bugs." + "left": "issue:45310", + "right": "issue:45478", + "accept": true, + "reason": "Same exact bug report: Qwen3.5 Moe `from_pretrained` error on transformers>=5.4.0; `45478` is a later duplicate of `45310`." }, { - "left": "issue:45290", - "right": "issue:45381", + "left": "issue:43606", + "right": "issue:44857", "accept": false, - "reason": "Chat-template tool-call handling and Qwen2.5-VL video position IDs are different multimodal code paths." + "reason": "Both are runtime failures, but CPU offload device mismatch and float16 AMP crash in LwDetrImageLoss are different code paths." }, { - "left": "issue:45920", - "right": "issue:46192", + "left": "issue:44448", + "right": "issue:44779", "accept": false, - "reason": "A broad tokenizer regression is not the same as `return_tensors` being ignored when `text_kwargs` is passed." + "reason": "Both mention v5 regressions, but one is Pegasus output drift and the other is Deepseek tokenizer correctness; not the same underlying bug." }, { - "left": "issue:44448", - "right": "issue:44977", + "left": "issue:44617", + "right": "issue:46697", "accept": false, - "reason": "Pegasus v4/v5 output drift and Qwen3.5 flash-attention generation failures are unrelated model-specific issues." + "reason": "CUDA OOM on Sam3Video is unrelated to Swin2SR over-padding." }, { - "left": "issue:27957", - "right": "issue:34689", + "left": "issue:39692", + "right": "issue:44493", "accept": false, - "reason": "Flash Attention 2 support for XLM-Roberta and Llama 3.2 Vision loading failures are different problems." + "reason": "Doc-example issues and position-id key warnings are not the same problem." }, { - "left": "issue:40444", - "right": "issue:43526", + "left": "issue:28218", + "right": "issue:32090", "accept": false, - "reason": "Multimodal finetuning with iterable datasets is unrelated to a BeitImageProcessorFast label-reduction bug." + "reason": "Tokenizer spacing after added tokens is unrelated to Trainer broadcast type errors." }, { - "left": "issue:43278", - "right": "issue:43329", + "left": "issue:44206", + "right": "issue:45823", "accept": false, - "reason": "Training/eval dtype drift is unrelated to undefined video helper names in `_get_num_multimodal_tokens`." + "reason": "Feature-extractor argument regression and Gemma4 PLE device mismatch are different failures." }, { - "left": "issue:44977", - "right": "issue:45229", + "left": "issue:44360", + "right": "issue:44561", "accept": false, - "reason": "Flash-attention generation failure and multi-GPU CUDA OOM are different symptoms with different likely fixes." + "reason": "DSA ReLU discussion and `is_torch_fx_available` removal breakage are unrelated." }, { - "left": "issue:45290", - "right": "issue:46655", + "left": "issue:44857", + "right": "issue:45072", "accept": false, - "reason": "Chat-template tool-call crash is unrelated to SmolVLM video padding calculation." + "reason": "Different models and different dtype/device mismatch manifestations; not the same bug." }, { - "left": "issue:44855", - "right": "issue:46399", + "left": "issue:43701", + "right": "issue:45083", "accept": false, - "reason": "Python 3.13 IndentationError from scripted model import is not the same as a missing decorator registration error." + "reason": "Checkpoint key mismatch and qwen3_omni_moe output-length helper behavior are unrelated." }, { - "left": "issue:10000", - "right": "issue:15354", + "left": "issue:32090", + "right": "issue:33357", "accept": false, - "reason": "These are unrelated legacy issues: German DistilBert model error vs GeneratorExp unsupported by `torch.jit.script`." + "reason": "Trainer GPU broadcast NoneType error is unrelated to the MacOS CLIP bus error." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet: issue:45920 is a broad tokenizer-regression report, and the only clearly duplicate-looking pair is the two identical Qwen3.5 MoE `from_pretrained` issues (45310/45478). The remaining soft-edge rejections are conservative and consistent with the titles/body signals provided." + "feedback": "The summary is grounded in the packet: the only clear duplicate pair is issue:45310 / issue:45478, which have identical titles and the earlier report is correctly chosen as the canonical anchor. The soft-edge rejections are conservative and do not overstate equivalence." }, "evaluator_used": true, "retried": false, @@ -9364,68 +9189,82 @@ "component_index": 1, "component_count": 22, "review_unit_index": 30, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ + "issue:10000", + "issue:15354", "issue:25251", - "issue:29127", + "issue:27957", + "issue:29942", "issue:32090", "issue:33290", - "issue:33357", - "issue:34567", + "issue:34689", + "issue:35141", "issue:37428", - "issue:39692", + "issue:38175", + "issue:40444", "issue:41720", "issue:42175", "issue:42490", - "issue:42915", - "issue:42994", "issue:43278", + "issue:43329", "issue:43381", - "issue:43412", + "issue:43526", + "issue:43550", "issue:43575", - "issue:43611", - "issue:43653", - "issue:43701", "issue:43746", "issue:43873", "issue:43927", "issue:43976", - "issue:44186", - "issue:44361", "issue:44448", "issue:44466", + "issue:44493", "issue:44561", "issue:44568", - "issue:44743", "issue:44779", + "issue:44855", "issue:44857", - "issue:44871", "issue:44977", - "issue:45198", + "issue:45083", "issue:45229", - "issue:45372", - "issue:45406", + "issue:45290", + "issue:45381", "issue:45440", "issue:45491", - "issue:45563", "issue:45588", "issue:45698", - "issue:45803", "issue:45823", - "issue:46121", - "issue:46209" + "issue:45920", + "issue:46192", + "issue:46209", + "issue:46399", + "issue:46655" ], "soft_pairs": [ - "issue:43927|issue:45698", + "issue:42490|issue:43575", + "issue:38175|issue:44493", + "issue:43550|issue:45083", + "issue:29942|issue:35141", + "issue:45920|issue:46192", + "issue:27957|issue:34689", + "issue:40444|issue:43526", + "issue:43278|issue:43329", + "issue:44448|issue:44977", + "issue:45290|issue:45381", + "issue:44855|issue:46399", + "issue:44977|issue:45229", + "issue:45290|issue:46655", "issue:43278|issue:43381", - "issue:43746|issue:46209", + "issue:10000|issue:15354", + "issue:43927|issue:45698", "issue:44857|issue:45588", + "issue:43746|issue:46209", "issue:44561|issue:44779", "issue:37428|issue:42175", "issue:32090|issue:33290", @@ -9433,234 +9272,220 @@ "issue:25251|issue:32090", "issue:43976|issue:45440", "issue:44448|issue:44466", - "issue:41720|issue:45229", - "issue:44568|issue:45491", "issue:42490|issue:43873", + "issue:44568|issue:45491", + "issue:41720|issue:45229", "issue:45588|issue:45823", - "issue:44977|issue:45563", - "issue:45372|issue:45588", - "issue:43278|issue:43575", - "issue:33357|issue:39692", - "issue:45198|issue:45803", - "issue:29127|issue:34567", - "issue:44871|issue:44977", - "issue:43381|issue:43701", - "issue:43653|issue:44743", - "issue:43412|issue:45588", - "issue:44186|issue:45198", - "issue:43611|issue:46121", - "issue:44361|issue:45198", - "issue:45372|issue:45406", - "issue:42915|issue:42994" + "issue:43278|issue:43575" ], - "prepared_review_unit_hash": "05039e81b6c4dc1ebab974092e803c2fad35a008511bd0ed159d6c24ceba2e8c", + "prepared_review_unit_hash": "a33f8ca64dca7977dce823b11db2fe673f8ade2077a2a7151d767eea3347f408", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36809, - "estimated_input_tokens": 9203, - "estimated_eval_tokens": 18662 + "serialized_chars": 36906, + "estimated_input_tokens": 9227, + "estimated_eval_tokens": 18710 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "A heterogeneous cluster of unrelated issues spanning tokenizers, loading, quantization, training, attention, and docs. None of the soft pairs look like the same underlying bug/change.", - "confidence": 0.94, - "canonical_issue_reason": "No true canonical duplicate target stands out; issue 43873 is only the broadest active report with the most discussion/inbound references, so it is the least-bad placeholder representative.", + "summary": "This cluster is a loose thematic similarity set, not a true duplicate group: it mixes unrelated bugs across tokenizers, flash-attention, multimodal processors, loading/saving, device mapping, and OOMs. None of the soft pairs look like the same underlying bug/change, so all are rejected.", + "confidence": 0.89, + "canonical_issue_reason": "issue:45698 is the broadest open load/save regression in the set and has the highest discussion/inbound-reference signal, so it is the best anchor for triage.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43873 is the best single representative by activity and breadth, but it should not be treated as a duplicate anchor for the rest of the cluster.", + "best_issue_reason": "issue:45920 is the most general concrete v5 regression report, affecting multiple model families, so it is the most actionable representative issue.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43927", - "right": "issue:45698", + "left": "issue:42490", + "right": "issue:43575", "accept": false, - "reason": "Both involve save/load behavior, but one is about preserving custom token IDs while the other is about loading the wrong custom module after save_pretrained; different bugs." + "reason": "One is a path-trailing-slash import warning; the other is TP-related OOM on a specific model." }, { - "left": "issue:43278", - "right": "issue:43381", + "left": "issue:38175", + "right": "issue:44493", "accept": false, - "reason": "Embedding dtype drift in eval and gradient-checkpointing disallowed in eval are unrelated failure modes." + "reason": "Different failures: zero probabilities in SigLIP2 vs unexpected position-id keys." }, { - "left": "issue:43746", - "right": "issue:46209", + "left": "issue:43550", + "right": "issue:45083", "accept": false, - "reason": "PEFT local checkpoint loading and FP8 inference on B200 are different model/runtime paths with no shared concrete bug." + "reason": "Unrelated model/runtime bugs: torch.compile SDPA on Bamba vs feature-length helper logic in qwen3_omni_moe." }, { - "left": "issue:44857", - "right": "issue:45588", + "left": "issue:29942", + "right": "issue:35141", "accept": false, - "reason": "LwDetrImageLoss AMP crash and flash_attention s_aux=None crash affect different code paths and symptoms." + "reason": "Flash Attention 2 test failure is unrelated to embedding reinitialization during post-init." }, { - "left": "issue:44561", - "right": "issue:44779", + "left": "issue:45920", + "right": "issue:46192", "accept": false, - "reason": "Removal of is_torch_fx_available breaks trust_remote_code models, while the other is a DeepSeek tokenizer regression; not the same issue." + "reason": "Both touch tokenizer APIs, but one is a wrong-token-ID regression and the other is return_tensors being ignored." }, { - "left": "issue:37428", - "right": "issue:42175", + "left": "issue:27957", + "right": "issue:34689", "accept": false, - "reason": "ImportError for flash attention utils and missing tensorflow in the torch extra are unrelated packaging/import problems." + "reason": "Different subsystems and symptoms: Flash Attention 2 support vs a model-loading breakage." }, { - "left": "issue:32090", - "right": "issue:33290", + "left": "issue:40444", + "right": "issue:43526", "accept": false, - "reason": "Trainer NoneType broadcast and Adafactor OOM in deepspeed are distinct runtime problems." + "reason": "Different areas: multimodal finetuning with multiple images vs BeitImageProcessorFast label reduction." }, { - "left": "issue:43873", - "right": "issue:45229", + "left": "issue:43278", + "right": "issue:43329", "accept": false, - "reason": "Both mention memory/quantization, but one is offloading with quantization and the other is Gemma4 multi-GPU OOM; different underlying bugs." + "reason": "Embedding dtype drift is unrelated to undefined variables in a video token-count helper." }, { - "left": "issue:25251", - "right": "issue:32090", + "left": "issue:44448", + "right": "issue:44977", "accept": false, - "reason": "Pipeline top_k nesting behavior and Trainer _gpu_broadcast_one NoneType are unrelated." + "reason": "Pegasus v4/v5 output mismatch is not the same as Qwen3.5 flash-attention generation failure." }, { - "left": "issue:43976", - "right": "issue:45440", + "left": "issue:45290", + "right": "issue:45381", "accept": false, - "reason": "Python version compatibility and DeepSeekV3MoE divergence from remote are unrelated issues." + "reason": "Chat-template crash with tool-call messages is different from qwen2.5-vl video position-id handling." }, { - "left": "issue:44448", - "right": "issue:44466", + "left": "issue:44855", + "right": "issue:46399", "accept": false, - "reason": "Both are v4/v5 regressions, but one changes Pegasus generation output while the other is device-dependent lm_head.weight serialization." + "reason": "Import parsing/torch.jit comment issue is unrelated to a missing decorator around apply_rotary_pos_emb registration." }, { - "left": "issue:41720", + "left": "issue:44977", "right": "issue:45229", "accept": false, - "reason": "Qwen3 auto device-mapping cuda assert and Gemma4 multi-GPU OOM are different models and different failure modes." + "reason": "Flash-attention generation bug and multi-GPU CUDA OOM are different concrete problems." }, { - "left": "issue:44568", - "right": "issue:45491", + "left": "issue:45290", + "right": "issue:46655", "accept": false, - "reason": "Tokenizer BOS/EOS insertion bug and Gemma3 NaN embeddings with sliding-window attention are unrelated." + "reason": "Assistant tool-call chat templating crash is unrelated to SmolVLM video padding math." }, { - "left": "issue:42490", - "right": "issue:43873", + "left": "issue:43278", + "right": "issue:43381", "accept": false, - "reason": "Trailing-slash import warning and quantization offloading are not the same bug." + "reason": "Training/eval dtype mismatch is not the same bug as gradient checkpointing in eval mode." }, { - "left": "issue:45588", - "right": "issue:45823", + "left": "issue:10000", + "right": "issue:15354", "accept": false, - "reason": "flash_attention s_aux=None crash and Gemma4 PLE device mismatch are separate runtime problems." + "reason": "German DistilBertModel issue and GeneratorExp torch.jit.script/export issue are unrelated." }, { - "left": "issue:44977", - "right": "issue:45563", + "left": "issue:43927", + "right": "issue:45698", "accept": false, - "reason": "One is a Qwen3.5 flash-attention generation failure; the other is a stale warning in paged generate(). Different code paths." + "reason": "Both involve save/load, but one loses custom token IDs while the other loads the wrong custom module." }, { - "left": "issue:45372", + "left": "issue:44857", "right": "issue:45588", "accept": false, - "reason": "Gemma4 processor import failure from mistral_common and flash_attention s_aux=None crash are unrelated." + "reason": "Float16 AMP crash in LwDetrImageLoss is unrelated to flash_attention.py handling s_aux=None." }, { - "left": "issue:43278", - "right": "issue:43575", + "left": "issue:43746", + "right": "issue:46209", "accept": false, - "reason": "Embedding dtype switching and Qwen2-57B-A14B-Instruct TP OOM are unrelated." + "reason": "PEFT checkpoint loading and FP8 inference on B200 are different code paths and failure modes." }, { - "left": "issue:33357", - "right": "issue:39692", + "left": "issue:44561", + "right": "issue:44779", "accept": false, - "reason": "MacOS bus error with a CLIP model and SigLIP2 docs/example issues are not the same underlying bug." + "reason": "Removal of is_torch_fx_available breaks remote-code models; DeepSeek tokenizer regression is a separate tokenizer bug." }, { - "left": "issue:45198", - "right": "issue:45803", + "left": "issue:37428", + "right": "issue:42175", "accept": false, - "reason": "Wav2Vec2 save/tokenization failures and FuyuBatchFeature's bare except swallowing errors are unrelated." + "reason": "Missing flash_attention utility symbol and TensorFlow missing from a torch install are unrelated packaging/runtime issues." }, { - "left": "issue:29127", - "right": "issue:34567", + "left": "issue:32090", + "right": "issue:33290", "accept": false, - "reason": "LayoutLMv3 error-message clarity and TrainerState token-count tracking are unrelated." + "reason": "NoneType broadcast error in Trainer is unrelated to AdaFactor OOM in DeepSpeed." }, { - "left": "issue:44871", - "right": "issue:44977", + "left": "issue:43873", + "right": "issue:45229", "accept": false, - "reason": "Gemma3 eos_token_id mismatch and Qwen3.5 flash-attention generation bug are different model-specific issues." + "reason": "Quantization offloading misbehavior and Gemma4 multi-GPU OOM are different memory-management issues." }, { - "left": "issue:43381", - "right": "issue:43701", + "left": "issue:25251", + "right": "issue:32090", "accept": false, - "reason": "Gradient checkpointing in eval mode and resume_from_checkpoint key mismatch are unrelated training-state bugs." + "reason": "Pipeline top_k nesting behavior is unrelated to the Trainer _gpu_broadcast_one NoneType error." }, { - "left": "issue:43653", - "right": "issue:44743", + "left": "issue:43976", + "right": "issue:45440", "accept": false, - "reason": "BigBirdTokenizer special-token decode failure and Qwen3.5 recurrent-state reset with cache are unrelated." + "reason": "Python-version compatibility breakage is unrelated to DeepSeekV3MoE divergence from remote implementation." }, { - "left": "issue:43412", - "right": "issue:45588", + "left": "issue:44448", + "right": "issue:44466", "accept": false, - "reason": "Gemma3n executorch export failure and flash_attention s_aux=None crash are different subsystems and symptoms." + "reason": "Different concrete bugs: model output regression vs inconsistent lm_head.weight serialization." }, { - "left": "issue:44186", - "right": "issue:45198", + "left": "issue:42490", + "right": "issue:43873", "accept": false, - "reason": "LayoutLMv2Tokenizer NER/padding crash and Wav2Vec2 save/tokenization failure are not the same bug." + "reason": "Path-trailing-slash import warning does not match quantization offloading behavior." }, { - "left": "issue:43611", - "right": "issue:46121", + "left": "issue:44568", + "right": "issue:45491", "accept": false, - "reason": "base_model_prefix loading regression and rope-param list validation TypeError are unrelated." + "reason": "Tokenizer add_special_tokens regression is unrelated to Gemma3 NaN embeddings from sliding-window attention." }, { - "left": "issue:44361", - "right": "issue:45198", + "left": "issue:41720", + "right": "issue:45229", "accept": false, - "reason": "MLukeTokenizer AttributeError and Wav2Vec2 save/tokenization failure are different tokenizer/model bugs." + "reason": "Qwen3 cudaErrorAssert with auto device mapping is not the same as Gemma4 OOM." }, { - "left": "issue:45372", - "right": "issue:45406", + "left": "issue:45588", + "right": "issue:45823", "accept": false, - "reason": "Both involve Gemma4 processors, but one is an import/dependency problem and the other is a missing _tokenizer attribute in serve; not the same concrete code path." + "reason": "Flash-attention sinkless-model AttributeError and PLE device mismatch are different forward-path bugs." }, { - "left": "issue:42915", - "right": "issue:42994", + "left": "issue:43278", + "right": "issue:43575", "accept": false, - "reason": "FineGrainedFP8Config failure for Qwen3Moe and quantized model saving failure are different quantization issues." + "reason": "Embedding dtype change in eval and TP-related OOM on Qwen2-57B are unrelated." } ] }, @@ -9676,77 +9501,91 @@ "component_index": 1, "component_count": 22, "review_unit_index": 31, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:27957", "issue:29127", "issue:29942", "issue:30064", "issue:33357", + "issue:34567", "issue:35141", + "issue:39692", "issue:42490", "issue:42915", + "issue:42994", "issue:43065", "issue:43066", - "issue:43334", "issue:43381", + "issue:43412", "issue:43421", - "issue:43526", - "issue:43531", "issue:43575", + "issue:43611", "issue:43653", "issue:43701", "issue:43756", - "issue:43824", - "issue:43931", "issue:43994", "issue:44186", "issue:44206", "issue:44265", - "issue:44279", - "issue:44387", + "issue:44361", "issue:44442", "issue:44462", "issue:44479", - "issue:45072", + "issue:44743", + "issue:44871", + "issue:44977", "issue:45198", "issue:45245", - "issue:45250", "issue:45372", + "issue:45406", "issue:45412", "issue:45440", "issue:45459", "issue:45491", "issue:45563", + "issue:45588", "issue:45593", "issue:45712", - "issue:45834", + "issue:45803", "issue:46121", "issue:46369", - "issue:46399", - "issue:46640", "issue:46693" ], "soft_pairs": [ + "issue:45372|issue:45588", + "issue:33357|issue:39692", + "issue:45198|issue:45803", + "issue:44977|issue:45563", + "issue:29127|issue:34567", + "issue:44871|issue:44977", + "issue:43381|issue:43701", + "issue:43412|issue:45588", + "issue:43653|issue:44743", + "issue:44186|issue:45198", + "issue:43611|issue:46121", + "issue:44361|issue:45198", + "issue:45372|issue:45406", + "issue:42915|issue:42994", "issue:45459|issue:45712", "issue:29942|issue:30064", "issue:44206|issue:44479", "issue:43756|issue:45440", - "issue:43065|issue:43994", - "issue:43065|issue:45412", "issue:43653|issue:44442", + "issue:43065|issue:45412", + "issue:43065|issue:43994", "issue:45491|issue:46693", "issue:45372|issue:45459", "issue:45245|issue:45563", - "issue:43066|issue:44462", - "issue:44265|issue:45198", "issue:29127|issue:33357", + "issue:44265|issue:45198", + "issue:43066|issue:44462", "issue:27957|issue:29942", "issue:45412|issue:46369", "issue:43994|issue:45412", @@ -9754,246 +9593,246 @@ "issue:43575|issue:43701", "issue:43421|issue:45491", "issue:29127|issue:35141", - "issue:42490|issue:43381", - "issue:43575|issue:44387", - "issue:43526|issue:45834", - "issue:44442|issue:45072", - "issue:43526|issue:46693", - "issue:42915|issue:43931", - "issue:43824|issue:44279", - "issue:43994|issue:46369", - "issue:43526|issue:44186", - "issue:46121|issue:46399", - "issue:45250|issue:46640", - "issue:27957|issue:29127", - "issue:43334|issue:43531" + "issue:42490|issue:43381" ], - "prepared_review_unit_hash": "e5cbd5a3cd2786f584ef2029dd73aabdcfad6753c7df7052d2a54cf92afb7350", + "prepared_review_unit_hash": "20049dd8cb09397b1b5c46622156aa0863f0b93a0b6cf78b54d47cf6afae977a", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 33, - "serialized_chars": 37438, - "estimated_input_tokens": 9360, - "estimated_eval_tokens": 18976 + "soft_pair_count": 35, + "serialized_chars": 38215, + "estimated_input_tokens": 9554, + "estimated_eval_tokens": 19364 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The items are a broad mix of unrelated issues; the soft-similarity links are mostly superficial title overlap rather than true duplicates. No PRs are present. If forced to pick a representative issue, the RT-DETR memory-leak report is the clearest concrete bug.", - "confidence": 0.21, - "canonical_issue_reason": "issue:45412 is the most concrete, well-scoped bug report in the set, with reproducible behavior (models not releasing memory) and follow-up discussion/inbound references. There is no strong evidence of a single thematic cluster, so this is only a weak canonical choice.", + "summary": "No soft duplicates look safe to merge: the pairs mostly share broad themes (Flash Attention, tokenizers, Gemma/Qwen, memory, export) but describe different concrete failures and code paths.", + "confidence": 0.91, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:45412 is the best standalone issue because it describes a specific failure mode with clear user impact and likely actionable debugging value.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:45459", - "right": "issue:45712", + "left": "issue:45372", + "right": "issue:45588", "accept": false, - "reason": "Tokenizer error handling with missing protobuf is unrelated to leftover dummy classes and repo hygiene." + "reason": "Both involve Gemma/processor loading, but one is an import error from mistral_common and the other is a flash_attention.py AttributeError on s_aux=None; different failure modes and paths." }, { - "left": "issue:29942", - "right": "issue:30064", + "left": "issue:33357", + "right": "issue:39692", "accept": false, - "reason": "Flash Attention 2 test failures are not the same bug as void segmentation map processing in an image processor." + "reason": "Both mention vision/CLIP/SigLIP2, but one is a MacOS bus error in model loading and the other is a documentation example with model/processor mismatch and quantization issues." }, { - "left": "issue:44206", - "right": "issue:44479", + "left": "issue:45198", + "right": "issue:45803", "accept": false, - "reason": "A feature extractor crash from an unsupported argument is different from a Qwen video-input regression." + "reason": "One is save_pretrained/tokenization failure for a specific Wav2Vec2 model; the other is a broad Fuyu conversion bug caused by a bare except swallowing errors." }, { - "left": "issue:43756", - "right": "issue:45440", + "left": "issue:44977", + "right": "issue:45563", "accept": false, - "reason": "Smollm3 RoPE-layer mismatch and DeepseekV3 implementation divergence are separate model-specific correctness issues." + "reason": "Both touch generation/flash-attention, but one is a Qwen3.5 correctness regression and the other is a stale warning for paged generate(); not the same bug." }, { - "left": "issue:43065", - "right": "issue:43994", + "left": "issue:29127", + "right": "issue:34567", "accept": false, - "reason": "A dummy Conv2d in Sam3PixelDecoder is unrelated to SigLIP2 producing bad outputs through AutoModel/pipeline." + "reason": "Unrelated: LayoutLMv3 error-message clarity versus TrainerState token counter not updating." }, { - "left": "issue:43065", - "right": "issue:45412", + "left": "issue:44871", + "right": "issue:44977", "accept": false, - "reason": "Dummy layer scaffolding in Sam3PixelDecoder is not the same as RT-DETR memory not being released." + "reason": "Gemma eos_token_id configuration mismatch is unrelated to Qwen3.5 flash-attention generation failure." }, { - "left": "issue:43653", - "right": "issue:44442", + "left": "issue:43381", + "right": "issue:43701", "accept": false, - "reason": "BigBird tokenizer special-token registration and AutoTokenizer loading of FastSpeech2ConformerTokenizer are different tokenizer bugs." + "reason": "Gradient checkpointing in eval mode and resume_from_checkpoint key mismatch are separate trainer/runtime issues." }, { - "left": "issue:45491", - "right": "issue:46693", + "left": "issue:43412", + "right": "issue:45588", "accept": false, - "reason": "Gemma3 NaN embeddings from all-padding windows are a correctness bug, while the other issue is a performance regression." + "reason": "Gemma3n export failures versus a flash-attention AttributeError on sink-less models; different models and code paths." }, { - "left": "issue:45372", - "right": "issue:45459", + "left": "issue:43653", + "right": "issue:44743", "accept": false, - "reason": "mistral_common import failure for Gemma 4 processor loading is unrelated to protobuf-dependent tokenizer error masking." + "reason": "BigBirdTokenizer special-token registration/empty decode is unrelated to Qwen recurrent-state reset behavior." }, { - "left": "issue:45245", - "right": "issue:45563", + "left": "issue:44186", + "right": "issue:45198", "accept": false, - "reason": "A hard runtime limit on categories is not the same as a stale warning in paged generate()." + "reason": "Different tokenizers and bugs: LayoutLMv2 NER/padding crash versus Wav2Vec2 save/tokenization failure." }, { - "left": "issue:43066", - "right": "issue:44462", + "left": "issue:43611", + "right": "issue:46121", "accept": false, - "reason": "Wrong tokenizer decoder type in v5 is a different issue from AutoTokenizer ignoring tokenizer.json." + "reason": "Base_model_prefix loading breakage is unrelated to rope validation TypeError on JSON-loaded lists." }, { - "left": "issue:44265", + "left": "issue:44361", "right": "issue:45198", "accept": false, - "reason": "torch.export failure around torch_compilable_check is unrelated to save_pretrained/tokenization failures in Wav2Vec2." + "reason": "MLukeTokenizer task AttributeError and Wav2Vec2 save/tokenization failure are different tokenizer-specific issues." }, { - "left": "issue:29127", - "right": "issue:33357", + "left": "issue:45372", + "right": "issue:45406", "accept": false, - "reason": "A clearer LayoutLMv3 error message is not the same as a MacOS bus error in a community CLIP model." + "reason": "Gemma4 processor loading import failure and Gemma4Processor missing _tokenizer in serve are related only by model name, not the same defect." }, { - "left": "issue:27957", - "right": "issue:29942", + "left": "issue:42915", + "right": "issue:42994", "accept": false, - "reason": "XLMRoberta Flash Attention 2 support and failing FA2 tests are related only at a broad subsystem level, not the same bug." + "reason": "Qwen3Moe FP8 config failure and quantized model saving failure are distinct quantization problems." }, { - "left": "issue:45412", - "right": "issue:46369", + "left": "issue:45459", + "right": "issue:45712", "accept": false, - "reason": "RT-DETR memory not being freed and sam3-litetext memory overflow are distinct memory problems in different model paths." + "reason": "Tokenizer error masking when protobuf is absent is unrelated to leftover dummy classes in dummy_pt_objects.py." }, { - "left": "issue:43994", - "right": "issue:45412", + "left": "issue:29942", + "right": "issue:30064", "accept": false, - "reason": "SigLIP2 nonsense outputs and RT-DETR memory retention do not share the same code path or failure mode." + "reason": "Flash Attention 2 test failures and void segmentation map processing are separate subsystems and bug reports." }, { - "left": "issue:45440", - "right": "issue:45593", + "left": "issue:44206", + "right": "issue:44479", "accept": false, - "reason": "DeepseekV3 implementation divergence is unrelated to D-FINE auxiliary losses when denoising is off." + "reason": "Feature extractor center-arg crash and video input regression affect different APIs and model families." }, { - "left": "issue:43575", - "right": "issue:43701", + "left": "issue:43756", + "right": "issue:45440", "accept": false, - "reason": "Tensor-parallel OOM on loading a model is unrelated to a resume_from_checkpoint key mismatch." + "reason": "Smollm3 RoPE layer selection bug and DeepseekV3MoE divergence from remote implementation are different model-specific logic issues." }, { - "left": "issue:43421", - "right": "issue:45491", + "left": "issue:43653", + "right": "issue:44442", "accept": false, - "reason": "Runtime special-token post-processor updates and Gemma3 NaN embeddings are different tokenizer/model behaviors." + "reason": "BigBird tokenizer decode bug is unrelated to AutoTokenizer loading FastSpeech2ConformerTokenizer." }, { - "left": "issue:29127", - "right": "issue:35141", + "left": "issue:43065", + "right": "issue:45412", "accept": false, - "reason": "LayoutLMv3 box-validation messaging and embedding reinitialization after resize are unrelated bugs." + "reason": "Dummy Sam3PixelDecoder Conv2d and RT-DETR memory leak are different problems." }, { - "left": "issue:42490", - "right": "issue:43381", + "left": "issue:43065", + "right": "issue:43994", "accept": false, - "reason": "A trailing-slash local path import warning is not the same as gradient checkpointing being disallowed in eval mode." + "reason": "Dummy Sam3PixelDecoder Conv2d is unrelated to SigLIP2 nonsensical outputs with AutoModel/pipeline." }, { - "left": "issue:43575", - "right": "issue:44387", + "left": "issue:45491", + "right": "issue:46693", "accept": false, - "reason": "OOM on tensor-parallel load and increased reserved CUDA memory under int4 quantization are different memory issues." + "reason": "Gemma3 NaN embeddings from mixed-length batching differ from a performance regression in long-context generation." }, { - "left": "issue:43526", - "right": "issue:45834", + "left": "issue:45372", + "right": "issue:45459", "accept": false, - "reason": "A label-reduction bug in BeitImageProcessorFast is not the same as a Kosmos2.5 OCR index error." + "reason": "Gemma4 processor import failure and tokenizer protobuf exception handling are unrelated error paths." }, { - "left": "issue:44442", - "right": "issue:45072", + "left": "issue:45245", + "right": "issue:45563", "accept": false, - "reason": "AutoTokenizer loading failure and dtype mismatches in bfloat16 inference are separate problems." + "reason": "Category-cardinality RuntimeError and paged-generate stale warning are not the same defect." }, { - "left": "issue:43526", - "right": "issue:46693", + "left": "issue:29127", + "right": "issue:33357", "accept": false, - "reason": "reduce_labels returning the wrong shape/value is unrelated to long-context Flash Attention 2 performance regression." + "reason": "LayoutLMv3 error-message clarity and MacOS bus error on a CLIP model are unrelated." }, { - "left": "issue:42915", - "right": "issue:43931", + "left": "issue:44265", + "right": "issue:45198", "accept": false, - "reason": "FineGrainedFP8Config failure in Qwen3Moe is not the same as Qwen3-VL weight-shape mismatch." + "reason": "torch.export failure with torch_compilable_check is unrelated to Wav2Vec2 save/tokenization issues." }, { - "left": "issue:43824", - "right": "issue:44279", + "left": "issue:43066", + "right": "issue:44462", "accept": false, - "reason": "Missing Qwen2_5_VLForConditionalGeneration import and a generic dependency issue are not the same concrete bug." + "reason": "Wrong tokenizer decoder type in Transformers v5 and ignoring tokenizer.json are different tokenizer-loading regressions." }, { - "left": "issue:43994", + "left": "issue:27957", + "right": "issue:29942", + "accept": false, + "reason": "XLMRoberta FA2 support and failing FA2 tests are too broad to be the same concrete bug." + }, + { + "left": "issue:45412", "right": "issue:46369", "accept": false, - "reason": "Incorrect SigLIP2 model outputs and sam3-litetext memory overflow do not indicate the same root cause." + "reason": "RT-DETR GC/memory leak and sam3-litetext memory overflow are distinct memory issues in different models." }, { - "left": "issue:43526", - "right": "issue:44186", + "left": "issue:43994", + "right": "issue:45412", "accept": false, - "reason": "BeitImageProcessorFast label handling and LayoutLMv2Tokenizer NER/padding crashes are different tokenizer/image-processing bugs." + "reason": "SigLIP2 nonsensical results and RT-DETR memory release failure are unrelated." }, { - "left": "issue:46121", - "right": "issue:46399", + "left": "issue:45440", + "right": "issue:45593", "accept": false, - "reason": "RoPE validation with JSON-loaded lists and Qwen3.5 kernel registration errors are unrelated." + "reason": "DeepseekV3MoE divergence from remote implementation is unrelated to D-FINE auxiliary-loss behavior." }, { - "left": "issue:45250", - "right": "issue:46640", + "left": "issue:43575", + "right": "issue:43701", "accept": false, - "reason": "These are both Flash Attention 2 titles, but the reports are too generic to treat as the same underlying bug." + "reason": "Qwen2-57B-A14B-Instruct TP OOM and resume_from_checkpoint key mismatch are different bugs." }, { - "left": "issue:27957", - "right": "issue:29127", + "left": "issue:43421", + "right": "issue:45491", "accept": false, - "reason": "XLMRoberta Flash Attention 2 support and LayoutLMv3 box-validation messaging are unrelated." + "reason": "Runtime special-token post-processor updates and Gemma3 NaN embeddings are unrelated tokenizer versus attention/batching issues." }, { - "left": "issue:43334", - "right": "issue:43531", + "left": "issue:29127", + "right": "issue:35141", + "accept": false, + "reason": "LayoutLMv3 error messaging and output embedding reinitialization after resizing are unrelated." + }, + { + "left": "issue:42490", + "right": "issue:43381", "accept": false, - "reason": "Qwen3-VL missing pad_token_id and Qwen3-MoE sliding-window issues are different model-loading/runtime bugs." + "reason": "Local-path trailing-slash import warning and gradient-checkpointing eval-mode failure are different loading/runtime issues." } ] }, @@ -10009,65 +9848,77 @@ "component_index": 1, "component_count": 22, "review_unit_index": 32, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:27957", + "issue:29127", "issue:33290", "issue:34567", "issue:34689", "issue:35141", - "issue:42175", "issue:42490", - "issue:43232", + "issue:42915", "issue:43262", - "issue:43316", + "issue:43334", "issue:43381", - "issue:43388", - "issue:43493", - "issue:43525", + "issue:43526", + "issue:43531", + "issue:43575", "issue:43611", "issue:43646", "issue:43720", - "issue:43723", "issue:43761", - "issue:43827", + "issue:43824", "issue:43873", "issue:43931", - "issue:44062", + "issue:43994", "issue:44077", - "issue:44079", - "issue:44188", + "issue:44186", "issue:44206", + "issue:44279", + "issue:44387", "issue:44442", - "issue:44479", "issue:44514", "issue:44743", "issue:44811", - "issue:44871", "issue:44898", "issue:44977", "issue:45003", - "issue:45216", + "issue:45072", "issue:45245", - "issue:45325", + "issue:45250", "issue:45381", "issue:45440", - "issue:45479", "issue:45588", "issue:45600", + "issue:45834", "issue:45859", - "issue:45907", - "issue:46234", - "issue:46697" + "issue:46121", + "issue:46369", + "issue:46399", + "issue:46640", + "issue:46693" ], "soft_pairs": [ + "issue:43575|issue:44387", + "issue:43526|issue:45834", + "issue:43526|issue:46693", + "issue:44442|issue:45072", + "issue:42915|issue:43931", + "issue:43824|issue:44279", + "issue:43526|issue:44186", + "issue:43994|issue:46369", + "issue:46121|issue:46399", + "issue:45250|issue:46640", + "issue:27957|issue:29127", "issue:27957|issue:33290", + "issue:43334|issue:43531", "issue:43761|issue:45859", "issue:33290|issue:34689", "issue:33290|issue:35141", @@ -10081,224 +9932,212 @@ "issue:44743|issue:44811", "issue:44442|issue:44898", "issue:45003|issue:45440", - "issue:42490|issue:44743", "issue:34689|issue:43611", - "issue:43720|issue:44442", - "issue:42175|issue:46234", - "issue:43723|issue:45479", - "issue:45325|issue:46697", - "issue:43316|issue:43525", - "issue:44077|issue:44479", - "issue:43493|issue:43827", - "issue:44062|issue:45907", - "issue:43761|issue:44079", - "issue:43873|issue:44871", - "issue:43761|issue:44188", - "issue:43232|issue:43388", - "issue:43931|issue:45216" + "issue:42490|issue:44743", + "issue:43720|issue:44442" ], - "prepared_review_unit_hash": "c58e63def289462090fda04f675929e213cf19659ceff7fe4c15d72e710a2446", + "prepared_review_unit_hash": "4679bf83cdfd05bd7e21bfdb2e24dbe873e9cb2f9503c4d7beca595694c745c3", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36553, - "estimated_input_tokens": 9139, - "estimated_eval_tokens": 18534 + "serialized_chars": 36188, + "estimated_input_tokens": 9047, + "estimated_eval_tokens": 18350 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is very heterogeneous: it mixes unrelated bugs across loading, generation, audio/video processors, quantization/offload, and docs. I found no soft pair that clearly looks like the same underlying bug/change; most are only loosely similar by subsystem or model family.", - "confidence": 0.16, - "canonical_issue_reason": "If one issue must represent the set, 43646 is the broadest regression-style report and the most generic user-facing breakage. That said, the cluster is not actually cohesive, so this is a weak canonical choice.", + "summary": "This cluster is highly heterogeneous: most items are separate model-specific bugs or regressions that only share broad themes like OOM, Flash Attention, or loading failures. None of the soft pairs looks like the same concrete bug/change, so they should stay separate.", + "confidence": 0.94, + "canonical_issue_reason": "No single canonical issue stands out; the items span unrelated subsystems and model families, and the apparent overlaps are too broad to treat as one duplicate cluster.", "canonical_pr_reason": null, - "best_issue_reason": "43646 is the best single representative because it is the most general and broadly framed failure among these issues. It is still not a true duplicate of the others.", + "best_issue_reason": "No global best issue for deduping: none of these reports is a clear superset or central tracker for the others.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:27957", - "right": "issue:33290", + "left": "issue:43575", + "right": "issue:44387", "accept": false, - "reason": "Different problems and code paths: Flash Attention support for XLM-Roberta vs OOM with Adafactor/DeepSpeed." + "reason": "Both mention OOM, but one is a Qwen2-57B TP loading issue and the other is an int4 quantization regression in Transformers 5.x; different causes and code paths." }, { - "left": "issue:43761", - "right": "issue:45859", + "left": "issue:43526", + "right": "issue:45834", "accept": false, - "reason": "Both involve model internals, but one is a CLIP forward-output regression and the other is a Qwen3.5 MoE rotary/offload incompatibility." + "reason": "Different models and failures: BEiT image processor label reduction vs Kosmos2.5 long OCR input memory overflow." }, { - "left": "issue:33290", - "right": "issue:34689", + "left": "issue:43526", + "right": "issue:46693", "accept": false, - "reason": "Unrelated failures: optimizer memory use vs model loading regression for Llama 3.2 Vision." + "reason": "Unrelated bugs: BEiT label handling vs a Flash Attention 2 long-context performance regression." }, { - "left": "issue:33290", - "right": "issue:35141", + "left": "issue:44442", + "right": "issue:45072", "accept": false, - "reason": "Different bug classes: optimizer OOM vs embedding resize/post_init reinitialization." + "reason": "Different components and symptoms: tokenizer loading failure for FastSpeech2ConformerTokenizer vs dtype mismatch in SwitchTransformers/TimmWrapperModel bfloat16 inference." }, { - "left": "issue:44514", - "right": "issue:45381", + "left": "issue:42915", + "right": "issue:43931", "accept": false, - "reason": "Both are Qwen2.5-VL related, but one is batched chat-template padding behavior and the other is wrong video position IDs; not the same concrete bug." + "reason": "Different model families and failure modes: Qwen3Moe FP8 config issue vs Qwen3-VL weight shape mismatch on load." }, { - "left": "issue:45588", - "right": "issue:45600", + "left": "issue:43824", + "right": "issue:44279", "accept": false, - "reason": "Different areas entirely: flash_attention sink handling vs stale auto_mappings references in CI." + "reason": "Too vague on the right and no concrete shared bug; an import error for Qwen2.5-VL is not the same as a generic dependency issue." }, { - "left": "issue:33290", - "right": "issue:34567", + "left": "issue:43526", + "right": "issue:44186", "accept": false, - "reason": "Optimizer memory issue vs TrainerState token counter not updating." + "reason": "Different tokenizer/processor paths: BEiT Fast image processor label reduction vs LayoutLMv2 tokenizer crashes on NER padding/truncation." }, { - "left": "issue:43381", - "right": "issue:43873", + "left": "issue:43994", + "right": "issue:46369", "accept": false, - "reason": "Gradient-checkpointing eval-mode behavior is unrelated to quantization offloading." + "reason": "Different subsystems: SigLIP2 inference corruption vs Sam3-LiteText memory overflow." }, { - "left": "issue:44977", - "right": "issue:45245", + "left": "issue:46121", + "right": "issue:46399", "accept": false, - "reason": "Both are user-visible regressions, but one is Qwen3.5 flash-attention generation and the other is a categories cardinality limit error." + "reason": "Distinct rope-related bugs: JSON-loaded validation list TypeError vs missing decorator for apply_rotary_pos_emb registration in Qwen3.5 kernels." }, { - "left": "issue:43262", - "right": "issue:44206", + "left": "issue:45250", + "right": "issue:46640", "accept": false, - "reason": "Both touch audio processing, but apply_chat_template sampling-rate defaulting and a feature extractor center-argument crash are different bugs." + "reason": "Only a broad Flash Attention 2 overlap; one is a generic title, the other is a sink-less model crash in flash_attention.py." }, { - "left": "issue:43646", - "right": "issue:44077", + "left": "issue:27957", + "right": "issue:29127", "accept": false, - "reason": "General custom initialization breakage vs a model-specific post_init policy issue; not the same fix." + "reason": "Completely different problems: Flash Attention 2 support for XLM-Roberta vs a LayoutLMv3 box/clarity error message." }, { - "left": "issue:44743", - "right": "issue:44811", + "left": "issue:27957", + "right": "issue:33290", "accept": false, - "reason": "Qwen3.5 recurrent-state reset bug vs Whisper batch_decode skip_special_tokens handling." + "reason": "Different bug classes: model/attention support vs DeepSpeed Adafactor OOM." }, { - "left": "issue:44442", - "right": "issue:44898", + "left": "issue:43334", + "right": "issue:43531", "accept": false, - "reason": "Tokenizer loading failure and Perceiver image-classification padding/interpolation failure are unrelated." + "reason": "Both are Qwen-related, but one is a pad_token_id load failure and the other is a sliding_window issue in Qwen3-MoE." }, { - "left": "issue:45003", - "right": "issue:45440", + "left": "issue:43761", + "right": "issue:45859", "accept": false, - "reason": "sys.modules access safety issue vs DeepseekV3MoE behavior divergence; different code paths and symptoms." + "reason": "Different model areas and bugs: CLIPVisionModel hidden_states regression vs Qwen3.5 rotary embedding CPU offload incompatibility." }, { - "left": "issue:42490", - "right": "issue:44743", + "left": "issue:33290", + "right": "issue:34689", "accept": false, - "reason": "Local-path slash warning is unrelated to Qwen3.5 cache-state reset." + "reason": "Both are OOM/loading complaints, but one is optimizer memory in DeepSpeed and the other is a specific model-loading regression for Llama 3.2 Vision." }, { - "left": "issue:34689", - "right": "issue:43611", + "left": "issue:33290", + "right": "issue:35141", "accept": false, - "reason": "Both are model-loading regressions, but they affect different mechanisms and versions; too broad to merge." + "reason": "Unrelated: Adafactor OOM in DeepSpeed vs embedding reinitialization after resizing token embeddings." }, { - "left": "issue:43720", - "right": "issue:44442", + "left": "issue:44514", + "right": "issue:45381", "accept": false, - "reason": "BitNet packed-weight loading bug vs tokenizer class loading failure; no shared underlying defect." + "reason": "Both involve Qwen2.5-VL, but one crashes in batched apply_chat_template with padding=False and the other has wrong vision_position_ids for video input." }, { - "left": "issue:42175", - "right": "issue:46234", + "left": "issue:45588", + "right": "issue:45600", "accept": false, - "reason": "Packaging/backend dependency issue vs missing import_utils helper; unrelated." + "reason": "Different files and failures: flash_attention.py AttributeError on s_aux=None vs auto_mappings.py referencing removed Sam3LiteText configs." }, { - "left": "issue:43723", - "right": "issue:45479", + "left": "issue:33290", + "right": "issue:34567", "accept": false, - "reason": "Tokenizer-loading regression and zero-loss classification bug are not the same underlying issue." + "reason": "Different trainer/optimizer issues: DeepSpeed Adafactor OOM vs TrainerState.num_input_tokens_seen not updating." }, { - "left": "issue:45325", - "right": "issue:46697", + "left": "issue:43381", + "right": "issue:43873", "accept": false, - "reason": "Qwen2.5-VL rope index scaling bug vs Swin2SR over-padding bug; different processor/model paths." + "reason": "Different behavior: gradient checkpointing disallowed in eval mode vs offloading not working as expected with quantization." }, { - "left": "issue:43316", - "right": "issue:43525", + "left": "issue:44977", + "right": "issue:45245", "accept": false, - "reason": "Both mention config/API discrepancies, but Gemma3TextConfig inconsistency and missing Llama4Config pad_token_id are distinct schema issues." + "reason": "Unrelated failures: Qwen3.5 flash-attention generation bug vs category-count runtime error." }, { - "left": "issue:44077", - "right": "issue:44479", + "left": "issue:43262", + "right": "issue:44206", "accept": false, - "reason": "Model post_init policy and Qwen video-input regression are unrelated." + "reason": "Different audio preprocessing bugs: apply_chat_template sampling-rate default vs unsupported center argument in LasrFeatureExtractor." }, { - "left": "issue:43493", - "right": "issue:43827", + "left": "issue:43646", + "right": "issue:44077", "accept": false, - "reason": "SigLIP2 implementation mismatch vs documentation still referencing removed pipeline() APIs." + "reason": "Related to initialization only at a broad level, but not the same concrete bug: one breaks custom model init in v5, the other is about patchtsmixer optional post_init handling." }, { - "left": "issue:44062", - "right": "issue:45907", + "left": "issue:44743", + "right": "issue:44811", "accept": false, - "reason": "AddedToken special-flag construction bug vs list_repo_files revision forwarding; unrelated." + "reason": "Different pathways and symptoms: Qwen3_5 recurrent state reset bug vs Whisper batch_decode ignoring skip_special_tokens." }, { - "left": "issue:43761", - "right": "issue:44079", + "left": "issue:44442", + "right": "issue:44898", "accept": false, - "reason": "Both concern outputs, but hidden_states being None and ModelOutput key assignment are different defects with different fixes." + "reason": "Different models and failures: FastSpeech2ConformerTokenizer loading vs Perceiver image classification failure at non-default resolution." }, { - "left": "issue:43873", - "right": "issue:44871", + "left": "issue:45003", + "right": "issue:45440", "accept": false, - "reason": "Quantization offload behavior and Gemma-3 eos_token_id configuration are unrelated." + "reason": "Distinct issues: unsafe sys.modules access in modeling_utils vs divergence of native DeepseekV3MoE from remote implementation." }, { - "left": "issue:43761", - "right": "issue:44188", + "left": "issue:34689", + "right": "issue:43611", "accept": false, - "reason": "CLIP hidden_states regression vs attention-kernel divergence under torch.compile are different code paths." + "reason": "Both are model-loading regressions, but they affect different releases and different root causes: Llama 3.2 Vision vs base_model_prefix handling in Transformers 5.0.0." }, { - "left": "issue:43232", - "right": "issue:43388", + "left": "issue:42490", + "right": "issue:44743", "accept": false, - "reason": "Generation kwargs synchronization bug vs gather_for_metrics label truncation bug." + "reason": "No real overlap: trailing-slash local model import warning vs Qwen3_5 recurrent-state reset bug." }, { - "left": "issue:43931", - "right": "issue:45216", + "left": "issue:43720", + "right": "issue:44442", "accept": false, - "reason": "Both involve Qwen3/VL save-load behavior, but one is weight-shape mismatch on loading and the other is incorrect save_pretrained output; not mergeable as one PR." + "reason": "Different domains: BitNet packed-weight loading in accelerate vs tokenizer loading failure for FastSpeech2ConformerTokenizer." } ] }, @@ -10314,16 +10153,14 @@ "component_index": 1, "component_count": 22, "review_unit_index": 33, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ - "issue:16998", - "issue:22355", "issue:25251", "issue:27957", "issue:28282", @@ -10331,286 +10168,260 @@ "issue:42175", "issue:42617", "issue:42907", - "issue:43295", + "issue:43232", "issue:43316", - "issue:43452", - "issue:43550", + "issue:43388", + "issue:43493", + "issue:43525", "issue:43575", - "issue:43606", "issue:43643", - "issue:43646", "issue:43701", - "issue:43716", + "issue:43723", "issue:43742", "issue:43749", - "issue:43825", + "issue:43761", + "issue:43827", + "issue:43873", "issue:43931", + "issue:44062", "issue:44077", - "issue:44162", + "issue:44079", + "issue:44188", "issue:44190", "issue:44442", "issue:44464", + "issue:44479", "issue:44496", - "issue:44534", - "issue:44792", "issue:44855", "issue:44871", "issue:44913", "issue:44987", "issue:45137", - "issue:45584", + "issue:45216", + "issue:45325", + "issue:45479", "issue:45698", "issue:45759", "issue:45803", - "issue:46050", - "issue:46170", + "issue:45907", "issue:46178", "issue:46234", - "issue:46291", - "issue:46365", "issue:46421", - "issue:46506" + "issue:46506", + "issue:46697" ], "soft_pairs": [ + "issue:42175|issue:46234", + "issue:43723|issue:45479", + "issue:45325|issue:46697", + "issue:43316|issue:43525", + "issue:44077|issue:44479", + "issue:43493|issue:43827", + "issue:44062|issue:45907", + "issue:43761|issue:44079", + "issue:43873|issue:44871", + "issue:43761|issue:44188", + "issue:43232|issue:43388", "issue:43749|issue:43931", + "issue:43931|issue:45216", "issue:43316|issue:45759", - "issue:43643|issue:44913", "issue:44855|issue:45137", - "issue:44987|issue:46178", + "issue:43643|issue:44913", "issue:44442|issue:45803", + "issue:44987|issue:46178", "issue:44190|issue:44496", "issue:42907|issue:45698", "issue:43701|issue:46506", "issue:42617|issue:43575", - "issue:44464|issue:46421", "issue:43575|issue:43742", "issue:25251|issue:28282", "issue:27957|issue:32090", - "issue:43550|issue:43701", - "issue:46234|issue:46291", - "issue:43606|issue:43701", - "issue:43295|issue:43825", - "issue:16998|issue:22355", - "issue:43295|issue:44871", - "issue:44871|issue:45584", - "issue:43316|issue:43716", - "issue:42175|issue:43316", - "issue:44077|issue:44534", - "issue:44871|issue:46050", - "issue:45803|issue:46365", - "issue:43646|issue:44913", - "issue:43452|issue:44496", - "issue:44162|issue:44792", - "issue:45803|issue:46170" + "issue:44464|issue:46421" ], - "prepared_review_unit_hash": "94bb14fd24884230321cbed183e04c11f8ab20fd1f40b125d3aee64d8c7902ad", + "prepared_review_unit_hash": "24c549e40539e693cf552a7c262bcbbac5e79cf8d7315e59e9050d65b8007eaf", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36503, - "estimated_input_tokens": 9126, - "estimated_eval_tokens": 18508 + "soft_pair_count": 26, + "serialized_chars": 35650, + "estimated_input_tokens": 8913, + "estimated_eval_tokens": 18082 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous rather than a duplicate set. The soft pairs mostly share broad themes like loading, config, or runtime regressions, but they point to different code paths and different models/APIs, so none should be merged as duplicates.", - "confidence": 0.96, - "canonical_issue_reason": "issue:45698 is the broadest and most central loader/regression report in the set, with active discussion and multiple inbound references, while most other issues are narrower, model-specific failures.", + "summary": "This cluster is mostly a set of superficially similar but actually different bug reports around loading, configs, generation, and model-specific regressions. I don\u2019t see any soft edge that looks like the same underlying fix.", + "confidence": 0.23, + "canonical_issue_reason": "issue:45698 is the broadest and most actively discussed loading/saving report in the set, so it\u2019s the best placeholder canonical even though the cluster is heterogeneous.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45698 is the strongest representative issue overall because it is open, actively discussed, and describes a general from_pretrained/save_pretrained custom-module loading failure.", + "best_issue_reason": "issue:45698 has the widest scope and strongest discussion signal among these issues, but it should be treated as a representative, not a true duplicate hub.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43749", - "right": "issue:43931", + "left": "issue:42175", + "right": "issue:46234", "accept": false, - "reason": "Different bugs: FSDP CPU RAM-efficient loading vs Qwen3-VL weight-shape mismatch on load." + "reason": "One is a backend/installation packaging problem; the other is a missing import-utils availability helper. Same general area, different bug." }, { - "left": "issue:43316", - "right": "issue:45759", + "left": "issue:43723", + "right": "issue:45479", "accept": false, - "reason": "Both involve config APIs, but one is a Gemma3TextConfig discrepancy and the other is AutoModelForCausalLM.from_config unwrapping text_config for Qwen multimodal configs." + "reason": "Tokenizer loading in v5 and a sequence-classification loss bug are unrelated code paths." }, { - "left": "issue:43643", - "right": "issue:44913", + "left": "issue:45325", + "right": "issue:46697", "accept": false, - "reason": "Both are config-related reload issues, but the symptoms and fields differ: missing fields from trust_remote_code vs rotary_pct resetting on GPTNeoX reload." + "reason": "Qwen2.5-VL rope/temporal position indexing is not the same bug as Swin2SR image over-padding." }, { - "left": "issue:44855", - "right": "issue:45137", + "left": "issue:43316", + "right": "issue:43525", "accept": false, - "reason": "Unrelated failures: DebertaV2 import/JIT parsing on Python 3.13 vs DeepSpeed ZeRO3 deque underflow." + "reason": "Both mention config API inconsistencies, but they affect different models and different missing fields." }, { - "left": "issue:44987", - "right": "issue:46178", + "left": "issue:44077", + "right": "issue:44479", "accept": false, - "reason": "Different failures in different subsystems: a model-loading regression vs a backend-mapping error for tensorflow_text." + "reason": "PatchTSMixer post-init policy and Qwen video-input regression are unrelated model-specific issues." }, { - "left": "issue:44442", - "right": "issue:45803", + "left": "issue:43493", + "right": "issue:43827", "accept": false, - "reason": "Tokenizer loading failure and a bare-except bug in FuyuBatchFeature are not the same underlying issue." + "reason": "A model implementation discrepancy and docs still referencing pipeline() are different kinds of problems." }, { - "left": "issue:44190", - "right": "issue:44496", + "left": "issue:44062", + "right": "issue:45907", "accept": false, - "reason": "Local dataset loading in an example script is unrelated to an unrecognized model/config loading error." + "reason": "Tokenizer AddedToken duplication and list_repo_files revision forwarding are unrelated API bugs." }, { - "left": "issue:42907", - "right": "issue:45698", + "left": "issue:43761", + "right": "issue:44079", "accept": false, - "reason": "Both touch save/load, but one is quantization-specific dequantized save failure and the other is wrong custom module selection after save_pretrained/from_pretrained." + "reason": "Hidden-states missing from a model forward pass is not the same as ModelOutput key assignment when a value was previously None." }, { - "left": "issue:43701", - "right": "issue:46506", + "left": "issue:43873", + "right": "issue:44871", "accept": false, - "reason": "Checkpoint resume key mismatch and CPU/CUDA output differences are different problems." + "reason": "Quantization/offloading behavior and Gemma eos_token_id configuration are separate issues." }, { - "left": "issue:42617", - "right": "issue:43575", + "left": "issue:43761", + "right": "issue:44188", "accept": false, - "reason": "3d_parallel.py execution failure and Qwen2-57B tensor-parallel OOM are distinct runtime problems." + "reason": "CLIPVision hidden_states regression and torch.compile attention-kernel branching are different failures." }, { - "left": "issue:44464", - "right": "issue:46421", + "left": "issue:43232", + "right": "issue:43388", "accept": false, - "reason": "Both involve generation/compilation performance, but one is inconsistent chunked generation outputs and the other is spurious recompilation from cache lazy initialization." + "reason": "Generation kwargs update after sync_gpus and metric gathering on last-batch labels are different training/inference bugs." }, { - "left": "issue:43575", - "right": "issue:43742", + "left": "issue:43749", + "right": "issue:43931", "accept": false, - "reason": "Different model-specific loading failures: tensor-parallel OOM vs key error on MobileLLM-125M." + "reason": "FSDP CPU RAM-efficient loading breakage is not the same as a specific Qwen3-VL weight-shape mismatch." }, { - "left": "issue:25251", - "right": "issue:28282", + "left": "issue:43931", + "right": "issue:45216", "accept": false, - "reason": "Pipeline top_k output nesting and missing transformers.onnx are unrelated." + "reason": "Model load shape mismatch and save_pretrained checkpoint corruption are different failure modes." }, { - "left": "issue:27957", - "right": "issue:32090", + "left": "issue:43316", + "right": "issue:45759", "accept": false, - "reason": "FlashAttention2 support for XLMRoberta and a Trainer _gpu_broadcast_one NoneType error are different code paths." + "reason": "Gemma3 text-config API mismatch and Qwen multimodal from_config text_config unwrapping are not the same underlying bug." }, { - "left": "issue:43550", - "right": "issue:43701", + "left": "issue:44855", + "right": "issue:45137", "accept": false, - "reason": "torch.compile/SDPA failure on Bamba-9B-v2 is unrelated to resume_from_checkpoint key mismatch." + "reason": "Python 3.13 torch.jit parsing/indentation issue and DeepSpeed ZeRO3 deque underflow are unrelated." }, { - "left": "issue:46234", - "right": "issue:46291", + "left": "issue:43643", + "right": "issue:44913", "accept": false, - "reason": "Missing utility function in import_utils and kernels v0.15 dependency/version requirements are different issues." + "reason": "Missing fields from trust_remote_code configs and GPTNeoX rotary_pct not persisting on reload are different config-loading bugs." }, { - "left": "issue:43606", - "right": "issue:43701", + "left": "issue:44442", + "right": "issue:45803", "accept": false, - "reason": "CPU offload device mismatch for bark-small is not the same as checkpoint resume key mismatches." + "reason": "AutoTokenizer failing on one tokenizer class and a bare-except swallowing errors in Fuyu conversion are unrelated." }, { - "left": "issue:43295", - "right": "issue:43825", - "accept": false, - "reason": "One is a regression in custom model code accessing processor.tokenizer; the other is an incorrect pipeline error message about translation support." - }, - { - "left": "issue:16998", - "right": "issue:22355", - "accept": false, - "reason": "Model max length handling for DeBERTa-V3 is unrelated to the transformers.onnx import error." - }, - { - "left": "issue:43295", - "right": "issue:44871", - "accept": false, - "reason": "Processor/tokenizer access regression and Gemma eos_token_id inconsistency are different multimodal/config issues." - }, - { - "left": "issue:44871", - "right": "issue:45584", - "accept": false, - "reason": "Gemma eos_token_id mismatch and Whisper empty-transcription failure after align_special_tokens are separate bugs." - }, - { - "left": "issue:43316", - "right": "issue:43716", + "left": "issue:44987", + "right": "issue:46178", "accept": false, - "reason": "Gemma3TextConfig API discrepancy and Mistral-3 image preprocessor dtype mismatch affect different models and paths." + "reason": "Loading a specific model repo and missing BACKENDS_MAPPING support for tensorflow_text are different backend-resolution issues." }, { - "left": "issue:42175", - "right": "issue:43316", + "left": "issue:44190", + "right": "issue:44496", "accept": false, - "reason": "Packaging/backends dependency omission for pip install '.[torch]' is unrelated to the Gemma3TextConfig API discrepancy." + "reason": "Local dataset loading in an example script and an unrecognized model/config error are not the same problem." }, { - "left": "issue:44077", - "right": "issue:44534", + "left": "issue:42907", + "right": "issue:45698", "accept": false, - "reason": "patchtsmixer post_init policy and junk-filled non-persistent buffers are different issues." + "reason": "Saving dequantized models and loading the wrong custom module after save_pretrained both touch serialization, but they are different code paths and failures." }, { - "left": "issue:44871", - "right": "issue:46050", + "left": "issue:43701", + "right": "issue:46506", "accept": false, - "reason": "Gemma eos_token_id configuration mismatch is not the same as sinq quantized save/load breakage." + "reason": "Checkpoint resume key mismatch and CPU/CUDA divergence in PPDocLayoutV3 are unrelated." }, { - "left": "issue:45803", - "right": "issue:46365", + "left": "issue:42617", + "right": "issue:43575", "accept": false, - "reason": "Bare except swallowing errors in FuyuBatchFeature and Trainer import failure caused by datasets are unrelated." + "reason": "3d_parallel.py startup failure and TP-induced OOM when loading Qwen2-57B are different parallelism issues." }, { - "left": "issue:43646", - "right": "issue:44913", + "left": "issue:43575", + "right": "issue:43742", "accept": false, - "reason": "Transformers 5 custom model initialization breakage and GPTNeoX rotary_pct reload behavior are different regressions." + "reason": "Qwen2-57B tensor-parallel OOM and MobileLLM loading key error are unrelated loading failures." }, { - "left": "issue:43452", - "right": "issue:44496", + "left": "issue:25251", + "right": "issue:28282", "accept": false, - "reason": "gguf_file breaking from_pretrained loading and an unrecognized model/config error are both loading-related but not the same bug." + "reason": "Pipeline top_k output-shape behavior and missing PyTorch installation import error are completely different." }, { - "left": "issue:44162", - "right": "issue:44792", + "left": "issue:27957", + "right": "issue:32090", "accept": false, - "reason": "ESM2 breakage and a janus image-generation test failure are different model-specific problems." + "reason": "Flash Attention support for XLMRoberta and a Trainer broadcast NoneType error do not share the same code path." }, { - "left": "issue:45803", - "right": "issue:46170", + "left": "issue:44464", + "right": "issue:46421", "accept": false, - "reason": "Bare-except error swallowing in FuyuBatchFeature is unrelated to the Windows ROCm torch.distributed dependency/import issue." + "reason": "Compiled-forward chunked generation inconsistency and cache lazy initialization causing recompilation are related at a high level, but not the same concrete bug." } ] }, @@ -10626,268 +10437,282 @@ "component_index": 1, "component_count": 22, "review_unit_index": 34, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ - "issue:35141", - "issue:41950", - "issue:42490", + "issue:16998", + "issue:22355", + "issue:42175", "issue:43065", + "issue:43295", + "issue:43316", "issue:43352", "issue:43388", "issue:43404", "issue:43441", - "issue:43475", - "issue:43504", - "issue:43525", - "issue:43531", + "issue:43452", "issue:43540", + "issue:43550", "issue:43572", + "issue:43606", + "issue:43646", "issue:43650", - "issue:43673", "issue:43688", + "issue:43701", + "issue:43716", "issue:43761", - "issue:43931", - "issue:44060", - "issue:44190", - "issue:44242", + "issue:43825", + "issue:44077", + "issue:44162", "issue:44479", + "issue:44496", "issue:44514", "issue:44534", "issue:44610", "issue:44617", - "issue:44734", - "issue:44743", "issue:44792", + "issue:44871", "issue:44913", - "issue:44964", - "issue:44977", "issue:45127", "issue:45200", - "issue:45406", - "issue:45542", - "issue:45600", + "issue:45584", "issue:45632", - "issue:45698", - "issue:46082", + "issue:45803", + "issue:46050", + "issue:46170", "issue:46178", "issue:46202", "issue:46234", + "issue:46291", + "issue:46365", "issue:46369", "issue:46424", - "issue:46640", - "issue:46697" + "issue:46640" ], "soft_pairs": [ + "issue:43550|issue:43701", + "issue:46234|issue:46291", + "issue:43606|issue:43701", + "issue:43295|issue:43825", + "issue:16998|issue:22355", + "issue:43295|issue:44871", + "issue:44871|issue:45584", + "issue:43316|issue:43716", + "issue:42175|issue:43316", + "issue:44077|issue:44534", + "issue:44871|issue:46050", + "issue:45803|issue:46365", + "issue:43646|issue:44913", + "issue:43452|issue:44496", + "issue:44162|issue:44792", + "issue:45803|issue:46170", "issue:45632|issue:46424", - "issue:44610|issue:46369", "issue:43404|issue:43441", + "issue:44610|issue:46369", "issue:43650|issue:46202", "issue:43761|issue:44479", "issue:43065|issue:44617", "issue:43540|issue:44514", "issue:45127|issue:45200", "issue:43388|issue:43688", - "issue:43572|issue:46178", "issue:43352|issue:46640", - "issue:45542|issue:46234", - "issue:43504|issue:45698", - "issue:44792|issue:44977", - "issue:43673|issue:44743", - "issue:44792|issue:45600", - "issue:41950|issue:43475", - "issue:43525|issue:43531", - "issue:44734|issue:45406", - "issue:44190|issue:44964", - "issue:44534|issue:44913", - "issue:46082|issue:46697", - "issue:35141|issue:42490", - "issue:43761|issue:44242", - "issue:43931|issue:44060" + "issue:43572|issue:46178" ], - "prepared_review_unit_hash": "088643272ab78ea51a03e33b53301fa69adbb03f75db3e98fc4fe375f0c5f881", + "prepared_review_unit_hash": "0e65eb2586586f85c6b123b69d74c3aa078bf202ea6a4fc5f84f14c9469b693b", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 25, - "serialized_chars": 35137, - "estimated_input_tokens": 8785, - "estimated_eval_tokens": 17826 + "soft_pair_count": 27, + "serialized_chars": 35625, + "estimated_input_tokens": 8907, + "estimated_eval_tokens": 18070 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: the soft pairs share broad keywords (cache, loading, flash attention, processor, etc.) but not the same concrete bug or fix path. No PRs are present. I\u2019d treat issue 45698 as the best representative issue only because it has the most activity and inbound references, not because it duplicates the others.", - "confidence": 0.93, - "canonical_issue_reason": "issue:45698 has the strongest signal in the set (highest discussion activity, inbound refs, and an open regression around from_pretrained/save_pretrained custom-module loading), so it is the best representative issue if one must be chosen.", + "summary": "This cluster is a loose mix of unrelated Transformers issues; none of the soft pairs look like the same underlying bug or change. Most similarities are only at the subsystem/model-family level, not the concrete failure mode.", + "confidence": 0.9, + "canonical_issue_reason": "Issue 43646 is the broadest, most representative regression in the set: a general Transformers 5.0.0 custom model initialization breakage.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45698 is the most central/active item in this collection and is the only one with broad follow-on discussion and references; it is the best issue to anchor the cluster despite the cluster\u2019s overall heterogeneity.", + "best_issue_reason": "Issue 43646 is the best standalone representative because it describes a concrete, high-impact v5 regression with broad user-facing scope.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:45632", - "right": "issue:46424", + "left": "issue:43550", + "right": "issue:43701", "accept": false, - "reason": "Both mention cache/path behavior, but one is a local-model trust_remote_code path collision and the other is max_cache_length handling in quantized cache; different bugs." + "reason": "Different failures: torch.compile/SDPA runtime bug vs checkpoint key mismatch." }, { - "left": "issue:44610", - "right": "issue:46369", + "left": "issue:46234", + "right": "issue:46291", "accept": false, - "reason": "Image-size mismatch in OmDet-Turbo processor vs memory overflow in sam3-litetext; unrelated symptoms and code paths." + "reason": "Unrelated import problems: missing utility symbol vs pipelines import constraint from kernels." }, { - "left": "issue:43404", - "right": "issue:43441", + "left": "issue:43606", + "right": "issue:43701", "accept": false, - "reason": "Same model family, but one is tied lm_head weights and the other is Flash Attention support; not the same underlying fix." + "reason": "CPU offload device mismatch is unrelated to resume-from-checkpoint key mapping." }, { - "left": "issue:43650", - "right": "issue:46202", + "left": "issue:43295", + "right": "issue:43825", "accept": false, - "reason": "No concrete overlap: one is a vague data request, the other is a Gemma pipeline roundtrip tokenization bug." + "reason": "Both are regressions, but one is a processor/tokenizer API break and the other is a pipeline error-message issue." }, { - "left": "issue:43761", - "right": "issue:44479", + "left": "issue:16998", + "right": "issue:22355", "accept": false, - "reason": "Different models and failures: CLIPVision hidden_states regression vs Qwen VL video-input regression." + "reason": "Model max length question and missing transformers.onnx module are unrelated." }, { - "left": "issue:43065", - "right": "issue:44617", + "left": "issue:43295", + "right": "issue:44871", "accept": false, - "reason": "One is a dummy Conv2d in Sam3PixelDecoder, the other is Sam3Video OOM; different problems." + "reason": "Different regressions: processor/tokenizer handling vs Gemma eos token config inconsistency." }, { - "left": "issue:43540", - "right": "issue:44514", + "left": "issue:44871", + "right": "issue:45584", "accept": false, - "reason": "Both involve multimodal/video processing, but they are different models and different failure modes (ValueError vs batched chat-template crash)." + "reason": "Both involve generation/token settings, but the concrete bugs differ: eos_token_id mismatch vs empty-transcription failure after token alignment." }, { - "left": "issue:45127", - "right": "issue:45200", + "left": "issue:43316", + "right": "issue:43716", "accept": false, - "reason": "LoRA merge/collapse with tied embeddings is unrelated to mm_token_type_ids defaulting in Gemma 4 fine-tuning." + "reason": "Gemma3 config API discrepancy and Mistral-3 image dtype mismatch are different issues." }, { - "left": "issue:43388", - "right": "issue:43688", + "left": "issue:42175", + "right": "issue:43316", "accept": false, - "reason": "Training/metrics issues in different subsystems; one drops last-batch labels, the other misnormalizes auxiliary loss." + "reason": "Backend packaging/install issue is unrelated to Gemma3 config API shape." }, { - "left": "issue:43572", - "right": "issue:46178", + "left": "issue:44077", + "right": "issue:44534", "accept": false, - "reason": "Config attribute rename (`pad_token_idx`) vs backend registration error (`tensorflow_text`); not the same bug." + "reason": "One is an invalid optional post_init allowance; the other is a non-persistent buffer initialization regression." }, { - "left": "issue:43352", - "right": "issue:46640", + "left": "issue:44871", + "right": "issue:46050", "accept": false, - "reason": "Both mention Flash Attention, but one is a model-specific unsupported feature warning and the other is a generic flash-attn issue; insufficiently specific." + "reason": "Gemma eos_token_id config mismatch is unrelated to quantized save/load breakage." }, { - "left": "issue:45542", - "right": "issue:46234", + "left": "issue:45803", + "right": "issue:46365", "accept": false, - "reason": "TensorBoard/TensorFlow backend mismatch is unrelated to a missing import_utils helper for falcon mamba availability." + "reason": "Bare except in tensor conversion is unrelated to Trainer import failure caused by datasets." }, { - "left": "issue:43504", - "right": "issue:45698", + "left": "issue:43646", + "right": "issue:44913", "accept": false, - "reason": "Both touch model loading, but one is a legacy preset field for Beit segmentation and the other is wrong custom-module resolution after save/load." + "reason": "Both relate to model/config loading, but one is a broad initialization regression and the other is a specific config field not persisting on reload." }, { - "left": "issue:44792", - "right": "issue:44977", + "left": "issue:43452", + "right": "issue:44496", "accept": false, - "reason": "Different models and behaviors: janus image-generation test failure vs Qwen3.5 flash-attention generation regression." + "reason": "GGUF loading path breakage and unrecognized model config are different model-loading problems." }, { - "left": "issue:43673", - "right": "issue:44743", + "left": "issue:44162", + "right": "issue:44792", "accept": false, - "reason": "Both involve cache/state handling, but one is GenerationMixin chunked_prefill cache and the other is Qwen3.5 recurrent state reset; not mergeable as one bug." + "reason": "ESM2 model breakage and a janus image-generation test failure are unrelated." }, { - "left": "issue:44792", - "right": "issue:45600", + "left": "issue:45803", + "right": "issue:46170", "accept": false, - "reason": "Janus generation test failure and Sam3LiteText auto_mappings CI break are unrelated." + "reason": "Different concerns: exception handling in Fuyu vs Windows/ROCm import dependency issues." }, { - "left": "issue:41950", - "right": "issue:43475", + "left": "issue:45632", + "right": "issue:46424", "accept": false, - "reason": "Video-classification processor mismatch vs Sam3 video encoder output missing an attribute; different layers and fixes." + "reason": "Cache path collision for trust_remote_code is unrelated to max_cache_length behavior." }, { - "left": "issue:43525", - "right": "issue:43531", + "left": "issue:43404", + "right": "issue:43441", "accept": false, - "reason": "Missing pad_token_id in config and sliding_window behavior in Qwen3-MoE are separate configuration issues." + "reason": "Mistral3 lm_head tying bug and Ministral-3 FlashAttention support issue affect different code paths." }, { - "left": "issue:44734", - "right": "issue:45406", + "left": "issue:44610", + "right": "issue:46369", "accept": false, - "reason": "Serve KV-cache tensor indexing crash is unrelated to Gemma4Processor missing _tokenizer." + "reason": "Processor output size mismatch and Sam3-litetext memory overflow are different problems." }, { - "left": "issue:44190", - "right": "issue:44964", + "left": "issue:43650", + "right": "issue:46202", "accept": false, - "reason": "Local dataset loading in an image-classification script is unrelated to loading a multimodal Phi-4 model." + "reason": "Placeholder issue text and Gemma pipeline roundtrip failure do not match." }, { - "left": "issue:44534", - "right": "issue:44913", + "left": "issue:43761", + "right": "issue:44479", "accept": false, - "reason": "Both are reload/regression style issues, but one is non-persistent buffer initialization and the other is GPTNeoX config persistence; different fixes." + "reason": "CLIPVision hidden_states regression and Qwen video-input regression are unrelated." }, { - "left": "issue:46082", - "right": "issue:46697", + "left": "issue:43065", + "right": "issue:44617", "accept": false, - "reason": "LlamaConfig head_dim validation and Swin2SR over-padding are completely unrelated." + "reason": "Dummy Conv2d in Sam3PixelDecoder is not the same as Sam3Video OOM." }, { - "left": "issue:35141", - "right": "issue:42490", + "left": "issue:43540", + "right": "issue:44514", "accept": false, - "reason": "Token-embedding reinit during post_init is unrelated to local-path trailing-slash import warnings." + "reason": "Both are Qwen multimodal bugs, but they hit different APIs and failure modes: video input processing vs batched chat-template padding." }, { - "left": "issue:43761", - "right": "issue:44242", + "left": "issue:45127", + "right": "issue:45200", "accept": false, - "reason": "CLIPVision hidden_states regression and missing load-balancing loss when output_router_logits=False are different model/subsystem bugs." + "reason": "LoRA merge collapse with tied embeddings is unrelated to Gemma 4 mm_token_type_ids defaults." }, { - "left": "issue:43931", - "right": "issue:44060", + "left": "issue:43388", + "right": "issue:43688", + "accept": false, + "reason": "Metric gathering label truncation and auxiliary loss normalization are different training bugs." + }, + { + "left": "issue:43352", + "right": "issue:46640", + "accept": false, + "reason": "Both mention Flash Attention, but 43352 is model-specific unsupported behavior while 46640 is too generic to be the same bug." + }, + { + "left": "issue:43572", + "right": "issue:46178", "accept": false, - "reason": "Qwen3-VL weight-shape mismatch during loading vs Qwen3-Next tied-weights warning; related model family but not the same concrete defect." + "reason": "StableLmConfig pad_token_idx regression and tensorflow_text backend mapping error are unrelated." } ] }, @@ -10903,64 +10728,78 @@ "component_index": 1, "component_count": 22, "review_unit_index": 35, - "review_unit_count": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:25251", "issue:27957", - "issue:39692", - "issue:41720", - "issue:43065", - "issue:43295", + "issue:35141", + "issue:41950", + "issue:42490", "issue:43335", - "issue:43526", - "issue:43540", - "issue:43550", - "issue:43644", - "issue:43673", + "issue:43475", + "issue:43504", + "issue:43525", + "issue:43531", + "issue:43550", + "issue:43644", + "issue:43673", + "issue:43761", "issue:43827", + "issue:43931", "issue:43994", - "issue:44077", - "issue:44112", + "issue:44060", "issue:44190", - "issue:44464", + "issue:44242", "issue:44479", - "issue:44521", - "issue:44610", + "issue:44534", "issue:44617", "issue:44734", - "issue:44737", - "issue:44779", + "issue:44743", + "issue:44792", "issue:44855", - "issue:44877", "issue:44913", "issue:44933", - "issue:45127", - "issue:45245", - "issue:45276", - "issue:45357", + "issue:44964", + "issue:44977", "issue:45362", "issue:45381", + "issue:45406", + "issue:45542", "issue:45563", "issue:45584", "issue:45588", "issue:45600", "issue:45632", - "issue:45803", + "issue:45698", + "issue:46082", + "issue:46234", "issue:46369", "issue:46421", - "issue:46489", "issue:46617", - "issue:46655", "issue:46697" ], "soft_pairs": [ + "issue:45542|issue:46234", + "issue:43504|issue:45698", + "issue:43673|issue:44743", + "issue:44792|issue:45600", + "issue:44792|issue:44977", + "issue:41950|issue:43475", + "issue:43525|issue:43531", + "issue:44190|issue:44964", + "issue:44734|issue:45406", + "issue:44534|issue:44913", + "issue:46082|issue:46697", + "issue:35141|issue:42490", "issue:44855|issue:45588", + "issue:43931|issue:44060", + "issue:43761|issue:44242", "issue:45632|issue:46617", "issue:43827|issue:43994", "issue:44617|issue:45584", @@ -10968,240 +10807,513 @@ "issue:45563|issue:45584", "issue:45381|issue:46369", "issue:43644|issue:44913", - "issue:44479|issue:44734", "issue:43335|issue:44855", + "issue:44479|issue:44734", "issue:44933|issue:45362", "issue:25251|issue:27957", - "issue:43550|issue:44190", - "issue:41720|issue:45245", - "issue:44077|issue:44877", - "issue:43065|issue:44610", - "issue:44464|issue:45563", - "issue:43335|issue:44190", - "issue:45381|issue:46655", - "issue:44617|issue:46369", - "issue:45127|issue:45357", - "issue:44779|issue:46489", - "issue:43526|issue:46697", - "issue:39692|issue:44190", - "issue:44112|issue:45600", - "issue:44737|issue:45803", - "issue:43295|issue:45276", - "issue:44521|issue:45381", - "issue:43540|issue:44734" + "issue:43550|issue:44190" ], - "prepared_review_unit_hash": "61bee999a2346d7069cf4531606447d9ac9b124ff366c88becd02394af4ab969", + "prepared_review_unit_hash": "bc7697b5d53c2b89b03280207f66a2f535c78c351a1c9530b1b135ca5d58c3cb", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 29, - "serialized_chars": 35825, - "estimated_input_tokens": 8957, - "estimated_eval_tokens": 18170 + "soft_pair_count": 27, + "serialized_chars": 35209, + "estimated_input_tokens": 8803, + "estimated_eval_tokens": 17862 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab bag of unrelated issue reports; there is no strong overall duplicate cluster. The only plausible duplicate-like pair is the DeepSeek tokenizer regression pair (44779/46489).", - "confidence": 0.84, - "canonical_issue_reason": "No single canonical issue cleanly represents the whole set because most items are unrelated. If one representative is needed, issue 46489 is the best anchor for the only clearly related topic: the DeepSeek v5 tokenizer regression with an explicit link to an incomplete prior fix.", + "summary": "This cluster is highly heterogeneous: most soft pairs only share broad themes like loading, caching, generation, or pipeline behavior, but they describe different bugs and code paths. No PRs are present.", + "confidence": 0.95, + "canonical_issue_reason": "No single issue truly represents this set because the items are not one duplicate cluster; if forced, #45698 is the broadest and most active loading-related issue, but it does not semantically cover the rest.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 46489 is the most actionable representative for the only meaningful duplicate candidate: it is specific, concrete, and explicitly ties the regression to PR #44801\u2019s incomplete fix.", + "best_issue_reason": "Not a strong global best for duplicate triage because the pairings are mostly unrelated; #45698 is only the most active generic loading bug, not a true cluster anchor.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:45542", + "right": "issue:46234", + "accept": false, + "reason": "Different failures: TensorBoard/TF backend import path vs a missing import-utils availability helper." + }, + { + "left": "issue:43504", + "right": "issue:45698", + "accept": false, + "reason": "Both involve model loading, but one is a legacy BEiT preset field bug and the other is wrong custom-module resolution after save/load." + }, + { + "left": "issue:43673", + "right": "issue:44743", + "accept": false, + "reason": "Cache-related, but different bugs: missing GenerationMixin cache in chunked prefill vs recurrent states resetting with seq_len>1." + }, + { + "left": "issue:44792", + "right": "issue:45600", + "accept": false, + "reason": "Unrelated: a Janus image-generation test failure vs auto_mappings referencing removed Sam3LiteText configs." + }, + { + "left": "issue:44792", + "right": "issue:44977", + "accept": false, + "reason": "Different models and paths: Janus image generation test failure vs Qwen3.5 flash-attention generation bug." + }, + { + "left": "issue:41950", + "right": "issue:43475", + "accept": false, + "reason": "Pipeline processor-selection bug vs a missing output attribute in Sam3 vision encoder output." + }, + { + "left": "issue:43525", + "right": "issue:43531", + "accept": false, + "reason": "Different config issues: missing pad_token_id on Llama4Config vs Qwen3-MoE sliding_window behavior." + }, + { + "left": "issue:44190", + "right": "issue:44964", + "accept": false, + "reason": "Different loading contexts: local dataset handling in a training script vs loading a multimodal model with latest transformers." + }, + { + "left": "issue:44734", + "right": "issue:45406", + "accept": false, + "reason": "Both touch serve, but one is KV-cache continuation indexing and the other is Gemma4Processor missing _tokenizer." + }, + { + "left": "issue:44534", + "right": "issue:44913", + "accept": false, + "reason": "Distinct bugs: non-persistent buffers being junk-filled vs GPTNeoXConfig rotary_pct not persisting on reload." + }, + { + "left": "issue:46082", + "right": "issue:46697", + "accept": false, + "reason": "Unrelated model/config validation vs image-processor padding behavior." + }, + { + "left": "issue:35141", + "right": "issue:42490", + "accept": false, + "reason": "Different areas: output embedding reinitialization after resize vs local-path trailing-slash import warning." + }, { "left": "issue:44855", "right": "issue:45588", "accept": false, - "reason": "Different subsystems and failures: Python 3.13 torch.jit parsing error vs flash-attention sink handling crash." + "reason": "Python 3.13 import parsing failure vs a flash_attention crash on s_aux=None." + }, + { + "left": "issue:43931", + "right": "issue:44060", + "accept": false, + "reason": "Different Qwen issues: VL weight-shape mismatch vs a tied-weights warning involving dt_bias." + }, + { + "left": "issue:43761", + "right": "issue:44242", + "accept": false, + "reason": "Both are v5/model behavior bugs, but one is CLIP hidden_states output and the other is MoE load-balancing loss gating." }, { "left": "issue:45632", "right": "issue:46617", "accept": false, - "reason": "Both involve trust_remote_code/cache, but one is a leaf-directory cache collision and the other is a FileNotFoundError while loading from cache; not the same concrete bug." + "reason": "Related cache/loading area, but not the same bug: cache-path collision for local leaf names vs FileNotFoundError when loading from cache." }, { "left": "issue:43827", "right": "issue:43994", "accept": false, - "reason": "One is docs stale after pipeline removal; the other is a runtime correctness issue with SigLIP2 outputs." + "reason": "Docs still referencing pipeline() is not the same as siglip2 producing bad outputs with AutoModel/pipeline." }, { "left": "issue:44617", "right": "issue:45584", "accept": false, - "reason": "Unrelated models and symptoms: Sam3Video OOM vs Whisper empty transcription after token alignment." + "reason": "Completely different symptoms: Sam3Video CUDA OOM vs Whisper empty-transcription generation failure." }, { "left": "issue:43673", "right": "issue:46421", "accept": false, - "reason": "Both mention chunked prefill/cache, but one reports a missing cache in v5.0.0 and the other spurious recompilation from lazy init; not clearly the same bug." + "reason": "Both mention cache/chunked prefill, but one is a missing cache regression and the other is lazy cache init causing recompilation." }, { "left": "issue:45563", "right": "issue:45584", "accept": false, - "reason": "A stale warning about num_return_sequences is unrelated to Whisper generation failing on empty transcription." + "reason": "Different generation bugs: stale warning for num_return_sequences vs Whisper empty-transcription handling." }, { "left": "issue:45381", "right": "issue:46369", "accept": false, - "reason": "Both are multimodal/video-related in a broad sense, but one is a Qwen2.5-VL position-id bug and the other is Sam3-lite text memory overflow." + "reason": "Different regressions: Qwen2.5-VL video position IDs vs Sam3-LiteText memory overflow." }, { "left": "issue:43644", "right": "issue:44913", "accept": false, - "reason": "Non-persistent buffer initialization junk vs GPTNeoX rotary_pct reload bug are different configuration/state issues." + "reason": "Different model-state issues: junk non-persistent buffers vs rotary_pct reverting after reload." }, { - "left": "issue:44479", - "right": "issue:44734", + "left": "issue:43335", + "right": "issue:44855", "accept": false, - "reason": "Different code paths: a video-input regression for Qwen models vs a serve KV-cache indexing crash." + "reason": "Unrelated: SwitchTransformers sparse-layer creation bug vs a DebertaV2 Python 3.13 indentation/parsing error." }, { - "left": "issue:43335", - "right": "issue:44855", + "left": "issue:44479", + "right": "issue:44734", "accept": false, - "reason": "Sparse-layer construction bug in SwitchTransformers is unrelated to DebertaV2 import parsing on Python 3.13." + "reason": "Both are regressions, but one is video-input handling and the other is serve KV-cache continuation indexing." }, { "left": "issue:44933", "right": "issue:45362", "accept": false, - "reason": "No shared underlying bug; one is a missing import from image_utils, the other is a Qwen chat crash." + "reason": "Different failures: nonexistent image_utils import vs Qwen3.5 chat generation crash." }, { "left": "issue:25251", "right": "issue:27957", "accept": false, - "reason": "Pipeline top_k nesting behavior and XLMRoberta Flash Attention 2 support are unrelated." + "reason": "Unrelated: pipeline top_k returning a nested list vs XLMRoberta Flash Attention 2 support." + }, + { + "left": "issue:43550", + "right": "issue:44190", + "accept": false, + "reason": "Different code paths: Bamba SDPA/torch.compile failure vs a local dataset-loading script 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": 22, + "review_unit_index": 36, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:39692", + "issue:41720", + "issue:42175", + "issue:42796", + "issue:43065", + "issue:43295", + "issue:43335", + "issue:43526", + "issue:43540", + "issue:43854", + "issue:44077", + "issue:44112", + "issue:44190", + "issue:44246", + "issue:44279", + "issue:44462", + "issue:44464", + "issue:44521", + "issue:44610", + "issue:44617", + "issue:44734", + "issue:44737", + "issue:44779", + "issue:44877", + "issue:44898", + "issue:44938", + "issue:44964", + "issue:45003", + "issue:45127", + "issue:45245", + "issue:45276", + "issue:45357", + "issue:45381", + "issue:45542", + "issue:45563", + "issue:45584", + "issue:45600", + "issue:45632", + "issue:45803", + "issue:45907", + "issue:46082", + "issue:46121", + "issue:46234", + "issue:46369", + "issue:46489", + "issue:46655", + "issue:46697" + ], + "soft_pairs": [ + "issue:41720|issue:45245", + "issue:44077|issue:44877", + "issue:43065|issue:44610", + "issue:43335|issue:44190", + "issue:44464|issue:45563", + "issue:45381|issue:46655", + "issue:45127|issue:45357", + "issue:44617|issue:46369", + "issue:44779|issue:46489", + "issue:43526|issue:46697", + "issue:39692|issue:44190", + "issue:44112|issue:45600", + "issue:44737|issue:45803", + "issue:43295|issue:45276", + "issue:43540|issue:44734", + "issue:44521|issue:45381", + "issue:44938|issue:45542", + "issue:45276|issue:45584", + "issue:44462|issue:45907", + "issue:45632|issue:46234", + "issue:44898|issue:46697", + "issue:44246|issue:45003", + "issue:46082|issue:46121", + "issue:43854|issue:44964", + "issue:44964|issue:45600", + "issue:45245|issue:45584", + "issue:42796|issue:44521", + "issue:42175|issue:44279", + "issue:44112|issue:44964" + ], + "prepared_review_unit_hash": "d9a8730effbba1dc2688b4fd791c2ce8ab868d3b4d5107e3b718cfee1da7e574", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35806, + "estimated_input_tokens": 8952, + "estimated_eval_tokens": 18160 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose soft-similarity set of unrelated issues across different models, processors, configs, and tooling. I would not merge any of the soft edges as duplicates.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue stands out: the items span different subsystems and symptoms, so the cluster is too heterogeneous for duplicate consolidation.", + "canonical_pr_reason": null, + "best_issue_reason": "No best issue for duplicate triage; if forced to pick a representative, none of these is meaningfully representative of the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41720", + "right": "issue:45245", + "accept": false, + "reason": "Qwen3 auto device-mapping CUDA assert and a category-count limit error are unrelated failures." + }, + { + "left": "issue:44077", + "right": "issue:44877", + "accept": false, + "reason": "Different problems: optional post_init on PatchTSMixer vs strict granite_speech config loading." + }, + { + "left": "issue:43065", + "right": "issue:44610", + "accept": false, + "reason": "Sam3PixelDecoder dummy Conv2d and OmDet-Turbo processor/model size mismatch are separate bugs." + }, + { + "left": "issue:43335", + "right": "issue:44190", + "accept": false, + "reason": "SwitchTransformers sparse-layer construction and local dataset loading in a training script do not match." + }, + { + "left": "issue:44464", + "right": "issue:45563", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency and a stale paged-generate warning are different issues." + }, + { + "left": "issue:45381", + "right": "issue:46655", + "accept": false, + "reason": "Wrong vision_position_ids in Qwen2.5-VL video input vs SmolVLM video padding calculation." + }, + { + "left": "issue:45127", + "right": "issue:45357", + "accept": false, + "reason": "LoRA merge collapse with tied embeddings is unrelated to incorrect visual encoder keys on save_pretrained." + }, + { + "left": "issue:44617", + "right": "issue:46369", + "accept": false, + "reason": "Same family, but one is Sam3Video OOM and the other is Sam3-LiteText memory overflow; not the same bug." + }, + { + "left": "issue:44779", + "right": "issue:46489", + "accept": false, + "reason": "Both involve DeepSeek tokenizers, but they target different models/versions and different incorrect-output causes." + }, + { + "left": "issue:43526", + "right": "issue:46697", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and Swin2SR over-padding are unrelated image-processing bugs." + }, + { + "left": "issue:39692", + "right": "issue:44190", + "accept": false, + "reason": "SigLIP2 docs example errors and a local-dataset training script issue are unrelated." + }, + { + "left": "issue:44112", + "right": "issue:45600", + "accept": false, + "reason": "Stale GraniteSpeech device-override test vs auto_mappings referencing removed Sam3LiteText configs." }, { - "left": "issue:43550", - "right": "issue:44190", + "left": "issue:44737", + "right": "issue:45803", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba-9B-v2 is unrelated to loading a local dataset in a training script." + "reason": "XLNet device placement in torch.arange and a bare-except swallowing errors are different code paths." }, { - "left": "issue:41720", - "right": "issue:45245", + "left": "issue:43295", + "right": "issue:45276", "accept": false, - "reason": "Different errors and domains: Qwen3 device-mapping cudaErrorAssert vs category-count limit in dataset/processing." + "reason": "Processor.tokenizer regression and Gemma4 embedding-resize propagation are separate API bugs." }, { - "left": "issue:44077", - "right": "issue:44877", + "left": "issue:43540", + "right": "issue:44734", "accept": false, - "reason": "Both involve config handling, but they are different models and different strictness/optional-field problems." + "reason": "Qwen3OmniMoe video input validation and serve KV-cache continuation indexing are unrelated crashes." }, { - "left": "issue:43065", - "right": "issue:44610", + "left": "issue:44521", + "right": "issue:45381", "accept": false, - "reason": "Sam3PixelDecoder dummy Conv2d and OmDet-Turbo processor size mismatch are unrelated model/processor issues." + "reason": "Both are multimodal, but assistant-mask generation and video vision_position_ids are distinct functions/issues." }, { - "left": "issue:44464", - "right": "issue:45563", + "left": "issue:44938", + "right": "issue:45542", "accept": false, - "reason": "Compiled forward chunked generation inconsistency is unrelated to a stale warning about num_return_sequences." + "reason": "Python 3.14 import failure and a missing TensorFlow backend when only tensorboard is installed are different packaging/runtime problems." }, { - "left": "issue:43335", - "right": "issue:44190", + "left": "issue:45276", + "right": "issue:45584", "accept": false, - "reason": "SwitchTransformers sparse-layer bug and local dataset loading failure are unrelated." + "reason": "Gemma4 resize_token_embeddings propagation and Whisper empty-transcription handling are unrelated." }, { - "left": "issue:45381", - "right": "issue:46655", + "left": "issue:44462", + "right": "issue:45907", "accept": false, - "reason": "Both are video-processing related, but they affect different models and different bugs (vision_position_ids vs padding calculation)." + "reason": "AutoTokenizer ignoring tokenizer.json and list_repo_files not forwarding revision are different repository-access bugs." }, { - "left": "issue:44617", - "right": "issue:46369", + "left": "issue:45632", + "right": "issue:46234", "accept": false, - "reason": "Same family name does not imply duplicate: one is video OOM, the other is litetext memory overflow, likely different code paths." + "reason": "Cache-path collision for trust_remote_code and a missing import_utils helper are unrelated." }, { - "left": "issue:45127", - "right": "issue:45357", + "left": "issue:44898", + "right": "issue:46697", "accept": false, - "reason": "LoRA merge with tied embeddings is unrelated to Qwen3.5 save_pretrained visual encoder key serialization." + "reason": "Perceiver interpolate_pos_encoding failure and Swin2SR over-padding are separate image-processor issues." }, { - "left": "issue:44779", - "right": "issue:46489", - "accept": true, - "reason": "Both describe DeepSeek tokenizer regressions in v5+, and 46489 explicitly says the issue is a gap in PR #44801\u2019s fix, making them the same underlying bug." + "left": "issue:44246", + "right": "issue:45003", + "accept": false, + "reason": "Slow import timing and unsafe sys.modules access are not the same problem." }, { - "left": "issue:43526", - "right": "issue:46697", + "left": "issue:46082", + "right": "issue:46121", "accept": false, - "reason": "BeitImageProcessorFast reduce_labels bug and Swin2SR over-padding bug are unrelated image-processing issues." + "reason": "Llama head_dim validation and JSON-loaded rope-key list handling are unrelated config parsing bugs." }, { - "left": "issue:39692", - "right": "issue:44190", + "left": "issue:43854", + "right": "issue:44964", "accept": false, - "reason": "SigLIP2 documentation/example failures are unrelated to local dataset loading in a training example." + "reason": "Both are model-loading reports, but they concern different models and likely different failure causes." }, { - "left": "issue:44112", + "left": "issue:44964", "right": "issue:45600", "accept": false, - "reason": "Both mention CI, but one is a stale GraniteSpeech device override test and the other is broken auto_mappings references to removed configs." + "reason": "Phi-4 multimodal load failure and Sam3LiteText auto-mapping CI break are unrelated." }, { - "left": "issue:44737", - "right": "issue:45803", + "left": "issue:45245", + "right": "issue:45584", "accept": false, - "reason": "XLNet device placement bug and a bare-except swallowing errors in FuyuBatchFeature are unrelated." + "reason": "Category-count overflow and Whisper empty transcription have no shared code path." }, { - "left": "issue:43295", - "right": "issue:45276", + "left": "issue:42796", + "right": "issue:44521", "accept": false, - "reason": "Processor/tokenizer regression and Gemma4 resize_token_embeddings behavior are different bugs." + "reason": "Tokenizer BPE-marker handling and multimodal assistant-mask generation are different subsystems." }, { - "left": "issue:44521", - "right": "issue:45381", + "left": "issue:42175", + "right": "issue:44279", "accept": false, - "reason": "All-zero assistant masks in apply_chat_template are not the same as wrong vision_position_ids in Qwen2.5-VL video inputs." + "reason": "TensorFlow missing from pip extras and a generic dependency issue are too vague and not the same concrete bug." }, { - "left": "issue:43540", - "right": "issue:44734", + "left": "issue:44112", + "right": "issue:44964", "accept": false, - "reason": "Qwen3OmniMoe video ValueError and serve KV-cache continuation indexing are different failures in different paths." + "reason": "GraniteSpeech CI device override and Phi-4 model loading are unrelated." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative about the soft-edge pairs. The only accepted duplicate-like pair, issue:44779 with issue:46489, is well supported by the shared DeepSeek tokenizer regression context and 46489\u2019s explicit reference to a gap in PR #44801. No other pair is overstated as a duplicate." - }, - "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 }, @@ -11209,14 +11321,14 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 36, - "review_unit_count": 56, + "review_unit_index": 37, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:27957", "issue:28282", @@ -11226,61 +11338,47 @@ "issue:38617", "issue:41628", "issue:41950", - "issue:42175", - "issue:42796", + "issue:42907", + "issue:43295", + "issue:43381", "issue:43404", + "issue:43441", "issue:43479", + "issue:43526", "issue:43531", "issue:43643", "issue:43650", "issue:43673", - "issue:43854", + "issue:43824", + "issue:43901", + "issue:43976", "issue:44016", - "issue:44112", + "issue:44062", "issue:44186", "issue:44220", - "issue:44246", "issue:44279", "issue:44295", - "issue:44462", + "issue:44336", "issue:44496", - "issue:44521", + "issue:44683", "issue:44734", - "issue:44898", - "issue:44938", - "issue:44964", - "issue:45003", + "issue:44792", + "issue:44933", + "issue:44987", + "issue:45042", "issue:45083", "issue:45216", - "issue:45245", - "issue:45276", + "issue:45335", "issue:45375", - "issue:45542", - "issue:45584", + "issue:45446", + "issue:45588", + "issue:45593", "issue:45600", - "issue:45632", - "issue:45907", - "issue:46082", - "issue:46121", - "issue:46234", + "issue:45759", "issue:46424", - "issue:46506", - "issue:46697" + "issue:46506" ], "soft_pairs": [ - "issue:44938|issue:45542", - "issue:44462|issue:45907", - "issue:45276|issue:45584", - "issue:45632|issue:46234", - "issue:44898|issue:46697", - "issue:44246|issue:45003", - "issue:46082|issue:46121", - "issue:42796|issue:44521", - "issue:43854|issue:44964", - "issue:44964|issue:45600", - "issue:45245|issue:45584", - "issue:42175|issue:44279", - "issue:44112|issue:44964", "issue:43643|issue:44496", "issue:27957|issue:35141", "issue:43404|issue:45216", @@ -11289,217 +11387,250 @@ "issue:43650|issue:44220", "issue:36246|issue:38617", "issue:44295|issue:46506", - "issue:41950|issue:44016", - "issue:43479|issue:45375", "issue:32090|issue:35141", + "issue:43479|issue:45375", + "issue:41950|issue:44016", "issue:28282|issue:35141", "issue:43673|issue:46424", "issue:43531|issue:45600", "issue:44016|issue:44220", - "issue:45083|issue:46506" + "issue:45083|issue:46506", + "issue:43295|issue:45335", + "issue:43976|issue:44987", + "issue:44683|issue:45446", + "issue:43526|issue:43901", + "issue:44496|issue:45042", + "issue:42907|issue:43441", + "issue:43673|issue:44336", + "issue:43381|issue:45593", + "issue:43901|issue:44496", + "issue:43976|issue:44279", + "issue:43441|issue:45588", + "issue:44016|issue:44062", + "issue:44496|issue:45759", + "issue:44792|issue:46424", + "issue:43824|issue:44933" ], - "prepared_review_unit_hash": "75907eed1a521c5bbf14185da9f7d709515064f57a52b971a0a70b5807d7b248", + "prepared_review_unit_hash": "287b6e4c060a4d22c933096cccfbea291f581553da60b151489fa9f031af44ad", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36466, - "estimated_input_tokens": 9117, - "estimated_eval_tokens": 18490 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36484, + "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 heterogeneous: mostly superficially similar Transformers issues, but the soft pairs describe different bugs, code paths, or model families. I would reject all soft edges as duplicates.", - "confidence": 0.95, - "canonical_issue_reason": null, + "summary": "These are mostly unrelated Transformers issues that only overlap at a high level (imports, configs, cache, attention, or model-loading). None of the proposed soft pairs look like the same underlying bug or change, so I would not merge any of them.", + "confidence": 0.91, + "canonical_issue_reason": "issue:44496 is the broadest, most reusable representative of the cluster\u2019s config/load-path failures and has a clear, user-facing reproduction.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a clear canonical representative; the cluster spans unrelated import errors, model-specific regressions, and notebook/CI bugs.", + "best_issue_reason": "issue:44496 is the clearest general loading/configuration failure with broad impact, making it the best anchor issue in this loose cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44938", - "right": "issue:45542", + "left": "issue:43643", + "right": "issue:44496", "accept": false, - "reason": "Different failure modes: Python 3.14 import/load problem vs missing TensorFlow backend when installing the torch extra." + "reason": "Both are loading/configuration failures, but one is about missing fields with `trust_remote_code=True` and the other is an unrecognized model/config `model_type`; different bugs." }, { - "left": "issue:44462", - "right": "issue:45907", + "left": "issue:27957", + "right": "issue:35141", "accept": false, - "reason": "One is AutoTokenizer ignoring tokenizer.json; the other is list_repo_files not forwarding revision. Different API paths." + "reason": "FlashAttention/model-support behavior and embedding reinitialization after resize are unrelated code paths." }, { - "left": "issue:45276", - "right": "issue:45584", + "left": "issue:43404", + "right": "issue:45216", "accept": false, - "reason": "Unrelated bugs: gemma4 resize_token_embeddings/output embeddings vs Whisper empty-transcription generation after align_special_tokens." + "reason": "Both touch weight tying, but they affect different model families and different failure modes." }, { - "left": "issue:45632", - "right": "issue:46234", + "left": "issue:44186", + "right": "issue:44734", "accept": false, - "reason": "Cache-path collision for trust_remote_code local models is unrelated to a missing import_utils availability symbol." + "reason": "Tokenizer padding/NER crashes are unrelated to the serve KV-cache tensor-indexing bug." }, { - "left": "issue:44898", - "right": "issue:46697", + "left": "issue:36246", + "right": "issue:41628", "accept": false, - "reason": "Different image-processing bugs in different models: Perceiver interpolation vs Swin2SR over-padding." + "reason": "Different missing exports: `Qwen2_5_VLImageProcessor` versus `AutoImageProcessor`." }, { - "left": "issue:44246", - "right": "issue:45003", + "left": "issue:43650", + "right": "issue:44220", "accept": false, - "reason": "Slow import timing issue vs unsafe sys.modules access in modeling_utils; not the same code-path bug." + "reason": "`ADD THE DATA` is not the same bug as `_torch_extract_fbank_features()` misbehavior." }, { - "left": "issue:46082", - "right": "issue:46121", + "left": "issue:36246", + "right": "issue:38617", "accept": false, - "reason": "LlamaConfig head_dim validation and rope-parameter list handling are distinct config-validation problems." + "reason": "Both are import errors, but they are for different symbols in different modules." }, { - "left": "issue:42796", - "right": "issue:44521", + "left": "issue:44295", + "right": "issue:46506", "accept": false, - "reason": "Tokenizer BPE marker handling is unrelated to multimodal assistant_mask generation in apply_chat_template." + "reason": "A buffer read/access issue is not the same as CPU/CUDA output divergence." }, { - "left": "issue:43854", - "right": "issue:44964", + "left": "issue:32090", + "right": "issue:35141", "accept": false, - "reason": "Different model-loading failures for different models; not the same broken loader or code path." + "reason": "Distributed broadcast `NoneType` handling and embedding reinitialization are unrelated." }, { - "left": "issue:44964", - "right": "issue:45600", + "left": "issue:43479", + "right": "issue:45375", "accept": false, - "reason": "Phi-4 multimodal loading failure vs stale auto_mappings references to removed configs; distinct failures." + "reason": "Default config initialization with `None` is not the same as a field being dropped by `@strict`." }, { - "left": "issue:45245", - "right": "issue:45584", + "left": "issue:41950", + "right": "issue:44016", "accept": false, - "reason": "Category-count overflow and Whisper empty-transcription generation are unrelated bugs." + "reason": "Pipeline processor lookup is unrelated to a notebook syntax error." }, { - "left": "issue:42175", - "right": "issue:44279", + "left": "issue:28282", + "right": "issue:35141", "accept": false, - "reason": "Packaging/backend dependency complaint vs generic dependency issue; too broad and not the same underlying defect." + "reason": "Missing PyTorch dependency detection and token-embedding resize behavior are different bugs." }, { - "left": "issue:44112", - "right": "issue:44964", + "left": "issue:43673", + "right": "issue:46424", "accept": false, - "reason": "CI stale-device override test in GraniteSpeech is unrelated to Phi-4 multimodal model loading." + "reason": "Both mention cache, but one is missing cache support in chunked prefill and the other is `max_cache_length` only applying to Quantized Cache." }, { - "left": "issue:43643", - "right": "issue:44496", + "left": "issue:43531", + "right": "issue:45600", "accept": false, - "reason": "trust_remote_code missing fields and unrecognized-model/config.json model_type errors are different config-loading failures." + "reason": "Sliding-window handling in Qwen3-MoE is unrelated to stale `auto_mappings.py` references." }, { - "left": "issue:27957", - "right": "issue:35141", + "left": "issue:44016", + "right": "issue:44220", "accept": false, - "reason": "FlashAttention2 support for XLM-Roberta is unrelated to post_init reinitializing embeddings when resizing token embeddings." + "reason": "A notebook syntax error is not the same as a feature-extraction helper issue." }, { - "left": "issue:43404", - "right": "issue:45216", + "left": "issue:45083", + "right": "issue:46506", "accept": false, - "reason": "Mistral3 lm_head tying bug and Qwen3.5 save_pretrained regression are different model-specific serialization/training issues." + "reason": "Helper output-length semantics and CPU/CUDA result differences are not the same concrete defect." }, { - "left": "issue:44186", - "right": "issue:44734", + "left": "issue:43295", + "right": "issue:45335", "accept": false, - "reason": "LayoutLMv2 tokenizer padding/NER crash is unrelated to transformers serve KV-cache continuation indexing." + "reason": "Processor/tokenizer regression and `decoder.embed_tokens` not updating after resize are different paths." }, { - "left": "issue:36246", - "right": "issue:41628", + "left": "issue:43976", + "right": "issue:44987", "accept": false, - "reason": "Missing Qwen2.5 VLImageProcessor export vs missing AutoImageProcessor import; different symbols and fixes." + "reason": "Python version compatibility and a specific model-loading failure are unrelated." }, { - "left": "issue:43650", - "right": "issue:44220", + "left": "issue:44683", + "right": "issue:45446", "accept": false, - "reason": "The 'ADD THE DATA' issue is unrelated to _torch_extract_fbank_features()." + "reason": "A compiled `flex_attention` failure on newer torch and an incorrect version guard for `AuxRequest` are different bugs." }, { - "left": "issue:36246", - "right": "issue:38617", + "left": "issue:43526", + "right": "issue:43901", + "accept": false, + "reason": "Image-label reduction behavior and documentation wording about `return_all_scores` do not describe the same issue." + }, + { + "left": "issue:44496", + "right": "issue:45042", + "accept": false, + "reason": "Generic model/config loading failure and an unnecessary torchvision dependency in PIL image processors are different bugs." + }, + { + "left": "issue:42907", + "right": "issue:43441", + "accept": false, + "reason": "Saving dequantized Ministrals/Devstrals and FlashAttention failures in Ministral-3 are related only at the model family level, not the same bug." + }, + { + "left": "issue:43673", + "right": "issue:44336", "accept": false, - "reason": "Different missing imports: Qwen2.5 image processor export vs layer_type_validation from configuration_utils." + "reason": "Cache behavior and ANSI escape handling in loading_report are unrelated." }, { - "left": "issue:44295", - "right": "issue:46506", + "left": "issue:43381", + "right": "issue:45593", "accept": false, - "reason": "position_ids buffer access error is not the same as CPU/CUDA output divergence in PPDocLayoutV3." + "reason": "Gradient checkpointing in eval mode and missing auxiliary losses for D-FINE are different behaviors." }, { - "left": "issue:41950", - "right": "issue:44016", + "left": "issue:43901", + "right": "issue:44496", "accept": false, - "reason": "Video-classification pipeline image-processor lookup bug vs a notebook syntax error; not duplicates." + "reason": "A docs/runtime mismatch about `return_all_scores` is not the same as a config/model-type import error." }, { - "left": "issue:43479", - "right": "issue:45375", + "left": "issue:43976", + "right": "issue:44279", "accept": false, - "reason": "Phi4MultimodalConfig default initialization with None is unrelated to Qwen3_5MoeVisionConfig dropping a field." + "reason": "Python 3.9+ compatibility and a vague dependency issue are not the same underlying bug." }, { - "left": "issue:32090", - "right": "issue:35141", + "left": "issue:43441", + "right": "issue:45588", "accept": false, - "reason": "Trainer GPU broadcast NoneType error and embedding reinitialization after resize_token_embeddings are different code paths." + "reason": "Both involve flash attention, but one is a model-level RC failure and the other is a specific `s_aux=None` AttributeError; too different to merge." }, { - "left": "issue:28282", - "right": "issue:35141", + "left": "issue:44016", + "right": "issue:44062", "accept": false, - "reason": "Missing PyTorch ImportError for AutoModel is unrelated to the tied-embedding reinit bug." + "reason": "Notebook syntax error and `AddedToken(..., special=...)` argument collision are unrelated." }, { - "left": "issue:43673", - "right": "issue:46424", + "left": "issue:44496", + "right": "issue:45759", "accept": false, - "reason": "GenerationMixin cache missing in chunked_prefill vs max_cache_length only honored for Quantized Cache are related to caching broadly, but not the same concrete defect." + "reason": "Unrecognized `model_type` and failure to unwrap `text_config` in composite multimodal configs are different loading bugs." }, { - "left": "issue:43531", - "right": "issue:45600", + "left": "issue:44792", + "right": "issue:46424", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and removed Sam3LiteText references in auto_mappings are unrelated." + "reason": "A janus image-generation test failure is not the same as cache-length handling for quantized cache." }, { - "left": "issue:44016", - "right": "issue:44220", + "left": "issue:43824", + "right": "issue:44933", "accept": false, - "reason": "Notebook syntax error vs fbank feature extraction bug; no shared underlying issue." + "reason": "Missing export of `Qwen2_5_VLForConditionalGeneration` and non-existent import from `image_utils` are different import errors." }, { - "left": "issue:45083", + "left": "issue:45042", "right": "issue:46506", "accept": false, - "reason": "Helper length calculation behavior in qwen3_omni_moe is unrelated to CPU/CUDA numerical differences in PPDocLayoutV3." + "reason": "Processor torchvision dependency regression and CPU/CUDA discrepancy are unrelated." } ] }, @@ -11514,83 +11645,67 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 37, - "review_unit_count": 56, + "review_unit_index": 38, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:39692", + "issue:41950", "issue:42796", - "issue:42907", "issue:43295", "issue:43299", - "issue:43381", - "issue:43441", - "issue:43526", + "issue:43479", + "issue:43525", "issue:43577", "issue:43650", - "issue:43673", - "issue:43824", + "issue:43653", + "issue:43854", "issue:43901", "issue:43976", "issue:44016", - "issue:44062", "issue:44075", + "issue:44242", "issue:44279", - "issue:44336", + "issue:44291", "issue:44351", "issue:44368", - "issue:44496", + "issue:44462", + "issue:44466", "issue:44521", "issue:44683", "issue:44779", - "issue:44792", - "issue:44933", - "issue:44987", - "issue:45042", + "issue:45020", + "issue:45081", + "issue:45200", "issue:45216", + "issue:45276", "issue:45278", "issue:45335", "issue:45357", "issue:45405", - "issue:45446", + "issue:45542", "issue:45584", - "issue:45588", "issue:45593", - "issue:45759", "issue:45812", + "issue:45823", "issue:46082", "issue:46123", + "issue:46209", "issue:46369", "issue:46399", - "issue:46424", "issue:46512", "issue:46515", "issue:46693" ], "soft_pairs": [ - "issue:43295|issue:45335", - "issue:43976|issue:44987", - "issue:44683|issue:45446", - "issue:43526|issue:43901", - "issue:44496|issue:45042", - "issue:42907|issue:43441", - "issue:43381|issue:45593", - "issue:43673|issue:44336", - "issue:43901|issue:44496", - "issue:43976|issue:44279", - "issue:43441|issue:45588", - "issue:44016|issue:44062", - "issue:44496|issue:45759", - "issue:44792|issue:46424", "issue:46123|issue:46369", - "issue:43824|issue:44933", - "issue:44521|issue:46693", "issue:43299|issue:45216", + "issue:44521|issue:46693", "issue:44279|issue:46515", "issue:45335|issue:45584", "issue:46082|issue:46399", @@ -11599,227 +11714,220 @@ "issue:44075|issue:44368", "issue:39692|issue:43901", "issue:43295|issue:44779", - "issue:42796|issue:43650", "issue:43976|issue:45405", + "issue:42796|issue:43650", "issue:44351|issue:45278", "issue:46082|issue:46512", "issue:43299|issue:45357", - "issue:44779|issue:45812" + "issue:44779|issue:45812", + "issue:44242|issue:45200", + "issue:41950|issue:45020", + "issue:43299|issue:43854", + "issue:43479|issue:43525", + "issue:44466|issue:45276", + "issue:41950|issue:45823", + "issue:43650|issue:46209", + "issue:43653|issue:45593", + "issue:44683|issue:45542", + "issue:44462|issue:45081", + "issue:44016|issue:44291" ], - "prepared_review_unit_hash": "5d0bbfd109ca3a0e5f079c34c3144fbc69a98435fbf2da0c64ca0072f986c0e3", + "prepared_review_unit_hash": "ef27cdcc7f63d5ea04ece6c6d7c701a9561e33c8142bebe9e1b83d0aca99e5e2", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 37215, - "estimated_input_tokens": 9304, - "estimated_eval_tokens": 18864 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35631, + "estimated_input_tokens": 8908, + "estimated_eval_tokens": 18072 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: the items cover unrelated tokenizer, model-loading, cache, attention, docs, and regression reports. I do not see any pair that describes the same concrete bug/change, so no soft edges should be merged.", - "confidence": 0.98, - "canonical_issue_reason": "No single canonical issue stands out; the issues are mostly unrelated and do not form a duplicate set.", + "summary": "This cluster is highly heterogeneous: the items cover unrelated regressions in tokenizers, multimodal configs, pipelines, docs, runtime/version compatibility, and model-specific save/load paths. None of the soft pairs look like the same concrete bug or change, so I would not collapse any of them together.", + "confidence": 0.96, + "canonical_issue_reason": "No canonical issue: the cluster spans many unrelated reports, and the soft pairs are not true same-bug duplicates.", "canonical_pr_reason": null, - "best_issue_reason": "No strong global best issue for deduplication purposes; the cluster does not cohere around one underlying defect.", + "best_issue_reason": "No single issue stands out as the global representative; the reports are too diverse in model, subsystem, and failure mode.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43295", - "right": "issue:45335", - "accept": false, - "reason": "Different bugs: multimodal processor/tokenizer regression vs resize_token_embeddings not updating decoder.embed_tokens." - }, - { - "left": "issue:43976", - "right": "issue:44987", - "accept": false, - "reason": "Python version compatibility problem vs failure loading a specific model." - }, - { - "left": "issue:44683", - "right": "issue:45446", + "left": "issue:46123", + "right": "issue:46369", "accept": false, - "reason": "Both mention flex_attention/PyTorch, but one is a compile failure on torch>=2.9 and the other is a version-check/import bug; not the same defect." + "reason": "Different bugs: partial-batch result dropping in MaskGenerationPipeline vs memory overflow in sam3-litetext." }, { - "left": "issue:43526", - "right": "issue:43901", + "left": "issue:43299", + "right": "issue:45216", "accept": false, - "reason": "BEiT image processor label reduction bug vs pipeline docs mentioning return_all_scores." + "reason": "Both are Qwen-related regressions, but one is model loading for Qwen3VL MoE and the other is save_pretrained checkpoint serialization for Qwen3.5." }, { - "left": "issue:44496", - "right": "issue:45042", + "left": "issue:44521", + "right": "issue:46693", "accept": false, - "reason": "Unrecognized model/config loading error vs PIL image processors incorrectly requiring torchvision." + "reason": "Unrelated: multimodal assistant_masks are all zero vs a long-context FA2 performance regression." }, { - "left": "issue:42907", - "right": "issue:43441", + "left": "issue:44279", + "right": "issue:46515", "accept": false, - "reason": "Saving dequantized Ministrals/Devstrals fails vs Ministral-3 flash-attention failure; different code paths." + "reason": "Different compatibility problems: a transformers dependency issue vs torch 2.4.1 incompatibility." }, { - "left": "issue:43381", - "right": "issue:45593", + "left": "issue:45335", + "right": "issue:45584", "accept": false, - "reason": "Gradient checkpointing eval-mode issue vs D-FINE auxiliary-loss behavior when denoising is off." + "reason": "Different code paths: resize_token_embeddings not updating t5gemma embeddings vs Whisper failing on empty transcription after align_special_tokens." }, { - "left": "issue:43673", - "right": "issue:44336", + "left": "issue:46082", + "right": "issue:46399", "accept": false, - "reason": "GenerationMixin cache/chunked_prefill bug vs ANSI escape codes in loading_report." + "reason": "Separate issues: LlamaConfig head_dim validation vs Qwen3.5 use_kernels rotary registration crash." }, { - "left": "issue:43901", - "right": "issue:44496", + "left": "issue:44279", + "right": "issue:45405", "accept": false, - "reason": "Docs mismatch for text-classification pipeline vs model/config loading error; unrelated." + "reason": "Dependency/version complaints, but not the same bug: generic transformers dependency issue vs an unsupported MIN_PEFT_VERSION bump." }, { - "left": "issue:43976", - "right": "issue:44279", + "left": "issue:43577", + "right": "issue:45216", "accept": false, - "reason": "Specific Python version incompatibility vs generic dependency issue." + "reason": "Different failures: BLIP2 dtype propagation bug vs Qwen3.5 save_pretrained regression." }, { - "left": "issue:43441", - "right": "issue:45588", + "left": "issue:44075", + "right": "issue:44368", "accept": false, - "reason": "Both are in flash_attention, but one is a Ministral flash-attention crash and the other is an s_aux=None AttributeError for sink-less models." + "reason": "Unrelated: SGD optimizer args ignored vs a Qwen3.5 LoRA warning about tie_word_embeddings." }, { - "left": "issue:44016", - "right": "issue:44062", + "left": "issue:39692", + "right": "issue:43901", "accept": false, - "reason": "Notebook syntax error vs AddedToken keyword argument conflict." + "reason": "Both are docs issues, but one is a SigLIP2 example error and the other is stale TextClassificationPipeline docs about return_all_scores." }, { - "left": "issue:44496", - "right": "issue:45759", + "left": "issue:43295", + "right": "issue:44779", "accept": false, - "reason": "Unrecognized model/config error vs AutoModelForCausalLM.from_config not unwrapping text_config for composite Qwen configs." + "reason": "Different tokenizer/processor problems: processor.tokenizer regression vs Deepseek tokenizer output corruption." }, { - "left": "issue:44792", - "right": "issue:46424", + "left": "issue:43976", + "right": "issue:45405", "accept": false, - "reason": "Janus image-generation test failure vs cache max length only honored for quantized cache." + "reason": "Different topics: Python 3.9+ support regression vs a PEFT minimum-version pin issue." }, { - "left": "issue:46123", - "right": "issue:46369", + "left": "issue:42796", + "right": "issue:43650", "accept": false, - "reason": "Partial-batch is_last bookkeeping bug vs SAM3-lite-text memory overflow." + "reason": "No relation: tokenizer BPE marker handling vs an empty/nonspecific issue titled 'ADD THE DATA'." }, { - "left": "issue:43824", - "right": "issue:44933", + "left": "issue:44351", + "right": "issue:45278", "accept": false, - "reason": "Missing Qwen2.5-VL import vs nonexistent image_utils import; different symbols and paths." + "reason": "Import failures, but not the same concrete problem: missing HybridCache export vs broad import errors after upgrade." }, { - "left": "issue:44521", - "right": "issue:46693", + "left": "issue:46082", + "right": "issue:46512", "accept": false, - "reason": "Multimodal assistant_mask generation bug vs FA2 long-context performance regression." + "reason": "Different behavior: config validation for head_dim vs from_config leaving model.config.dtype stale." }, { "left": "issue:43299", - "right": "issue:45216", - "accept": false, - "reason": "Qwen3VL MoE loading regression vs Qwen3.5 save_pretrained checkpoint corruption." - }, - { - "left": "issue:44279", - "right": "issue:46515", + "right": "issue:45357", "accept": false, - "reason": "Generic dependency issue vs torch 2.4.1 incompatibility; too vague and not the same bug." + "reason": "Separate Qwen regressions: loading Qwen3VL MoE vs saving incorrect visual encoder keys for Qwen3.5." }, { - "left": "issue:45335", - "right": "issue:45584", + "left": "issue:44779", + "right": "issue:45812", "accept": false, - "reason": "T5Gemma embedding resize bug vs Whisper empty-transcription failure after align_special_tokens." + "reason": "Both tokenizer regressions, but for different model families and different symptoms: Deepseek tokenization vs Granite token IDs." }, { - "left": "issue:46082", - "right": "issue:46399", + "left": "issue:44242", + "right": "issue:45200", "accept": false, - "reason": "LlamaConfig head_dim validation vs Qwen3.5 use_kernels registration/decorator crash." + "reason": "Different architecture bugs: missing load-balancing loss vs Gemma 4 mm_token_type_ids defaulting." }, { - "left": "issue:44279", - "right": "issue:45405", + "left": "issue:41950", + "right": "issue:45020", "accept": false, - "reason": "Generic dependency issue vs PEFT minimum-version bump to an unreleased version." + "reason": "Unrelated: video-classification pipeline using image processors vs remote_code models breaking in recent versions." }, { - "left": "issue:43577", - "right": "issue:45216", + "left": "issue:43299", + "right": "issue:43854", "accept": false, - "reason": "BLIP2 dtype propagation issue vs Qwen3.5 save_pretrained visual-key regression." + "reason": "Different models and failure modes: Qwen3VL loading regression vs GLM-4.7-Flash failing in unit tests." }, { - "left": "issue:44075", - "right": "issue:44368", + "left": "issue:43479", + "right": "issue:43525", "accept": false, - "reason": "SGD args being ignored vs Qwen warning about tie_word_embeddings; unrelated." + "reason": "Different config bugs: Phi4MultimodalConfig default handling vs Llama4Config missing pad_token_id." }, { - "left": "issue:39692", - "right": "issue:43901", + "left": "issue:44466", + "right": "issue:45276", "accept": false, - "reason": "SigLIP2 example errors vs pipeline docs mentioning return_all_scores; both docs-related but different topics." + "reason": "Different serialization vs embedding-resize issues: lm_head.weight tied-weight inconsistency vs Gemma4 resize_token_embeddings not propagating." }, { - "left": "issue:43295", - "right": "issue:44779", + "left": "issue:41950", + "right": "issue:45823", "accept": false, - "reason": "Processor/tokenizer regression for multimodal custom code vs Deepseek tokenizer output regression." + "reason": "Different subsystems: pipeline processor lookup vs Gemma4 PLE device mismatch during forward." }, { - "left": "issue:42796", - "right": "issue:43650", + "left": "issue:43650", + "right": "issue:46209", "accept": false, - "reason": "Tokenizer BPE-marker handling bug vs unrelated placeholder issue titled 'ADD THE DATA'." + "reason": "No overlap: a vague 'ADD THE DATA' issue vs FP8 inference failure on B200." }, { - "left": "issue:44351", - "right": "issue:45278", + "left": "issue:43653", + "right": "issue:45593", "accept": false, - "reason": "HybridCache import missing vs broad import errors after an update; not the same concrete failure." + "reason": "Different model bugs: BigBirdTokenizer special-token registration vs D-FINE auxiliary losses with denoising off." }, { - "left": "issue:46082", - "right": "issue:46512", + "left": "issue:44683", + "right": "issue:45542", "accept": false, - "reason": "LlamaConfig head_dim rejection vs from_config(dtype=...) leaving config.dtype stale." + "reason": "Unrelated runtime issues: compiled flex_attention on torch>=2.9 vs tensorboard-only install causing TF backend errors." }, { - "left": "issue:45357", - "right": "issue:45584", + "left": "issue:44462", + "right": "issue:45081", "accept": false, - "reason": "Qwen3.5 save_pretrained visual-key regression vs Whisper generation failure after align_special_tokens." + "reason": "Both involve tokenizers, but one is AutoTokenizer ignoring tokenizer.json and the other is Mistral regex patching crashing on backend_tokenizer access." }, { - "left": "issue:44779", - "right": "issue:45812", + "left": "issue:44016", + "right": "issue:44291", "accept": false, - "reason": "Deepseek tokenizer regression vs Granite tokenizer ID regression; different tokenizer/model families and failure modes." + "reason": "Completely different: a notebook syntax error vs init_empty_weights passing an unexpected _is_hf_initialized argument." } ] }, @@ -11834,52 +11942,50 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 38, - "review_unit_count": 56, + "review_unit_index": 39, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:33290", - "issue:41950", + "issue:41628", "issue:42175", "issue:42994", - "issue:43299", "issue:43317", "issue:43441", - "issue:43479", "issue:43525", - "issue:43650", - "issue:43653", + "issue:43643", "issue:43673", "issue:43723", "issue:43761", - "issue:43854", + "issue:43824", + "issue:43881", "issue:43901", "issue:43976", "issue:43994", "issue:44016", "issue:44162", - "issue:44242", + "issue:44188", + "issue:44246", "issue:44261", "issue:44263", - "issue:44291", + "issue:44297", + "issue:44336", "issue:44361", - "issue:44462", - "issue:44466", + "issue:44373", + "issue:44589", "issue:44623", "issue:44671", "issue:44683", "issue:44749", "issue:44779", - "issue:45020", "issue:45081", "issue:45103", - "issue:45200", - "issue:45276", + "issue:45362", "issue:45372", "issue:45397", "issue:45542", @@ -11887,28 +11993,19 @@ "issue:45600", "issue:45656", "issue:45712", + "issue:45800", "issue:45823", "issue:45834", "issue:45907", - "issue:46209" + "issue:46291", + "issue:46399" ], "soft_pairs": [ - "issue:44242|issue:45200", - "issue:41950|issue:45020", - "issue:43299|issue:43854", - "issue:43479|issue:43525", - "issue:43650|issue:46209", - "issue:44466|issue:45276", - "issue:41950|issue:45823", - "issue:43653|issue:45593", - "issue:44683|issue:45542", - "issue:44462|issue:45081", - "issue:44016|issue:44291", "issue:43441|issue:45834", "issue:45103|issue:45372", "issue:44361|issue:45593", - "issue:45081|issue:45907", "issue:33290|issue:45656", + "issue:45081|issue:45907", "issue:44261|issue:44263", "issue:42994|issue:43317", "issue:45600|issue:45712", @@ -11919,198 +12016,216 @@ "issue:43901|issue:43994", "issue:43441|issue:44683", "issue:43673|issue:43761", - "issue:43723|issue:44162" + "issue:43723|issue:44162", + "issue:43901|issue:44246", + "issue:46291|issue:46399", + "issue:45542|issue:45800", + "issue:43525|issue:43881", + "issue:43824|issue:43976", + "issue:44016|issue:45823", + "issue:44188|issue:44336", + "issue:43525|issue:43643", + "issue:44336|issue:44373", + "issue:44589|issue:45362", + "issue:41628|issue:43976", + "issue:44297|issue:45397" ], - "prepared_review_unit_hash": "dd48051f5f41812d33b4127dd8e7f9b8bb2d431dfb4df6a178a068772e455f29", + "prepared_review_unit_hash": "9f41bc2fd015c20deb9f90ecf2d62050f8aedd9a574ef01b434849302d46bf8d", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 27, - "serialized_chars": 35795, - "estimated_input_tokens": 8949, - "estimated_eval_tokens": 18154 + "soft_pair_count": 28, + "serialized_chars": 35990, + "estimated_input_tokens": 8998, + "estimated_eval_tokens": 18252 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "No duplicate cluster emerges here: the items span unrelated bugs, regressions, docs, and dependency issues. I rejected every soft edge as a false match.", - "confidence": 0.92, - "canonical_issue_reason": null, + "summary": "Mostly unrelated issue reports with a loose Transformers v5/regression theme; none of the soft pairs look like the same underlying bug or change. I\u2019d use the broad tokenizer-loading regression as the closest canonical issue.", + "confidence": 0.41, + "canonical_issue_reason": "issue:43723 is the broadest, most umbrella-like report here: a v5 tokenizer-loading failure via AutoTokenizer.from_pretrained that best represents the shared loading/regression theme.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44261 is the clearest standalone bug report: it names a specific code path (MLA q_a_layernorm), a concrete missing-config cause, and a measurable user-facing impact.", + "best_issue_reason": "issue:43723 is the strongest standalone representative because it is generic, user-facing, and closest to the cluster\u2019s recurring v5 loading/import regressions.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44242", - "right": "issue:45200", + "left": "issue:43441", + "right": "issue:45834", "accept": false, - "reason": "Different bugs and code paths: MoE load-balancing loss vs Gemma 4 text-only mm_token_type_ids defaulting." + "reason": "Different models and failure paths; not the same bug." }, { - "left": "issue:41950", - "right": "issue:45020", + "left": "issue:45103", + "right": "issue:45372", "accept": false, - "reason": "Video-classification processor lookup and remote_code regressions are unrelated subsystems." + "reason": "Docstring helper crash vs missing import in Gemma processor loading." }, { - "left": "issue:43299", - "right": "issue:43854", + "left": "issue:44361", + "right": "issue:45593", "accept": false, - "reason": "Different models and failures: Qwen3VL MoE loading vs GLM-4.7-Flash unit-test loading." + "reason": "Tokenizer AttributeError vs missing auxiliary losses; unrelated." }, { - "left": "issue:43479", - "right": "issue:43525", + "left": "issue:33290", + "right": "issue:45656", "accept": false, - "reason": "Separate config bugs: Phi4Multimodal default sub-config initialization vs Llama4 missing pad_token_id." + "reason": "Both mention deepspeed, but optimizer OOM and double-step are distinct bugs." }, { - "left": "issue:43650", - "right": "issue:46209", + "left": "issue:45081", + "right": "issue:45907", "accept": false, - "reason": "No shared underlying bug; one is an \u0905\u0938\u094d\u092a\u0937\u094d\u091f data request, the other is FP8 inference on B200." + "reason": "Tokenizer regex crash vs list_repo_files revision forwarding; unrelated." }, { - "left": "issue:44466", - "right": "issue:45276", + "left": "issue:44261", + "right": "issue:44263", "accept": false, - "reason": "Serialization of tied weights and resize_token_embeddings behavior are distinct issues." + "reason": "Different model internals and different symptoms." }, { - "left": "issue:41950", - "right": "issue:45823", + "left": "issue:42994", + "right": "issue:43317", "accept": false, - "reason": "Different model/pipeline problems: video processor lookup vs Gemma4 PLE device mismatch." + "reason": "Saving quantized models vs loading dequantized models are different code paths." }, { - "left": "issue:43653", - "right": "issue:45593", + "left": "issue:45600", + "right": "issue:45712", "accept": false, - "reason": "Tokenizer special-token registration and D-FINE auxiliary-loss behavior are unrelated." + "reason": "Different CI/repo hygiene problems, not one underlying defect." }, { - "left": "issue:44683", - "right": "issue:45542", + "left": "issue:44623", + "right": "issue:44749", "accept": false, - "reason": "Torch 2.9 flex_attention failure is unrelated to TensorBoard-installed-without-TensorFlow backend errors." + "reason": "Processor saving files issue vs performance regression; unrelated." }, { - "left": "issue:44462", - "right": "issue:45081", + "left": "issue:44623", + "right": "issue:45397", "accept": false, - "reason": "Both involve tokenizers, but one is repository tokenizer file selection and the other is a Mistral regex/backend attribute crash." + "reason": "Processor save bug vs gemma-4 Zero3 load bug; different failures." }, { - "left": "issue:44016", - "right": "issue:44291", + "left": "issue:42175", + "right": "issue:43976", "accept": false, - "reason": "Notebook syntax error and init_empty_weights argument mismatch are unrelated." + "reason": "Backend dependency packaging vs Python version incompatibility; unrelated." }, { - "left": "issue:43441", - "right": "issue:45834", + "left": "issue:44671", + "right": "issue:44779", "accept": false, - "reason": "Different model-specific failures: Ministral FlashAttention vs Kosmos2.5 OCR index error." + "reason": "Different models and different incorrect outputs." }, { - "left": "issue:45103", - "right": "issue:45372", + "left": "issue:43901", + "right": "issue:43994", "accept": false, - "reason": "auto_docstring future-annotations crash and missing ReasoningEffort import are separate bugs." + "reason": "Docs mismatch vs model/pipeline inference error; unrelated." }, { - "left": "issue:44361", - "right": "issue:45593", + "left": "issue:43441", + "right": "issue:44683", "accept": false, - "reason": "Tokenizer AttributeError on tasks is unrelated to D-FINE training-loss behavior." + "reason": "Both attention-related, but different kernels and distinct failures." }, { - "left": "issue:45081", - "right": "issue:45907", + "left": "issue:43673", + "right": "issue:43761", "accept": false, - "reason": "Tokenizer regex patch crash and list_repo_files revision forwarding are different code paths." + "reason": "Cache regression vs CLIP hidden_states bug; different code paths." }, { - "left": "issue:33290", - "right": "issue:45656", + "left": "issue:43723", + "right": "issue:44162", "accept": false, - "reason": "Both mention deepspeed/optimizer, but OOM with Adafactor and double optimizer.step() are different failures." + "reason": "Tokenizer loading issue vs ESM2 model breakage; not the same bug." }, { - "left": "issue:44261", - "right": "issue:44263", + "left": "issue:43901", + "right": "issue:44246", "accept": false, - "reason": "Completely different components: MLA layernorm epsilon handling vs GlmMoeDsaIndexer split outputs." + "reason": "Documentation issue vs slow import issue; unrelated." }, { - "left": "issue:42994", - "right": "issue:43317", + "left": "issue:46291", + "right": "issue:46399", "accept": false, - "reason": "Saving a quantized model and loading a dequantized model with device_map=auto are different stages/problems." + "reason": "Both kernels-related, but import requirement failure and missing decorator are separate issues." }, { - "left": "issue:45600", - "right": "issue:45712", + "left": "issue:45542", + "right": "issue:45800", "accept": false, - "reason": "Both are repo-maintenance/CI-ish, but they concern different files and defects." + "reason": "TensorBoard/TensorFlow backend packaging vs torch/transformers incompatibility; unrelated." }, { - "left": "issue:44623", - "right": "issue:44749", + "left": "issue:43525", + "right": "issue:43881", "accept": false, - "reason": "Processor save_pretrained missing files is unrelated to the reported slowdown after upgrading Transformers." + "reason": "Different model loading problems with different root causes." }, { - "left": "issue:44623", - "right": "issue:45397", + "left": "issue:43824", + "right": "issue:43976", "accept": false, - "reason": "Missing processor save files and gemma-4 zero3 from_pretrained are distinct issues." + "reason": "Missing export/import error vs Python version support issue; unrelated." }, { - "left": "issue:42175", - "right": "issue:43976", + "left": "issue:44016", + "right": "issue:45823", "accept": false, - "reason": "Package extra missing TensorFlow backend and Python-version compatibility are different dependency problems." + "reason": "Notebook syntax error vs device mismatch in Gemma4; unrelated." }, { - "left": "issue:44671", - "right": "issue:44779", + "left": "issue:44188", + "right": "issue:44336", "accept": false, - "reason": "Different models and symptoms: CamemBERT masked-LM predictions vs Deepseek tokenizer output correctness." + "reason": "Attention divergence bug vs ANSI logging output; unrelated." }, { - "left": "issue:43901", - "right": "issue:43994", + "left": "issue:43525", + "right": "issue:43643", "accept": false, - "reason": "Documentation mismatch for return_all_scores is unrelated to SigLIP2 model output quality." + "reason": "Missing config field in one model vs remote-code field loss in another path." }, { - "left": "issue:43441", - "right": "issue:44683", + "left": "issue:44336", + "right": "issue:44373", "accept": false, - "reason": "Model-specific FlashAttention regression and torch>=2.9 compiled flex_attention failure are not the same bug." + "reason": "Logging ANSI codes vs docstring text; unrelated." }, { - "left": "issue:43673", - "right": "issue:43761", + "left": "issue:44589", + "right": "issue:45362", "accept": false, - "reason": "Generation cache missing in chunked_prefill and CLIPVisionModel hidden_states regression are unrelated." + "reason": "Storage type import failure vs chat runtime crash; different bugs." }, { - "left": "issue:43723", - "right": "issue:44162", + "left": "issue:41628", + "right": "issue:43976", + "accept": false, + "reason": "Import error for AutoImageProcessor vs Python version incompatibility; unrelated." + }, + { + "left": "issue:44297", + "right": "issue:45397", "accept": false, - "reason": "AutoTokenizer loading regression and ESM2 breakage are different underlying problems." + "reason": "Tokenizer save metadata mismatch vs gemma-4 Zero3 loading failure; different issues." } ] }, @@ -12125,77 +12240,65 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 39, - "review_unit_count": 56, + "review_unit_index": 40, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:36246", - "issue:41628", "issue:41950", "issue:42491", "issue:42757", "issue:42886", + "issue:43257", "issue:43299", - "issue:43525", + "issue:43475", "issue:43526", "issue:43531", + "issue:43540", "issue:43550", - "issue:43643", + "issue:43673", "issue:43784", - "issue:43824", - "issue:43881", "issue:43901", - "issue:43976", - "issue:44016", - "issue:44188", "issue:44220", "issue:44242", "issue:44246", "issue:44295", - "issue:44297", "issue:44336", "issue:44351", - "issue:44373", + "issue:44479", "issue:44485", "issue:44496", - "issue:44589", + "issue:44561", + "issue:44734", + "issue:44779", + "issue:44857", "issue:44871", "issue:45003", "issue:45103", + "issue:45137", + "issue:45292", "issue:45362", "issue:45397", "issue:45542", "issue:45561", "issue:45563", "issue:45593", - "issue:45800", - "issue:45823", + "issue:45859", "issue:46132", - "issue:46291", + "issue:46369", "issue:46399", "issue:46424", "issue:46506", "issue:46614", + "issue:46655", "issue:46665" ], "soft_pairs": [ - "issue:43901|issue:44246", - "issue:46291|issue:46399", - "issue:43525|issue:43881", - "issue:45542|issue:45800", - "issue:44016|issue:45823", - "issue:43824|issue:43976", - "issue:44188|issue:44336", - "issue:43525|issue:43643", - "issue:44589|issue:45362", - "issue:44336|issue:44373", - "issue:41628|issue:43976", - "issue:44297|issue:45397", "issue:43784|issue:45542", "issue:44246|issue:44496", "issue:45561|issue:45563", @@ -12211,218 +12314,236 @@ "issue:42886|issue:46132", "issue:44336|issue:46424", "issue:43550|issue:46506", - "issue:41950|issue:44220", "issue:36246|issue:42757", - "issue:36246|issue:44351" + "issue:41950|issue:44220", + "issue:36246|issue:44351", + "issue:44779|issue:45292", + "issue:45859|issue:46399", + "issue:45003|issue:45561", + "issue:44734|issue:46424", + "issue:43901|issue:45003", + "issue:43540|issue:46655", + "issue:43257|issue:45137", + "issue:43673|issue:44561", + "issue:44479|issue:46655", + "issue:44496|issue:45003", + "issue:43475|issue:46369", + "issue:44242|issue:44857" ], - "prepared_review_unit_hash": "93c60e3cd7c321bc86474df588fbfacdfb11e7c1f8b4f0eb9f6889c863adae3b", + "prepared_review_unit_hash": "45488d8feb6367647cba632443c8782e1bfb36c16561b9b4482634cf0184dd2d", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36587, - "estimated_input_tokens": 9147, - "estimated_eval_tokens": 18550 + "serialized_chars": 36589, + "estimated_input_tokens": 9148, + "estimated_eval_tokens": 18552 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is not a true duplicate set: it mixes unrelated import errors, model-loading bugs, tokenizer/cache issues, docs, and test/utilities problems. None of the soft-edge pairs look like the same underlying bug or change.", - "confidence": 0.94, - "canonical_issue_reason": null, + "summary": "This cluster is heterogeneous: most items are distinct bug reports across unrelated subsystems, and none of the soft pairs look like the same underlying issue. The broadest video-regression report is the best representative, but not a duplicate match for the others.", + "confidence": 0.71, + "canonical_issue_reason": "issue:44479 is the broadest, most umbrella-like report in the set: it covers a v5.3.0 video-input regression across multiple Qwen VL/VL-MoE variants, making it the strongest canonical issue candidate among these unrelated reports.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a good canonical representative; the cluster is too heterogeneous, so any choice would collapse distinct bugs into one bucket.", + "best_issue_reason": "issue:44479 is the most general and centrally phrased issue here, with the widest model coverage and clearest regression framing; it best represents the only loosely related theme in the set.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43901", - "right": "issue:44246", + "left": "issue:43784", + "right": "issue:45542", "accept": false, - "reason": "Docs/behavior mismatch for text classification vs an intermittent import-time performance complaint; different problems." + "reason": "Different import failures in different packages; no shared code path or concrete bug." }, { - "left": "issue:46291", - "right": "issue:46399", + "left": "issue:44246", + "right": "issue:44496", "accept": false, - "reason": "One is a kernels dependency/version import failure; the other is a Qwen3.5 registration/decorator crash. Different code paths." + "reason": "One is an import-performance complaint, the other is model/config loading; unrelated." }, { - "left": "issue:43525", - "right": "issue:43881", + "left": "issue:45561", + "right": "issue:45563", "accept": false, - "reason": "Different model families and failures: Llama4 missing pad_token_id vs GLM-4V loading failure." + "reason": "Different bugs: xdist race on captured_info.txt vs a stale warning in paged generation." }, { - "left": "issue:45542", - "right": "issue:45800", + "left": "issue:46614", + "right": "issue:46665", "accept": false, - "reason": "Tensorboard-without-TensorFlow backend error is unrelated to torch/transformers version incompatibility." + "reason": "Different tokenizer/cache behavior versus from_pretrained behavior in Python classes; not the same defect." }, { - "left": "issue:44016", - "right": "issue:45823", + "left": "issue:45103", + "right": "issue:46399", "accept": false, - "reason": "Notebook syntax error is unrelated to Gemma4 PLE device-mapping forward mismatch." + "reason": "Unrelated areas: auto-docstring annotation handling vs Qwen3.5 kernel registration crash." }, { - "left": "issue:43824", - "right": "issue:43976", + "left": "issue:44242", + "right": "issue:45593", "accept": false, - "reason": "Qwen2.5-VL symbol import failure is not the same as a Python version compatibility break." + "reason": "Both mention missing auxiliary losses, but they affect different models and mechanisms; too broad to be the same bug." }, { - "left": "issue:44188", - "right": "issue:44336", + "left": "issue:42491", + "right": "issue:43299", "accept": false, - "reason": "Torch.compile attention-kernel branching bug vs ANSI codes in loading_report; unrelated subsystems." + "reason": "Both involve Qwen3 MoE loading/version compatibility, but the concrete failures and code paths differ." }, { - "left": "issue:43525", - "right": "issue:43643", + "left": "issue:43526", + "right": "issue:45003", "accept": false, - "reason": "Llama4 pad_token_id attribute error vs trust_remote_code missing fields; distinct config-loading issues." + "reason": "BeitImageProcessorFast label reduction vs sys.modules access in modeling_utils are unrelated." }, { - "left": "issue:44589", + "left": "issue:43531", "right": "issue:45362", "accept": false, - "reason": "Float8 storage lookup failure is unrelated to Qwen3.5 chat crashing." + "reason": "Same model family only at a high level; sliding-window behavior and chat crash are different issues." + }, + { + "left": "issue:44295", + "right": "issue:45397", + "accept": false, + "reason": "Different subsystems and symptoms: position_ids buffer access vs Gemma-4 ZeRO-3 from_pretrained." }, { "left": "issue:44336", - "right": "issue:44373", + "right": "issue:44485", "accept": false, - "reason": "Terminal ANSI output bug vs a wrong docstring; not the same bug." + "reason": "Loading-report ANSI output is unrelated to GLM-5 RoPE implementation." }, { - "left": "issue:41628", - "right": "issue:43976", + "left": "issue:44246", + "right": "issue:44871", "accept": false, - "reason": "AutoImageProcessor import failure and Python 3.9/3.10 compatibility break are different issues." + "reason": "Import slowness and Gemma eos_token_id mismatch are unrelated." }, { - "left": "issue:44297", - "right": "issue:45397", + "left": "issue:42886", + "right": "issue:46132", "accept": false, - "reason": "Tokenizer config mismatch is unrelated to Gemma-4 zero3 from_pretrained failure." + "reason": "Offline tokenizer cache loading vs AttentionInterface.register behavior change are unrelated." }, { - "left": "issue:43784", - "right": "issue:45542", + "left": "issue:44336", + "right": "issue:46424", "accept": false, - "reason": "sentence-transformers import NameError vs a missing TensorFlow backend error; different failure modes." + "reason": "ANSI handling in loading_report has nothing to do with cache max length behavior." }, { - "left": "issue:44246", - "right": "issue:44496", + "left": "issue:43550", + "right": "issue:46506", "accept": false, - "reason": "Import-performance complaint is unrelated to an unrecognized-model/config.json loading error." + "reason": "Different model/runtime failures: torch.compile+SDPA vs CPU/CUDA numerical divergence." }, { - "left": "issue:45561", - "right": "issue:45563", + "left": "issue:36246", + "right": "issue:42757", "accept": false, - "reason": "pytest-xdist file race vs stale generate() warning; different test/runtime behaviors." + "reason": "Both are import errors, but for unrelated symbols in different libraries." }, { - "left": "issue:46614", - "right": "issue:46665", + "left": "issue:41950", + "right": "issue:44220", "accept": false, - "reason": "Stale phonemizer language state vs from_pretrained behavior inside Python classes; unrelated state bugs." + "reason": "Video-classification processor lookup and fbank feature extraction are different bugs." }, { - "left": "issue:45103", - "right": "issue:46399", + "left": "issue:36246", + "right": "issue:44351", "accept": false, - "reason": "auto_docstring annotations crash is unrelated to the Qwen3.5 use_kernels registration bug." + "reason": "Different missing imports from transformers; no evidence of the same symbol or code path." }, { - "left": "issue:44242", - "right": "issue:45593", + "left": "issue:44779", + "right": "issue:45292", "accept": false, - "reason": "Missing load-balancing loss and missing auxiliary losses are different model-training conditions, not the same bug." + "reason": "Tokenizer regression and embedding-resize behavior are different defects." }, { - "left": "issue:42491", - "right": "issue:43299", + "left": "issue:45859", + "right": "issue:46399", "accept": false, - "reason": "Both mention Qwen3 MoE, but one is LoRA hf4.x vs hf5.x incompatibility and the other is a sliding_window loading issue; not the same bug." + "reason": "Both are Qwen3.5 rotary-related, but one is CPU offload incompatibility and the other is a kernel registration/decorator crash; not mergeable as one fix." }, { - "left": "issue:43526", - "right": "issue:45003", + "left": "issue:45003", + "right": "issue:45561", "accept": false, - "reason": "BeitImageProcessorFast label reduction bug vs unsafe sys.modules access in modeling_utils; unrelated code paths." + "reason": "Sys.modules access and a pytest-xdist file race are unrelated." }, { - "left": "issue:43531", - "right": "issue:45362", + "left": "issue:44734", + "right": "issue:46424", "accept": false, - "reason": "Qwen3-MoE sliding_window issue is unrelated to Qwen3.5 chat crashes." + "reason": "KV-cache continuation indexing and cache max-length configuration are distinct cache bugs." }, { - "left": "issue:44295", - "right": "issue:45397", + "left": "issue:43901", + "right": "issue:45003", "accept": false, - "reason": "position_ids buffer access error vs Gemma-4 zero3 from_pretrained failure; different mechanisms." + "reason": "Docs/behavior mismatch in TextClassificationPipeline is unrelated to sys.modules access." }, { - "left": "issue:44336", - "right": "issue:44485", + "left": "issue:43540", + "right": "issue:46655", "accept": false, - "reason": "ANSI-code terminal output bug vs GLM-5 RoPE implementation discussion; unrelated." + "reason": "Both are video-input failures, but they affect different models and likely different processors; too broad to treat as one bug." }, { - "left": "issue:44246", - "right": "issue:44871", + "left": "issue:43257", + "right": "issue:45137", "accept": false, - "reason": "Import-time performance issue is unrelated to Gemma-3 eos_token_id inconsistency." + "reason": "DeepSpeed-related issues, but one is Qwen3 MoE weight conversion and the other is a ZeRO3 deque error." }, { - "left": "issue:42886", - "right": "issue:46132", + "left": "issue:43673", + "right": "issue:44561", "accept": false, - "reason": "Offline tokenizer cache loading bug vs AttentionInterface.register changing function behavior; different areas." + "reason": "Generation cache/chunked_prefill issue and trust_remote_code FX helper removal are unrelated." }, { - "left": "issue:44336", - "right": "issue:46424", + "left": "issue:44479", + "right": "issue:46655", "accept": false, - "reason": "Loading-report ANSI codes and cache max length semantics are unrelated." + "reason": "Both are video regressions, but 44479 is a broad Qwen VL-family regression while 46655 is a SmolVLM padding bug; different concrete fixes." }, { - "left": "issue:43550", - "right": "issue:46506", + "left": "issue:43475", + "right": "issue:46369", "accept": false, - "reason": "Bamba torch.compile SDPA failure is unrelated to PPDocLayoutV3 CPU/CUDA divergence." + "reason": "SAM3 attribute error and SAM3 memory overflow are different failure modes." }, { - "left": "issue:41950", - "right": "issue:44220", + "left": "issue:44496", + "right": "issue:45003", "accept": false, - "reason": "Video-classification pipeline using image processors vs _torch_extract_fbank_features audio bug; different pipeline paths." + "reason": "Unrecognized model/config parsing and sys.modules access are unrelated." }, { - "left": "issue:36246", - "right": "issue:42757", + "left": "issue:44734", + "right": "issue:46424", "accept": false, - "reason": "Qwen2.5-VL import symbol failure vs huggingface_hub is_offline_mode import failure; different missing names/packages." + "reason": "Cache continuation crash and quantized-cache max length handling are not the same bug." }, { - "left": "issue:36246", - "right": "issue:44351", + "left": "issue:44242", + "right": "issue:44857", "accept": false, - "reason": "Qwen2.5-VL image processor import failure is unrelated to HybridCache import failure." + "reason": "Load-balancing loss omission and AMP/CUDA crash in LwDetrImageLoss are unrelated." } ] }, @@ -12437,297 +12558,289 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 40, - "review_unit_count": 56, + "review_unit_index": 41, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:36246", - "issue:43257", + "issue:42886", + "issue:43065", "issue:43262", "issue:43352", + "issue:43408", "issue:43475", "issue:43540", "issue:43596", - "issue:43673", - "issue:43901", + "issue:43653", + "issue:43931", "issue:44038", "issue:44060", "issue:44077", "issue:44162", - "issue:44242", "issue:44246", "issue:44295", + "issue:44368", "issue:44466", "issue:44479", - "issue:44496", "issue:44561", - "issue:44734", "issue:44749", - "issue:44779", - "issue:44857", "issue:44877", + "issue:44977", "issue:44987", - "issue:45003", + "issue:44995", "issue:45137", "issue:45200", + "issue:45229", "issue:45250", "issue:45292", "issue:45335", "issue:45375", + "issue:45397", "issue:45412", "issue:45459", - "issue:45561", + "issue:45563", "issue:45584", + "issue:45588", "issue:45600", + "issue:45656", "issue:45701", "issue:45759", - "issue:45859", "issue:46123", "issue:46170", - "issue:46369", "issue:46399", - "issue:46424", - "issue:46506", - "issue:46655" + "issue:46506" ], "soft_pairs": [ - "issue:44779|issue:45292", - "issue:45859|issue:46399", - "issue:45003|issue:45561", - "issue:44734|issue:46424", - "issue:43540|issue:46655", - "issue:43901|issue:45003", - "issue:43257|issue:45137", - "issue:44479|issue:46655", - "issue:43673|issue:44561", - "issue:44496|issue:45003", - "issue:43475|issue:46369", - "issue:44242|issue:44857", "issue:44162|issue:45200", "issue:44295|issue:44749", "issue:45375|issue:45759", "issue:43262|issue:44246", "issue:45292|issue:45584", - "issue:44561|issue:46399", - "issue:44877|issue:45375", "issue:45292|issue:45701", + "issue:44561|issue:46399", "issue:45137|issue:45600", + "issue:44877|issue:45375", "issue:36246|issue:43596", "issue:44038|issue:46506", "issue:44466|issue:45335", "issue:44987|issue:45459", - "issue:45412|issue:46123", "issue:43352|issue:45250", + "issue:45412|issue:46123", "issue:44479|issue:46170", - "issue:44060|issue:44077" + "issue:44060|issue:44077", + "issue:44995|issue:45200", + "issue:43653|issue:44162", + "issue:44368|issue:44977", + "issue:43475|issue:43540", + "issue:42886|issue:45229", + "issue:45397|issue:45656", + "issue:44877|issue:46170", + "issue:44995|issue:45588", + "issue:43065|issue:43408", + "issue:44368|issue:45563", + "issue:43931|issue:44368" ], - "prepared_review_unit_hash": "ea83c22ef4215f75f17a1ee1c4959938dbafd06bcd66ed1fe4e1c52899f7b0a3", + "prepared_review_unit_hash": "c96112fd6aa3e4edf6e4dc35f3f350ee601513ddbe0b8de3e10e8f038e0d9211", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36478, - "estimated_input_tokens": 9120, - "estimated_eval_tokens": 18496 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35602, + "estimated_input_tokens": 8901, + "estimated_eval_tokens": 18058 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is heterogeneous: the soft pairs mostly share only broad subsystem keywords (video, cache, config, DeepSpeed, embeddings) but not the same concrete bug or code path, so none should be merged as duplicates. The strongest standalone issue is the Qwen2_5_VL/Qwen3_VL video regression, which is broad, actionable, and clearly scoped.", - "confidence": 0.91, - "canonical_issue_reason": "Issue 44479 is the broadest and most concrete bug report in the set: it describes a specific v5.3.0 video-input regression across several Qwen VL variants with a clear failure mode.", + "summary": "These items are mostly distinct bugs across unrelated models/subsystems; none of the soft pairs look like the same underlying issue, so all soft edges are rejected. The Gemma4 multi-GPU CUDA OOM report is the strongest anchor because it has the highest discussion activity and clear user impact.", + "confidence": 0.9, + "canonical_issue_reason": "issue:45229 has the highest discussion activity and inbound reference count in the set, with a concrete, reproducible failure and broad user impact.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44479 is the best representative issue because it names the affected model families, version range, and regression class clearly, making it the most useful anchor for triage.", + "best_issue_reason": "issue:45229 is the best overall representative of the cluster because it is the most active, clearly described, and high-impact issue among otherwise unrelated reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44779", - "right": "issue:45292", - "accept": false, - "reason": "Different bugs: tokenizer output mismatch vs resize_token_embeddings not updating output embeddings." - }, - { - "left": "issue:45859", - "right": "issue:46399", + "left": "issue:44162", + "right": "issue:45200", "accept": false, - "reason": "Both involve Qwen3.5, but the failures are different code paths: CPU offload rotary embedding vs kernel registration for apply_rotary_pos_emb." + "reason": "ESM2 breakage and Gemma4 mm_token_type_ids defaulting are different models and different failure modes." }, { - "left": "issue:45003", - "right": "issue:45561", + "left": "issue:44295", + "right": "issue:44749", "accept": false, - "reason": "Unrelated topics: sys.modules access in modeling_utils vs pytest-xdist file race in testing utilities." + "reason": "One is a buffer/position_ids error; the other is a tokenization slowdown regression after version upgrade." }, { - "left": "issue:44734", - "right": "issue:46424", + "left": "issue:45375", + "right": "issue:45759", "accept": false, - "reason": "Both mention cache behavior, but one is a tensor-indexing crash in /v1/responses continuation and the other is max_cache_length handling for quantized cache only." + "reason": "Missing Qwen3.5MoeVisionConfig field vs AutoModelForCausalLM text_config unwrapping are separate config-path bugs." }, { - "left": "issue:43540", - "right": "issue:46655", + "left": "issue:43262", + "right": "issue:44246", "accept": false, - "reason": "Both are video-processing issues, but they affect different models and different failure points (Qwen3OmniMoe input processing vs SmolVLM padding calculation)." + "reason": "Audio chat-template sampling-rate handling is unrelated to intermittent import latency." }, { - "left": "issue:43901", - "right": "issue:45003", + "left": "issue:45292", + "right": "issue:45584", "accept": false, - "reason": "Docs mismatch for TextClassificationPipeline is unrelated to unsafe sys.modules access." + "reason": "resize_token_embeddings mismatch is not the same as Whisper empty-transcription failure after align_special_tokens." }, { - "left": "issue:43257", - "right": "issue:45137", + "left": "issue:45292", + "right": "issue:45701", "accept": false, - "reason": "Both involve DeepSpeed, but one is Qwen3 MoE weight conversion and the other is a Zero3 deque underflow." + "reason": "Embedding resize behavior and tokenization changes are different bugs." }, { - "left": "issue:44479", - "right": "issue:46655", + "left": "issue:44561", + "right": "issue:46399", "accept": false, - "reason": "Video-related, but not the same bug: Qwen VL video regression across several models vs SmolVLM padding error." + "reason": "Removed helper breaking trust_remote_code is unrelated to Qwen3.5 kernel registration/decorator failure." }, { - "left": "issue:43673", - "right": "issue:44561", + "left": "issue:45137", + "right": "issue:45600", "accept": false, - "reason": "Different areas: chunked_prefill cache handling vs removal of is_torch_fx_available breaking trust_remote_code models." + "reason": "DeepSpeed ZeRO3 deque crash and stale auto_mappings config references do not share a concrete code path." }, { - "left": "issue:44496", - "right": "issue:45003", + "left": "issue:44877", + "right": "issue:45375", "accept": false, - "reason": "Unrelated: model config loading failure vs sys.modules access issue." + "reason": "Granite speech strict-config loading and Qwen3.5MoeVisionConfig field loss are distinct config issues." }, { - "left": "issue:43475", - "right": "issue:46369", + "left": "issue:36246", + "right": "issue:43596", "accept": false, - "reason": "Same SAM3 family, but distinct problems: missing output attribute vs memory overflow." + "reason": "Missing Qwen2_5_VLImageProcessor import is unrelated to the ZeRO3/BertModel index error." }, { - "left": "issue:44242", - "right": "issue:44857", + "left": "issue:44038", + "right": "issue:46506", "accept": false, - "reason": "Different model and loss bugs; no shared concrete code path." + "reason": "Qwen3-VL-Moe loading issues and PPDocLayoutV3 CPU/CUDA divergence are different subsystems and symptoms." }, { - "left": "issue:44162", - "right": "issue:45200", + "left": "issue:44466", + "right": "issue:45335", "accept": false, - "reason": "Unrelated model breakage: ESM2 regression vs Gemma 4 multimodal token IDs defaulting." + "reason": "Tied-weight serialization inconsistency is not the same bug as resize_token_embeddings not updating decoder.embed_tokens." }, { - "left": "issue:44295", - "right": "issue:44749", + "left": "issue:44987", + "right": "issue:45459", "accept": false, - "reason": "Different failure modes: position_ids buffer access error vs general slowdown after upgrade." + "reason": "Model loading failure with transformers>=5.1.0 is separate from tokenizer error masking when protobuf is absent." }, { - "left": "issue:45375", - "right": "issue:45759", + "left": "issue:43352", + "right": "issue:45250", "accept": false, - "reason": "Both concern Qwen 3.5/3.6 composite configs, but one is a missing strict-serialized field and the other is from_config not unwrapping text_config." + "reason": "A model-specific Flash Attention 2 unsupported error is not the same as a generic flash-attention topic." }, { - "left": "issue:43262", - "right": "issue:44246", + "left": "issue:45412", + "right": "issue:46123", "accept": false, - "reason": "Audio chat-template sampling-rate bug is unrelated to import performance." + "reason": "RT-DETR memory retention and MaskGenerationPipeline stale partial-batch handling are unrelated bugs." }, { - "left": "issue:45292", - "right": "issue:45584", + "left": "issue:44479", + "right": "issue:46170", "accept": false, - "reason": "resize_token_embeddings propagation bug is unrelated to Whisper empty-transcription generation failure." + "reason": "Video-input regression in Qwen models is unrelated to the Windows/ROCm torch.distributed import problem." }, { - "left": "issue:44561", - "right": "issue:46399", + "left": "issue:44060", + "right": "issue:44077", "accept": false, - "reason": "Different regressions: removed helper breaking remote-code models vs missing decorator for Qwen3.5 kernel registration." + "reason": "Incorrect tied-weights warnings and patchtsmixer post_init policy are different issues." }, { - "left": "issue:44877", - "right": "issue:45375", + "left": "issue:44995", + "right": "issue:45200", "accept": false, - "reason": "Both are config-related, but they involve different models and different strict-config behaviors." + "reason": "GlmMoeDsa stale cache on second forward pass is unrelated to Gemma4 mm_token_type_ids defaults." }, { - "left": "issue:45292", - "right": "issue:45701", + "left": "issue:43653", + "right": "issue:44162", "accept": false, - "reason": "Embedding resize bug is unrelated to tokenization changes across versions." + "reason": "BigBirdTokenizer special-token registration and ESM2 breakage are not the same underlying defect." }, { - "left": "issue:45137", - "right": "issue:45600", + "left": "issue:44368", + "right": "issue:44977", "accept": false, - "reason": "DeepSpeed Zero3 deque issue is unrelated to auto_mappings referencing removed Sam3LiteText configs." + "reason": "A tied-embeddings warning during LoRA fine-tuning is not the same as a Flash Attention generation bug." }, { - "left": "issue:36246", - "right": "issue:43596", + "left": "issue:43475", + "right": "issue:43540", "accept": false, - "reason": "ImportError for a missing Qwen2.5 VL processor is unrelated to DeepSpeed Zero3 with BertModel." + "reason": "SAM3 video encoder output attribute loss and Qwen3OmniMoe video processing errors are separate model paths." }, { - "left": "issue:44038", - "right": "issue:46506", + "left": "issue:42886", + "right": "issue:45229", "accept": false, - "reason": "Different models and problems: Qwen3-VL-Moe bug vs CPU/CUDA result discrepancy in PPDocLayoutV3." + "reason": "Offline tokenizer cache loading and Gemma4 multi-GPU CUDA OOM are unrelated." }, { - "left": "issue:44466", - "right": "issue:45335", + "left": "issue:45397", + "right": "issue:45656", "accept": false, - "reason": "Both involve embeddings/tied weights, but one is serialization inconsistency and the other is resize_token_embeddings not updating decoder.embed_tokens." + "reason": "Gemma-4 ZeRO3 from_pretrained failure and duplicate optimizer.step calls are different DeepSpeed bugs." }, { - "left": "issue:44987", - "right": "issue:45459", + "left": "issue:44877", + "right": "issue:46170", "accept": false, - "reason": "Unrelated: model loading failure vs tokenizer error masking when protobuf is absent." + "reason": "Granite_speech config strictness has nothing to do with the Windows/ROCm distributed dependency failure." }, { - "left": "issue:45412", - "right": "issue:46123", + "left": "issue:44995", + "right": "issue:45588", "accept": false, - "reason": "Memory-release behavior for RT-DETR is unrelated to MaskGenerationPipeline batch handling." + "reason": "Stale indexer cache on second forward and flash_attention AttributeError on s_aux=None are different failure modes." }, { - "left": "issue:43352", - "right": "issue:45250", + "left": "issue:43065", + "right": "issue:43408", "accept": false, - "reason": "General Flash Attention topic is too broad; one is a specific Nemotron model support error, the other is only a generic FA2 issue." + "reason": "Dummy Conv2d in Sam3PixelDecoder and model-type mismatch warning in sam3_tracker are not the same bug." }, { - "left": "issue:44479", - "right": "issue:46170", + "left": "issue:44368", + "right": "issue:45563", "accept": false, - "reason": "Video regression in Qwen VL models is unrelated to Windows/ROCm torch.distributed import failure." + "reason": "Tied-embedding warnings and paged generate stale num_return_sequences warning are unrelated." }, { - "left": "issue:44060", - "right": "issue:44077", + "left": "issue:43931", + "right": "issue:44368", "accept": false, - "reason": "Different Qwen/architecture issues: incorrect tied-weights warning vs patchtsmixer post_init validation." + "reason": "Qwen3-VL-30B weight-shape mismatch and tied-embeddings warnings are distinct issues." } ] }, @@ -12742,52 +12855,52 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 41, - "review_unit_count": 56, + "review_unit_index": 42, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ + "issue:28282", + "issue:30990", + "issue:41628", "issue:42371", - "issue:42886", "issue:42994", "issue:43065", "issue:43232", - "issue:43408", "issue:43475", "issue:43482", - "issue:43540", + "issue:43525", "issue:43643", "issue:43646", - "issue:43653", "issue:43673", "issue:43701", "issue:43906", - "issue:43931", + "issue:44038", "issue:44162", "issue:44188", "issue:44297", "issue:44361", - "issue:44368", "issue:44483", "issue:44623", "issue:44749", "issue:44869", "issue:44871", - "issue:44877", - "issue:44977", - "issue:44995", + "issue:44933", + "issue:44964", "issue:45092", - "issue:45200", - "issue:45229", + "issue:45137", + "issue:45230", + "issue:45310", "issue:45356", + "issue:45375", + "issue:45381", "issue:45397", - "issue:45563", + "issue:45478", "issue:45584", - "issue:45588", "issue:45656", "issue:45685", "issue:45987", @@ -12800,19 +12913,8 @@ "issue:46431" ], "soft_pairs": [ - "issue:44995|issue:45200", - "issue:43653|issue:44162", - "issue:44368|issue:44977", - "issue:43475|issue:43540", - "issue:45397|issue:45656", - "issue:42886|issue:45229", - "issue:44877|issue:46170", - "issue:44995|issue:45588", - "issue:43065|issue:43408", - "issue:44368|issue:45563", - "issue:43906|issue:46202", - "issue:43931|issue:44368", "issue:45987|issue:46082", + "issue:43906|issue:46202", "issue:44188|issue:46202", "issue:44483|issue:46202", "issue:45685|issue:46431", @@ -12830,222 +12932,219 @@ "issue:42994|issue:43482", "issue:44869|issue:45356", "issue:43475|issue:43643", - "issue:42371|issue:45092" + "issue:42371|issue:45092", + "issue:45375|issue:45478", + "issue:28282|issue:30990", + "issue:41628|issue:44933", + "issue:45230|issue:45397", + "issue:44038|issue:44623", + "issue:45310|issue:45375", + "issue:44964|issue:45137", + "issue:43475|issue:45381", + "issue:42994|issue:43525" ], - "prepared_review_unit_hash": "403fbfa03d30896b4878067ea341d952e4d51a7d10e128d43e31e2a8794a44ce", + "prepared_review_unit_hash": "e483375eb980969ba647bd02f321661ac0d257500ad5660946e75bb0630f6c4d", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 31, - "serialized_chars": 36387, - "estimated_input_tokens": 9097, - "estimated_eval_tokens": 18450 + "soft_pair_count": 29, + "serialized_chars": 35554, + "estimated_input_tokens": 8889, + "estimated_eval_tokens": 18034 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly a grab-bag of unrelated closed issues across tokenizers, generation, model loading, SAM3, Gemma, Qwen, and backend/runtime regressions. I did not find any soft pair that clearly describes the exact same underlying bug/change. If a single issue must represent the cluster, 43646 is the broadest v5-regression umbrella.", - "confidence": 0.67, - "canonical_issue_reason": "issue:43646 is the broadest, most general regression report in the set: Transformers 5.0.0 breaking custom model initialization. It can act as an umbrella for several v5-related breakage reports, even though the rest are largely model-specific.", + "summary": "Mostly a bag of unrelated Transformers issues. Only one clear duplicate pair stands out: the two identical Qwen3.5-Moe from_pretrained regression reports.", + "confidence": 0.86, + "canonical_issue_reason": "issue:45310 is the earliest clear report of the exact Qwen3.5-Moe from_pretrained failure and has an exact duplicate in issue:45478, making it the best canonical issue anchor.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43646 is the best single representative because it is broad, version-oriented, and closer to a cross-cutting compatibility regression than the narrower model-specific tickets.", + "best_issue_reason": "issue:45310 is the strongest representative issue because it names the concrete regression precisely and matches an exact duplicate, while the rest of the cluster is largely unrelated.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44995", - "right": "issue:45200", - "accept": false, - "reason": "Different failure modes: stale indexer cache vs. Gemma4 mm_token_type_ids defaults. No shared code path." - }, - { - "left": "issue:43653", - "right": "issue:44162", - "accept": false, - "reason": "Tokenizer special-token registration bug vs. ESM2 breakage; same broad area, but not the same defect." - }, - { - "left": "issue:44368", - "right": "issue:44977", - "accept": false, - "reason": "A warning about tie_word_embeddings in LoRA fine-tuning is unrelated to flash-attention generation failure." - }, - { - "left": "issue:43475", - "right": "issue:43540", - "accept": false, - "reason": "SAM3 vision encoder attribute error and Qwen3OmniMoe video input validation are different bugs." - }, - { - "left": "issue:45397", - "right": "issue:45656", - "accept": false, - "reason": "ZeRO/from_pretrained loading issue is unrelated to deepspeed double optimizer stepping." - }, - { - "left": "issue:42886", - "right": "issue:45229", - "accept": false, - "reason": "Offline tokenizer cache loading and Gemma4 multi-GPU OOM are unrelated." - }, - { - "left": "issue:44877", - "right": "issue:46170", - "accept": false, - "reason": "Granite speech config strictness and Windows/ROCm torch.distributed import failure are different subsystems." - }, - { - "left": "issue:44995", - "right": "issue:45588", - "accept": false, - "reason": "Stale indexer cache vs. flash_attention s_aux=None AttributeError; no common underlying bug." - }, - { - "left": "issue:43065", - "right": "issue:43408", - "accept": false, - "reason": "Dummy Conv2d implementation in Sam3PixelDecoder is not the same issue as the sam3_video/sam3_tracker config warning." - }, - { - "left": "issue:44368", - "right": "issue:45563", + "left": "issue:45987", + "right": "issue:46082", "accept": false, - "reason": "LoRA tie_word_embeddings warning is unrelated to the stale num_return_sequences warning in paged generate()." + "reason": "Different bugs: cache length return-type mismatch vs LlamaConfig head_dim validation." }, { "left": "issue:43906", "right": "issue:46202", "accept": false, - "reason": "Isolated reproduction of an external issue and Gemma pipeline roundtrip on \u2581 are unrelated." - }, - { - "left": "issue:43931", - "right": "issue:44368", - "accept": false, - "reason": "Qwen3-VL weight-shape mismatch is unrelated to the Qwen3.5 tie_word_embeddings warning." - }, - { - "left": "issue:45987", - "right": "issue:46082", - "accept": false, - "reason": "StaticCache return type mismatch and LlamaConfig head_dim validation are different problems." + "reason": "No shared code-path or symptom; one is a reproduction of an older issue, the other is a Gemma tokenizer roundtrip failure." }, { "left": "issue:44188", "right": "issue:46202", "accept": false, - "reason": "torch.compile branching causing diverging attention kernels is unrelated to Gemma roundtrip token handling." + "reason": "Different failures: torch.compile attention-kernel divergence vs Gemma roundtrip on a special token." }, { "left": "issue:44483", "right": "issue:46202", "accept": false, - "reason": "Chat completions API compatibility failure is not the same as Gemma pipeline roundtrip failure." + "reason": "Chat completions API rejection is unrelated to Gemma tokenizer roundtrip behavior." }, { "left": "issue:45685", "right": "issue:46431", "accept": false, - "reason": "MPS histogram error in moe code is unrelated to missing chat template in a processor." + "reason": "Unrelated MPS histogram error vs missing chat template in a processor." }, { "left": "issue:44188", "right": "issue:46421", "accept": false, - "reason": "Attention-kernel divergence under torch.compile and cache lazy init causing recompilation are different defects." + "reason": "Both mention compile/cache, but the concrete bugs differ: attention-kernel divergence vs spurious recompilation from lazy cache init." }, { "left": "issue:45584", "right": "issue:46170", "accept": false, - "reason": "Whisper empty-transcription failure is unrelated to the Windows/ROCm distributed import issue." + "reason": "Different subsystems and failures: Whisper empty-transcription generation vs Windows ROCm import/torch.distributed dependency issues." }, { "left": "issue:43065", "right": "issue:46123", "accept": false, - "reason": "SAM3 dummy Conv2d and MaskGenerationPipeline dropping final partial batch are unrelated." + "reason": "SAM3 pixel-decoder dummy layer issue is unrelated to MaskGenerationPipeline dropping partial-batch results." }, { "left": "issue:44297", "right": "issue:44623", "accept": false, - "reason": "Tokenizer save_pretrained metadata mismatch and processor save_pretrained missing files are both serialization regressions, but not the same code path." + "reason": "Both involve save_pretrained, but one is tokenizer config metadata and the other is processor file emission; not the same bug." }, { "left": "issue:44361", "right": "issue:44749", "accept": false, - "reason": "MLukeTokenizer task AttributeError and a 10x filtering slowdown after upgrade are unrelated." + "reason": "Tokenizer AttributeError on tasks is unrelated to the reported performance regression in data filtering." }, { "left": "issue:44749", "right": "issue:45656", "accept": false, - "reason": "Data filtering slowdown and deepspeed double optimizer stepping are unrelated." + "reason": "Performance slowdown during filtering is not the same as DeepSpeed causing optimizer.step to run twice." }, { "left": "issue:43646", "right": "issue:44162", "accept": false, - "reason": "Both mention v5 breakage, but custom model initialization and ESM2 breakage are not the same concrete bug." + "reason": "Custom config initialization/field loss is unrelated to ESM2 being broken." }, { "left": "issue:43701", "right": "issue:44749", "accept": false, - "reason": "resume_from_checkpoint key mismatch is unrelated to dataset filtering performance regression." + "reason": "Checkpoint key mismatch is a different failure mode from the filtering slowdown report." }, { "left": "issue:44623", "right": "issue:45656", "accept": false, - "reason": "Missing processor save_pretrained files and double optimizer steps are unrelated." + "reason": "Processor.save_pretrained missing files is unrelated to duplicate optimizer stepping under DeepSpeed." }, { "left": "issue:44871", "right": "issue:46132", "accept": false, - "reason": "Gemma eos_token_id mismatch is unrelated to AttentionInterface.register behavior changes." + "reason": "Gemma eos_token_id mismatch is unrelated to AttentionInterface.register changing registered-function behavior." }, { "left": "issue:43232", "right": "issue:43673", "accept": false, - "reason": "sync_gpus generation kwargs update issue and missing GenerationMixin cache during chunked_prefill are different generation bugs." + "reason": "Both are generation-related, but the concrete bugs differ: sync_gpus/update_model_kwargs vs chunked_prefill cache missing." }, { "left": "issue:42994", "right": "issue:43482", "accept": false, - "reason": "Quantized model saving failure and Qwen2.5-GGUF loading failure are opposite-direction serialization/loading issues, not the same bug." + "reason": "Quantized model saving failure is unrelated to Qwen2.5-GGUF loading under transformers v5." }, { "left": "issue:44869", "right": "issue:45356", "accept": false, - "reason": "Whisper word-timestamp decode crash and Kimi-K2.5 tokenizer regression are distinct tokenizer/decoding problems." + "reason": "Whisper timestamp decoding crash is unrelated to Kimi-K2.5 tokenizer codec regression." }, { "left": "issue:43475", "right": "issue:43643", "accept": false, - "reason": "SAM3 vision encoder missing attribute and AutoConfig missing fields with trust_remote_code are unrelated." + "reason": "SAM3 vision encoder missing attribute is a different bug from trust_remote_code returning incomplete configs." }, { "left": "issue:42371", "right": "issue:45092", "accept": false, - "reason": "TF32 settings guidance and InternVL2 meta-init incompatibility are unrelated." + "reason": "TF32 API guidance and InternVL2 meta-init incompatibility are unrelated." + }, + { + "left": "issue:45375", + "right": "issue:45478", + "accept": false, + "reason": "Related model family, but one is a specific missing config field and the other is a broader from_pretrained failure; not enough to confirm same underlying bug." + }, + { + "left": "issue:28282", + "right": "issue:30990", + "accept": false, + "reason": "AutoModel missing PyTorch import error is unrelated to Sentence Transformers hanging on load." + }, + { + "left": "issue:41628", + "right": "issue:44933", + "accept": false, + "reason": "Importing AutoImageProcessor and a non-existent import from image_utils are different import-path issues." + }, + { + "left": "issue:45230", + "right": "issue:45397", + "accept": false, + "reason": "One item is too generic to establish identity; the Gemma-4 ZeRO3 from_pretrained bug is specific and distinct." + }, + { + "left": "issue:44038", + "right": "issue:44623", + "accept": false, + "reason": "Qwen3-VL-Moe loading problems are not the same as processor.save_pretrained missing files." + }, + { + "left": "issue:45310", + "right": "issue:45375", + "accept": false, + "reason": "Same model family, but the first is a generic loading error while the second is a specific config-field omission; not clearly the same bug." + }, + { + "left": "issue:44964", + "right": "issue:45137", + "accept": false, + "reason": "Phi-4 multimodal loading failure and DeepSpeed ZeRO3 deque pop error are unrelated." + }, + { + "left": "issue:43475", + "right": "issue:45381", + "accept": false, + "reason": "Different models and symptoms: SAM3 vision encoder attribute error vs Qwen2.5-VL video position-id mismatch." + }, + { + "left": "issue:42994", + "right": "issue:43525", + "accept": false, + "reason": "Quantized model save failure and Llama4Config pad_token_id access are unrelated config/model issues." } ] }, @@ -13060,30 +13159,30 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 42, - "review_unit_count": 56, + "review_unit_index": 43, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ - "issue:28282", "issue:30333", "issue:30990", "issue:33290", + "issue:35141", "issue:36010", "issue:41628", "issue:42175", "issue:42673", "issue:42831", + "issue:42898", "issue:42994", "issue:43232", "issue:43278", - "issue:43475", + "issue:43377", "issue:43504", - "issue:43525", "issue:43704", "issue:43723", "issue:43784", @@ -13093,291 +13192,277 @@ "issue:44038", "issue:44188", "issue:44222", + "issue:44242", "issue:44351", "issue:44496", + "issue:44568", "issue:44623", "issue:44792", "issue:44933", "issue:44945", - "issue:44964", "issue:45081", "issue:45103", "issue:45137", - "issue:45230", "issue:45310", - "issue:45375", - "issue:45381", - "issue:45397", "issue:45406", - "issue:45478", "issue:45561", "issue:45600", "issue:45712", "issue:45753", + "issue:46077", + "issue:46123", "issue:46202", "issue:46326", - "issue:46399" + "issue:46399", + "issue:46693", + "issue:46701" ], "soft_pairs": [ - "issue:45375|issue:45478", - "issue:28282|issue:30990", - "issue:45230|issue:45397", - "issue:41628|issue:44933", - "issue:44038|issue:44623", - "issue:45310|issue:45375", - "issue:44964|issue:45137", - "issue:43475|issue:45381", - "issue:42994|issue:43525", "issue:43723|issue:46326", "issue:43784|issue:46399", - "issue:45406|issue:46202", "issue:43232|issue:44188", + "issue:45406|issue:46202", "issue:43873|issue:44945", "issue:44945|issue:45753", - "issue:45081|issue:45103", + "issue:46202|issue:46701", "issue:41628|issue:44351", + "issue:45081|issue:45103", "issue:42673|issue:43704", "issue:42994|issue:43824", "issue:44038|issue:45310", "issue:44222|issue:45310", "issue:36010|issue:41628", - "issue:42831|issue:44222", "issue:43278|issue:45137", + "issue:42831|issue:44222", "issue:43232|issue:44792", "issue:43504|issue:45561", "issue:42175|issue:44933", "issue:44496|issue:45561", "issue:45561|issue:45600", - "issue:43976|issue:44933", "issue:30333|issue:33290", - "issue:42175|issue:45712" + "issue:43976|issue:44933", + "issue:42175|issue:45712", + "issue:46077|issue:46123", + "issue:30990|issue:35141", + "issue:42898|issue:44568", + "issue:43377|issue:44242", + "issue:44623|issue:45310", + "issue:42673|issue:46693" ], - "prepared_review_unit_hash": "65bf83f5a8a85a5171d04d79a35669115861b76ca3d6cff25dd3833517a53add", + "prepared_review_unit_hash": "182b2fe847c41573675544740f26dd3261710f4ce0b4d1fed199fd7dda4371e4", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 36974, - "estimated_input_tokens": 9244, - "estimated_eval_tokens": 18744 + "soft_pair_count": 30, + "serialized_chars": 36559, + "estimated_input_tokens": 9140, + "estimated_eval_tokens": 18536 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab bag of unrelated Transformers bug reports. The only clear duplicate pair is the repeated Qwen3ForCausalLM VRAM leak report; the rest share broad keywords but do not look like the same underlying bug or change.", - "confidence": 0.95, - "canonical_issue_reason": "issue:42673 is an exact duplicate of issue:43704 (same title and same Qwen3ForCausalLM multi-dataloader VRAM leak). It is the earlier report and the better canonical representative.", + "summary": "This cluster is mostly a set of unrelated bug reports that happen to share broad themes like import/load regressions and generation issues. Only one pair is a clear duplicate: the Qwen3ForCausalLM VRAM leak report.", + "confidence": 0.89, + "canonical_issue_reason": "Issue 42673 is the clearest duplicate anchor: it is the older, more detailed report of the same Qwen3ForCausalLM VRAM leak in multi-dataloader-thread usage, and it matches issue 43704 exactly by title and problem description.", "canonical_pr_reason": null, - "best_issue_reason": "issue:42673 is the strongest issue to represent the cluster because it is a precise duplicate pair with a concrete, well-scoped bug report and earlier discussion.", + "best_issue_reason": "Issue 42673 is the best issue representative for the only accepted duplicate pair because it is the earlier and fuller report, with the strongest discussion context for the same concrete bug.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:45375", - "right": "issue:45478", - "accept": false, - "reason": "Same Qwen3.5/Moe family, but one is a missing config field and the other is a broader from_pretrained failure; not clearly the same underlying bug." - }, - { - "left": "issue:28282", - "right": "issue:30990", - "accept": false, - "reason": "Both are loading-related, but one is a missing PyTorch import error and the other is Sentence Transformers getting stuck loading; different failure modes." - }, - { - "left": "issue:45230", - "right": "issue:45397", - "accept": false, - "reason": "No meaningful evidence of the same bug; one is a placeholder issue title and the other is a ZeRO3 from_pretrained problem." - }, - { - "left": "issue:41628", - "right": "issue:44933", - "accept": false, - "reason": "Both mention missing imports, but they concern different symbols and different code paths; too broad to be the same bug." - }, - { - "left": "issue:44038", - "right": "issue:44623", - "accept": false, - "reason": "One is a Qwen3-VL-Moe bug and the other is processor.save_pretrained missing files; not the same underlying change or defect." - }, - { - "left": "issue:45310", - "right": "issue:45375", - "accept": false, - "reason": "Related Qwen3.5/Moe area, but the reports are about different errors and likely separate fixes." - }, - { - "left": "issue:44964", - "right": "issue:45137", - "accept": false, - "reason": "A model-loading failure and a DeepSpeed ZeRO3 deque IndexError are unrelated bugs." - }, - { - "left": "issue:43475", - "right": "issue:45381", - "accept": false, - "reason": "Different multimodal model families and different symptoms; no sign of a shared root cause." - }, - { - "left": "issue:42994", - "right": "issue:43525", - "accept": false, - "reason": "Quantized model saving and a missing Llama4Config attribute are distinct issues." - }, { "left": "issue:43723", "right": "issue:46326", "accept": false, - "reason": "Both are tokenizer-related, but one is an AutoTokenizer loading issue and the other is continue_final_message logic for mistral_common tokenizers." + "reason": "Both are tokenizer-related, but one is a tokenizer loading/import issue and the other is a continue_final_message behavior bug for mistral_common tokenizers; different code paths." }, { "left": "issue:43784", "right": "issue:46399", "accept": false, - "reason": "Importing sentence-transformers with latest transformers and a Qwen3.5 kernel registration error are unrelated." + "reason": "Unrelated failures: one is a sentence-transformers import NameError, the other is a Qwen3.5 kernel registration error." }, { - "left": "issue:45406", - "right": "issue:46202", + "left": "issue:43232", + "right": "issue:44188", "accept": false, - "reason": "Gemma4 processor attribute error and Gemma pipeline roundtrip failure are different bugs." + "reason": "Different generation bugs: sync_gpus/_update_model_kwargs_for_generation versus attention-kernel divergence under torch.compile." }, { - "left": "issue:43232", - "right": "issue:44188", + "left": "issue:45406", + "right": "issue:46202", "accept": false, - "reason": "One concerns generation kwargs after sync_gpus, the other attention kernel divergence under torch.compile; different code paths." + "reason": "Different subsystems and symptoms: Gemma4Processor serve crash versus Gemma pipeline roundtrip/tokenization issue." }, { "left": "issue:43873", "right": "issue:44945", "accept": false, - "reason": "Offloading/quantization behavior and incorrect output under pipeline parallelism are not the same defect." + "reason": "Offloading/quantization behavior is not the same as incorrect outputs under pipeline parallelism." }, { "left": "issue:44945", "right": "issue:45753", "accept": false, - "reason": "Pipeline parallelism output corruption and an infinite loop on a specific image are separate model-behavior bugs." + "reason": "Pipeline-parallel output corruption is distinct from a Qwen3_5 infinite-loop-on-image issue." }, { - "left": "issue:45081", - "right": "issue:45103", + "left": "issue:46202", + "right": "issue:46701", "accept": false, - "reason": "Tokenizer regex patch crash and auto_docstring annotation crash are unrelated." + "reason": "Both involve Gemma, but one is a tokenization roundtrip bug and the other is an image-text-to-text pipeline failure; not the same defect." }, { "left": "issue:41628", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but for different missing names and likely different release regressions." + "reason": "Both are import errors, but they concern different symbols and likely different missing exports." + }, + { + "left": "issue:45081", + "right": "issue:45103", + "accept": false, + "reason": "Completely different bugs: Mistral tokenizer regex patching versus auto-docstring processing with future annotations." }, { "left": "issue:42673", "right": "issue:43704", "accept": true, - "reason": "Exact duplicate reports with the same title and same Qwen3ForCausalLM VRAM leak in multiple dataloader threads." + "reason": "Exact duplicate: same Qwen3ForCausalLM VRAM leak when used across multiple dataloader threads, with matching title and failure mode." }, { "left": "issue:42994", "right": "issue:43824", "accept": false, - "reason": "Quantized model saving and a missing Qwen2_5_VLForConditionalGeneration import are different problems." + "reason": "Quantized model saving failure and an import error for Qwen2_5_VLForConditionalGeneration are unrelated." }, { "left": "issue:44038", "right": "issue:45310", "accept": false, - "reason": "Different Qwen model variants and different failure descriptions; too broad to merge." + "reason": "Related model family, but different versions and different reported failures; not enough evidence they are the same concrete bug." }, { "left": "issue:44222", "right": "issue:45310", "accept": false, - "reason": "FP8 save_pretrained on moe and Qwen3.5 Moe from_pretrained failure do not clearly share a single root cause." + "reason": "FP8 save_pretrained failure is a save-path bug, while the Qwen3.5 Moe report is a from_pretrained/load-path error." }, { "left": "issue:36010", "right": "issue:41628", "accept": false, - "reason": "Both are import errors from transformers, but they involve different missing APIs and are not the same bug." + "reason": "Both are import failures, but they affect different APIs and likely different export points." }, { - "left": "issue:42831", - "right": "issue:44222", + "left": "issue:43278", + "right": "issue:45137", "accept": false, - "reason": "Accuracy issues with FineGrainedFP8 and FP8 save_pretrained failures are unrelated." + "reason": "Embedding dtype drift in evaluate is unrelated to the DeepSpeed ZeRO3 deque IndexError." }, { - "left": "issue:43278", - "right": "issue:45137", + "left": "issue:42831", + "right": "issue:44222", "accept": false, - "reason": "Embedding dtype drift during evaluate and a DeepSpeed ZeRO3 deque error are unrelated." + "reason": "One is an accuracy regression in FineGrainedFP8; the other is a model saving bug with FP8/MoE." }, { "left": "issue:43232", "right": "issue:44792", "accept": false, - "reason": "Different functionality entirely: generation kwargs after sync_gpus versus a janus image-generation test failure." + "reason": "Generation-kwargs synchronization bug and janus image generation test failure are different issues." }, { "left": "issue:43504", "right": "issue:45561", "accept": false, - "reason": "A pretrained-model loading bug and a pytest-xdist file race are unrelated." + "reason": "Model-loading preset failure and a pytest-xdist race in testing utilities are unrelated." }, { "left": "issue:42175", "right": "issue:44933", "accept": false, - "reason": "Packaging/backend dependency issue versus a missing import from image_utils; different bugs." + "reason": "Backend dependency packaging issue versus a missing import from image_utils; different problems." }, { "left": "issue:44496", "right": "issue:45561", "accept": false, - "reason": "Unrecognized model config and a test utility race condition are unrelated." + "reason": "Model config loading failure and a test-utils file race are unrelated." }, { "left": "issue:45561", "right": "issue:45600", "accept": false, - "reason": "Both are CI/testing related, but one is an xdist race and the other is stale auto_mappings references; not the same defect." + "reason": "One is an xdist race on captured_info.txt; the other is stale auto_mappings references breaking CI." }, { - "left": "issue:43976", - "right": "issue:44933", + "left": "issue:30333", + "right": "issue:33290", "accept": false, - "reason": "Python version compatibility and a missing import from image_utils are unrelated." + "reason": "Training-job MLFlow reporting failure and DeepSpeed/Adafactor OOM are different failure modes." }, { - "left": "issue:30333", - "right": "issue:33290", + "left": "issue:43976", + "right": "issue:44933", "accept": false, - "reason": "MLFlow reporting after failed training and OOM with Adafactor in DeepSpeed are different issues." + "reason": "Python version compatibility problem versus an import-from-image_utils issue; no common underlying bug." }, { "left": "issue:42175", "right": "issue:45712", "accept": false, - "reason": "Torch backend packaging and leftover dummy class cleanup are separate maintenance problems." + "reason": "Different packaging/test concerns: missing TensorFlow backend in extra install versus leftover dummy classes in dummy_pt_objects." + }, + { + "left": "issue:46077", + "right": "issue:46123", + "accept": false, + "reason": "Different behavior bugs: bidirectional attention masking versus pipeline final-batch result dropping." + }, + { + "left": "issue:30990", + "right": "issue:35141", + "accept": false, + "reason": "Sentence-transformers loading hang is unrelated to embedding reinitialization after resize_token_embeddings." + }, + { + "left": "issue:42898", + "right": "issue:44568", + "accept": false, + "reason": "Both are tokenizer regressions in v5, but one is cleanup-space behavior and the other is BOS/EOS insertion; different code paths." + }, + { + "left": "issue:43377", + "right": "issue:44242", + "accept": false, + "reason": "MIMI padding-mask batching bug and missing load-balancing loss when router logits are disabled are unrelated." + }, + { + "left": "issue:44623", + "right": "issue:45310", + "accept": false, + "reason": "processor.save_pretrained missing files is a serialization bug, not the same as Qwen3.5 Moe from_pretrained failure." + }, + { + "left": "issue:42673", + "right": "issue:46693", + "accept": false, + "reason": "Both mention generation performance, but VRAM leakage in threads and FA2 long-context regression are different bugs." } ] }, "evaluator_result": { "accept": true, - "feedback": "The cluster summary is grounded in the packet and stays conservative on the soft-edge pairs. The only clear merge/canonical choice is the identical Qwen3ForCausalLM VRAM leak pair (42673/43704), and the rest are reasonably treated as unrelated or at least not clearly the same bug based on the available titles." + "feedback": "Grounded overall. The accepted duplicate pair is well supported by the packet: the titles match exactly, and 42673 is earlier and has more discussion/body content than 43704. The non-duplicate verdicts are conservative based on the provided titles." }, "evaluator_used": true, "retried": false, @@ -13389,30 +13474,30 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 43, - "review_unit_count": 56, + "review_unit_index": 44, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:22355", "issue:27957", "issue:28282", "issue:30333", - "issue:30990", "issue:34689", "issue:35141", "issue:36331", "issue:42222", - "issue:42673", + "issue:42796", + "issue:42831", "issue:42898", "issue:43116", "issue:43295", - "issue:43377", "issue:43479", + "issue:43644", "issue:43723", "issue:43784", "issue:43824", @@ -13423,9 +13508,9 @@ "issue:44079", "issue:44190", "issue:44222", - "issue:44242", "issue:44295", "issue:44297", + "issue:44448", "issue:44568", "issue:44623", "issue:44734", @@ -13436,41 +13521,34 @@ "issue:45103", "issue:45125", "issue:45200", - "issue:45310", + "issue:45245", "issue:45356", "issue:45405", "issue:45656", "issue:45712", "issue:45812", - "issue:46077", - "issue:46123", + "issue:46132", "issue:46231", - "issue:46693" + "issue:46665" ], "soft_pairs": [ - "issue:46077|issue:46123", - "issue:30990|issue:35141", - "issue:42898|issue:44568", - "issue:43377|issue:44242", - "issue:44623|issue:45310", - "issue:42673|issue:46693", "issue:43479|issue:44869", "issue:42222|issue:45200", "issue:44016|issue:46231", - "issue:43723|issue:44568", "issue:42898|issue:45812", + "issue:43723|issue:44568", "issue:43874|issue:45092", "issue:43479|issue:44079", "issue:42898|issue:44779", "issue:44734|issue:45103", "issue:43784|issue:43824", "issue:42898|issue:45356", + "issue:27957|issue:30333", "issue:45405|issue:45712", "issue:42898|issue:43295", - "issue:27957|issue:30333", - "issue:44295|issue:45656", - "issue:44060|issue:45125", "issue:30333|issue:34689", + "issue:44060|issue:45125", + "issue:44295|issue:45656", "issue:43116|issue:44190", "issue:44933|issue:45712", "issue:30333|issue:36331", @@ -13478,222 +13556,235 @@ "issue:44075|issue:46231", "issue:30333|issue:35141", "issue:22355|issue:28282", - "issue:44297|issue:45656" + "issue:44297|issue:45656", + "issue:42831|issue:44623", + "issue:44448|issue:46132", + "issue:42796|issue:43723", + "issue:44222|issue:44623", + "issue:43479|issue:43644", + "issue:45245|issue:46132", + "issue:43479|issue:46665" ], - "prepared_review_unit_hash": "1065b48e055df1114ae44f8a49473aced54c6abdcb3d3d8dfd686393b4cc6d27", + "prepared_review_unit_hash": "a137d930d0e12b10aec45bf06331ad151d2ae5e8835cae547a6962ea58ad21e1", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 37217, - "estimated_input_tokens": 9305, - "estimated_eval_tokens": 18866 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 32, + "serialized_chars": 36886, + "estimated_input_tokens": 9222, + "estimated_eval_tokens": 18700 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These items span many unrelated Transformers bugs/regressions across tokenizers, model loading, training, multimodal configs, serving, and example scripts. The soft-similarity pairs are mostly broad theme matches, but they do not describe the same concrete underlying bug or mergeable fix.", - "confidence": 0.96, - "canonical_issue_reason": "No single issue stands out as a duplicate anchor; the set is heterogeneous and the reported failures affect different subsystems and code paths.", - "canonical_pr_reason": "No pull requests are present.", - "best_issue_reason": "Issue 42898 is the clearest standalone bug report in the set: it describes a specific v5 tokenizer behavior change with direct user impact and a concrete regression surface.", - "best_pr_reason": "No pull requests are present.", + "summary": "This cluster is heterogeneous: it mixes unrelated tokenizer, model-loading, training, config, docs, and utility bugs. None of the soft pairs looks like the same underlying issue, so no duplicate merge is justified.", + "confidence": 0.92, + "canonical_issue_reason": "No single canonical issue stands out because the items do not share one concrete bug or code path; they are mostly separate regressions that only share broad Transformers terminology.", + "canonical_pr_reason": "No PRs are present in this cluster.", + "best_issue_reason": "No global best issue: the set is too mixed to choose one representative fix target without collapsing distinct bugs into one bucket.", + "best_pr_reason": "No PRs are present in this cluster.", "soft_edge_verdicts": [ - { - "left": "issue:46077", - "right": "issue:46123", - "accept": false, - "reason": "Different bugs: bidirectional attention mask construction vs final partial-batch handling in a pipeline." - }, - { - "left": "issue:30990", - "right": "issue:35141", - "accept": false, - "reason": "Unrelated failures: loading hang in sentence-transformers vs embedding resize/post_init reinitialization." - }, - { - "left": "issue:42898", - "right": "issue:44568", - "accept": false, - "reason": "Both are tokenizer regressions, but one is clean-up spacing behavior and the other is BOS/EOS insertion for a specific model." - }, - { - "left": "issue:43377", - "right": "issue:44242", - "accept": false, - "reason": "Different code paths: padding-mask consistency in MIMI vs missing load-balancing loss when router logits are disabled." - }, - { - "left": "issue:44623", - "right": "issue:45310", - "accept": false, - "reason": "Different issues: processor save/export file omission vs Qwen3.5 MoE from_pretrained failure." - }, - { - "left": "issue:42673", - "right": "issue:46693", - "accept": false, - "reason": "VRAM leak in multi-threaded dataloading is not the same as FA2 long-context generation slowdown/regression." - }, { "left": "issue:43479", "right": "issue:44869", "accept": false, - "reason": "Config initialization bug vs Whisper timestamp decode crash; separate components and fixes." + "reason": "Different multimodal config init bug vs Whisper decode crash; unrelated code paths." }, { "left": "issue:42222", "right": "issue:45200", "accept": false, - "reason": "Different multimodal model problems: broken VitPose models vs Gemma 4 token-type defaults for text-only fine-tuning." + "reason": "VitPose model breakage vs Gemma 4 token-type defaults; different failures." }, { "left": "issue:44016", "right": "issue:46231", "accept": false, - "reason": "Notebook syntax error is unrelated to a documentation parameter-name mismatch." + "reason": "Notebook syntax error vs docs parameter-name mismatch; not the same bug." }, { - "left": "issue:43723", - "right": "issue:44568", + "left": "issue:42898", + "right": "issue:45812", "accept": false, - "reason": "AutoTokenizer load issue in v5 is distinct from add_special_tokens behavior on mdeberta-v3." + "reason": "Both are tokenizer regressions, but different models and symptoms." }, { - "left": "issue:42898", - "right": "issue:45812", + "left": "issue:43723", + "right": "issue:44568", "accept": false, - "reason": "Different tokenizer regressions affecting different models and failure modes." + "reason": "Loading failure vs missing BOS/EOS insertion; distinct tokenizer issues." }, { "left": "issue:43874", "right": "issue:45092", "accept": false, - "reason": "Missing image-patch method in a processor is separate from InternVL2 meta-init incompatibility." + "reason": "Missing image-processor method vs remote-code/meta-init incompatibility; unrelated." }, { "left": "issue:43479", "right": "issue:44079", "accept": false, - "reason": "Config defaulting bug and ModelOutput key assignment bug are unrelated." + "reason": "Config defaulting bug vs ModelOutput key assignment bug; different components." }, { "left": "issue:42898", "right": "issue:44779", "accept": false, - "reason": "Both affect tokenizers, but the reported defects are different regressions for different models." + "reason": "Tokenizer regression for different models; no shared concrete fix path." }, { "left": "issue:44734", "right": "issue:45103", "accept": false, - "reason": "Serving KV-cache indexing crash and auto_docstring future-annotations crash are unrelated." + "reason": "Serve/KV-cache indexing crash vs auto-docstring annotations crash; unrelated." }, { "left": "issue:43784", "right": "issue:43824", "accept": false, - "reason": "Import error for sentence-transformers internals is unrelated to a missing Qwen2.5-VL export." + "reason": "Different import errors with different missing symbols/dependencies." }, { "left": "issue:42898", "right": "issue:45356", "accept": false, - "reason": "Different tokenizer regressions: spacing cleanup vs codec-handling/warning behavior for Kimi-K2.5." + "reason": "Separate tokenizer regressions for different models/code paths." + }, + { + "left": "issue:27957", + "right": "issue:30333", + "accept": false, + "reason": "Flash Attention model support vs MLflow reporting on failed training; unrelated." }, { "left": "issue:45405", "right": "issue:45712", "accept": false, - "reason": "A dependency version bump is unrelated to leftover dummy class cleanup." + "reason": "Dependency-version bump vs leftover dummy classes/check-repo cleanup; different issues." }, { "left": "issue:42898", "right": "issue:43295", "accept": false, - "reason": "Tokenizer clean-up behavior change is not the same as processor.tokenizer/image-passing regression." - }, - { - "left": "issue:27957", - "right": "issue:30333", - "accept": false, - "reason": "XLM-R FA2 behavior and MLflow reporting on failed jobs are entirely different subsystems." + "reason": "Both involve v5 behavior changes, but one is token IDs and the other is processor/tokenizer API access." }, { - "left": "issue:44295", - "right": "issue:45656", + "left": "issue:30333", + "right": "issue:34689", "accept": false, - "reason": "Position_ids buffer access error is unrelated to Deepspeed causing optimizer.step to run twice." + "reason": "Training-job reporting bug vs model-loading regression; unrelated." }, { "left": "issue:44060", "right": "issue:45125", "accept": false, - "reason": "Incorrect tied-weights warning and missing tensor-parallel plan are distinct Qwen3.5 MoE issues." + "reason": "Wrong tied-weights warning vs missing tensor-parallel plan; different Qwen3 bugs." }, { - "left": "issue:30333", - "right": "issue:34689", + "left": "issue:44295", + "right": "issue:45656", "accept": false, - "reason": "Training-job MLflow reporting bug is unrelated to a model-loading regression for Llama 3.2 Vision." + "reason": "Buffer read error vs double optimizer step with deepspeed; unrelated." }, { "left": "issue:43116", "right": "issue:44190", "accept": false, - "reason": "Example script returning empty multi-label results is not the same as local dataset loading failure." + "reason": "Different example-script failures in different pipelines." }, { "left": "issue:44933", "right": "issue:45712", "accept": false, - "reason": "Missing import from image_utils and leftover dummy classes are separate packaging/import cleanup issues." + "reason": "Missing import from image_utils vs leftover dummy classes; not the same defect." }, { "left": "issue:30333", "right": "issue:36331", "accept": false, - "reason": "MLflow job-state reporting and CustomTrainer compute_loss signature mismatch are different bugs." + "reason": "MLflow tracking on failure vs CustomTrainer signature mismatch; distinct training bugs." }, { "left": "issue:44222", "right": "issue:44297", "accept": false, - "reason": "FP8 MoE save_pretrained and tokenizer_class mismatch in tokenizer config are unrelated save/export issues." + "reason": "FP8 MoE save_pretrained issue vs tokenizer metadata mismatch; different save paths." }, { "left": "issue:44075", "right": "issue:46231", "accept": false, - "reason": "Optimizer argument handling and training-doc parameter naming are not the same underlying defect." + "reason": "Optimizer argument usage bug vs documentation name mismatch; unrelated." }, { "left": "issue:30333", "right": "issue:35141", "accept": false, - "reason": "MLflow reporting failure and embedding reinitialization are unrelated." + "reason": "Failed-job reporting vs embedding resize/post_init reinit bug; different code paths." }, { "left": "issue:22355", "right": "issue:28282", "accept": false, - "reason": "Different import/setup errors: missing transformers.onnx module vs missing PyTorch dependency for AutoModel." + "reason": "Different missing-import/dependency errors; no shared root cause." }, { "left": "issue:44297", "right": "issue:45656", "accept": false, - "reason": "Tokenizer save_pretrained metadata mismatch is unrelated to Deepspeed double optimizer stepping." + "reason": "Tokenizer save metadata bug vs optimizer double-step bug; unrelated." + }, + { + "left": "issue:42831", + "right": "issue:44623", + "accept": false, + "reason": "FP8 accuracy regression vs processor save_pretrained missing files; different issues." + }, + { + "left": "issue:44448", + "right": "issue:46132", + "accept": false, + "reason": "Model output difference vs attention-registration behavior change; unrelated." + }, + { + "left": "issue:42796", + "right": "issue:43723", + "accept": false, + "reason": "Tokenizer encoding bug vs tokenizer-loading bug; distinct symptoms and fixes." + }, + { + "left": "issue:44222", + "right": "issue:44623", + "accept": false, + "reason": "Model save bug vs processor save bug; separate artifacts and code paths." + }, + { + "left": "issue:43479", + "right": "issue:43644", + "accept": false, + "reason": "Config None-handling bug vs non-persistent-buffer initialization bug; unrelated." + }, + { + "left": "issue:45245", + "right": "issue:46132", + "accept": false, + "reason": "Category-limit runtime error vs attention-registration behavior change; unrelated." + }, + { + "left": "issue:43479", + "right": "issue:46665", + "accept": false, + "reason": "Specific multimodal config default bug vs generic from_pretrained-in-classes inconsistency; not the same bug." } ] }, @@ -13708,26 +13799,24 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 44, - "review_unit_count": 56, + "review_unit_index": 45, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:30990", "issue:38175", "issue:41628", "issue:42757", - "issue:42796", "issue:42831", "issue:42907", "issue:43122", "issue:43232", "issue:43475", - "issue:43479", "issue:43644", "issue:43723", "issue:44190", @@ -13735,60 +13824,54 @@ "issue:44222", "issue:44261", "issue:44263", + "issue:44279", "issue:44295", "issue:44297", "issue:44360", "issue:44373", - "issue:44448", "issue:44484", "issue:44485", "issue:44623", "issue:44734", + "issue:44829", "issue:44871", "issue:44977", "issue:45071", "issue:45092", "issue:45103", - "issue:45245", "issue:45310", "issue:45357", "issue:45381", "issue:45834", + "issue:46002", "issue:46050", "issue:46082", "issue:46132", "issue:46209", "issue:46305", + "issue:46328", "issue:46421", "issue:46424", "issue:46431", "issue:46655", - "issue:46665", "issue:46693" ], "soft_pairs": [ - "issue:42831|issue:44623", - "issue:44448|issue:46132", - "issue:42796|issue:43723", - "issue:44222|issue:44623", - "issue:43479|issue:43644", - "issue:45245|issue:46132", - "issue:43479|issue:46665", "issue:41628|issue:42757", - "issue:44297|issue:45310", "issue:45071|issue:46082", + "issue:44297|issue:45310", "issue:44373|issue:44485", "issue:44263|issue:45092", "issue:43723|issue:44977", "issue:46132|issue:46693", "issue:44222|issue:44484", "issue:44206|issue:45103", - "issue:43122|issue:46209", "issue:44190|issue:46082", "issue:43232|issue:46424", - "issue:46209|issue:46431", + "issue:43122|issue:46209", "issue:43232|issue:45834", "issue:44734|issue:46421", + "issue:46209|issue:46431", "issue:46305|issue:46655", "issue:43475|issue:46305", "issue:43723|issue:44871", @@ -13801,246 +13884,218 @@ "issue:44263|issue:44373", "issue:43644|issue:46305", "issue:42907|issue:46050", - "issue:42831|issue:45310" + "issue:42831|issue:45310", + "issue:44829|issue:46002", + "issue:44279|issue:46328" ], - "prepared_review_unit_hash": "a2f76af5718cde29057fb90f78dfd95d2d38d86ca5e5c79422b09d8cd2b5a58d", + "prepared_review_unit_hash": "e2bf1dfc3418d0dada1bfe11271efaaa3ae77ce1486132bf3676757717991fce", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 35, - "serialized_chars": 37872, - "estimated_input_tokens": 9468, - "estimated_eval_tokens": 19192 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 35828, + "estimated_input_tokens": 8957, + "estimated_eval_tokens": 18170 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is mostly a loose similarity cluster of unrelated Transformers issues; none of the soft pairs look like true duplicates. I\u2019d use the broad v5 buffer-regression report as the best issue anchor, but only as a representative, not a merged duplicate target.", - "confidence": 0.68, - "canonical_issue_reason": "issue:43644 is the broadest, most cross-cutting regression here (v5.0 non-persistent buffer initialization) and is a reasonable anchor for nearby v5 regression reports.", + "summary": "Loose similarity cluster of mostly Transformers v5 regression reports across loading, config validation, generation, quantization, multimodal, and cache paths; no true duplicate hub emerges.", + "confidence": 0.93, + "canonical_issue_reason": "No single canonical duplicate: the items span unrelated bugs across tokenization, config validation, model saving/loading, generation, and performance.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43644 best represents the cluster because it describes a general Transformers v5 runtime regression rather than a narrow model-specific symptom.", + "best_issue_reason": "issue:45071 is the broadest version-regression report and the closest thing to a central infrastructure bug, but it still does not subsume the others.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:42831", - "right": "issue:44623", - "accept": false, - "reason": "FP8 accuracy issue vs processor.save_pretrained missing files; same quantization/save area, but different bugs and code paths." - }, - { - "left": "issue:44448", - "right": "issue:46132", - "accept": false, - "reason": "Different-output regression for Pegasus vs AttentionInterface behavior change; both are behavior regressions, not the same defect." - }, - { - "left": "issue:42796", - "right": "issue:43723", - "accept": false, - "reason": "Tokenizer BPE-marker handling vs AutoTokenizer.from_pretrained load failure; related subsystem, but not the same bug." - }, - { - "left": "issue:44222", - "right": "issue:44623", - "accept": false, - "reason": "FP8 moe save_pretrained bug vs processor save_pretrained missing files; both involve saving, but on different objects and failure modes." - }, - { - "left": "issue:43479", - "right": "issue:43644", - "accept": false, - "reason": "Phi4Multimodal config default initialization vs v5.0 buffer junk; both are initialization issues, but not the same underlying problem." - }, - { - "left": "issue:45245", - "right": "issue:46132", - "accept": false, - "reason": "Category-limit RuntimeError vs AttentionInterface.register behavior change; no shared concrete code path." - }, - { - "left": "issue:43479", - "right": "issue:46665", - "accept": false, - "reason": "Config None-handling bug vs from_pretrained behavior inconsistency inside Python classes; both are construction/loading issues, but distinct." - }, { "left": "issue:41628", "right": "issue:42757", "accept": false, - "reason": "Import errors from different symbols and packages; similar symptom, but different API breakages." - }, - { - "left": "issue:44297", - "right": "issue:45310", - "accept": false, - "reason": "tokenizer.save_pretrained mismatch vs Qwen3.5 Moe from_pretrained error; both involve config/save/load, but not the same bug." + "reason": "Different missing imports from different modules: AutoImageProcessor vs is_offline_mode." }, { "left": "issue:45071", "right": "issue:46082", "accept": false, - "reason": "PretrainedConfig type checking regression vs LlamaConfig head_dim validation; both config-related, but unrelated defects." + "reason": "Related to config handling, but one is a type-checking regression and the other is a head_dim validation error on LlamaConfig." + }, + { + "left": "issue:44297", + "right": "issue:45310", + "accept": false, + "reason": "Tokenizer save metadata mismatch vs MoE from_pretrained failure; different code paths and symptoms." }, { "left": "issue:44373", "right": "issue:44485", "accept": false, - "reason": "Wrong docstring for position_ids vs GLM-5 RoPE implementation discussion; documentation issue vs implementation issue." + "reason": "Docstring issue for position_ids versus a GLM-5 RoPE implementation discussion." }, { "left": "issue:44263", "right": "issue:45092", "accept": false, - "reason": "DSA indexer torch.split return issue vs InternVL2 meta-init incompatibility; different models and different failure points." + "reason": "torch.split return handling in a DSA indexer is unrelated to InternVL2 meta-initialization incompatibility." }, { "left": "issue:43723", "right": "issue:44977", "accept": false, - "reason": "Tokenizer loading bug vs Qwen3.5 flash-attention generation bug; same broad library area, not the same defect." + "reason": "Tokenizer loading failure and flash-attention generation failure are distinct bugs." }, { "left": "issue:46132", "right": "issue:46693", "accept": false, - "reason": "AttentionInterface.register behavior change vs FA2 long-context performance regression; not the same concrete code-path problem." + "reason": "API behavior change in AttentionInterface.register versus a long-context FA2 performance regression." }, { "left": "issue:44222", "right": "issue:44484", "accept": false, - "reason": "FP8 save_pretrained moe bug vs max_shard_size default question; one is a correctness bug, the other is a configuration/design question." + "reason": "Saving a quantized model is not the same as asking why save_pretrained defaults to 50GB shard size." }, { "left": "issue:44206", "right": "issue:45103", "accept": false, - "reason": "feature extractor center-arg crash vs auto_docstring crash under future annotations; both are crashes, but unrelated subsystems." - }, - { - "left": "issue:43122", - "right": "issue:46209", - "accept": false, - "reason": "Tokenization differences across versions vs FP8 inference failure on B200; both version/hardware regressions, but different issues." + "reason": "Feature extractor center-arg crash and auto_docstring future-annotations crash are unrelated." }, { "left": "issue:44190", "right": "issue:46082", "accept": false, - "reason": "Local dataset loading in an example script vs LlamaConfig head_dim validation; no shared underlying bug." + "reason": "Local dataset loading in a training script is unrelated to LlamaConfig head_dim validation." }, { "left": "issue:43232", "right": "issue:46424", "accept": false, - "reason": "Generation kwargs update after sync_gpus vs cache max_cache_length handling; both generation/cache adjacent, but distinct behaviors." + "reason": "Generation after sync_gpus and cache max_cache_length semantics are different cache-related issues." }, { - "left": "issue:46209", - "right": "issue:46431", + "left": "issue:43122", + "right": "issue:46209", "accept": false, - "reason": "FP8 inference failure on B200 vs missing chat template error; completely different failure domains." + "reason": "Tokenizer version-difference bug versus FP8 inference failure on B200; different subsystems." }, { "left": "issue:43232", "right": "issue:45834", "accept": false, - "reason": "Generation kwargs update bug vs Kosmos2.5 OCR index error; unrelated model paths." + "reason": "Generation kwargs update after sync_gpus does not match a Kosmos2.5 OCR index error." }, { "left": "issue:44734", "right": "issue:46421", "accept": false, - "reason": "KV-cache continuation crash in serve vs chunked-prefill cache lazy-init recompilation; both cache-related, but not the same bug." + "reason": "KV-cache continuation tensor indexing and cache lazy initialization recompilation are not the same defect." + }, + { + "left": "issue:46209", + "right": "issue:46431", + "accept": false, + "reason": "FP8 hardware inference failure is unrelated to missing chat-template support in a processor." }, { "left": "issue:46305", "right": "issue:46655", "accept": false, - "reason": "Sliding-window off-by-one in masking vs video padding calculation; both sequence-layout bugs, but different implementations." + "reason": "Different sliding-window/padding bugs in different models and processors." }, { "left": "issue:43475", "right": "issue:46305", "accept": false, - "reason": "Missing fpn_position_embeddings attribute vs sliding-window mask off-by-one; different model components and symptoms." + "reason": "Missing SAM3 output attribute is unrelated to a Moonshine streaming mask off-by-one." }, { "left": "issue:43723", "right": "issue:44871", "accept": false, - "reason": "Tokenizer loading failure vs Gemma-3 eos_token_id inconsistency; both tokenizer/config issues, but not the same bug." + "reason": "Both involve loading/config, but tokenizer import failure and eos_token_id mismatch are different bugs." }, { "left": "issue:44263", "right": "issue:44360", "accept": false, - "reason": "torch.split return handling in DSA indexer vs missing ReLU in DSA indexer; same area, but clearly different defects." + "reason": "Different DSA indexer bugs: torch.split outputs versus missing ReLU." }, { "left": "issue:44295", "right": "issue:44623", "accept": false, - "reason": "position_ids buffer read error vs processor save_pretrained missing files; unrelated save/load behaviors." + "reason": "Position_ids buffer read error and processor.save_pretrained missing files are separate save/load issues." }, { "left": "issue:44261", "right": "issue:45092", "accept": false, - "reason": "MLA rms_norm_eps precision issue vs remote-code checkpoint/meta-init incompatibility; both model-loading/regression reports, but different root causes." + "reason": "MLA layernorm epsilon precision issue is unrelated to old InternVL2 meta-init incompatibility." }, { "left": "issue:45381", "right": "issue:46305", "accept": false, - "reason": "qwen2.5-vl vision_position_ids bug vs MoonshineStreaming sliding-window bug; both position-related, but different models and logic." + "reason": "Qwen2.5-VL vision_position_ids bug and Moonshine mask off-by-one are different model-specific logic errors." }, { "left": "issue:45092", "right": "issue:45357", "accept": false, - "reason": "Meta-initialization incompatibility for old InternVL2 checkpoints vs Qwen3.5 save_pretrained visual-key regression; different checkpoints and different failure modes." + "reason": "Old remote-code checkpoint meta-init incompatibility is not the same as incorrect visual encoder keys on save_pretrained." }, { "left": "issue:30990", "right": "issue:38175", "accept": false, - "reason": "Sentence Transformers loading hang vs SIGLIP2 zero probabilities; both model-usage issues, but unrelated bugs." + "reason": "Sentence Transformers loading hang and SIGLIP2 zero probabilities are unrelated." }, { "left": "issue:44263", "right": "issue:44373", "accept": false, - "reason": "DSA indexer return-value issue vs wrong position_ids docstring; code bug vs documentation bug." + "reason": "Code bug in DSA indexer versus a wrong docstring for position_ids." }, { "left": "issue:43644", "right": "issue:46305", "accept": false, - "reason": "v5.0 non-persistent buffer junk vs MoonshineStreaming masking off-by-one; both implementation regressions, but different components and causes." + "reason": "Non-persistent buffer junk in v5.0 is unrelated to a streaming sliding-window mask error." }, { "left": "issue:42907", "right": "issue:46050", "accept": false, - "reason": "Saving dequantized Ministrals/Devstrals vs sinq quantization save/load breakage; both save-related quantization bugs, but not the same artifact." + "reason": "Both are save/load quantization issues, but they concern different quantization methods and different failure modes." }, { "left": "issue:42831", "right": "issue:45310", "accept": false, - "reason": "FineGrainedFP8 accuracy issue vs Qwen3.5 Moe from_pretrained error; same quantized-model theme, but different symptoms and paths." + "reason": "FineGrainedFP8 accuracy regression is not the same as a Qwen3.5 from_pretrained load error." + }, + { + "left": "issue:44829", + "right": "issue:46002", + "accept": false, + "reason": "Degenerate training with FA3 and NaNs at checkout/init are separate training stability bugs." + }, + { + "left": "issue:44279", + "right": "issue:46328", + "accept": false, + "reason": "A dependency issue is unrelated to a SequenceFeatureExtractor.pad performance problem." } ] }, @@ -14055,14 +14110,14 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 45, - "review_unit_count": 56, + "review_unit_index": 46, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:22355", "issue:27957", @@ -14087,6 +14142,7 @@ "issue:43756", "issue:43784", "issue:43824", + "issue:43937", "issue:43976", "issue:44038", "issue:44060", @@ -14094,7 +14150,6 @@ "issue:44188", "issue:44261", "issue:44263", - "issue:44279", "issue:44295", "issue:44297", "issue:44373", @@ -14102,20 +14157,18 @@ "issue:44488", "issue:44568", "issue:44704", - "issue:44829", "issue:44945", + "issue:44964", "issue:45278", "issue:45406", "issue:45593", + "issue:45600", "issue:45753", "issue:46002", - "issue:46328", "issue:46369", "issue:46402" ], "soft_pairs": [ - "issue:44829|issue:46002", - "issue:44279|issue:46328", "issue:44483|issue:45406", "issue:22355|issue:27957", "issue:42831|issue:44038", @@ -14131,8 +14184,8 @@ "issue:43232|issue:45593", "issue:44188|issue:44373", "issue:43408|issue:46369", - "issue:42796|issue:43122", "issue:44261|issue:44373", + "issue:42796|issue:43122", "issue:42371|issue:44263", "issue:43646|issue:46002", "issue:42907|issue:44945", @@ -14142,210 +14195,226 @@ "issue:44295|issue:44297", "issue:43976|issue:44945", "issue:39401|issue:44488", - "issue:44117|issue:44704" + "issue:44117|issue:44704", + "issue:42907|issue:44964", + "issue:41628|issue:45278", + "issue:43784|issue:45600", + "issue:42907|issue:43937" ], - "prepared_review_unit_hash": "b3c415e091c64b10a648ef6212bcd4c39ae698280d85575e067451b749e8062b", + "prepared_review_unit_hash": "cbfde06f5ebc747af8d82e2f41f1be5205b77b4518ac076440621981b97d7add", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 35875, - "estimated_input_tokens": 8969, - "estimated_eval_tokens": 18194 + "soft_pair_count": 31, + "serialized_chars": 36388, + "estimated_input_tokens": 9097, + "estimated_eval_tokens": 18450 }, "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 overlap (tokenizers, imports, model loading) but not the same concrete bug. I would not merge any of the proposed pairs as duplicates.", - "confidence": 0.94, - "canonical_issue_reason": "No single canonical issue stands out because the items span many unrelated failures and none of the soft pairs describe the same underlying defect.", + "summary": "These pairs mostly overlap only at a broad theme level (tokenizers, loading, version regressions, or import errors) but they describe different concrete bugs/models/code paths, so I would not merge any of them as duplicates.", + "confidence": 0.56, + "canonical_issue_reason": "issue:45278 is the broadest and most central representative of the version-regression/import-failure subset in this set; it is the best umbrella issue to anchor related reports, even though it is not a safe duplicate match for the specific pairs here.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44297 is the most self-contained and actionable report: a narrow tokenizer save/load mismatch with a clear symptom and likely localized fix.", + "best_issue_reason": "issue:45278 has the clearest cluster-level relevance because it captures the broad upgrade-regression/import breakage pattern that several other issues loosely resemble.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:44829", - "right": "issue:46002", - "accept": false, - "reason": "Different failure modes: flash_attention_3 training degeneration vs NaNs after checkout/init. Same broad area, not the same bug." - }, - { - "left": "issue:44279", - "right": "issue:46328", - "accept": false, - "reason": "Unrelated performance/memory concerns: a dependency issue vs a padding conversion inefficiency." - }, { "left": "issue:44483", "right": "issue:45406", "accept": false, - "reason": "Different APIs and symptoms: chat/completions request handling vs serve crashing on Gemma4Processor _tokenizer." + "reason": "Different failures: REST chat-completions request rejection vs Gemma4Processor attribute error in transformers serve." }, { "left": "issue:22355", "right": "issue:27957", "accept": false, - "reason": "One is an import/module packaging error; the other is a Flash Attention 2 model-specific compatibility issue." + "reason": "Unrelated problems: missing transformers.onnx module vs XLMRoberta Flash Attention 2 support." }, { "left": "issue:42831", "right": "issue:44038", "accept": false, - "reason": "Both mention precision/quantization, but one is a FineGrainedFP8 accuracy problem and the other is a Qwen3-VL-Moe bug in transformers 5.0." + "reason": "Both are model-specific, but one is FineGrainedFP8 accuracy and the other is a Qwen3-VL-Moe bug; not the same code path." }, { "left": "issue:41628", "right": "issue:43824", "accept": false, - "reason": "Both are import errors, but they involve different missing symbols and different code paths." + "reason": "Both are import errors, but for different classes/models: AutoImageProcessor vs Qwen2.5-VL generation class." }, { "left": "issue:43408", "right": "issue:46402", "accept": false, - "reason": "Different model-mismatch problems: sam3_video/sam3_tracker warning vs sam3 text-encoder weights not loading." + "reason": "Different Sam3 symptoms: model-type warning vs missing text-encoder weights loading." }, { "left": "issue:42831", "right": "issue:45753", "accept": false, - "reason": "Different issues despite both involving model behavior: FP8 accuracy degradation vs a Qwen3_5 image-specific infinite loop." + "reason": "Different model bugs: FP8 accuracy regression vs Qwen3_5 infinite loop on a specific image." }, { "left": "issue:22355", "right": "issue:28218", "accept": false, - "reason": "Completely different bugs: missing transformers.onnx module vs tokenizer adding an extra space after an added token." + "reason": "No shared underlying bug: missing transformers.onnx module vs tokenizer adding extra spaces after added tokens." }, { "left": "issue:39401", "right": "issue:44568", "accept": false, - "reason": "Both are tokenizer correctness bugs, but one is wrong offset_mapping and the other is add_special_tokens not adding BOS/EOS for a specific model." + "reason": "Both tokenizer-related, but one is wrong offset_mapping and the other is add_special_tokens not adding BOS/EOS." }, { "left": "issue:42757", "right": "issue:45278", "accept": false, - "reason": "The latter is a broad upgrade-related import breakage, while the former is a specific offline-mode import failure from huggingface_hub." + "reason": "The first is a specific is_offline_mode import failure; the second is a broad set of import errors after upgrade." }, { "left": "issue:42886", "right": "issue:43066", "accept": false, - "reason": "Different tokenizer problems: cache/offline loading failure vs wrong decoder type selected in v5." + "reason": "Offline cache loading failure and wrong tokenizer decoder type are different tokenizer issues." }, { "left": "issue:27957", "right": "issue:30990", "accept": false, - "reason": "One is XLMRoberta Flash Attention 2 behavior, the other is Sentence Transformers hanging during load." + "reason": "Flash Attention 2 support for XLMRoberta is unrelated to Sentence Transformers getting stuck loading." }, { "left": "issue:42886", "right": "issue:42994", "accept": false, - "reason": "One is offline cache loading; the other is quantized model saving failure. No shared code-path bug." + "reason": "One is offline tokenizer cache loading; the other is quantized model saving failure." }, { "left": "issue:43232", "right": "issue:45593", "accept": false, - "reason": "Generation kwargs handling and auxiliary-loss behavior are unrelated problems in different subsystems." + "reason": "Different generation/training concerns: sync_gpus kwargs update vs auxiliary losses omitted when denoising is off." }, { "left": "issue:44188", "right": "issue:44373", "accept": false, - "reason": "torch.compile attention-kernel divergence is a runtime bug; wrong docstring for position_ids is documentation-only." + "reason": "Runtime attention-kernel divergence is unrelated to a docstring issue for position_ids." }, { "left": "issue:43408", "right": "issue:46369", "accept": false, - "reason": "Both mention sam3, but one is a model-type warning and the other is memory overflow when calling a specific model." + "reason": "Both mention Sam3, but one is a model-type warning and the other is memory overflow in sam3-litetext." }, { - "left": "issue:42796", - "right": "issue:43122", + "left": "issue:44261", + "right": "issue:44373", "accept": false, - "reason": "Tokenizer marker handling for Ministral-3-8B-Instruct is distinct from version-to-version tokenization differences." + "reason": "Precision bug in MLA q_a_layernorm is unrelated to a wrong docstring for position_ids." }, { - "left": "issue:44261", - "right": "issue:44373", + "left": "issue:42796", + "right": "issue:43122", "accept": false, - "reason": "Precision mismatch from missing rms_norm_eps is a runtime/config bug, not a docstring issue." + "reason": "Both concern tokenization, but one is BPE marker handling for a specific model and the other is cross-version tokenization drift." }, { "left": "issue:42371", "right": "issue:44263", "accept": false, - "reason": "TF32 settings warning/breakage is unrelated to torch.split return values in GlmMoeDsaIndexer." + "reason": "TF32 configuration guidance is unrelated to torch.split return values in GlmMoeDsaIndexer." }, { "left": "issue:43646", "right": "issue:46002", "accept": false, - "reason": "Custom model initialization breakage and NaNs in classification heads are different defects." + "reason": "Custom model initialization breakage and NaNs in classification heads are different failure modes." }, { "left": "issue:42907", "right": "issue:44945", "accept": false, - "reason": "Quantized model saving and incorrect outputs under pipeline parallelism are unrelated." + "reason": "Quantized/dequantized model saving failure is unrelated to incorrect LLM output under pipeline parallelism." }, { "left": "issue:43756", "right": "issue:44060", "accept": false, - "reason": "Different Qwen/Smollm bugs: RoPE layer dropping vs tied-weights warning misbinding dt_bias." + "reason": "Smollm3 RoPE-layer mismatch and Qwen3-Next tied-weights warning are different model-architecture bugs." }, { "left": "issue:30990", "right": "issue:33357", "accept": false, - "reason": "Different load/runtime failures: Sentence Transformers loading hang vs MacOS bus error on a CLIP community model." + "reason": "Sentence-Transformers load hang and MacOS bus error with a CLIP model are not the same issue." }, { "left": "issue:43576", "right": "issue:43784", "accept": false, - "reason": "Broken env command and a NameError while importing sentence-transformers are separate issues." + "reason": "Broken transformers env command and a NameError importing sentence-transformers are unrelated." }, { "left": "issue:44295", "right": "issue:44297", "accept": false, - "reason": "Both concern tokenizers, but one is reading a bufferized position_ids tensor and the other is tokenizer_class metadata mismatch on save_pretrained." + "reason": "Reading registered position_ids as a buffer and tokenizer_class mismatch on save_pretrained are different bugs." }, { "left": "issue:43976", "right": "issue:44945", "accept": false, - "reason": "Python version compatibility and pipeline-parallel output correctness are different failure classes." + "reason": "Python version support regression is unrelated to incorrect LLM output from pipeline parallelism." }, { "left": "issue:39401", "right": "issue:44488", "accept": false, - "reason": "Wrong offset_mapping and failing to load a specific model are unrelated tokenizer vs model-loading bugs." + "reason": "Wrong offset_mapping in Qwen3 is unrelated to loading cjvt/sleng-bert." }, { "left": "issue:44117", "right": "issue:44704", "accept": false, - "reason": "Mapping-name None handling in from_pretrained is unrelated to AutoProcessor not forwarding kwargs to cached_file." + "reason": "A None handling bug in TOKENIZER_MAPPING_NAMES is different from AutoProcessor not forwarding kwargs to cached_file." + }, + { + "left": "issue:42907", + "right": "issue:44964", + "accept": false, + "reason": "Saving dequantized Ministrals/Devstrals and loading Phi-4 multimodal instruct are different model-loading/saving problems." + }, + { + "left": "issue:41628", + "right": "issue:45278", + "accept": false, + "reason": "Both are import-related, but this is a specific AutoImageProcessor import error versus a broad many-import-errors regression." + }, + { + "left": "issue:43784", + "right": "issue:45600", + "accept": false, + "reason": "Importing sentence-transformers fails with NameError, while the Sam3LiteText issue is broken auto_mappings references." + }, + { + "left": "issue:42907", + "right": "issue:43937", + "accept": false, + "reason": "Quantized model saving failure is unrelated to GLM-5 invalid GenerationConfig." } ] }, @@ -14360,14 +14429,14 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 46, - "review_unit_count": 56, + "review_unit_index": 47, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:22355", "issue:29127", @@ -14375,8 +14444,8 @@ "issue:36296", "issue:38617", "issue:39401", - "issue:41628", "issue:42175", + "issue:42673", "issue:42757", "issue:42898", "issue:42907", @@ -14388,7 +14457,7 @@ "issue:43452", "issue:43475", "issue:43493", - "issue:43784", + "issue:43611", "issue:43824", "issue:43937", "issue:44038", @@ -14406,23 +14475,19 @@ "issue:44829", "issue:44938", "issue:44945", - "issue:44964", + "issue:45072", "issue:45200", "issue:45216", "issue:45278", "issue:45290", "issue:45478", - "issue:45600", "issue:45753", + "issue:46002", "issue:46326", "issue:46328", "issue:46614" ], "soft_pairs": [ - "issue:42907|issue:44964", - "issue:41628|issue:45278", - "issue:42907|issue:43937", - "issue:43784|issue:45600", "issue:44360|issue:44485", "issue:44704|issue:45290", "issue:44625|issue:44829", @@ -14447,231 +14512,221 @@ "issue:22355|issue:29127", "issue:43493|issue:45753", "issue:44188|issue:46328", - "issue:42898|issue:46614", "issue:43937|issue:44485", + "issue:42898|issue:46614", "issue:38617|issue:45278", - "issue:43066|issue:44779" + "issue:43066|issue:44779", + "issue:42673|issue:45072", + "issue:43611|issue:46002" ], - "prepared_review_unit_hash": "db9c966c8b89a30465dfefd949481a5afc6b836db3fb9e8707a421c9508f02fd", + "prepared_review_unit_hash": "910995c7ac3af1b97a9b88e77c87a3334e262c92f08449f7f9093b1503513bda", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 36985, - "estimated_input_tokens": 9247, - "estimated_eval_tokens": 18750 + "soft_pair_count": 30, + "serialized_chars": 36360, + "estimated_input_tokens": 9090, + "estimated_eval_tokens": 18436 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly unrelated issues: several are broad v5 regressions or import failures, but the concrete bugs and code paths differ too much to treat as duplicates. No PRs are present; the best umbrella issue here is the broad import-regression report.", - "confidence": 0.56, - "canonical_issue_reason": "Issue 45278 is the closest umbrella candidate because it explicitly describes multiple import failures after upgrading to Transformers 5.5.0, which makes it the broadest representative of the import-regression subset in this set.", + "summary": "This cluster is mostly a loose set of Transformers issues with a few tokenizer/import-regression themes, but the soft pairs are still too heterogeneous to treat as duplicates. I would not merge any of the proposed soft edges.", + "confidence": 0.34, + "canonical_issue_reason": "If a representative issue is needed, #43122 is the broadest tokenizer-regression report (v4.57.3 to v5.0) and best matches the general theme of the cluster, but it is not a true duplicate anchor for most other items.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45278 is the best global issue candidate since it is the most general and could reasonably cover several import-related reports, unlike the more narrowly scoped model/tokenizer bugs.", + "best_issue_reason": "#43122 is the most umbrella-like issue in the set: it describes a general tokenizer behavior regression across versions, which is more central than the narrower model-specific reports.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:42907", - "right": "issue:44964", - "accept": false, - "reason": "Different failure modes and code paths: saving dequantized Ministrals/Devstrals vs loading Phi-4 multimodal." - }, - { - "left": "issue:41628", - "right": "issue:45278", - "accept": false, - "reason": "45278 is a broad umbrella import-regression report, but 41628 is a specific AutoImageProcessor import error; not the same concrete bug." - }, - { - "left": "issue:42907", - "right": "issue:43937", - "accept": false, - "reason": "Unrelated model-specific save/load issues: dequantized save failure vs invalid GenerationConfig." - }, - { - "left": "issue:43784", - "right": "issue:45600", - "accept": false, - "reason": "Both are import-time failures, but they involve different symbols and causes: sentence-transformers nn NameError vs stale auto_mappings references." - }, { "left": "issue:44360", "right": "issue:44485", "accept": false, - "reason": "Completely different topics: DSA indexer ReLU discussion vs GLM-5 RoPE implementation." + "reason": "Different bugs: a DSA indexer discussion versus GLM-5 RoPE implementation; same broad area, not the same defect." }, { "left": "issue:44704", "right": "issue:45290", "accept": false, - "reason": "Different APIs and bugs: cached_file kwargs forwarding vs apply_chat_template crashing on tool-call messages." + "reason": "Different code paths: cached_file kwargs forwarding versus chat template crashes with tool-call assistant messages." }, { "left": "issue:44625", "right": "issue:44829", "accept": false, - "reason": "Different model/config paths: num_labels propagation vs flash_attention_3 training degeneration." + "reason": "Unrelated failures: num_labels propagation in Qwen3.5 config versus degenerate training with flash_attention_3." }, { "left": "issue:39401", "right": "issue:44521", "accept": false, - "reason": "Different tokenizer issues: offset_mapping bug vs assistant_masks being all zeros for multimodal chat templates." + "reason": "Both involve tokenization-related behavior, but one is wrong offset mappings and the other is multimodal assistant masks; not the same bug." }, { "left": "issue:43122", "right": "issue:46326", "accept": false, - "reason": "Both are tokenizer-related regressions, but one is a general v4-to-v5 behavior change and the other is a specific continue_final_message bug." + "reason": "Both are tokenizer regressions, but one is general version-to-version tokenization drift and the other is continue_final_message handling for mistral_common tokenizers." }, { "left": "issue:43122", "right": "issue:44521", "accept": false, - "reason": "Different subsystems: generic tokenization regression vs multimodal chat-template masking." + "reason": "Different features and symptoms: general tokenization changes versus multimodal chat template masks." }, { "left": "issue:43408", "right": "issue:43475", "accept": false, - "reason": "Same SAM3 video area, but one is a model-type warning and the other is a missing attribute on a vision encoder output." + "reason": "Same SAM 3 family, but one is a model-type warning and the other is a missing attribute on vision encoder output." }, { "left": "issue:44038", "right": "issue:44297", "accept": false, - "reason": "Different bugs: Qwen3-VL-Moe compatibility issue vs tokenizer_class mismatch in tokenizer_config.json." + "reason": "Different issues: Qwen3-VL-Moe loading bug versus tokenizer save_pretrained class-name mismatch." }, { "left": "issue:43116", "right": "issue:45200", "accept": false, - "reason": "Different examples and failure points: classification example returning empty results vs mm_token_type_ids defaulting behavior." + "reason": "One is a multi-label classification example script issue; the other is Gemma 4 multimodal token type IDs defaulting to zeros." }, { "left": "issue:39401", "right": "issue:46614", "accept": false, - "reason": "Different tokenizer implementations and bugs: Qwen3 offset mapping vs Wav2Vec2PhonemeCTCTokenizer phonemizer language state." + "reason": "Both tokenizer-adjacent, but offset_mapping correctness and stale phonemizer language backend are distinct defects." }, { "left": "issue:42907", "right": "issue:43452", "accept": false, - "reason": "Different model-loading/saving paths: dequantized save_pretrained vs gguf_file loading failures." + "reason": "Different save/load failures: dequantized model saving versus gguf_file breaking from_pretrained resolution." }, { "left": "issue:44038", "right": "issue:44295", "accept": false, - "reason": "Unrelated issues: Qwen3-VL-Moe behavior vs position_ids buffer read error." + "reason": "Different regressions: Qwen3-VL-Moe loading versus position_ids buffer access." }, { "left": "issue:44038", "right": "issue:44222", "accept": false, - "reason": "Same broad model family, but one is a generic bug report and the other is specifically FP8 save_pretrained; not clearly the same defect." + "reason": "Both mention FP8/MoE contexts, but the bugs are different: loading behavior versus save_pretrained on moe." }, { "left": "issue:22355", "right": "issue:29159", "accept": false, - "reason": "No module transformers.onnx is an import/path issue, while this is slow-vs-fast tokenizer inconsistency." + "reason": "An import error for transformers.onnx is unrelated to slow/fast tokenizer inconsistency." }, { "left": "issue:45216", "right": "issue:45478", "accept": false, - "reason": "Related Qwen3.5/version regression context, but save_pretrained corruption and from_pretrained import error are different concrete problems." + "reason": "Related to Qwen3.5/Qwen3.5-Moe version regressions, but one is save_pretrained checkpoint correctness and the other is from_pretrained loading error." }, { "left": "issue:43824", "right": "issue:44351", "accept": false, - "reason": "Both are missing imports, but for different symbols and packages; not the same underlying bug." + "reason": "Both are import errors, but they concern different missing symbols from different modules." }, { "left": "issue:43122", "right": "issue:44779", "accept": false, - "reason": "Both concern tokenizer regressions in v5, but they affect different models and code paths." + "reason": "Both are tokenizer regressions, but one is a broad version drift report and the other is a specific Deepseek tokenizer failure; not enough evidence they are the same bug." }, { "left": "issue:36296", "right": "issue:43116", "accept": false, - "reason": "Different domains entirely: tensor-parallel training bug vs multi-label classification example output." + "reason": "Tensor parallel training bug and example-script multi-label classification failure are unrelated." }, { "left": "issue:43278", "right": "issue:44945", "accept": false, - "reason": "Different training/evaluation and parallelism issues: dtype drift vs incorrect output under pipeline parallelism." + "reason": "Different distributed/evaluation behaviors: dtype change in evaluate versus incorrect output under pipeline parallelism." }, { "left": "issue:42757", "right": "issue:44351", "accept": false, - "reason": "Different import failures from different libraries/symbols: huggingface_hub offline mode vs transformers HybridCache." + "reason": "Both are import errors, but they are for different missing imports and different dependency surfaces." }, { "left": "issue:42175", "right": "issue:44938", "accept": false, - "reason": "TensorFlow backend packaging issue and Python 3.14 load failure are unrelated." + "reason": "Backend packaging/install issue versus Python 3.14 load failure; not the same underlying problem." }, { "left": "issue:22355", "right": "issue:29127", "accept": false, - "reason": "No module transformers.onnx is unrelated to LayoutLMv3 error-message clarity." + "reason": "No meaningful overlap: ONNX import failure versus LayoutLMv3 error-message clarity." }, { "left": "issue:43493", "right": "issue:45753", "accept": false, - "reason": "Different bugs: SigLIP2 implementation discrepancy vs Qwen3_5 infinite loop on a specific image." + "reason": "Both involve Qwen3.5, but one is implementation discrepancy versus infinite-loop behavior on a specific image." }, { "left": "issue:44188", "right": "issue:46328", "accept": false, - "reason": "Different code paths: torch.compile attention-kernel divergence vs a pad-time conversion inefficiency." + "reason": "Attention-kernel divergence under torch.compile is unrelated to SequenceFeatureExtractor.pad performance." }, { - "left": "issue:42898", - "right": "issue:46614", + "left": "issue:43937", + "right": "issue:44485", "accept": false, - "reason": "Different behaviors: tokenization cleanup change in v5 vs stale phonemizer language state." + "reason": "GenerationConfig validation for GLM-5 is unrelated to GLM-5 RoPE implementation discussion." }, { - "left": "issue:43937", - "right": "issue:44485", + "left": "issue:42898", + "right": "issue:46614", "accept": false, - "reason": "Both are GLM-5-related, but one is GenerationConfig validation and the other is RoPE implementation; not the same bug." + "reason": "v5 tokenization-space cleanup behavior and phonemizer language state are different mechanisms." }, { "left": "issue:38617", "right": "issue:45278", "accept": false, - "reason": "Both are import-related, but 38617 is a single missing export while 45278 is a broad multi-import regression report." + "reason": "Different import breakages: layer_type_validation missing versus many import errors after upgrading to 5.5.0." }, { "left": "issue:43066", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer regressions in v5, but they affect different tokenizers and failure signatures." + "reason": "Both are tokenizer regressions, but one is decoder-type mismatch in v5 and the other is Deepseek tokenization output changes; not the same concrete bug." + }, + { + "left": "issue:42673", + "right": "issue:45072", + "accept": false, + "reason": "VRAM leak in Qwen3ForCausalLM multi-threaded use is unrelated to dtype mismatches in CI inference." + }, + { + "left": "issue:43611", + "right": "issue:46002", + "accept": false, + "reason": "base_model_prefix loading regression and NaNs in classification heads are distinct failure modes." } ] }, @@ -14686,14 +14741,14 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 47, - "review_unit_count": 56, + "review_unit_index": 48, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:28348", "issue:33453", @@ -14706,7 +14761,6 @@ "issue:43116", "issue:43441", "issue:43493", - "issue:43611", "issue:43673", "issue:43825", "issue:43827", @@ -14720,13 +14774,13 @@ "issue:44291", "issue:44355", "issue:44380", + "issue:44466", "issue:44625", "issue:44704", "issue:44779", "issue:44936", "issue:44995", "issue:45071", - "issue:45072", "issue:45083", "issue:45356", "issue:45357", @@ -14745,8 +14799,6 @@ "issue:46697" ], "soft_pairs": [ - "issue:42673|issue:45072", - "issue:43611|issue:46002", "issue:44355|issue:46231", "issue:43116|issue:43493", "issue:42371|issue:43874", @@ -14756,234 +14808,229 @@ "issue:45357|issue:45478", "issue:45083|issue:45446", "issue:42222|issue:43441", - "issue:43116|issue:43825", "issue:44075|issue:44936", + "issue:43116|issue:43825", "issue:43992|issue:44704", "issue:28348|issue:46640", "issue:44995|issue:46697", "issue:43064|issue:43673", - "issue:43937|issue:45563", "issue:42673|issue:42796", "issue:43827|issue:45446", "issue:33453|issue:39401", - "issue:45356|issue:46614", + "issue:43937|issue:45563", "issue:42673|issue:46655", + "issue:45356|issue:46614", "issue:44380|issue:46077", "issue:43673|issue:46328", "issue:44117|issue:45588", - "issue:44291|issue:45071", "issue:45071|issue:46614", + "issue:44291|issue:45071", + "issue:45405|issue:45446", "issue:39401|issue:44779", - "issue:45405|issue:45446" + "issue:44466|issue:46002" ], - "prepared_review_unit_hash": "1f6907624067fbcd6adbe6ee38f6293cb5641ed132b6c82162c8327cbe2b92ce", + "prepared_review_unit_hash": "d7c976b2609a53d24401d94a2ba3551fbf5f02201aafb649f26d50feb8874463", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36523, - "estimated_input_tokens": 9131, - "estimated_eval_tokens": 18518 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35688, + "estimated_input_tokens": 8922, + "estimated_eval_tokens": 18100 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: the soft pairs mostly share broad themes (tokenizers, v5 regressions, attention, loaders) but not the same concrete bug or change. I reject all soft edges; none look like true duplicates or mergeable PR-equivalents.", - "confidence": 0.66, - "canonical_issue_reason": "No single canonical duplicate stands out; the items span unrelated bugs across tokenization, generation, training, attention, and docs, so the soft pairs are not true duplicates.", + "summary": "This cluster is mostly a grab-bag of unrelated transformer bugs and regressions. None of the soft pairs look like the same underlying issue strongly enough to merge.", + "confidence": 0.24, + "canonical_issue_reason": "No single issue cleanly canonicalizes this cluster; the items span unrelated subsystems (tokenizers, attention, training, vision, version checks) with no clear duplicate nucleus.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45478 is the strongest standalone report: it names a specific failing path (`from_pretrained`) plus a concrete model/version combination, making it the most actionable representative issue.", + "best_issue_reason": "If one issue must stand in for the cluster, 33453 is the broadest and most general tokenizer-regression report, but it is still not a true duplicate of the other items.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:42673", - "right": "issue:45072", - "accept": false, - "reason": "Different failures: VRAM leak across dataloader threads vs bfloat16 dtype mismatches in inference." - }, - { - "left": "issue:43611", - "right": "issue:46002", - "accept": false, - "reason": "Different code paths: model loading with `base_model_prefix` vs NaNs in classification heads after init/checkout." - }, { "left": "issue:44355", "right": "issue:46231", "accept": false, - "reason": "Unrelated: compiled Python-file errors vs a docs/parameter name mismatch." + "reason": "Unrelated problems: compiled Python file handling vs set_training() docs parameter mismatch." }, { "left": "issue:43116", "right": "issue:43493", "accept": false, - "reason": "Different bugs: empty multi-label example output vs SigLIP2 implementation discrepancy." + "reason": "Different bugs: classification example returns empty results vs SigLIP2 implementation discrepancy." }, { "left": "issue:42371", "right": "issue:43874", "accept": false, - "reason": "One is TF32 API guidance; the other is a missing image-processor method causing an AttributeError." + "reason": "Different subsystems and failures: TF32 API settings vs missing image-patch method in GLM46V fast processor." }, { "left": "issue:43976", "right": "issue:45446", "accept": false, - "reason": "Both involve version checks, but they affect different imports/platform constraints and different subsystems." + "reason": "Both involve version checks, but for different imports/components and different failure modes." }, { "left": "issue:44038", "right": "issue:45478", "accept": false, - "reason": "Same family name, but different concrete failures and versions; not the same bug." + "reason": "Related model family, but not the same concrete bug; Qwen3-VL-Moe vs Qwen3.5 Moe from_pretrained errors are not clearly the same code path." }, { "left": "issue:44625", "right": "issue:46002", "accept": false, - "reason": "`num_labels` propagation issue is distinct from NaNs in classification heads." + "reason": "Different behaviors: num_labels propagation vs NaNs in classification heads." }, { "left": "issue:45357", "right": "issue:45478", "accept": false, - "reason": "Save-path visual-encoder key corruption vs load-time `from_pretrained` failure are different problems." + "reason": "Different regression reports for different Qwen3.5 issues; not enough evidence of one shared bug." }, { "left": "issue:45083", "right": "issue:45446", "accept": false, - "reason": "Feature-extractor helper behavior is unrelated to a PyTorch version guard bug." + "reason": "Different functionality: qwen3_omni_moe feature-extract length helper vs flex_attention PyTorch version gating." }, { "left": "issue:42222", "right": "issue:43441", "accept": false, - "reason": "Different models and failures: vitpose breakage vs Ministral FlashAttention error." + "reason": "Both touch Ministral/vitpose model support, but the failures are unrelated model-specific breakages." }, { - "left": "issue:43116", - "right": "issue:43825", + "left": "issue:44075", + "right": "issue:44936", "accept": false, - "reason": "Example-script classification behavior is unrelated to a pipeline error message/docs issue." + "reason": "Different training/optimizer bug vs trainer.evaluate() post-train failure." }, { - "left": "issue:44075", - "right": "issue:44936", + "left": "issue:43116", + "right": "issue:43825", "accept": false, - "reason": "Optimizer-argument handling and post-train evaluation failure are separate training bugs." + "reason": "Both mention pipeline/example behavior, but the reported issues are different (classification output vs translation-task error messaging)." }, { "left": "issue:43992", "right": "issue:44704", "accept": false, - "reason": "`from_pretrained` weight loading vs `AutoProcessor` forwarding kwargs to `cached_file` are different call paths." + "reason": "Different from_pretrained/caching issues affecting different APIs and code paths." }, { "left": "issue:28348", "right": "issue:46640", "accept": false, - "reason": "Both mention FlashAttention, but one is GPT-2 support request and the other is a generic crash report." + "reason": "One is a request to add FA2 support for GPT2LMHeadModel; the other is a generic flash-attention failure with no clear shared bug." }, { "left": "issue:44995", "right": "issue:46697", "accept": false, - "reason": "Stale cache crash in GLM MoE vs image over-padding in Swin2SR are unrelated." + "reason": "Unrelated model bugs: stale indexer cache on second forward pass vs image over-padding." }, { "left": "issue:43064", "right": "issue:43673", "accept": false, - "reason": "Optimizer-state corruption under FSDP/PEFT is unrelated to missing generation cache in chunked prefill." - }, - { - "left": "issue:43937", - "right": "issue:45563", - "accept": false, - "reason": "Invalid generation config and a stale warning for `num_return_sequences` are different generation issues." + "reason": "Different training-state optimizer issue vs GenerationMixin cache missing during chunked prefill." }, { "left": "issue:42673", "right": "issue:42796", "accept": false, - "reason": "VRAM leak in threaded loading is unrelated to tokenizer BPE-marker handling." + "reason": "Different problems: VRAM leak in Qwen3ForCausalLM threads vs tokenizer BPE marker handling." }, { "left": "issue:43827", "right": "issue:45446", "accept": false, - "reason": "Docs still referencing removed pipelines vs a PyTorch version check bug are not the same issue." + "reason": "Documentation/pipeline deprecation issue vs PyTorch version check bug; not the same artifact." }, { "left": "issue:33453", "right": "issue:39401", "accept": false, - "reason": "Generic tokenizer loading regression and Qwen3 offset mapping bug are different concrete failures." + "reason": "Both are tokenizer-related, but one is a loading regression and the other is wrong offset_mapping on Qwen3." }, { - "left": "issue:45356", - "right": "issue:46614", + "left": "issue:43937", + "right": "issue:45563", "accept": false, - "reason": "Kimi tokenizer regression and stale phonemizer backend state are unrelated tokenizer/phoneme bugs." + "reason": "Different generation issues: invalid GenerationConfig vs stale warning for num_return_sequences." }, { "left": "issue:42673", "right": "issue:46655", "accept": false, - "reason": "VRAM leak in multi-threaded inference is unrelated to video padding logic." + "reason": "Different model/component: VRAM leak in Qwen3ForCausalLM vs video padding bug in SmolVLM." + }, + { + "left": "issue:45356", + "right": "issue:46614", + "accept": false, + "reason": "Different tokenization/runtime issues: Kimi-K2.5 codec regression vs stale phonemizer backend state." }, { "left": "issue:44380", "right": "issue:46077", "accept": false, - "reason": "Different attention semantics and models: GPT2 scaling config vs Gemma4 bidirectional mask construction." + "reason": "Both are attention-related, but one is GPT2 scaling under SDPA/FA and the other is Gemma4 bidirectional mask construction." }, { "left": "issue:43673", "right": "issue:46328", "accept": false, - "reason": "Missing generation cache and slow feature-extractor padding are different performance/logic issues." + "reason": "Different areas: generation cache behavior vs feature extractor padding performance." }, { "left": "issue:44117", "right": "issue:45588", "accept": false, - "reason": "Tokenizer mapping fallback bug vs flash-attention crash on `s_aux=None` are unrelated." + "reason": "Tokenizer mapping None bug vs flash_attention.py s_aux=None AttributeError; unrelated code paths." + }, + { + "left": "issue:45071", + "right": "issue:46614", + "accept": false, + "reason": "Different issues: PretrainedConfig type checking vs tokenizer phonemizer backend state." }, { "left": "issue:44291", "right": "issue:45071", "accept": false, - "reason": "`init_empty_weights` argument mismatch and `PretrainedConfig` type checking are distinct API breakages." + "reason": "Different regressions: init_empty_weights argument handling vs PretrainedConfig type checking." }, { - "left": "issue:45071", - "right": "issue:46614", + "left": "issue:45405", + "right": "issue:45446", "accept": false, - "reason": "Config type checking and phonemizer language caching are unrelated." + "reason": "Different versioning problems in different dependencies; not the same bug." }, { "left": "issue:39401", "right": "issue:44779", "accept": false, - "reason": "Different tokenizer families and failure modes: offset mapping vs incorrect overall results." + "reason": "Both tokenizer regressions, but one is offset_mapping and the other is general incorrect results for Deepseek; not clearly the same defect." }, { - "left": "issue:45405", - "right": "issue:45446", + "left": "issue:44466", + "right": "issue:46002", "accept": false, - "reason": "PEFT minimum-version packaging issue is unrelated to a flex_attention import/version guard." + "reason": "Inconsistent lm_head serialization vs NaNs in classification heads are separate problems." } ] }, @@ -14998,14 +15045,14 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 48, - "review_unit_count": 56, + "review_unit_index": 49, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:38617", "issue:39401", @@ -15014,6 +15061,7 @@ "issue:42371", "issue:42617", "issue:43066", + "issue:43097", "issue:43122", "issue:43408", "issue:43450", @@ -15028,9 +15076,9 @@ "issue:44261", "issue:44263", "issue:44410", - "issue:44466", "issue:44484", "issue:44485", + "issue:44493", "issue:44496", "issue:44829", "issue:44877", @@ -15047,7 +15095,6 @@ "issue:45542", "issue:45561", "issue:45656", - "issue:46002", "issue:46178", "issue:46209", "issue:46231", @@ -15057,9 +15104,8 @@ "issue:46617" ], "soft_pairs": [ - "issue:44466|issue:46002", - "issue:43937|issue:44977", "issue:44410|issue:45310", + "issue:43937|issue:44977", "issue:43450|issue:43986", "issue:45341|issue:45561", "issue:45341|issue:46328", @@ -15068,8 +15114,8 @@ "issue:44117|issue:45081", "issue:42617|issue:45341", "issue:38617|issue:41628", - "issue:43122|issue:46298", "issue:42371|issue:44261", + "issue:43122|issue:46298", "issue:42222|issue:46178", "issue:45362|issue:46328", "issue:43408|issue:43742", @@ -15082,220 +15128,221 @@ "issue:44410|issue:45478", "issue:44075|issue:45656", "issue:38617|issue:44220", - "issue:43066|issue:43122", "issue:44263|issue:44485", + "issue:43066|issue:43122", "issue:42222|issue:45542", + "issue:43874|issue:44261", "issue:43673|issue:45341", - "issue:43874|issue:44261" + "issue:43097|issue:44493" ], - "prepared_review_unit_hash": "2b94dee4cb326234c3038f1c0643fd4d0d629d04c9bd9e2b008b2502de5c31c8", + "prepared_review_unit_hash": "c68a91efbdb52eb9c0f8cfbfa1c7f9c61705096dc740a3839c51ead6aedddf1a", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36332, - "estimated_input_tokens": 9083, - "estimated_eval_tokens": 18422 + "serialized_chars": 36277, + "estimated_input_tokens": 9070, + "estimated_eval_tokens": 18396 }, "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. I found one strong duplicate pair: the two identical Qwen3.5 MoE `from_pretrained` error reports (#45310 and #45478). The rest are only loosely related by subsystem/theme and should not be merged.", - "confidence": 0.87, - "canonical_issue_reason": "issue:45310 is the best canonical issue because it is the earlier report and exactly matches issue:45478 in title and bug description (Qwen3.5 MoE `from_pretrained` failure on transformers>=5.4.0).", + "summary": "This cluster is mostly a grab bag of unrelated Transformers regressions and support issues, with a common surface theme of version-5 breakages, tokenizer/config/import problems, and a few model-specific bugs. I do not see a true duplicate core spanning the set.", + "confidence": 0.44, + "canonical_issue_reason": "issue:45278 is the broadest umbrella report here, covering multiple import failures after upgrading to 5.5.0, so it is the best representative of the cluster\u2019s shared upgrade/regression theme.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45310 is the strongest representative of the cluster\u2019s only clear duplicate pair, with a concrete, reproducible model-loading regression and an exact twin in issue:45478.", + "best_issue_reason": "issue:45278 best represents the cluster because it is the most general and highest-level regression report; however, the rest of the items are too heterogeneous to treat as one duplicate family.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44466", - "right": "issue:46002", + "left": "issue:44410", + "right": "issue:45310", "accept": false, - "reason": "Different bugs: serialization of `lm_head.weight` vs NaNs in classification heads after checkout/init." + "reason": "Both are Qwen-family failures, but one is missing layer projections and the other is a from_pretrained error for Qwen3.5 MoE; different concrete bugs." }, { "left": "issue:43937", "right": "issue:44977", "accept": false, - "reason": "Both involve generation/model behavior, but they describe different failures in different models and code paths." - }, - { - "left": "issue:44410", - "right": "issue:45310", - "accept": false, - "reason": "Different Qwen-family issues: missing projections in qwen3next layer 0 vs Qwen3.5 MoE `from_pretrained` error." + "reason": "GLM generation-config validation and Qwen3.5 flash-attention generation issues are unrelated code paths." }, { "left": "issue:43450", "right": "issue:43986", "accept": false, - "reason": "Different root causes: batched video processor shape bug vs crash loading video model without torchvision." + "reason": "One is a video-processor shape bug for batched input; the other is a dependency-related AutoProcessor crash without torchvision." }, { "left": "issue:45341", "right": "issue:45561", "accept": false, - "reason": "Both touch testing utils, but one is a general bug while the other is a specific xdist race on `captured_info.txt`." + "reason": "Both mention testing utils, but one is a generic bug report and the other is a pytest-xdist race on captured_info.txt; not the same bug." }, { "left": "issue:45341", "right": "issue:46328", "accept": false, - "reason": "Unrelated: testing utils bug vs `SequenceFeatureExtractor.pad` performance issue." + "reason": "testing_utils versus SequenceFeatureExtractor.pad performance; unrelated issues." }, { "left": "issue:45071", "right": "issue:46512", "accept": false, - "reason": "Both concern config handling, but one is type checking in v5.4.0 and the other is stale `config.dtype` after `from_config`." + "reason": "Both concern config/dtype behavior, but one is type-checking breakage and the other is stale model.config.dtype after from_config; different fixes." }, { "left": "issue:44829", "right": "issue:45216", "accept": false, - "reason": "Different failures: flash_attention_3 causes degenerate training, while `save_pretrained` has a checkpoint serialization regression." + "reason": "Flash-attention training degeneration and Qwen3.5 save_pretrained regression are distinct problems." }, { "left": "issue:44117", "right": "issue:45081", "accept": false, - "reason": "Both are tokenizer-related, but one is a mapping fallback bug and the other is a Mistral regex patch crash." + "reason": "Tokenizer mapping None-handling and Mistral regex patch backend_tokenizer crash are separate tokenizer-loading defects." }, { "left": "issue:42617", "right": "issue:45341", "accept": false, - "reason": "Completely different issues: 3d_parallel execution failure vs testing utils bug." + "reason": "3d_parallel.py execution failure is unrelated to a testing_utils bug." }, { "left": "issue:38617", "right": "issue:41628", "accept": false, - "reason": "Both are import errors, but they concern different missing symbols and different failure points." + "reason": "Two different import errors for different symbols; no shared underlying bug shown." }, { - "left": "issue:43122", - "right": "issue:46298", + "left": "issue:42371", + "right": "issue:44261", "accept": false, - "reason": "Different scope: tokenizer output regression vs a noisy logging message." + "reason": "TF32 settings guidance and MLA rms_norm_eps precision issue do not overlap in code path." }, { - "left": "issue:42371", - "right": "issue:44261", + "left": "issue:43122", + "right": "issue:46298", "accept": false, - "reason": "No same underlying bug: TF32 API behavior vs MLA layernorm epsilon precision error." + "reason": "Tokenizer behavior regression and a noisy logging message are unrelated." }, { "left": "issue:42222", "right": "issue:46178", "accept": false, - "reason": "Different problems: broken vitpose models vs backend mapping error for `tensorflow_text`." + "reason": "VitPose model breakage and missing tensorflow_text backend mapping are different issues." }, { "left": "issue:45362", "right": "issue:46328", "accept": false, - "reason": "Unrelated: Qwen3.5 chat crash vs a padding performance inefficiency." + "reason": "Chat crash for Qwen3.5-35B and a pad() performance issue are unrelated." }, { "left": "issue:43408", "right": "issue:43742", "accept": false, - "reason": "Different model-loading issues with different symptoms and affected models." + "reason": "Model-type warning for sam3_* and MobileLLM key error are not the same defect." }, { "left": "issue:44877", "right": "issue:46231", "accept": false, - "reason": "Different artifacts entirely: strict config loading vs documentation parameter mismatch." + "reason": "Strict config loading for granite_speech and a docs parameter-name mismatch are unrelated." }, { "left": "issue:42371", "right": "issue:46617", "accept": false, - "reason": "Different bugs: TF32 settings vs cache FileNotFoundError for trust-remote-code models." + "reason": "TF32 backend settings and trust_remote_code cache FileNotFoundError are unrelated." }, { "left": "issue:44484", "right": "issue:45310", "accept": false, - "reason": "Unrelated: a `max_shard_size` question vs a Qwen3.5 MoE loading error." + "reason": "save_pretrained shard-size question and Qwen3.5 MoE from_pretrained failure are unrelated." }, { "left": "issue:39401", "right": "issue:46209", "accept": false, - "reason": "Different model areas and failures: tokenizer offset mapping vs FP8 inference on B200." + "reason": "Tokenizer offset mapping and FP8 inference on B200 are completely different bugs." }, { "left": "issue:43937", "right": "issue:44496", "accept": false, - "reason": "Different errors: invalid GenerationConfig vs unrecognized model/config `model_type` missing." + "reason": "Invalid generation config for GLM-5 and unrecognized model config for Olmo-Hybrid are separate issues." }, { "left": "issue:44933", "right": "issue:45278", "accept": false, - "reason": "Both mention import problems, but one is a specific nonexistent import and the other is a broad upgrade regression." + "reason": "Both are import-related, but one is a specific missing image_utils import while the other is a broad upgrade regression; not the same concrete failure." }, { "left": "issue:44410", "right": "issue:45478", "accept": false, - "reason": "Different Qwen reports: missing projections in qwen3next vs the exact Qwen3.5 MoE `from_pretrained` error." + "reason": "Different Qwen model families and different failure symptoms." }, { "left": "issue:44075", "right": "issue:45656", "accept": false, - "reason": "Both concern optimizers, but one is unused SGD args and the other is double optimizer stepping with deepspeed." + "reason": "SGD args ignored and optimizer.step called twice with deepspeed are different optimizer bugs." }, { "left": "issue:38617", "right": "issue:44220", "accept": false, - "reason": "Different failures: import error in configuration utils vs audio feature extraction issue." + "reason": "ImportError for configuration_utils and _torch_extract_fbank_features are unrelated." }, { - "left": "issue:43066", - "right": "issue:43122", + "left": "issue:44263", + "right": "issue:44485", "accept": false, - "reason": "Related to tokenizer behavior in v5, but not clearly the same concrete bug or fixable in one change." + "reason": "GLM tensor split issue and GLM-5 RoPE implementation issue are separate model-specific bugs." }, { - "left": "issue:44263", - "right": "issue:44485", + "left": "issue:43066", + "right": "issue:43122", "accept": false, - "reason": "Different GLM-5 issues: `torch.split` return handling vs RoPE implementation discussion." + "reason": "Both involve tokenizer changes in v5, but one is a decoder-type mismatch and the other is changed tokenization output; not the same bug." }, { "left": "issue:42222", "right": "issue:45542", "accept": false, - "reason": "Different subsystems and symptoms: vitpose model breakage vs backend detection error for TensorFlow." + "reason": "VitPose breakage and TensorFlow backend detection are unrelated." + }, + { + "left": "issue:43874", + "right": "issue:44261", + "accept": false, + "reason": "GLM46V image-processor attribute error and MLA rms_norm_eps precision issue do not share a code path." }, { "left": "issue:43673", "right": "issue:45341", "accept": false, - "reason": "Generation cache regression vs an unrelated testing utils bug." + "reason": "Chunked prefill cache issue and testing_utils bug are unrelated." }, { - "left": "issue:43874", - "right": "issue:44261", + "left": "issue:43097", + "right": "issue:44493", "accept": false, - "reason": "Different GLM-related bugs: missing image patch count method vs missing `rms_norm_eps` in q_a_layernorm." + "reason": "Both mention config/key changes, but one is removal of a config option and the other is unexpected position-id keys; not the same underlying bug." } ] }, @@ -15310,14 +15357,14 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 49, - "review_unit_count": 56, + "review_unit_index": 50, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:27957", "issue:38617", @@ -15347,7 +15394,6 @@ "issue:44483", "issue:44484", "issue:44485", - "issue:44493", "issue:44829", "issue:44933", "issue:44987", @@ -15355,6 +15401,7 @@ "issue:45071", "issue:45341", "issue:45372", + "issue:45375", "issue:45440", "issue:45464", "issue:45478", @@ -15369,15 +15416,14 @@ "issue:46664" ], "soft_pairs": [ - "issue:43097|issue:44493", "issue:42175|issue:42222", "issue:41628|issue:45071", "issue:45071|issue:45542", - "issue:44483|issue:45464", - "issue:43931|issue:44410", "issue:44263|issue:46172", + "issue:43931|issue:44410", "issue:44360|issue:46172", "issue:43012|issue:43577", + "issue:44483|issue:45464", "issue:44484|issue:45478", "issue:27957|issue:43352", "issue:43441|issue:44829", @@ -15390,217 +15436,218 @@ "issue:45753|issue:46369", "issue:44261|issue:44485", "issue:44117|issue:44987", - "issue:43531|issue:45341", "issue:45440|issue:46172", "issue:43673|issue:46123", - "issue:43986|issue:45372", + "issue:43531|issue:45341", "issue:43097|issue:45588", "issue:43064|issue:44315", + "issue:43986|issue:45372", "issue:38617|issue:42757", - "issue:46655|issue:46664" + "issue:46655|issue:46664", + "issue:44410|issue:45375" ], - "prepared_review_unit_hash": "672c5d98198c710bfa420ed9c3ecc4cde8890d23190b8a4ee130489ef8339bda", + "prepared_review_unit_hash": "11fc709e71e003df22f92aebcb0a41e8908f3998fc962653dc12c69dd29ea260", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36159, - "estimated_input_tokens": 9040, - "estimated_eval_tokens": 18336 + "serialized_chars": 36193, + "estimated_input_tokens": 9049, + "estimated_eval_tokens": 18354 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These are mostly separate, model-specific regressions or dependency/import failures rather than duplicates. I rejected all soft pairs because they do not appear to share the same concrete bug path. For a representative issue, 43097 is the strongest cluster anchor due to its broad v5 regression impact and downstream references.", - "confidence": 0.58, - "canonical_issue_reason": "issue:43097 is the best representative because it\u2019s a concrete, broadly relevant v5 breaking-change report with downstream references, making it a stronger duplicate anchor than the more niche model-specific bugs here.", + "summary": "This cluster is mostly heterogeneous: the soft pairs are broad topical matches (same model family, same subsystem, or same class of error) rather than the same concrete bug/change. I don\u2019t see any pair that should be merged as true duplicates.", + "confidence": 0.82, + "canonical_issue_reason": "No clear canonical issue: the items span unrelated import regressions, model-specific loading bugs, training/runtime issues, and documentation problems. The soft links are too broad to indicate one shared underlying defect.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43097 is the most suitable issue to keep as the representative report for this set; it is actionable, broadly impactful, and better positioned as a canonical duplicate target than the other isolated failures.", + "best_issue_reason": "issue:45372 is the strongest standalone issue in the set: it has a clear regression, concrete import error, and active discussion, making it the most useful representative issue even though it is not a duplicate of the others.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43097", - "right": "issue:44493", - "accept": false, - "reason": "Both are v5-era model-loading regressions, but the failure modes differ: removed embeddings tie logic vs unexpected position-id keys." - }, { "left": "issue:42175", "right": "issue:42222", "accept": false, - "reason": "One is a packaging/dependency-extra issue; the other is a VitPose model breakage. Same area, not the same bug." + "reason": "Both are model/backend complaints, but one is a TensorFlow packaging/backend issue and the other is a ViTPose model breakage; different code paths and fixes." }, { "left": "issue:41628", "right": "issue:45071", "accept": false, - "reason": "AutoImageProcessor import failure and PretrainedConfig type-checking break are unrelated code paths." + "reason": "An AutoImageProcessor import failure and a PreTrainedConfig type-checking regression are unrelated bugs." }, { "left": "issue:45071", "right": "issue:45542", "accept": false, - "reason": "Both involve backend/import behavior, but one is config typing and the other is TensorBoard/TensorFlow backend detection." + "reason": "These both involve API/backend validation, but one is a config typing regression and the other is a TensorBoard/TensorFlow backend install problem." }, { - "left": "issue:44483", - "right": "issue:45464", + "left": "issue:44263", + "right": "issue:46172", "accept": false, - "reason": "Both mention chat/completions, but one is request acceptance and the other is streaming inference on Qwen3.5; not the same concrete defect." + "reason": "Different model internals and failure modes: torch.split indexing vs a dtype/transpose bug in deepseek-v4." }, { "left": "issue:43931", "right": "issue:44410", "accept": false, - "reason": "Both are Qwen-family model issues, but one is a VL weight-shape mismatch and the other is missing projections in Qwen3Next." - }, - { - "left": "issue:44263", - "right": "issue:46172", - "accept": false, - "reason": "Different math/code paths: torch.split return handling in GlmMoeDsaIndexer vs a DeepSeek comb transpose bug." + "reason": "Both mention Qwen models, but one is a weight-shape load error and the other is missing projections in layer 0; not the same defect." }, { "left": "issue:44360", "right": "issue:46172", "accept": false, - "reason": "DSA indexer missing ReLU and DeepSeek transpose behavior are unrelated implementation bugs." + "reason": "A missing ReLU in a DSA indexer is unrelated to a tensor dtype/transpose issue." }, { "left": "issue:43012", "right": "issue:43577", "accept": false, - "reason": "Both are dtype/precision complaints, but one is a PyTorch warning during bfloat16 compilation and the other is a loading bug leaving Blip2 components in float32." + "reason": "Both are dtype-related, but one is a compilation warning and the other is incorrect dtype propagation during Blip2 loading." + }, + { + "left": "issue:44483", + "right": "issue:45464", + "accept": false, + "reason": "Both touch chat/completions, but one is a general endpoint request rejection and the other is a Qwen3.5 streaming inference failure." }, { "left": "issue:44484", "right": "issue:45478", "accept": false, - "reason": "A max_shard_size question and a Qwen3.5-MoE from_pretrained error are unrelated." + "reason": "A save_pretrained shard-size question is unrelated to a Qwen3.5 MoE from_pretrained error." }, { "left": "issue:27957", "right": "issue:43352", "accept": false, - "reason": "Both concern Flash Attention 2 support, but they are different models with different support/status issues." + "reason": "Both are Flash Attention 2 support issues, but they concern different models and likely different implementation gaps." }, { "left": "issue:43441", "right": "issue:44829", "accept": false, - "reason": "Different FlashAttention problems: unsupported/failed inference for Ministral-3 vs degenerate training with flash_attention_3." + "reason": "Both mention flash attention, but one is a model load failure in v5 RC and the other is degenerate training with FA3; different problems." }, { "left": "issue:43931", "right": "issue:44038", "accept": false, - "reason": "Same model family, but one is a loading shape mismatch and the other is a Qwen3-VL-Moe issue; not the same failure path." + "reason": "Same broad Qwen3-VL/MoE area, but one is a concrete weight-shape mismatch and the other is a general version/model bug report." }, { "left": "issue:43097", "right": "issue:43452", "accept": false, - "reason": "A removed-tie-embeddings regression and a gguf_file loading regression are separate issues." + "reason": "A removed config method and a gguf_file tokenizer/model loading break are different API regressions." }, { "left": "issue:45003", "right": "issue:45341", "accept": false, - "reason": "sys.modules access in modeling_utils and a testing_utils bug do not share the same underlying defect." + "reason": "Unsafe sys.modules access in modeling_utils is unrelated to a small testing_utils bug." }, { "left": "issue:43986", "right": "issue:45341", "accept": false, - "reason": "AutoProcessor/torchvision loading failure is unrelated to the testing_utils issue." + "reason": "A video-model AutoProcessor crash without torchvision is unrelated to a testing utility bug." }, { "left": "issue:43874", "right": "issue:44263", "accept": false, - "reason": "Missing get_number_of_image_patches and torch.split return handling are different multimodal helper bugs." + "reason": "Different model families and failures: missing image-patch counting vs torch.split return-value handling." }, { "left": "issue:44933", "right": "issue:46231", "accept": false, - "reason": "A nonexistent import from image_utils and a docs parameter-name mismatch are unrelated." + "reason": "A missing import from image_utils and a documentation parameter mismatch are not the same bug." }, { "left": "issue:45753", "right": "issue:46369", "accept": false, - "reason": "Qwen3_5 infinite-loop behavior and SAM3 LiteText memory overflow are different model failures." + "reason": "Infinite-loop behavior on a specific image is unrelated to a memory overflow in another model." }, { "left": "issue:44261", "right": "issue:44485", "accept": false, - "reason": "Both are model-implementation precision issues, but one is missing rms_norm_eps in MLA layernorm and the other is a GLM-5 RoPE implementation question." + "reason": "A missing rms_norm_eps in MLA q_a_layernorm is different from a GLM-5 RoPE implementation discussion." }, { "left": "issue:44117", "right": "issue:44987", "accept": false, - "reason": "TOKENIZER_MAPPING_NAMES returning None and loading physical-intelligence/fast are not the same bug." - }, - { - "left": "issue:43531", - "right": "issue:45341", - "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and the testing_utils bug are unrelated." + "reason": "A tokenizer-mapping None issue and a model-specific loading failure overlap only at a high level; no shared concrete code path is evident." }, { "left": "issue:45440", "right": "issue:46172", "accept": false, - "reason": "Different DeepSeek generations and different defects: remote parity divergence vs a dtype/transpose expression bug." + "reason": "Both are DeepSeek-related, but one is divergence from the remote implementation and the other is a tensor dtype/transpose bug." }, { "left": "issue:43673", "right": "issue:46123", "accept": false, - "reason": "GenerationMixin cache handling during chunked prefill and MaskGenerationPipeline final-batch handling are distinct pipeline bugs." + "reason": "A generation cache regression during chunked prefill is unrelated to MaskGenerationPipeline's final partial-batch flag handling." }, { - "left": "issue:43986", - "right": "issue:45372", + "left": "issue:43531", + "right": "issue:45341", "accept": false, - "reason": "Both are import/dependency-related, but they fail through different missing symbols and different loading paths." + "reason": "A Qwen3-MoE sliding_window bug is unrelated to a testing utility issue." }, { "left": "issue:43097", "right": "issue:45588", "accept": false, - "reason": "Tie-embeddings removal and flash_attention.py crashing on s_aux=None are unrelated issues." + "reason": "A removed embedding-tie method and a flash_attention AttributeError on sink-less models are different code paths." }, { "left": "issue:43064", "right": "issue:44315", "accept": false, - "reason": "FSDP2/PEFT optimizer-state corruption and Liger Kernel not applied with model_init are different training-stack problems." + "reason": "FSDP2/PEFT optimizer-state corruption and Liger Kernel application during model_init are unrelated training bugs." + }, + { + "left": "issue:43986", + "right": "issue:45372", + "accept": false, + "reason": "A torchvision-dependent video processor crash and a mistral_common import error affecting Gemma 4 processor loading are separate dependency issues." }, { "left": "issue:38617", "right": "issue:42757", "accept": false, - "reason": "Both are import errors, but they involve different symbols from different packages and different breakages." + "reason": "Both are ImportErrors from dependency churn, but they involve different missing symbols in different packages." }, { "left": "issue:46655", "right": "issue:46664", "accept": false, - "reason": "Both hit SmolVLM video processing, but one is padding calculation and the other is the meaning of num_frames; not the same bug." + "reason": "Both are SmolVLM video-processor issues, but one is padding calculation and the other is the meaning of num_frames; related area, not the same bug." + }, + { + "left": "issue:44410", + "right": "issue:45375", + "accept": false, + "reason": "Both are Qwen3.5/MoE vision-related, but one is missing attention projections and the other is a missing config field silently dropped by strict validation." } ] }, @@ -15615,14 +15662,14 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 50, - "review_unit_count": 56, + "review_unit_index": 51, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:36683", "issue:38617", @@ -15659,7 +15706,6 @@ "issue:44936", "issue:44938", "issue:45103", - "issue:45375", "issue:45478", "issue:45479", "issue:45584", @@ -15671,16 +15717,17 @@ "issue:46178", "issue:46298", "issue:46365", - "issue:46421" + "issue:46421", + "issue:46701" ], "soft_pairs": [ - "issue:44410|issue:45375", "issue:43097|issue:46050", "issue:44038|issue:44410", "issue:43986|issue:44484", "issue:42890|issue:46298", "issue:38617|issue:44351", "issue:44062|issue:44117", + "issue:44117|issue:46701", "issue:43097|issue:43873", "issue:44936|issue:46365", "issue:43502|issue:43992", @@ -15690,8 +15737,8 @@ "issue:43986|issue:44117", "issue:43986|issue:44016", "issue:43208|issue:43550", - "issue:43825|issue:44655", "issue:42907|issue:44655", + "issue:43825|issue:44655", "issue:42907|issue:45753", "issue:43012|issue:43550", "issue:45712|issue:46298", @@ -15709,232 +15756,238 @@ "issue:44380|issue:45479", "issue:36683|issue:43582" ], - "prepared_review_unit_hash": "5587e734875a329815994b5244c8a8360bb837479de6545aaaae6a520bf692b4", + "prepared_review_unit_hash": "3b1d945546bac57eadab231f378b8a55d78f8643d5f02c19cc398371adaf91d0", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 37652, - "estimated_input_tokens": 9413, - "estimated_eval_tokens": 19082 + "serialized_chars": 37631, + "estimated_input_tokens": 9408, + "estimated_eval_tokens": 19072 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a loose collection of unrelated Transformers bug reports; the soft links are driven by broad topical similarity (tokenizers/config/imports/save-load/backend quirks) rather than true duplicate-level sameness. I did not find any PRs in the set.", - "confidence": 0.93, - "canonical_issue_reason": "issue:43873 is the most active open report with the highest discussion/inbound traffic, so it is the best representative anchor among these issues, though it does not actually subsume the rest.", + "summary": "These items are mostly unrelated standalone Transformers issues spanning different models, loaders, serialization paths, and runtime backends. I found no soft-edge pair that clearly represents the same underlying bug/change, so no pair is accepted.", + "confidence": 0.31, + "canonical_issue_reason": "issue:43873 has the strongest general signal in this set (high discussion and inbound references) and is a broad, active bug report in a core feature area, making it the best representative placeholder canonical issue.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43873 has the strongest engagement and is user-facing, making it the most suitable single issue to anchor triage for this loose cluster.", + "best_issue_reason": "issue:43873 is the most prominent issue here by activity and scope; among otherwise unrelated reports, it is the most suitable single issue to anchor the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:44410", - "right": "issue:45375", - "accept": false, - "reason": "Different bugs: missing layer projections in qwen3next vs missing config field dropped by strict config handling." - }, { "left": "issue:43097", "right": "issue:46050", "accept": false, - "reason": "One is a v5 tie_embeddings API/removal complaint; the other is quantized save/load corruption after from_pretrained." + "reason": "Both involve v5/serialization-adjacent behavior, but one is an API removal warning and the other is quantized save/load breakage; different failure modes and code paths." }, { "left": "issue:44038", "right": "issue:44410", "accept": false, - "reason": "Generic Qwen3-VL-Moe/v5 issue vs a specific qwen3next missing-projection bug; not the same code path." + "reason": "Both mention Qwen/VL-family models, but one is a transformers 5.0 compatibility bug and the other is missing projections in qwen3next; not the same defect." }, { "left": "issue:43986", "right": "issue:44484", "accept": false, - "reason": "AutoProcessor/torchvision import crash vs a save_pretrained shard-size question." + "reason": "One is a loading crash without torchvision, the other is a save_pretrained shard-size question; unrelated problems." }, { "left": "issue:42890", "right": "issue:46298", "accept": false, - "reason": "Test flakiness from missing set_seed vs a noisy logging message; unrelated symptoms and fixes." + "reason": "A flaky test due to missing set_seed is unrelated to a noisy logging message." }, { "left": "issue:38617", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but for different missing symbols and different API regressions." + "reason": "Both are import errors, but they concern different missing symbols from different migration paths; not one shared bug." }, { "left": "issue:44062", "right": "issue:44117", "accept": false, - "reason": "AddedToken kwarg collision vs tokenizer mapping returning None; distinct tokenizer loader failures." + "reason": "Different tokenizer bugs: AddedToken argument handling versus TOKENIZER_MAPPING_NAMES returning None." + }, + { + "left": "issue:44117", + "right": "issue:46701", + "accept": false, + "reason": "Tokenizer mapping lookup failure is unrelated to a DiffusionGemma image-text pipeline failure." }, { "left": "issue:43097", "right": "issue:43873", "accept": false, - "reason": "API removal/tie-embeddings warning is not the same concrete bug as quantization offloading behavior." + "reason": "Tie-embeddings removal and offloading-with-quantization are separate regressions with different root causes." }, { "left": "issue:44936", "right": "issue:46365", "accept": false, - "reason": "trainer.evaluate failure after train is unrelated to failing to import Trainer because of datasets." + "reason": "trainer.evaluate() after train() is unrelated to import-time Trainer breakage caused by datasets." }, { "left": "issue:43502", "right": "issue:43992", "accept": false, - "reason": "local_files_only still causing network requests vs missing embed_tokens.weight on UMT5Encoder load." + "reason": "One is incorrect network access under local_files_only, the other is a missing UMT5Encoder weight during loading." }, { "left": "issue:44704", "right": "issue:45103", "accept": false, - "reason": "cached_file kwargs forwarding bug vs docstring processing crash under future annotations." + "reason": "AutoProcessor kwargs forwarding and auto_docstring future-annotations crash are different subsystems." }, { "left": "issue:44380", "right": "issue:46421", "accept": false, - "reason": "GPT2 attention scaling ignored by backend vs cache lazy init causing recompilation; different mechanisms." + "reason": "GPT2 attention scaling with SDPA/FlashAttention and cache lazy-init recompilation are separate backend issues." }, { "left": "issue:44938", "right": "issue:46178", "accept": false, - "reason": "Python 3.14 import/load failure vs unsupported backend tensorflow_text; not the same bug." + "reason": "Python 3.14 import failure and tensorflow_text backend validation are unrelated compatibility errors." }, { "left": "issue:43986", "right": "issue:44117", "accept": false, - "reason": "Video-model AutoProcessor crash without torchvision is unrelated to tokenizer mapping None handling." + "reason": "Video-model loading without torchvision does not match TOKENIZER_MAPPING_NAMES returning None." }, { "left": "issue:43986", "right": "issue:44016", "accept": false, - "reason": "AutoProcessor crash vs notebook syntax error; different layers of the stack." + "reason": "A missing torchvision dependency crash is unrelated to a notebook syntax error." }, { "left": "issue:43208", "right": "issue:43550", "accept": false, - "reason": "xLSTM training bugs and Bamba torch.compile/SDPA failure are separate model-specific issues." + "reason": "xLSTM training bugs and Bamba torch.compile/SDPA failure are different model/backend problems." }, { - "left": "issue:43825", + "left": "issue:42907", "right": "issue:44655", "accept": false, - "reason": "Wrong pipeline error message vs inability to save Pipeline objects; not the same change." + "reason": "Dequantized model saving failure is not the same as pipeline save_pretrained limitations." }, { - "left": "issue:42907", + "left": "issue:43825", "right": "issue:44655", "accept": false, - "reason": "Dequantized model save failure vs Pipeline save_pretrained failure; different objects and code paths." + "reason": "A v5 pipeline error message issue is unrelated to saving Pipeline objects." }, { "left": "issue:42907", "right": "issue:45753", "accept": false, - "reason": "Model save/load bug vs Qwen3_5 image infinite loop; unrelated failures." + "reason": "Model dequantized-save failure and Qwen3_5 infinite loop on a specific image are different bugs." }, { "left": "issue:43012", "right": "issue:43550", "accept": false, - "reason": "PyTorch bfloat16 warning during compile vs a Bamba SDPA compile error; different failure modes." + "reason": "bf16 compile warning and Bamba SDPA compile failure are not the same concrete defect." }, { "left": "issue:45712", "right": "issue:46298", "accept": false, - "reason": "Leftover dummy classes in dir(transformers) vs a noisy logging message; unrelated cleanup/logging issues." + "reason": "Leftover dummy classes leaking into dir(transformers) is unrelated to a noisy logging message." }, { "left": "issue:43012", "right": "issue:44368", "accept": false, - "reason": "Both mention warnings, but one is compile precision and the other is tie_word_embeddings; not the same bug." + "reason": "PyTorch precision warning and a tie_word_embeddings warning are distinct warning sources." }, { "left": "issue:43066", "right": "issue:43906", "accept": false, - "reason": "Wrong tokenizer decoder type in v5 vs an isolated reproduction of a different issue." + "reason": "Tokenizer decoder type mismatch and a reproduction of issue 38071 do not appear to be the same bug." }, { "left": "issue:43992", "right": "issue:44484", "accept": false, - "reason": "Missing pretrained weight during load vs why save_pretrained uses 50GB shard size; different concerns." + "reason": "Missing embedding weights on load and a max_shard_size question are unrelated." }, { "left": "issue:43352", "right": "issue:44380", "accept": false, - "reason": "Flash Attention 2 not supported yet vs GPT2 scaling ignored under SDPA/FlashAttention; unrelated backend/model bugs." + "reason": "Nemotron FA2 unsupported and GPT2 attention scaling under SDPA/FlashAttention are different backend/model issues." }, { "left": "issue:43352", "right": "issue:43976", "accept": false, - "reason": "Model-specific FA2 support issue vs Python version compatibility regression." + "reason": "Unsupported Flash Attention 2 for Nemotron is unrelated to the Python version compatibility report." }, { "left": "issue:43097", "right": "issue:46298", "accept": false, - "reason": "Tie_embeddings API removal/warning is unrelated to a noisy logging message." + "reason": "A removed config method and a noisy log message are not the same underlying issue." }, { "left": "issue:44162", "right": "issue:46172", "accept": false, - "reason": "ESM2 breakage vs DeepSeek tensor dtype/transpose issue; different models and code paths." + "reason": "ESM2 breakage in workflows and a DeepSeek-V4 tensor transpose bug are unrelated model-specific failures." }, { "left": "issue:45584", "right": "issue:46365", "accept": false, - "reason": "Whisper empty transcription bug is unrelated to importing Trainer with datasets installed." + "reason": "Whisper empty-transcription generation failure is unrelated to importing Trainer." }, { "left": "issue:44380", "right": "issue:46132", "accept": false, - "reason": "GPT2 backend scaling bug vs AttentionInterface.register behavior change; different APIs." + "reason": "GPT2 backend-specific attention scaling and AttentionInterface.register behavior change are different code paths." }, { "left": "issue:44077", "right": "issue:46365", "accept": false, - "reason": "patchtsmixer post_init policy issue vs Trainer import failure; no shared underlying defect." + "reason": "patchtsmixer post_init validation and Trainer import failure are unrelated." }, { "left": "issue:44222", "right": "issue:45478", "accept": false, - "reason": "Both involve Qwen3.5/Moe save/load, but the versions, failure points, and concrete bugs differ." + "reason": "FP8 save_pretrained for MoE and Qwen3.5 MoE from_pretrained error are both MoE-related but not the same concrete bug or mergeable fix." + }, + { + "left": "issue:44380", + "right": "issue:45479", + "accept": false, + "reason": "Attention scaling ignored in specific backends is unrelated to single-label-classification zero-loss degeneration." }, { "left": "issue:36683", "right": "issue:43582", "accept": false, - "reason": "Gemma3Config missing vocab_size vs Apple Silicon caching_allocator_warmup TypeError; unrelated issues." + "reason": "Gemma3Config missing vocab_size and Apple Silicon caching_allocator_warmup TypeError are unrelated." } ] }, @@ -15949,14 +16002,14 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 51, - "review_unit_count": 56, + "review_unit_index": 52, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:36010", "issue:36683", @@ -16014,8 +16067,8 @@ "issue:42757|issue:43824", "issue:43012|issue:44683", "issue:43208|issue:43526", - "issue:44483|issue:44514", "issue:45600|issue:46298", + "issue:44483|issue:44514", "issue:43618|issue:44380", "issue:45698|issue:46172", "issue:44038|issue:44230", @@ -16028,11 +16081,11 @@ "issue:44625|issue:46365", "issue:45276|issue:46077", "issue:42898|issue:43066", - "issue:44483|issue:45290", "issue:36010|issue:42757", - "issue:43901|issue:44655", + "issue:44483|issue:45290", "issue:43352|issue:43937", "issue:43611|issue:44655", + "issue:43901|issue:44655", "issue:36683|issue:44077", "issue:43976|issue:44938", "issue:43867|issue:45216", @@ -16042,232 +16095,232 @@ "issue:43618|issue:45216", "issue:36683|issue:44625" ], - "prepared_review_unit_hash": "2884808e298c410320613dab9c3933823f3d7bab567fc7df2ba86bead4ca518b", + "prepared_review_unit_hash": "04efc25ef308e3e23522f86126aa95033fe71510d5d27e368cce2209ae6c1ec6", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 33, - "serialized_chars": 37199, - "estimated_input_tokens": 9300, - "estimated_eval_tokens": 18856 + "serialized_chars": 37204, + "estimated_input_tokens": 9301, + "estimated_eval_tokens": 18858 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is heterogeneous and does not form a clean duplicate cluster. Most soft pairs are only superficially similar (same subsystem, same model family, or same broad API area) but describe different bugs or feature requests, so I rejected all of them.", - "confidence": 0.82, - "canonical_issue_reason": "Issue 45698 is the strongest representative by discussion volume and inbound references, and it describes a concrete user-facing save/load regression. There is no single duplicate hub across the full set, but this is the best anchor among the listed issues.", + "summary": "This cluster is mostly a set of superficial transformer-regression lookalikes (import errors, save/load bugs, config mismatches, chat-template issues, backend-specific model bugs). I don\u2019t see any soft pair that is clearly the same underlying bug/change, so all candidate merges should be rejected. If one issue must act as a representative, the save/load regression in 45698 is the broadest fit, but it is still not a true duplicate umbrella.", + "confidence": 0.89, + "canonical_issue_reason": "45698 is the closest broad representative because it centers on save_pretrained/from_pretrained serialization behavior, which is the only theme that recurs across multiple items; however, the cluster is too heterogeneous for it to be a true canonical duplicate target.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45698 is the most suitable global representative: it has the most activity/inbound references and a clear, actionable bug in the save/load path. It is still not a duplicate of the rest of the set.", + "best_issue_reason": "45698 is the best single issue to represent the cluster because it is active, broad, and aligned with the most repeated save/load-related theme in the set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:42175", "right": "issue:46365", "accept": false, - "reason": "Both mention import/backend setup, but one is about torch extra installs not pulling TensorFlow and the other is about Trainer import breaking with datasets installed; different failures." + "reason": "Both mention import/training availability, but one is backend packaging and the other is a datasets-installed Trainer import conflict; unrelated root causes." }, { "left": "issue:43064", "right": "issue:44380", "accept": false, - "reason": "Different code paths: FSDP2/PEFT optimizer state mismatch vs GPT-2 attention scaling being ignored under SDPA/FlashAttention." + "reason": "Different code paths: distributed optimizer-state corruption vs attention-scaling being ignored under SDPA/FlashAttention." }, { "left": "issue:42490", "right": "issue:43012", "accept": false, - "reason": "Local-model-path slash warning is unrelated to a bfloat16 compilation precision warning." + "reason": "One is a local-path import warning, the other is a PyTorch precision warning while compiling bfloat16 models." }, { "left": "issue:42757", "right": "issue:43824", "accept": false, - "reason": "Both are ImportErrors, but for different missing symbols from different packages and different root causes." + "reason": "Both are import errors, but for different symbols and different dependency surfaces (huggingface_hub vs transformers model export)." }, { "left": "issue:43012", "right": "issue:44683", "accept": false, - "reason": "One is a PyTorch warning about float32 precision in bfloat16 compilation; the other is a torch>=2.9 flex_attention failure." + "reason": "Both involve PyTorch/compile behavior, but one is a warning and the other is a hard failure on newer torch; not the same bug." }, { "left": "issue:43208", "right": "issue:43526", "accept": false, - "reason": "xLSTM training bugs and BeitImageProcessorFast label reduction are unrelated subsystems and failure modes." + "reason": "xLSTM training bugs and BeitImageProcessorFast label reduction are unrelated model/data-processing issues." }, { - "left": "issue:44483", - "right": "issue:44514", + "left": "issue:45600", + "right": "issue:46298", "accept": false, - "reason": "One is server-side /v1/chat/completions request rejection; the other is client-side chat template batching crash with padding=False." + "reason": "auto_mappings CI breakage from removed config references is not the same as a noisy logging message." }, { - "left": "issue:45600", - "right": "issue:46298", + "left": "issue:44483", + "right": "issue:44514", "accept": false, - "reason": "Sam3LiteText auto-mapping breakage is not the same as a noisy logging message." + "reason": "Both touch chat/completions, but one is an API acceptance problem and the other is a processor crash on batched tool-call inputs." }, { "left": "issue:43618", "right": "issue:44380", "accept": false, - "reason": "CLIP output field regression and GPT-2 attention scaling under alternative attention backends are different bugs." + "reason": "CLIP output-attention regression and GPT2 attention scaling under backend kernels are different model behaviors." }, { "left": "issue:45698", "right": "issue:46172", "accept": false, - "reason": "save_pretrained/from_pretrained custom-module mixup is unrelated to the deepseek-v4 transpose/dtype bug." + "reason": "Save/load custom-module resolution is unrelated to the deepseek-v4 transpose/dtype bug." }, { "left": "issue:44038", "right": "issue:44230", "accept": false, - "reason": "Same Qwen3-VL/Moe family, but one is a loading regression and the other is an fp8 support request; not the same bug." + "reason": "Both concern Qwen3-VL/MoE, but one is a loading bug and the other is fp8 inference support; not the same fix." }, { "left": "issue:43618", "right": "issue:43761", "accept": false, - "reason": "Both are CLIP v5 regressions, but they affect different returned fields (attentions vs hidden_states) and are not clearly the same defect." + "reason": "Both are CLIP regressions, but one loses attentions and the other loses hidden_states; they are separate outputs and fixes." }, { "left": "issue:38617", "right": "issue:43502", "accept": false, - "reason": "Missing import from configuration_utils and local_files_only still making API requests are unrelated issues." + "reason": "Importing layer_type_validation and ignoring local_files_only are unrelated dependency vs network-policy bugs." }, { "left": "issue:43756", "right": "issue:44485", "accept": false, - "reason": "Both mention RoPE, but one is about Smollm3 dropping RoPE layers and the other is a GLM-5 RoPE implementation discussion." + "reason": "Smollm3 RoPE-layer counting and GLM-5 RoPE implementation are different model-specific architecture questions." }, { "left": "issue:43299", "right": "issue:43867", "accept": false, - "reason": "Both are model-loading problems, but with different symptoms and causes; not enough evidence for a single underlying bug." + "reason": "Qwen3VL MoE loading regression and sorted-state_dict load failure are distinct loading problems." }, { "left": "issue:42994", "right": "issue:44655", "accept": false, - "reason": "Quantized model saving failure and Pipeline.save_pretrained failure are different object types and save paths." + "reason": "Quantized-model save failure and inability to save Pipeline objects both involve saving, but the affected objects and failure modes differ." }, { "left": "issue:36683", "right": "issue:43475", "accept": false, - "reason": "Missing Gemma3Config.vocab_size and missing Sam3VisionEncoderOutput.fpn_position_embeddings are unrelated missing-attribute bugs." + "reason": "Gemma3Config missing vocab_size and SAM3 vision output missing fpn_position_embeddings are separate missing-attribute bugs in different models." }, { "left": "issue:44625", "right": "issue:46365", "accept": false, - "reason": "Qwen3.5 config propagation and Trainer import interference with datasets are unrelated." + "reason": "Qwen3.5 num_labels propagation and Trainer import conflicts with datasets are unrelated." }, { "left": "issue:45276", "right": "issue:46077", "accept": false, - "reason": "Both are Gemma4 issues, but one is resizing embeddings and the other is bidirectional attention mask construction; different code paths." + "reason": "Both are Gemma4-related, but one is embedding resize propagation and the other is attention-mask construction; separate code paths." }, { "left": "issue:42898", "right": "issue:43066", "accept": false, - "reason": "Tokenizer v5 behavior changes are broad, but these are distinct regressions: clean_up_tokenization_spaces vs decoder type selection." - }, - { - "left": "issue:44483", - "right": "issue:45290", - "accept": false, - "reason": "Chat completions request validation and apply_chat_template tokenization crash are different layers and different failure conditions." + "reason": "Tokenizer cleanup behavior in v5 and wrong tokenizer decoder type are different tokenizer regressions." }, { "left": "issue:36010", "right": "issue:42757", "accept": false, - "reason": "Different import symbols from different modules; no shared underlying bug." + "reason": "Different missing imports from different modules and different dependency updates." }, { - "left": "issue:43901", - "right": "issue:44655", + "left": "issue:44483", + "right": "issue:45290", "accept": false, - "reason": "Documentation mismatch for return_all_scores is unrelated to pipeline serialization failure." + "reason": "Both involve chat-template / chat-completions, but one is request validation and the other is tokenization crash on assistant tool-call messages." }, { "left": "issue:43352", "right": "issue:43937", "accept": false, - "reason": "FlashAttention 2 unsupported for Nemotron and GLM-5 generation config validation are different issues." + "reason": "Flash Attention 2 support for NemotronH and invalid GenerationConfig for GLM-5 are unrelated model-specific issues." }, { "left": "issue:43611", "right": "issue:44655", "accept": false, - "reason": "Base_model_prefix loading regression and pipeline save_pretrained failure are unrelated." + "reason": "Model loading with base_model_prefix and saving Pipelines are different serialization behaviors." + }, + { + "left": "issue:43901", + "right": "issue:44655", + "accept": false, + "reason": "Docs mismatch for return_all_scores and Pipeline save_pretrained failure are not the same issue." }, { "left": "issue:36683", "right": "issue:44077", "accept": false, - "reason": "Different configs and different failures: missing vocab_size vs patchtsmixer post_init policy." + "reason": "Missing Gemma config attribute and patchtsmixer post_init validity are unrelated model config constraints." }, { "left": "issue:43976", "right": "issue:44938", "accept": false, - "reason": "Both are Python-version compatibility complaints, but they target different interpreter versions and likely different packaging issues." + "reason": "Python version support breakage in transformers 5.1.0 and Python 3.14 load failures are different compatibility issues." }, { "left": "issue:43867", "right": "issue:45216", "accept": false, - "reason": "A sorted-state_dict load error is not the same as a Qwen3.5 checkpoint-save regression." + "reason": "Sorted state_dict loading error and incorrect Qwen3.5 saved checkpoint are both save/load-related, but the concrete failure modes do not match." }, { "left": "issue:44625", "right": "issue:46077", "accept": false, - "reason": "Qwen3.5 num_labels propagation and Gemma4 attention masks are unrelated." + "reason": "Different Qwen3.5/Gemma4 model families and different bugs." }, { "left": "issue:46077", "right": "issue:46082", "accept": false, - "reason": "Gemma4 attention-mask logic and LlamaConfig head_dim validation are separate config behaviors." + "reason": "Gemma4 attention-mask behavior and LlamaConfig head_dim validation are unrelated config/model-shape issues." }, { "left": "issue:36683", "right": "issue:43525", "accept": false, - "reason": "Both are missing config attributes, but in different model families and on different fields." + "reason": "Two AttributeErrors, but for different config fields in different models." }, { "left": "issue:43618", "right": "issue:45216", "accept": false, - "reason": "CLIP output field regression is unrelated to Qwen3.5 checkpoint corruption." + "reason": "CLIP attention output regression and Qwen3.5 checkpoint corruption are separate bugs." }, { "left": "issue:36683", "right": "issue:44625", "accept": false, - "reason": "Different model families and different missing config propagation bugs." + "reason": "Gemma3Config missing vocab_size and Qwen3.5 num_labels propagation are different config-synchronization bugs." } ] }, @@ -16282,14 +16335,14 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 52, - "review_unit_count": 56, + "review_unit_index": 53, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:28282", "issue:33453", @@ -16341,14 +16394,14 @@ "issue:46664" ], "soft_pairs": [ - "issue:43064|issue:44492", "issue:43992|issue:45310", + "issue:43064|issue:44492", "issue:44075|issue:46365", "issue:43010|issue:45278", "issue:44355|issue:44855", "issue:44938|issue:45071", - "issue:43867|issue:46202", "issue:43867|issue:45375", + "issue:43867|issue:46202", "issue:36683|issue:43064", "issue:42890|issue:45561", "issue:41669|issue:44246", @@ -16360,32 +16413,32 @@ "issue:43992|issue:45478", "issue:28282|issue:33453", "issue:43618|issue:46285", - "issue:43618|issue:44857", "issue:44360|issue:45440", + "issue:43618|issue:44857", "issue:42890|issue:44964", "issue:42890|issue:44112", "issue:42898|issue:46664", "issue:42890|issue:45600", - "issue:43874|issue:46617", "issue:43502|issue:44062", + "issue:43874|issue:46617", "issue:43723|issue:44556", "issue:44222|issue:45125", "issue:44734|issue:46617" ], - "prepared_review_unit_hash": "e51d9b18f95750252831f8f368453b663fb4f82df44aa4c73e42f4f1aadd64e5", + "prepared_review_unit_hash": "ff9ba39342a7407dc307aa18a48639a5d48533bd9d82c58be1df391966d23ec7", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36256, + "serialized_chars": 36255, "estimated_input_tokens": 9064, "estimated_eval_tokens": 18384 }, @@ -16394,192 +16447,186 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is mostly a collection of unrelated transformer issues that are only loosely similar by topic. I found one true duplicate pair: the two Qwen3.5 MoE from_pretrained regression reports. The rest should stay separate.", - "confidence": 0.94, - "canonical_issue_reason": "Issue 45310 is the cleanest canonical for the only exact duplicate pair: it states the concrete Qwen3.5 MoE from_pretrained regression, while 45478 is essentially the same report later on.", + "summary": "This is a broad cluster of mostly unrelated v5-era import/load/regression reports. The soft pairs are mostly only loosely similar on surface wording, and none meet a strict duplicate threshold.", + "confidence": 0.57, + "canonical_issue_reason": "Issue 45278 is the broadest representative symptom in the set: a general breakage report after upgrading from 4.57.0 to 5.5.0, matching the cluster\u2019s overall import/load regression theme better than the narrower model-specific bugs.", "canonical_pr_reason": null, - "best_issue_reason": "45310 is the best representative issue overall because it is specific, actionable, and matches the only confirmed duplicate edge in the cluster.", + "best_issue_reason": "Issue 45278 is the best single anchor for the cluster because it captures the widest user-visible regression among the listed issues and is most likely to subsume nearby upgrade/import failures.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43064", - "right": "issue:44492", + "left": "issue:43992", + "right": "issue:45310", "accept": false, - "reason": "Different bugs: one is a Trainer/FSDP2 optimizer-state problem, the other is a typo in cache strategy docs/code." + "reason": "Both are loading errors, but they concern different models and different failure modes (missing embed_tokens.weight vs Qwen3.5 Moe from_pretrained error)." }, { - "left": "issue:43992", - "right": "issue:45310", + "left": "issue:43064", + "right": "issue:44492", "accept": false, - "reason": "Different model/code paths: UMT5Encoder missing embed_tokens.weight versus Qwen3.5 MoE from_pretrained failure." + "reason": "Optimizer-state corruption in training is unrelated to a cache-strategy typo." }, { "left": "issue:44075", "right": "issue:46365", "accept": false, - "reason": "Unrelated symptoms: SGD args ignored versus Trainer import depending on datasets presence." + "reason": "One is about ignored SGD args; the other is an import dependency interaction with datasets." }, { "left": "issue:43010", "right": "issue:45278", "accept": false, - "reason": "Both are broad v5 complaints, but they describe different problems and no shared concrete code path." + "reason": "Cache/update no_grad behavior is a concrete code-path issue; the other is a broad import regression report, not the same bug." }, { "left": "issue:44355", "right": "issue:44855", "accept": false, - "reason": "Both involve loading/importing Python files, but one is compiled-file errors and the other is a JIT parsing/IndentationError issue." + "reason": "Generic compiled-file import failures are different from the specific DebertaV2 JIT parsing/comment issue." }, { "left": "issue:44938", "right": "issue:45071", "accept": false, - "reason": "Different failures: Python 3.14 import/load breakage versus PretrainedConfig type-checking regression." + "reason": "Python 3.14 load failure and PretrainedConfig type-checking regression are different problems." }, { "left": "issue:43867", - "right": "issue:46202", + "right": "issue:45375", "accept": false, - "reason": "State_dict loading versus Gemma pipeline roundtrip token handling are different bugs." + "reason": "A state_dict loading-order problem is not the same as a missing vision-config field being dropped by strict validation." }, { "left": "issue:43867", - "right": "issue:45375", + "right": "issue:46202", "accept": false, - "reason": "One is a model load/state_dict issue; the other is a missing config field silently dropped by strict handling." + "reason": "Model load failure and Gemma token roundtrip failure are unrelated." }, { "left": "issue:36683", "right": "issue:43064", "accept": false, - "reason": "Gemma3Config vocab_size attribute error and FSDP2 optimizer-state corruption are unrelated." + "reason": "Gemma3Config missing vocab_size is unrelated to the FSDP2/PEFT optimizer-state bug." }, { "left": "issue:42890", "right": "issue:45561", "accept": false, - "reason": "One is missing set_seed in integration tests; the other is a pytest-xdist file race in testing utilities." + "reason": "Both are test failures, but one is seed-related flakiness and the other is an xdist file race." }, { "left": "issue:41669", "right": "issue:44246", "accept": false, - "reason": "Both concern import performance, but one targets import * bloat and the other reports intermittent slow imports more generally." + "reason": "Both mention import slowness, but one is a specific import-star cleanup proposal and the other is an intermittent import slowdown report without the same concrete cause." }, { "left": "issue:43010", "right": "issue:45850", "accept": false, - "reason": "Generic v5 regressions, but not the same underlying defect." + "reason": "A cache no_grad fix and a broad 'too many models break' complaint are not the same underlying change or bug." }, { "left": "issue:36296", "right": "issue:43575", "accept": false, - "reason": "Tensor-parallel training bug and Qwen2-57B-A14B-Instruct OOM during tp load are not the same code-path problem." + "reason": "Tensor-parallel training bug and tensor-parallel OOM on model load are distinct symptoms with different likely causes." }, { "left": "issue:43867", "right": "issue:45357", "accept": false, - "reason": "Different issues: load failure with sorted state_dict versus incorrect visual encoder keys on save_pretrained." + "reason": "Loading error from sorted state_dict keys is not the same as saving incorrect visual encoder keys." }, { "left": "issue:43618", "right": "issue:44077", "accept": false, - "reason": "Different model behaviors: CLIPOutput attentions regression versus patchtsmixer post_init policy change." + "reason": "CLIPOutput attentions assignment and patchtsmixer post_init policy are unrelated." }, { "left": "issue:43010", "right": "issue:43976", "accept": false, - "reason": "No shared bug: cache/layer no_grad decoration versus Python version compatibility regression." + "reason": "Cache no_grad behavior is unrelated to a Python-version compatibility failure." }, { "left": "issue:43992", "right": "issue:45478", "accept": false, - "reason": "Different model families: UMT5Encoder missing weights versus Qwen3.5 MoE from_pretrained error." + "reason": "Different models and different load errors; not the same duplicate." }, { "left": "issue:28282", "right": "issue:33453", "accept": false, - "reason": "PyTorch missing ImportError for AutoModel is unrelated to tokenizer-loading regression." + "reason": "Missing PyTorch causes an ImportError, while tokenizer loading regression is a separate issue." }, { "left": "issue:43618", "right": "issue:46285", "accept": false, - "reason": "Different CLIP-related issues: attentions assignment versus inconsistent class structure after v5." + "reason": "Both are CLIP-related, but one is about attentions fields and the other is an architectural inconsistency in text_model presence." }, { - "left": "issue:43618", - "right": "issue:44857", + "left": "issue:44360", + "right": "issue:45440", "accept": false, - "reason": "CLIP attentions regression and LwDetrImageLoss AMP/CUDA crash are unrelated." + "reason": "A DSA/ReLU discussion is unrelated to DeepseekV3MoE divergence from remote implementation." }, { - "left": "issue:44360", - "right": "issue:45440", + "left": "issue:43618", + "right": "issue:44857", "accept": false, - "reason": "DSA indexer ReLU discussion and DeepseekV3MoE divergence are different implementation issues." + "reason": "CLIP attentions handling and a float16 AMP crash in LwDetrImageLoss are unrelated." }, { "left": "issue:42890", "right": "issue:44964", "accept": false, - "reason": "Test seeding failure is unrelated to loading Phi-4 multimodal model." + "reason": "Seed-related test flakiness is not the same as a Phi-4 multimodal loading failure." }, { "left": "issue:42890", "right": "issue:44112", "accept": false, - "reason": "Different CI/test problems: missing set_seed versus stale device override test on GraniteSpeech." + "reason": "Both are CI/test issues, but they have different roots and different failing components." }, { "left": "issue:42898", "right": "issue:46664", "accept": false, - "reason": "Tokenization cleanup behavior in v5 and SmolVLM num_frames semantics are unrelated." + "reason": "Tokenization cleanup behavior and video num_frames semantics are different API changes." }, { - "left": "issue:42890", - "right": "issue:45600", + "left": "issue:43502", + "right": "issue:44062", "accept": false, - "reason": "Different test/CI bugs with no shared failing code path." + "reason": "Offline-loading API requests leaking is unrelated to AddedToken receiving duplicate special kwargs." }, { "left": "issue:43874", "right": "issue:46617", "accept": false, - "reason": "Image processor token counting bug versus trust_remote_code cache FileNotFoundError are unrelated." - }, - { - "left": "issue:43502", - "right": "issue:44062", - "accept": false, - "reason": "local_files_only network leakage and AddedToken keyword collision are distinct tokenizer/loading bugs." + "reason": "A missing image-patch-count method and a cache FileNotFoundError are different load failures." }, { "left": "issue:43723", "right": "issue:44556", "accept": false, - "reason": "Both are loading-related, but one is tokenizer loading in v5 and the other is checkpoint reload incompatibility after upgrade." + "reason": "Both are v5 loading regressions, but one is tokenizer loading and the other is checkpoint reload compatibility; too broad to merge." }, { "left": "issue:44222", "right": "issue:45125", "accept": false, - "reason": "FP8 save_pretrained for MoE and missing _tp_plan for tensor parallelism are different features and failures." + "reason": "FP8 save_pretrained behavior and missing tensor-parallel plan are unrelated." }, { "left": "issue:44734", "right": "issue:46617", "accept": false, - "reason": "KV-cache indexing crash in transformers serve is unrelated to loading a trust_remote_code model from cache." + "reason": "KV-cache continuation tensor indexing and trust_remote_code cache file lookup are different bugs." } ] }, @@ -16594,14 +16641,14 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 53, - "review_unit_count": 56, + "review_unit_index": 54, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:27957", "issue:30333", @@ -16664,8 +16711,8 @@ "issue:42548|issue:44351", "issue:43010|issue:43232", "issue:30990|issue:31795", - "issue:43502|issue:44961", "issue:43502|issue:44222", + "issue:43502|issue:44961", "issue:43827|issue:44961", "issue:30333|issue:36296", "issue:43299|issue:44556", @@ -16692,154 +16739,154 @@ "issue:41628|issue:42548", "issue:44485|issue:46664" ], - "prepared_review_unit_hash": "c6361879d710beb6b1fd277f0aaf232c5d548dfc1a06ae063c1bb2673626d5b4", + "prepared_review_unit_hash": "9aaba30b02073f4b5300869b96fc9c49eb8793237766bae54a988205265d4910", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 39, - "serialized_chars": 38113, - "estimated_input_tokens": 9529, - "estimated_eval_tokens": 19314 + "serialized_chars": 38108, + "estimated_input_tokens": 9527, + "estimated_eval_tokens": 19310 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These are mostly soft title-similarity false positives: import errors, loading regressions, docs, quantization, tensor-parallel, and model-specific bugs. None of the candidate pairs looks like the same concrete underlying bug or a single mergeable fix.", - "confidence": 0.96, - "canonical_issue_reason": "Issue 43576 is the broadest, most user-facing regression in the set and has some inbound references, so it is the best representative if one must be chosen. It is not actually a duplicate of the others.", + "summary": "No convincing duplicate cluster emerges here. The titles overlap mostly at a generic level (v5, loading, import errors, docs), but the underlying bugs and code paths are different, so I would not merge any soft pair.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue is justified: this set spans unrelated regressions across imports, model loading, docs, training, tensor parallelism, FP8, and vision/audio components.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43576 is the strongest general triage candidate: clear regression report, broad impact, and enough discussion/reference signal to anchor follow-up. The rest are largely unrelated, model-specific, or API/import issues.", + "best_issue_reason": "If forced to pick a representative issue, 43576 is the broadest v5 regression report with the clearest cross-cutting impact, but it still is not a duplicate hub for the rest.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44387", "right": "issue:44556", "accept": false, - "reason": "Both mention v5 regressions and loading, but one is an int4 memory/OOM problem while the other is checkpoint reload compatibility. Different failure modes and likely different fixes." + "reason": "Both mention v5 regressions, but one is int4 OOM/memory growth and the other is checkpoint reload failure; different failure modes and code paths." }, { "left": "issue:43010", "right": "issue:43065", "accept": false, - "reason": "One is about decorating cache/layer update methods with no_grad; the other is about a dummy Conv2d in Sam3PixelDecoder. Unrelated code paths." + "reason": "Cache no_grad decoration and Sam3PixelDecoder dummy Conv2d are unrelated component bugs." }, { "left": "issue:43824", "right": "issue:46617", "accept": false, - "reason": "Both are import/loading themed, but 43824 is a missing exported class, while 46617 is a FileNotFoundError when loading a trust_remote_code model from cache. Different root causes." + "reason": "ImportError for a specific Qwen2.5-VL class is not the same as a FileNotFoundError when loading a trust_remote_code model from cache." }, { "left": "issue:43502", "right": "issue:44485", "accept": false, - "reason": "43502 is about unwanted network API requests despite local_files_only=True; 44485 is a RoPE implementation discussion. No shared bug." + "reason": "local_files_only network leakage and GLM-5 RoPE implementation are different issues." }, { "left": "issue:43576", "right": "issue:46132", "accept": false, - "reason": "43576 is a broken CLI env command in v5; 46132 is an attention registration behavior change. Different features and fixes." + "reason": "Broken env command and AttentionInterface.register behavior are separate CLI vs API problems." }, { "left": "issue:45125", "right": "issue:45478", "accept": false, - "reason": "Both involve Qwen3.5 MoE, but 45125 is missing tensor-parallel plan metadata while 45478 is a from_pretrained error in transformers>=5.4.0. Not the same concrete defect." + "reason": "Both involve Qwen3.5 MoE, but one is missing _tp_plan for tensor parallelism while the other is a from_pretrained loading regression." }, { "left": "issue:30333", "right": "issue:33453", "accept": false, - "reason": "30333 is MLFlow reporting after training failure; 33453 is tokenizer loading regression. Only generic failure wording overlaps." + "reason": "MLflow reporting on failed training and tokenizer loading regression are unrelated." }, { "left": "issue:44355", "right": "issue:44749", "accept": false, - "reason": "Both mention errors after running compiled/code paths or upgrading, but one is compiled Python file handling and the other is a slowdown in data filtering. Different bugs." + "reason": "Compiled Python file errors and the Chinese report about slower filtering after upgrade are not the same bug." }, { "left": "issue:45125", "right": "issue:45375", "accept": false, - "reason": "Both are Qwen3.5 MoE-related, but one is tensor-parallel support and the other is a missing config field silently dropped by strict parsing. Separate issues." + "reason": "Tensor-parallel plan missing and strict-dropping of a Qwen3.5MoeVisionConfig field are different model/config issues." }, { "left": "issue:42548", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but one is missing PreTrainedModel and the other is missing HybridCache. Different symbols and likely different export paths." + "reason": "Missing PreTrainedModel export and missing HybridCache export are distinct import errors." }, { "left": "issue:43010", "right": "issue:43232", "accept": false, - "reason": "Cache update decoration versus generation kwargs propagation after sync_gpus are distinct generation internals." + "reason": "Both touch generation internals, but cache update no_grad and post-sync_gpus kwargs updating are different code paths." }, { "left": "issue:30990", "right": "issue:31795", "accept": false, - "reason": "Sentence Transformers loading hang and documentation confusion around input_ids/past_key_values are unrelated." + "reason": "Sentence Transformers loading hang and documentation confusion about forward arguments are unrelated." }, { "left": "issue:43502", - "right": "issue:44961", + "right": "issue:44222", "accept": false, - "reason": "API requests despite local_files_only=True has nothing to do with a placeholder/nonsense issue title." + "reason": "Unexpected API requests with local_files_only and FP8 save_pretrained behavior are different concerns." }, { "left": "issue:43502", - "right": "issue:44222", + "right": "issue:44961", "accept": false, - "reason": "Local-only network leakage and FP8 save_pretrained moe are different subsystems and bugs." + "reason": "A network-usage bug and an empty-title issue are not related." }, { "left": "issue:43827", "right": "issue:44961", "accept": false, - "reason": "Docs still referencing pipeline() after removals is a documentation cleanup issue; 'racoon' is not a meaningful matching bug report." + "reason": "Docs still referencing pipeline() and the unrelated 'racoon' issue do not match." }, { "left": "issue:30333", "right": "issue:36296", "accept": false, - "reason": "MLFlow reporting failure and tensor parallel training bug are unrelated training concerns." + "reason": "MLflow reporting and tensor-parallel training are separate problems." }, { "left": "issue:43299", "right": "issue:44556", "accept": false, - "reason": "Both concern v5 model compatibility, but 43299 is Qwen3VL MoE loading and 44556 is reload of older checkpoints after upgrade. Different model families and failure conditions." + "reason": "Both are upgrade/loading regressions, but they affect different models and symptoms, so they are not the same concrete bug." }, { "left": "issue:44355", "right": "issue:45071", "accept": false, - "reason": "Compiled Python file errors and PretrainedConfig type checking regression are unrelated." + "reason": "Compiled-file load errors and PretrainedConfig type-checking breakage are unrelated." }, { "left": "issue:44355", "right": "issue:44655", "accept": false, - "reason": "Compiled file loading errors versus inability to save Pipeline objects are different code paths." + "reason": "Compiled Python file failures and inability to save Pipeline objects are different issues." }, { "left": "issue:33453", "right": "issue:34689", "accept": false, - "reason": "Tokenizer loading regression and Llama 3.2 90B Vision model loading breakage are both loading-related but not the same bug." + "reason": "Tokenizer-loading regression and Llama 3.2 Vision model loading breakage are separate load bugs." }, { "left": "issue:27957", @@ -16851,61 +16898,61 @@ "left": "issue:42757", "right": "issue:43576", "accept": false, - "reason": "Missing is_offline_mode import from huggingface_hub and broken transformers env command are separate CLI/import issues." + "reason": "Missing huggingface_hub export and broken transformers env command are different API/CLI regressions." }, { "left": "issue:43576", "right": "issue:45071", "accept": false, - "reason": "Broken env command and PretrainedConfig type checking regression do not share a code path or fix." + "reason": "Broken env command and PretrainedConfig type checking are unrelated v5 issues." }, { "left": "issue:44315", "right": "issue:44360", "accept": false, - "reason": "Liger Kernel not applied with model_init and a DSA indexer lacking ReLU are unrelated." + "reason": "Liger Kernel application in model_init and DSA indexer ReLU discussion are unrelated." }, { "left": "issue:31795", "right": "issue:38617", "accept": false, - "reason": "Docs confusion in model.forward and an ImportError for layer_type_validation are different classes of problems." + "reason": "Documentation confusion and an import error for layer_type_validation are not the same bug." }, { "left": "issue:43576", "right": "issue:43824", "accept": false, - "reason": "CLI env regression and missing Qwen2.5-VL export/import symbol are unrelated." + "reason": "CLI env regression and Qwen2.5-VL import error are different failures." }, { "left": "issue:44961", "right": "issue:45468", "accept": false, - "reason": "The generic/placeholder issue title does not match a Gemma-4 audio positional encoding bug." + "reason": "The 'racoon' issue contains no evidence of sharing the Gemma-4 audio positional encoding bug." }, { "left": "issue:44485", "right": "issue:44961", "accept": false, - "reason": "RoPE implementation discussion and an unrelated placeholder title are not duplicates." + "reason": "GLM-5 RoPE implementation and the unrelated 'racoon' issue are not duplicates." }, { "left": "issue:44351", "right": "issue:46617", "accept": false, - "reason": "Missing HybridCache import and cache loading FileNotFoundError are both cache-related only at a very high level; the bugs are different." + "reason": "Importing HybridCache from transformers and loading a trust_remote_code model from cache are different error classes and code paths." }, { "left": "issue:44908", "right": "issue:44961", "accept": false, - "reason": "Scheduler kwargs bug and an unrelated placeholder title are not the same issue." + "reason": "Scheduler kwarg handling and the unrelated 'racoon' issue do not match." }, { "left": "issue:43576", "right": "issue:44938", "accept": false, - "reason": "Broken env command in v5 and Python 3.14 loading failure are different compatibility problems." + "reason": "Broken env command in v5 and Python 3.14 load failure are distinct compatibility issues." }, { "left": "issue:36296", @@ -16917,43 +16964,43 @@ "left": "issue:46143", "right": "issue:46144", "accept": false, - "reason": "RoFormer kwargs not passed through methods and RoFormer attention not using attention interface are adjacent but distinct bugs; the fixes would not plausibly be one PR." + "reason": "Both are RoFormer-related, but one is about kwargs propagation and the other about attention-interface usage; not the same concrete bug." }, { "left": "issue:44222", "right": "issue:44230", "accept": false, - "reason": "FP8 save_pretrained bug and FP8 inference support request are related to FP8 but concern different operations and failure points." + "reason": "FP8 save_pretrained/moe and FP8 inference support for Qwen3-VL/Qwen3.5 MoE are related theme-wise but not the same code-path problem." }, { "left": "issue:36296", "right": "issue:43526", "accept": false, - "reason": "Tensor parallel training bug and BeitImageProcessorFast reduce_labels behavior are unrelated." + "reason": "Tensor parallel training and BeitImageProcessorFast label reduction are unrelated." }, { "left": "issue:43475", "right": "issue:46617", "accept": false, - "reason": "SAM3 video attribute error and trust_remote_code cache FileNotFoundError are different loading/runtime issues." + "reason": "Missing Sam3VisionEncoderOutput attribute and cache FileNotFoundError are different bugs." }, { "left": "issue:44315", "right": "issue:46209", "accept": false, - "reason": "Liger Kernel application under model_init and FP8 inference on B200 are unrelated accelerator/backend issues." + "reason": "Liger Kernel application and FP8 inference on B200 are unrelated issues." }, { "left": "issue:41628", "right": "issue:42548", "accept": false, - "reason": "Missing AutoImageProcessor import and missing PreTrainedModel import are both export/import problems, but for different symbols and likely different packaging fixes." + "reason": "AutoImageProcessor import failure and PreTrainedModel import failure are different missing exports." }, { "left": "issue:44485", "right": "issue:46664", "accept": false, - "reason": "GLM-5 RoPE implementation and SmolVLM num_frames semantics are unrelated model-specific bugs." + "reason": "GLM-5 RoPE implementation and SmolVLM num_frames semantics are unrelated." } ] }, @@ -16968,14 +17015,14 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 54, - "review_unit_count": 56, + "review_unit_index": 55, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:31795", "issue:33453", @@ -17034,16 +17081,16 @@ "issue:44623|issue:45230", "issue:44038|issue:45125", "issue:44273|issue:45003", - "issue:44393|issue:45071", "issue:44393|issue:45753", + "issue:44393|issue:45071", "issue:45381|issue:46664", "issue:43576|issue:44556", "issue:41669|issue:43673", "issue:43525|issue:45850", - "issue:43572|issue:45850", "issue:44556|issue:45092", - "issue:44230|issue:46095", + "issue:43572|issue:45850", "issue:45278|issue:45850", + "issue:44230|issue:46095", "issue:44492|issue:45850", "issue:42371|issue:44492", "issue:41669|issue:46218", @@ -17069,280 +17116,280 @@ "issue:44315|issue:44829", "issue:43576|issue:44273" ], - "prepared_review_unit_hash": "81e96a71e8d0df65da989fa4827159f6c52eb90415ef3fbff5db23479e1d7b4e", + "prepared_review_unit_hash": "a49ae9501034ade8cae99cc43ca1f6a338b074cf3548ac6354887173573d5cbd", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 41, - "serialized_chars": 39439, - "estimated_input_tokens": 9860, - "estimated_eval_tokens": 19976 + "serialized_chars": 39443, + "estimated_input_tokens": 9861, + "estimated_eval_tokens": 19978 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is broadly mixed: most items are separate v5 regressions, model-specific loading issues, or documentation/test cleanup reports. The only coherent theme is import/lazy-loading and model loading behavior, but most soft pairs are not the same underlying bug.", - "confidence": 0.39, - "canonical_issue_reason": "issue:44273 is the most representative and actionable issue here: it is open, concrete, and sits closest to the import/lazy-loading regression theme that overlaps several other candidates.", + "summary": "This cluster is mostly a mix of unrelated Transformers v5 regressions, model-specific failures, and import/load issues. I do not see any soft pair that is clearly the same underlying bug/change, so I would not merge any of the candidate edges.", + "confidence": 0.62, + "canonical_issue_reason": "issue:45850 is the broadest umbrella report ('Since 5.0 version it breaks too many models') and best reflects the shared v5-regression theme across the cluster, even though it is not a strict duplicate of the others.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44273 is the best single issue to keep because it is specific enough to act on and broader/more central than the many vague version-level complaints.", + "best_issue_reason": "issue:45850 is the best cluster representative because it is the most general summary of the widespread v5 breakage reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45362", "right": "issue:45850", "accept": false, - "reason": "Different scope: a specific Qwen3.5 chat crash vs a vague broad complaint about v5 breaking many models." + "reason": "Generic v5 breakage is too broad; this is a specific Qwen3.5 chat crash, not the same bug." }, { "left": "issue:43673", "right": "issue:44492", "accept": false, - "reason": "Cache prefill regression and a cache-strategy typo are unrelated." + "reason": "Chunked-prefill cache failure and a cache-strategy typo are different issues." }, { "left": "issue:44273", "right": "issue:45542", "accept": false, - "reason": "Lazy loading regression is not the same bug as the TensorFlow/tensorboard backend import error." + "reason": "Lazy loading regressions and tensorboard-without-TF backend errors are unrelated." }, { "left": "issue:44360", "right": "issue:46664", "accept": false, - "reason": "DSA/ReLU discussion and video num_frames semantics are unrelated." + "reason": "DSA indexer/ReLU discussion is unrelated to SmolVLM video frame semantics." }, { "left": "issue:44623", "right": "issue:45230", "accept": false, - "reason": "Specific processor.save_pretrained file-loss bug vs an unspecified bug report." + "reason": "Missing processor save files is not the same as a vague bug report." }, { "left": "issue:44038", "right": "issue:45125", "accept": false, - "reason": "Both mention Qwen3.5 MoE, but one is a general bug and the other is a tensor-parallelism-specific missing _tp_plan issue." + "reason": "Both involve Qwen3/MoE, but one is a model bug and the other is tensor-parallel config support." }, { "left": "issue:44273", "right": "issue:45003", "accept": false, - "reason": "Lazy loading failure and sys.modules access safety are related to imports, but not the same concrete defect." + "reason": "Lazy loading and unsafe sys.modules access are different failure modes." }, { "left": "issue:44393", - "right": "issue:45071", + "right": "issue:45753", "accept": false, - "reason": "Qwen3-VL 2D box output issue is unrelated to PretrainedConfig type checking." + "reason": "Both are Qwen3-VL related, but one is 2D box output hallucination and the other is an infinite loop on a specific image." }, { "left": "issue:44393", - "right": "issue:45753", + "right": "issue:45071", "accept": false, - "reason": "Different symptoms in different Qwen3-VL/Qwen3.5 paths; not the same bug." + "reason": "A vision-model output bug is not the same as a PretrainedConfig type-checking regression." }, { "left": "issue:45381", "right": "issue:46664", "accept": false, - "reason": "Video position_ids mismatch and num_frames meaning mismatch are different processor bugs." + "reason": "Video position-id corruption and num_frames semantics are distinct processor bugs." }, { "left": "issue:43576", "right": "issue:44556", "accept": false, - "reason": "Broken env command and checkpoint reload incompatibility are separate v5 issues." + "reason": "Broken env command and checkpoint reload incompatibility are different v5 regressions." }, { "left": "issue:41669", "right": "issue:43673", "accept": false, - "reason": "Import-star/lazy-loading performance regression and chunked_prefill cache regression are different problems." + "reason": "Import-star/lazy-loading performance issues are not the same as cache-missing during chunked prefill." }, { "left": "issue:43525", "right": "issue:45850", "accept": false, - "reason": "Llama4Config pad_token_id error is model-specific, not the same as broad v5 breakage." - }, - { - "left": "issue:43572", - "right": "issue:45850", - "accept": false, - "reason": "StableLmConfig pad_token_idx regression is distinct from the general v5 complaint." + "reason": "Missing pad_token_id on Llama4Config is a specific config bug, not the broad v5 breakage report." }, { "left": "issue:44556", "right": "issue:45092", "accept": false, - "reason": "Both concern loading old checkpoints in v5, but the generic reload breakage and InternVL2 meta-init incompatibility are not the same concrete issue." + "reason": "Both mention v5 compatibility, but one is generic checkpoint reload failure and the other is remote-code/meta-init incompatibility." }, { - "left": "issue:44230", - "right": "issue:46095", + "left": "issue:43572", + "right": "issue:45850", "accept": false, - "reason": "FP8 support request and weight-conversion-script request are unrelated feature asks." + "reason": "StableLmConfig pad_token_idx missing is a model-specific config regression, not the umbrella v5 report." }, { "left": "issue:45278", "right": "issue:45850", "accept": false, - "reason": "Import errors after upgrade is too broad to be the same bug as the vague v5-breakage complaint." + "reason": "General import errors after upgrade are broader, but still not the same concrete issue as the umbrella complaint." + }, + { + "left": "issue:44230", + "right": "issue:46095", + "accept": false, + "reason": "FP8 inference support for Qwen models is unrelated to a DeepSeek weight-conversion script request." }, { "left": "issue:44492", "right": "issue:45850", "accept": false, - "reason": "A typo in cache strategies is not the same as broad model breakage." + "reason": "A cache-strategy typo is not the same as broad v5 breakage." }, { "left": "issue:42371", "right": "issue:44492", "accept": false, - "reason": "TF32 API guidance and a cache-strategy typo are unrelated." + "reason": "TF32 API guidance and a cache-strategy typo have no direct overlap." }, { "left": "issue:41669", "right": "issue:46218", "accept": false, - "reason": "Import-star/lazy-loading slowdown and a specific missing module import are different issues." + "reason": "Import-star removal/import slowdown is not the same as a specific Qwen3_5ForCausalLM import failure." }, { "left": "issue:42890", "right": "issue:44908", "accept": false, - "reason": "Test seeding flakiness and inverse_sqrt scheduler kwargs handling are unrelated." + "reason": "Missing set_seed in tests and inverse_sqrt scheduler kwargs are unrelated." }, { "left": "issue:43408", "right": "issue:44315", "accept": false, - "reason": "Model-type warning and Liger Kernel model_init behavior are unrelated." + "reason": "A model-type warning and Liger Kernel application in model_init are different problems." }, { "left": "issue:44246", "right": "issue:44273", "accept": false, - "reason": "Import-time slowdown is not the same as lazy-loading correctness failure." + "reason": "Import performance and lazy-loading correctness are related only loosely, not the same bug." }, { "left": "issue:43352", "right": "issue:44393", "accept": false, - "reason": "Nemotron Flash Attention support and Qwen3-VL bounding-box errors are unrelated." + "reason": "Flash Attention 2 support for Nemotron is unrelated to Qwen3-VL 2D box output errors." }, { "left": "issue:43576", "right": "issue:44908", "accept": false, - "reason": "Env command breakage and scheduler kwargs handling are separate code paths." + "reason": "Environment CLI breakage and scheduler kwargs handling are different code paths." }, { "left": "issue:44704", "right": "issue:44908", "accept": false, - "reason": "Processor.from_pretrained kwargs forwarding and scheduler kwargs handling are unrelated." + "reason": "AutoProcessor cached_file kwargs and scheduler kwargs are unrelated." }, { "left": "issue:43976", "right": "issue:44393", "accept": false, - "reason": "Python-version compatibility and Qwen3-VL output errors are unrelated." + "reason": "Python version support regression is unrelated to a vision-model output bug." }, { "left": "issue:44908", "right": "issue:45071", "accept": false, - "reason": "Scheduler kwargs and PretrainedConfig type checking are unrelated." + "reason": "Scheduler kwargs and config type checking do not share the same underlying defect." }, { "left": "issue:44295", "right": "issue:45230", "accept": false, - "reason": "position_ids buffer access bug and an unspecified bug report are not duplicates." + "reason": "Position_ids buffer handling is not the same as an unspecified bug report." }, { "left": "issue:44297", "right": "issue:45230", "accept": false, - "reason": "Tokenizer save_pretrained mismatch and an unspecified bug report are not duplicates." + "reason": "Tokenizer save metadata mismatch is a specific bug, not a generic report." }, { "left": "issue:41669", "right": "issue:44351", "accept": false, - "reason": "Import-star cleanup and HybridCache import failure are different issues." + "reason": "Import-star/lazy-loading performance problems are not the same as missing HybridCache exports." }, { "left": "issue:44492", "right": "issue:46328", "accept": false, - "reason": "Cache-strategy typo and SequenceFeatureExtractor.pad performance issue are unrelated." + "reason": "Cache-strategy typo and SequenceFeatureExtractor.pad inefficiency are unrelated." }, { "left": "issue:41669", "right": "issue:45341", "accept": false, - "reason": "Import-star/lazy-loading cleanup and a testing_utils bug are unrelated." + "reason": "Import-time slowdown is unrelated to a testing_utils bug." }, { "left": "issue:43976", "right": "issue:45850", "accept": false, - "reason": "Python support regression and broad v5 breakage are not the same concrete issue." + "reason": "Python compatibility regression is not the same as generic v5 model breakage." }, { "left": "issue:44393", "right": "issue:45478", "accept": false, - "reason": "Qwen3-VL bounding-box error and Qwen3.5 MoE from_pretrained failure are different model paths." + "reason": "Qwen3-VL bounding-box hallucination is not the same as Qwen3.5 MoE from_pretrained failure." }, { "left": "issue:43976", "right": "issue:46218", "accept": false, - "reason": "Python-version support and missing Qwen3_5ForCausalLM import are unrelated." + "reason": "Python support issues are unrelated to a missing Qwen3_5ForCausalLM import." }, { "left": "issue:44077", "right": "issue:44861", "accept": false, - "reason": "Optional post_init policy and tied-weights list crash are unrelated." + "reason": "Optional post_init policy and tied-weight-key attribute errors are different model-internals issues." }, { "left": "issue:43827", "right": "issue:44908", "accept": false, - "reason": "Docs still referencing pipeline() and scheduler kwargs handling are unrelated." + "reason": "Docs referencing pipeline() and scheduler kwargs handling are unrelated." }, { "left": "issue:31795", "right": "issue:33453", "accept": false, - "reason": "Documentation confusion and tokenizer-loading regression are unrelated." + "reason": "Documentation confusion and tokenizer loading regression are different issues." }, { "left": "issue:44315", "right": "issue:44829", "accept": false, - "reason": "Liger Kernel application with model_init and flash_attention_3 training degeneration are different bugs." + "reason": "Liger Kernel not applied in model_init and flash_attention_3 training degeneration are unrelated." }, { "left": "issue:43576", "right": "issue:44273", "accept": false, - "reason": "Broken env command and lazy-loading failure are related to v5 imports only at a high level, not the same bug." + "reason": "Broken env command and lazy loading correctness are different v5-related regressions, not one bug." } ] }, @@ -17357,14 +17404,14 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 55, - "review_unit_count": 56, + "review_unit_index": 56, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:28348", "issue:30333", @@ -17448,8 +17495,8 @@ "issue:44998|issue:45850", "issue:44938|issue:44998", "issue:44861|issue:45290", - "issue:43525|issue:44861", "issue:31515|issue:38617", + "issue:43525|issue:44861", "issue:43572|issue:44861", "issue:28348|issue:43352", "issue:44230|issue:44623", @@ -17459,14 +17506,14 @@ "issue:31515|issue:33453", "issue:43298|issue:44230" ], - "prepared_review_unit_hash": "ddd710366c25d2004227b111b1d3a1ee6371b58cb703ff13712306286d1cfa61", + "prepared_review_unit_hash": "612e541b6a4ad3789237a84ef682e150478c3b9c0c9f17d05296cf3f70e2a5f5", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 47, @@ -17481,274 +17528,271 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of unrelated issues that only share broad Transformers/v5 vocabulary. I found one plausible duplicate-like pair around tokenizer behavior in v5; the rest are too different in symptom or code path to merge as duplicates.", - "confidence": 0.58, - "canonical_issue_reason": "issue:43066 is the most concrete, self-contained tokenizer-v5 bug description here, and it best captures the likely root cause behind the nearby v4-to-v5 tokenizer behavior report.", + "summary": "No soft edge looks like a true duplicate: the set is a heterogeneous mix of model-specific bugs, loading regressions, docs issues, feature requests, and unrelated failures. I would not merge any of the suggested pairs.", + "confidence": 0.95, + "canonical_issue_reason": "issue:33453 is the clearest concrete regression among the loading-related subset, but the overall cluster is too heterogeneous for a strong canonical beyond that.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43066 is the strongest canonical anchor among these issues because it names a specific tokenizer-v5 defect rather than a broad symptom or a generic request.", + "best_issue_reason": "issue:33453 is the best available anchor because it is a concrete, self-contained loading regression and more reusable than the model-specific or unrelated reports; however, it is still only a loose fit for the full cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43519", "right": "issue:46655", "accept": false, - "reason": "Different models and different failure modes: Qwen3VL timestamp math vs SmolVLM video padding calculation." + "reason": "Different processor bugs in different models: timestamp calculation vs padding calculation." }, { "left": "issue:44360", "right": "issue:45468", "accept": false, - "reason": "Unrelated model internals: DSA indexer ReLU discussion vs Gemma-4 audio positional encoding." + "reason": "Unrelated subsystems and failures: DSA indexer ReLU vs Gemma-4 audio positional encoding." }, { "left": "issue:44655", "right": "issue:44861", "accept": false, - "reason": "Both involve save/load plumbing, but one is pipeline serialization and the other is tied-weight key handling; not the same bug." + "reason": "Different code paths: saving Pipeline objects vs _get_tied_weight_keys AttributeError." }, { "left": "issue:44273", "right": "issue:44908", "accept": false, - "reason": "Lazy loading behavior and inverse_sqrt scheduler kwargs are separate subsystems and separate defects." + "reason": "Lazy loading behavior is unrelated to inverse_sqrt scheduler kwargs handling." }, { "left": "issue:45230", "right": "issue:45310", "accept": false, - "reason": "Generic placeholder bug report vs a specific Qwen3.5 MoE from_pretrained failure; not enough to treat as the same issue." + "reason": "Generic bug report vs a specific Qwen3.5 MoE from_pretrained error; no shared concrete bug." }, { "left": "issue:43576", "right": "issue:46218", "accept": false, - "reason": "One is the env command, the other is importing Qwen3_5ForCausalLM; distinct command/import failures." + "reason": "CLI env command breakage is not the same as a model import failure." }, { "left": "issue:43519", "right": "issue:44485", "accept": false, - "reason": "Qwen3VL timestamp calculation and GLM-5 RoPE implementation are different model-specific bugs." + "reason": "Different models and bugs: Qwen3VL timestamp logic vs GLM-5 RoPE implementation." }, { "left": "issue:43519", "right": "issue:44908", "accept": false, - "reason": "Timestamp computation in a processor vs scheduler keyword propagation are unrelated." + "reason": "Processor timestamp math is unrelated to scheduler parameter plumbing." }, { "left": "issue:41669", "right": "issue:43704", "accept": false, - "reason": "Import-star cleanup/perf issue vs VRAM leak in Qwen3ForCausalLM; different code paths." + "reason": "Import-* cleanup/perf issue is unrelated to the VRAM leak in dataloader threads." }, { "left": "issue:46018", "right": "issue:46172", "accept": false, - "reason": "DeepSeek-V4 shared-expert gating and a comb.to(dtype).transpose bug are distinct model issues." + "reason": "DeepSeek-V4 shared-expert gating and a tensor transpose bug are different code-path problems." }, { "left": "issue:31795", "right": "issue:34689", "accept": false, - "reason": "Documentation confusion for forward args is not the same as a model-loading regression for Llama 3.2 Vision." + "reason": "Documentation confusion is not the same as a model-loading regression." }, { "left": "issue:31515", "right": "issue:31795", "accept": false, - "reason": "Slow checkpoint loading and forward-pass documentation are unrelated." + "reason": "Loading slowdown and documentation ambiguity are different issues." }, { "left": "issue:43704", "right": "issue:44485", "accept": false, - "reason": "Memory leak in Qwen3ForCausalLM vs GLM-5 RoPE math; different defects." + "reason": "VRAM leak in multi-threaded inference does not match a RoPE implementation discussion." }, { "left": "issue:44485", "right": "issue:45468", "accept": false, - "reason": "Different model families and different math layers; no shared underlying bug." + "reason": "Different models and different math paths; not the same bug." }, { "left": "issue:43502", "right": "issue:43519", "accept": false, - "reason": "local_files_only network leakage is unrelated to timestamp calculation in a processor." + "reason": "local_files_only network leakage is unrelated to timestamp calculation." }, { "left": "issue:42371", "right": "issue:43704", "accept": false, - "reason": "TF32 config guidance and a VRAM leak are unrelated." + "reason": "TF32 API guidance and a VRAM leak are unrelated." }, { "left": "issue:45850", "right": "issue:46218", "accept": false, - "reason": "Broad complaint that 5.0 breaks many models is not the same as a specific Qwen3_5 import failure." + "reason": "Too broad/vague v5 breakage is not the same concrete bug as a Qwen import error." }, { "left": "issue:43519", "right": "issue:44623", "accept": false, - "reason": "Processor save_pretrained missing files is a serialization bug, not timestamp computation." + "reason": "Timestamp calculation and missing save_pretrained files are different failures." }, { "left": "issue:43606", "right": "issue:45468", "accept": false, - "reason": "CPU offload device mismatch for bark-small is unrelated to Gemma-4 audio positional encoding." + "reason": "CPU offload device mismatch is unrelated to Gemma-4 audio positional encoding." }, { "left": "issue:44263", "right": "issue:45468", "accept": false, - "reason": "torch.split return handling in GlmMoeDsaIndexer is a different issue from Gemma-4 audio positional encoding." + "reason": "torch.split return values bug does not match Gemma-4 audio encoding." }, { "left": "issue:44261", "right": "issue:45468", "accept": false, - "reason": "Missing rms_norm_eps in MLA q_a_layernorm is a distinct precision/config bug." + "reason": "Precision error in MLA layernorm is unrelated to Gemma-4 audio positional encoding." }, { "left": "issue:43298", "right": "issue:46095", "accept": false, - "reason": "Continuous batching audio support and a DeepSeek-V4 weight conversion question are not the same change." + "reason": "Continuous batching audio support and a DeepSeek weight conversion question are different feature areas." }, { "left": "issue:43519", "right": "issue:43704", "accept": false, - "reason": "Processor timestamp bug and VRAM leak are separate problems." + "reason": "Timestamp calculation bug and VRAM leak are unrelated." }, { "left": "issue:44351", "right": "issue:46218", "accept": false, - "reason": "HybridCache import failure and Qwen3_5ForCausalLM import failure affect different symbols and likely different modules." + "reason": "Both are import errors, but for different symbols/modules and likely different fixes." }, { "left": "issue:43352", "right": "issue:46218", "accept": false, - "reason": "Flash Attention 2.0 support for NemotronHForCausalLM is a feature request, not the same as a Qwen3_5 import error." + "reason": "Flash Attention 2.0 support for Nemotron is unrelated to a Qwen import failure." }, { "left": "issue:42913", "right": "issue:43066", - "accept": true, - "reason": "Both point to tokenizer-v5 behavior regressions; the wrong decoder type is a plausible concrete root cause for the v4-to-v5 behavior change." + "accept": false, + "reason": "Both concern tokenizer behavior, but one is a behavior difference and the other a wrong decoder type; not the same bug." }, { "left": "issue:28348", "right": "issue:45250", "accept": false, - "reason": "GPT2LMHeadModel-specific Flash Attention 2.0 support request is broader but not the same concrete issue as a generic flash-attention title." + "reason": "Specific GPT2 Flash Attention 2.0 support request is not the same as a generic Flash Attention 2.0 issue." }, { "left": "issue:44908", "right": "issue:44998", "accept": false, - "reason": "Scheduler kwargs handling and an unrelated placeholder issue title are not duplicates." + "reason": "Scheduler kwargs bug is unrelated to the spam/invalid report." }, { "left": "issue:44492", "right": "issue:44998", "accept": false, - "reason": "A typo in cache strategies is unrelated to the placeholder issue." + "reason": "Typo in cache strategies is unrelated to the invalid title." }, { "left": "issue:43618", "right": "issue:44998", "accept": false, - "reason": "CLIPOutput attentions regression is unrelated to the placeholder issue." + "reason": "CLIPOutput attentions regression is unrelated to the invalid report." + }, + { + "left": "issue:44998", + "right": "issue:45850", + "accept": false, + "reason": "Invalid/spam report and a broad v5 breakage complaint are not duplicates." }, { "left": "issue:44938", "right": "issue:44998", "accept": false, - "reason": "Python 3.14 load failure is unrelated to the placeholder issue." + "reason": "Python 3.14 load failure is unrelated to the invalid report." }, { "left": "issue:44861", "right": "issue:45290", "accept": false, - "reason": "Tied-weights key handling and chat template crashes are different code paths." - }, - { - "left": "issue:43525", - "right": "issue:44861", - "accept": false, - "reason": "Missing pad_token_id in Llama4Config is unrelated to tied-weight key crashes." + "reason": "Tied-weight key crash and apply_chat_template tool-call crash are different code paths." }, { "left": "issue:31515", "right": "issue:38617", "accept": false, - "reason": "Slow from_pretrained checkpoint loading and an import error for layer_type_validation are different failure modes." + "reason": "Slow from_pretrained loading and a missing import name are different problems." }, { - "left": "issue:43572", + "left": "issue:43525", "right": "issue:44861", "accept": false, - "reason": "Missing pad_token_idx in StableLmConfig is a separate config bug." + "reason": "Missing pad_token_id in Llama4Config is unrelated to tied-weight key handling." }, { "left": "issue:28348", "right": "issue:43352", "accept": false, - "reason": "Both mention Flash Attention 2.0, but they target different models and different requests, so they should not be merged." + "reason": "Both mention Flash Attention 2.0, but they are different models and distinct support gaps." }, { "left": "issue:44230", "right": "issue:44623", "accept": false, - "reason": "fp8 inference support request and missing processor save files are unrelated." + "reason": "FP8 support request is unrelated to missing processor.save_pretrained files." }, { "left": "issue:30333", "right": "issue:31515", "accept": false, - "reason": "MLFlow reporting on failed jobs and slow checkpoint loading are unrelated." + "reason": "MLFlow reporting on failed jobs is unrelated to checkpoint loading performance." }, { "left": "issue:31515", "right": "issue:36296", "accept": false, - "reason": "from_pretrained slowness and tensor parallel training bugs are different issues." + "reason": "Loading slowness and tensor-parallel training bugs are not the same issue." }, { "left": "issue:44230", "right": "issue:45310", "accept": false, - "reason": "fp8 support request and Qwen3.5 MoE from_pretrained error are unrelated." + "reason": "FP8 support request and Qwen3.5 MoE from_pretrained failure are different problems." }, { "left": "issue:31515", "right": "issue:33453", "accept": false, - "reason": "Checkpoint loading slowness and tokenizer loading regression are separate bugs." + "reason": "Both are loading-related, but one is slow checkpoint loading and the other is a tokenizer-loading regression; not the same bug." }, { "left": "issue:43298", "right": "issue:44230", "accept": false, - "reason": "Audio continuous batching support and fp8 Qwen3-VL/Qwen3.5 MoE support are different feature requests." + "reason": "Continuous batching for audio models and FP8 support are unrelated." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative overall. The one accepted tokenizer pair is plausibly related from the titles alone, and the rest of the false pairs are reasonably distinguished by model/symptom/code path. The canonical anchor choice is also defensible." - }, - "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 }, @@ -17756,14 +17800,14 @@ "label": "issue", "component_index": 1, "component_count": 22, - "review_unit_index": 56, - "review_unit_count": 56, + "review_unit_index": 57, + "review_unit_count": 57, "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-475", + "cluster_id": "cluster-10000-476", "nodes": [ "issue:37934", "issue:42936", @@ -17789,20 +17833,20 @@ "issue:37934|issue:43668", "issue:42936|issue:43668" ], - "prepared_review_unit_hash": "660a9d09b4bc6daaa7ca54ca2ee1bf5c1de8adb02f2be11f721ed791f5ab207b", + "prepared_review_unit_hash": "1c55e55ec313330ef91410557d0892bb6d6ca92a284e6a9892e5a8bfc53ed1c2", "component_budget": { - "node_count": 475, - "item_count": 475, - "soft_pair_count": 1748, - "serialized_chars": 906167, - "estimated_input_tokens": 226542, - "estimated_eval_tokens": 453340 + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 }, "budget": { "node_count": 9, "item_count": 9, "soft_pair_count": 12, - "serialized_chars": 8609, + "serialized_chars": 8610, "estimated_input_tokens": 2153, "estimated_eval_tokens": 4562 }, @@ -17811,84 +17855,84 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: it mixes unrelated model, tokenizer, config, and typo reports with no clear shared underlying bug or change. All soft edges look like similarity noise rather than true duplicates.", - "confidence": 0.97, - "canonical_issue_reason": "No single canonical issue stands out because the items target different models and different failure modes; the cluster does not appear to represent one duplicate topic.", + "summary": "This is not a true duplicate cluster: the issues span unrelated models, configs, tokenizer warnings, and small correctness/typing fixes. None of the soft pairs describe the same underlying bug or change.", + "confidence": 0.96, + "canonical_issue_reason": "Issue #42936 is the strongest standalone bug report in the set: it describes a concrete implementation omission with an active discussion, while the other items are unrelated to it.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 42936 is the strongest standalone representative only because it is a concrete, actionable bug report with a specific model and clear symptom, but it is still not a duplicate anchor for the whole cluster.", + "best_issue_reason": "Issue #42936 is the best anchor for the cluster because it is the most concrete and actionable report; the rest are mostly unrelated or narrower cleanup items.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43717", "right": "issue:46144", "accept": false, - "reason": "Different models and different bugs: Mamba init-weight handling vs RoFormer attention interface usage. No shared code path or fix." + "reason": "Mamba/Mamba-2 init_weights behavior and RoFormer attention interface are different code paths and bugs." }, { "left": "issue:42936", "right": "issue:46018", "accept": false, - "reason": "Different architectures and distinct issues: Mask2former ignore_value handling vs DeepSeek-V4 shared expert gating. Not the same defect." + "reason": "Mask2Former ignore_value handling and DeepSeek-V4 shared-expert gating are unrelated model issues." }, { "left": "issue:43668", "right": "issue:44031", "accept": false, - "reason": "Config type-hint typo in ModernBERT is unrelated to tokenizer regex warnings. No common underlying change." + "reason": "A type-hint mistake in ModernBERTConfig is not the same as tokenizer regex warning behavior." }, { "left": "issue:42936", "right": "issue:43240", "accept": false, - "reason": "Both mention loss-related behavior, but one is Mask2former ignore_value usage and the other is kwargs propagation to loss functions. Different bugs." + "reason": "One is about Mask2Former ignore_value usage; the other is about kwargs not reaching a loss function." }, { "left": "issue:43240", "right": "issue:46144", "accept": false, - "reason": "Loss kwargs propagation and RoFormer attention-interface usage are unrelated code paths and fixes." + "reason": "Loss-kwargs propagation and RoFormer attention interface are different bugs in different subsystems." }, { "left": "issue:42936", "right": "issue:44031", "accept": false, - "reason": "Mask2former ignore_value handling is unrelated to tokenizer regex pattern warnings." + "reason": "Mask2Former loss masking and tokenizer regex warnings are unrelated failure modes." }, { "left": "issue:42936", "right": "issue:46144", "accept": false, - "reason": "Different model components and defects; no shared implementation path." + "reason": "Mask2Former ignore_value and RoFormer attention interface do not describe the same underlying problem." }, { "left": "issue:42936", "right": "issue:43813", "accept": false, - "reason": "Mask2former loss/config behavior is unrelated to a quantization-typo report. Not the same bug." + "reason": "A Mask2Former missing-use bug is unrelated to a quantization typo in another area." }, { "left": "issue:43240", "right": "issue:43668", "accept": false, - "reason": "A loss kwargs plumbing issue and a ModernBERT type-hint issue are unrelated." + "reason": "Passing kwargs to loss code and fixing an incorrect type hint are not the same change." }, { "left": "issue:43668", "right": "issue:43813", "accept": false, - "reason": "Both are small correctness/typing-style reports, but they concern different modules and different fixes." + "reason": "ModernBERTConfig typing and a quantization-operation typo are distinct cleanup issues." }, { "left": "issue:37934", "right": "issue:43668", "accept": false, - "reason": "Llama4TextL2Norm/RMS-norm question is unrelated to a ModernBERTConfig type hint typo." + "reason": "Llama4TextL2Norm RMS-norm behavior and ModernBERTConfig norm_eps typing concern different components and symptoms." }, { "left": "issue:42936", "right": "issue:43668", "accept": false, - "reason": "Mask2former ignore_value usage and ModernBERT norm_eps typing are unrelated issues with no shared code path." + "reason": "Mask2Former ignore_value usage and ModernBERTConfig norm_eps typing are unrelated issues." } ] }, @@ -17921,52 +17965,52 @@ "issue:10519|issue:11005", "issue:11005|issue:11307" ], - "prepared_review_unit_hash": "b61901931c039e9f1f5ed56850a04a78216484f7cc2fda85ec3ad509dbf862a0", + "prepared_review_unit_hash": "455ecc031685a3d30704ea0885304de515496f1de3785c20ab7fce581cf7338a", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 2876, - "estimated_input_tokens": 719, - "estimated_eval_tokens": 1694 + "serialized_chars": 2877, + "estimated_input_tokens": 720, + "estimated_eval_tokens": 1696 }, "budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 2876, - "estimated_input_tokens": 719, - "estimated_eval_tokens": 1694 + "serialized_chars": 2877, + "estimated_input_tokens": 720, + "estimated_eval_tokens": 1696 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The three issues are about unrelated feature requests in different areas (tokenizer truncation behavior, learning-rate scheduler behavior, and Wav2Vec2 word timing offsets), so none look like duplicates of the same underlying bug or change.", + "summary": "The three issues are about unrelated feature requests: tokenizer truncation behavior, ReduceLROnPlateau-like scheduling, and Wav2Vec2 word time offsets. There is no duplicate cluster here.", "confidence": 0.98, - "canonical_issue_reason": "No single issue is a true canonical duplicate for the cluster; if one must be chosen, issue #11307 is the strongest representative because it has the most discussion and the clearest concrete use case.", + "canonical_issue_reason": "No canonical issue: the items describe different products/features and do not share the same underlying bug or change.", "canonical_pr_reason": null, - "best_issue_reason": "Issue #11307 is the best standalone issue in this set: it is the most actively discussed and has a specific, well-scoped request. It still does not cover the other two topics.", + "best_issue_reason": "Issue 10519 is the most concrete and self-contained request, but it is not a duplicate of the others.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:10519", "right": "issue:11307", "accept": false, - "reason": "Different problem domains: tokenizer truncation direction vs Wav2Vec2 word boundary offsets. No shared code-path or concrete change target." + "reason": "Tokenizer truncation direction and Wav2Vec2 word timestamp extraction are unrelated features in different code paths." }, { "left": "issue:10519", "right": "issue:11005", "accept": false, - "reason": "One is about text preprocessing truncation options; the other is about optimizer/scheduler behavior. These are unrelated feature requests." + "reason": "Tokenizer truncation behavior and learning-rate scheduler functionality are completely different requests." }, { "left": "issue:11005", "right": "issue:11307", "accept": false, - "reason": "ReduceLROnPlateau-like scheduling and Wav2Vec2 time offsets are unrelated concerns with no evidence of the same underlying bug or fix." + "reason": "ReduceLROnPlateau-like training control and speech word boundary offsets do not describe the same problem." } ] }, @@ -17998,7 +18042,7 @@ "issue:13244|issue:16618", "issue:15980|issue:16618" ], - "prepared_review_unit_hash": "1a5a88b93de1f3529020af467d763dd9411eafa60b133c6689acdb16ee9a1331", + "prepared_review_unit_hash": "d4e0e7b79e724df758e2fa16f9f2872655bbb1b9d12b599fc03601d3bbd46598", "component_budget": { "node_count": 3, "item_count": 3, @@ -18020,24 +18064,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These three issues do not look like duplicates of one another. #16618 is the clearest concrete model-loading failure; #15980 is specifically about the error path for unauthenticated downloads of private models; #13244 is about tokenizer behavior differing from TensorFlow code. No soft matches are strong enough to merge.", + "summary": "This cluster does not look like a true duplicate set: one issue is about Tapas tokenization, one about private-model auth/download errors, and one about loading bert-base-uncased. The only partial overlap is between the two model-loading issues, but their titles point to different failure modes.", "confidence": 0.89, - "canonical_issue_reason": "#16618 is the most concrete and central bug report in the set: it targets a specific model-loading failure for 'bert-base-uncased' and is the best representative of the underlying user-visible problem among these issues.", + "canonical_issue_reason": "16618 is the broadest and most central model-loading report, but it still appears distinct from the private-download/auth issue and the Tapas tokenization bug.", "canonical_pr_reason": null, - "best_issue_reason": "#16618 is the best standalone issue because it describes a direct, reproducible failure to load a well-known public model, whereas the other issues are narrower and concern different failure modes.", + "best_issue_reason": "16618 is the best anchor for the cluster because it is the most general and most discussed issue, even though it is not a true duplicate of the others.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:13244", "right": "issue:16618", "accept": false, - "reason": "Different problems: tokenizer output mismatch versus model-loading failure. They involve different code paths and symptoms, so they are not the same underlying bug." + "reason": "Tapas tokenization mismatches are unrelated to a failure to load bert-base-uncased; different subsystem and different bug." }, { "left": "issue:15980", "right": "issue:16618", "accept": false, - "reason": "Not the same bug: #15980 is about an unauthenticated private-model download error message, while #16618 is about failing to load 'bert-base-uncased'. The shared area is model loading, but the concrete failure conditions differ." + "reason": "Both involve model loading/downloads, but one is specifically private-model access without login while the other is a generic bert-base-uncased load failure; not the same concrete bug." } ] }, @@ -18067,12 +18111,12 @@ "soft_pairs": [ "issue:26350|issue:33260" ], - "prepared_review_unit_hash": "97a2fa1c311e9a9e97c2a3c1012282e310171c279e40239f8a5c6045dcd73346", + "prepared_review_unit_hash": "db645671ddd09d1f8db6428429ce66a4b8c6a9b50f16a2fb4c2a7e1e89dcc76f", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1589, + "serialized_chars": 1590, "estimated_input_tokens": 398, "estimated_eval_tokens": 1052 }, @@ -18080,7 +18124,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1589, + "serialized_chars": 1590, "estimated_input_tokens": 398, "estimated_eval_tokens": 1052 }, @@ -18089,18 +18133,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two distinct community contribution requests: one for Flash Attention 2 support, the other for GGUF support. They touch different functionality and are not duplicates.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "These are two unrelated feature-request issues: one for Flash Attention 2 support on more architectures, the other for GGUF support on more architectures. They share only a broad \u201ccommunity contribution\u201d framing, not the same underlying bug or change.", + "confidence": 0.98, + "canonical_issue_reason": "Issue 26350 is the clearer canonical issue because it was created earlier and has substantially more discussion, indicating it is the more established thread for its own feature request.", "canonical_pr_reason": null, - "best_issue_reason": "issue:26350 is the older and more discussed thread, but it is not a true canonical choice for this cluster because it addresses a different feature than issue:33260.", + "best_issue_reason": "Issue 26350 is the best issue representative for the cluster only in the sense that it is the older, more active thread; however, the cluster is not truly duplicate-worthy since the requests are different.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:26350", "right": "issue:33260", "accept": false, - "reason": "Different feature requests with different targets: Flash Attention 2 support vs GGUF support. They are related only by being community architecture-support proposals, not the same underlying issue." + "reason": "Reject: these request different capabilities for different architectures (Flash Attention 2 vs GGUF). They are not the same underlying bug/change and should not be merged." } ] }, @@ -18159,187 +18203,187 @@ "issue:44370|issue:46226", "issue:46222|issue:46226" ], - "prepared_review_unit_hash": "9bb1c76095db79f39014a4db411fb5d211d8f550a8c97bc23cf89c5c36a009be", + "prepared_review_unit_hash": "4cd1acf5ac6a5bfdca6518cd29c9bcf6818f4eabc01886a36779f33b5baea9bf", "component_budget": { "node_count": 8, "item_count": 8, "soft_pair_count": 24, - "serialized_chars": 13530, - "estimated_input_tokens": 3383, - "estimated_eval_tokens": 7022 + "serialized_chars": 13527, + "estimated_input_tokens": 3382, + "estimated_eval_tokens": 7020 }, "budget": { "node_count": 8, "item_count": 8, "soft_pair_count": 24, - "serialized_chars": 13530, - "estimated_input_tokens": 3383, - "estimated_eval_tokens": 7022 + "serialized_chars": 13527, + "estimated_input_tokens": 3382, + "estimated_eval_tokens": 7020 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "There are two patterns here: Turkish/Spanish/Romanian translation requests are separate issues and should not be merged, while the generic placeholder / near-identical translation-request issues form a tight duplicate subgroup. 43834 is the best representative of that duplicate subgroup.", - "confidence": 0.84, - "canonical_issue_reason": "43834 is the earliest clean representative of the tightly duplicated placeholder translation-request subgroup; the Turkish, Spanish, and Romanian issues are distinct language-specific requests and should remain separate.", + "summary": "This cluster splits into two real duplicate pockets: the English placeholder-title issues and the identical Thai-title pair. The language-specific translation requests remain separate, and cross-pocket similarity is not enough to merge them into one duplicate set.", + "confidence": 0.9, + "canonical_issue_reason": "No single canonical issue for the whole cluster: the evidence supports two separate duplicate pockets rather than one unified set. If a representative is needed, it should be chosen within the English placeholder pocket, not across pockets.", "canonical_pr_reason": null, - "best_issue_reason": "43834 is the most central/representative node among the accepted near-identical placeholder issues.", + "best_issue_reason": "Issue 43834 is the best global representative because it is the earliest item in the strongest English placeholder subgroup and matches the other placeholder issues closely; the Thai pair should be handled as a separate duplicate pocket.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:27088", "right": "issue:44370", "accept": false, - "reason": "Different target languages; same broad translation theme, but not the same underlying issue." + "reason": "Turkish translation request vs generic placeholder template; similar theme, but not the same issue." }, { "left": "issue:27088", "right": "issue:44365", "accept": false, - "reason": "Different target languages; these are separate translation requests." + "reason": "Different concrete language request versus placeholder title; no evidence of a true duplicate." }, { "left": "issue:27088", "right": "issue:43834", "accept": false, - "reason": "Turkish request vs generic placeholder issue; not the same concrete task." + "reason": "Turkish request is distinct from the placeholder subgroup and should not be merged with it." }, { "left": "issue:27088", "right": "issue:38435", "accept": false, - "reason": "Different target languages (Turkish vs Romanian); not duplicates." + "reason": "Different target languages (Turkish vs Romanian), so these are separate requests." }, { "left": "issue:27088", "right": "issue:28936", "accept": false, - "reason": "Turkish vs Spanish translation requests are distinct issues." + "reason": "Different target languages (Turkish vs Spanish), so these are separate requests." }, { "left": "issue:28936", "right": "issue:43834", "accept": false, - "reason": "Spanish request vs generic placeholder issue; similar template, but not the same underlying request." + "reason": "Spanish translation request versus placeholder template; related area, but not the same request." }, { "left": "issue:28936", "right": "issue:44365", "accept": false, - "reason": "Spanish request vs generic placeholder issue; different concrete translation task." + "reason": "Spanish translation request versus placeholder template; not enough to call them duplicates." }, { "left": "issue:28936", "right": "issue:38435", "accept": false, - "reason": "Different target languages (Spanish vs Romanian); not duplicates." + "reason": "Different target languages (Spanish vs Romanian), so these are separate issues." }, { "left": "issue:28936", "right": "issue:44370", "accept": false, - "reason": "Spanish request vs generic placeholder issue; similar wording only." + "reason": "Spanish translation request versus placeholder template; similar wording alone is insufficient." }, { "left": "issue:38435", "right": "issue:44370", "accept": false, - "reason": "Romanian request vs generic placeholder issue; not the same concrete issue." + "reason": "Romanian request versus placeholder issue; same broad theme, but not the same concrete task." }, { "left": "issue:38435", "right": "issue:44365", "accept": false, - "reason": "Romanian request vs generic placeholder issue; different translation task." + "reason": "Romanian request versus placeholder issue; no direct evidence they are duplicates." }, { "left": "issue:38435", "right": "issue:43834", "accept": false, - "reason": "Romanian request vs generic placeholder issue; not a duplicate." + "reason": "Romanian-specific request is distinct from the generic placeholder subgroup." }, { "left": "issue:38435", "right": "issue:46222", "accept": false, - "reason": "Romanian request vs generic placeholder/gibberish issue; no evidence of the same underlying task." + "reason": "Romanian request versus Thai-language issue; soft similarity is not enough to merge." }, { "left": "issue:38435", "right": "issue:46226", "accept": false, - "reason": "Romanian request vs generic placeholder/gibberish issue; not the same issue." + "reason": "Romanian request versus Thai-language issue; same broad topic, different concrete issue." }, { "left": "issue:43834", "right": "issue:44370", "accept": true, - "reason": "Near-identical placeholder translation-request issues; bodies/titles look like the same underlying template issue." + "reason": "Near-identical placeholder translation issues with the same template and highly similar bodies." }, { "left": "issue:43834", "right": "issue:44365", "accept": true, - "reason": "Effectively identical generic placeholder translation request; plausible duplicate." + "reason": "Same placeholder-title template and almost identical body content; same underlying issue." }, { "left": "issue:43834", "right": "issue:46222", - "accept": true, - "reason": "Very high content overlap and same template-like request; looks like the same underlying issue." + "accept": false, + "reason": "Different duplicate pocket: English placeholder template versus Thai title; similarity alone is not enough." }, { "left": "issue:43834", "right": "issue:46226", - "accept": true, - "reason": "Very high content overlap and same template-like request; likely duplicate." + "accept": false, + "reason": "Different duplicate pocket: English placeholder template versus Thai title; treat as separate cluster." }, { "left": "issue:44365", "right": "issue:44370", "accept": true, - "reason": "Effectively identical generic placeholder translation-request issues; same underlying change request." + "reason": "Same placeholder subgroup with nearly identical content; clearly the same underlying request." }, { "left": "issue:44365", "right": "issue:46222", - "accept": true, - "reason": "Near-identical template issue content; same underlying request." + "accept": false, + "reason": "Cross-pocket match between English placeholder and Thai issue; too weak to merge conservatively." }, { "left": "issue:44365", "right": "issue:46226", - "accept": true, - "reason": "Near-identical template issue content; same underlying request." + "accept": false, + "reason": "Cross-pocket match between English placeholder and Thai issue; not enough evidence for a duplicate." }, { "left": "issue:44370", "right": "issue:46222", - "accept": true, - "reason": "Near-identical template issue content; same underlying request." + "accept": false, + "reason": "Cross-pocket match between English placeholder and Thai issue; identical body length is not sufficient." }, { "left": "issue:44370", "right": "issue:46226", - "accept": true, - "reason": "Near-identical template issue content; same underlying request." + "accept": false, + "reason": "Cross-pocket match between English placeholder and Thai issue; same theme, different duplicate pocket." }, { "left": "issue:46222", "right": "issue:46226", "accept": true, - "reason": "Exact/near-exact duplicate pair with identical title and body similarity." + "reason": "Exact same Thai title and identical body length; strongest duplicate match in the cluster." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The cluster summary is grounded in the packet: the Turkish, Spanish, and Romanian issues are clearly language-specific and should stay separate, while the remaining placeholder-style issues show extremely high soft similarity and form a plausible duplicate subgroup. The choice of 43834 as the representative is also consistent with its earliest creation date among that subgroup." + "accept": false, + "feedback": "The cluster is only well-supported as two separate near-duplicate pockets: the English placeholder-title issues (43834/44365/44370) and the identical Thai-title pair (46222/46226). The accepted cross-pairs between those two pockets overstate the evidence; the packet only shows soft similarity and identical body lengths, not enough to conservatively treat them as the same duplicate set. The language-specific issues (27088/28936/38435) should remain separate as you noted, but the cross-subgroup duplicate verdicts should be rejected." }, "evaluator_used": true, - "retried": false, + "retried": true, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -18418,12 +18462,12 @@ "issue:42503|issue:46653", "issue:45027|issue:46653" ], - "prepared_review_unit_hash": "632f38390a79586c05bff023742a30bdb6ae8587aca092beb3e769d3e3cbbe5c", + "prepared_review_unit_hash": "9e81a3001fc65823eac29f12ef2e6f1563885c092f93eb53f9c71bb533a69e80", "component_budget": { "node_count": 18, "item_count": 18, "soft_pair_count": 40, - "serialized_chars": 24143, + "serialized_chars": 24142, "estimated_input_tokens": 6036, "estimated_eval_tokens": 12328 }, @@ -18431,7 +18475,7 @@ "node_count": 18, "item_count": 18, "soft_pair_count": 40, - "serialized_chars": 24143, + "serialized_chars": 24142, "estimated_input_tokens": 6036, "estimated_eval_tokens": 12328 }, @@ -18440,252 +18484,252 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is a broad set of distinct model-request issues, not true duplicates. They all share the same generic pattern of asking for support for different models, so soft similarity is high but the underlying requests are separate.", - "confidence": 0.97, - "canonical_issue_reason": "No exact duplicate issue emerges. If a representative must be chosen, issue 43671 is a good cluster anchor because it is a clear, concrete support request with active discussion and fits the dominant 'add model support' theme.", + "summary": "This is a loose similarity cluster of many different model-add/support requests. They share broad wording, but they target different concrete models or features, so there are no true duplicates here.", + "confidence": 0.79, + "canonical_issue_reason": "issue:43671 is the most central and representative concrete request in this noisy similarity cluster, so it is the least-bad anchor even though the items are not duplicates.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43671 is the most suitable representative issue: it is open, specific, and discussed, while still reflecting the common pattern across the cluster. It is not a duplicate target, just the best exemplar.", + "best_issue_reason": "issue:43671 is the best standalone representative of the cluster because it is a clear model-support request and has the strongest connectivity, but the cluster is too broad for true deduping.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:36817", "right": "issue:41211", "accept": false, - "reason": "Different models and different integration targets; only the generic 'Add model' framing matches." + "reason": "Different models: EuroBert vs DEIMv2." }, { "left": "issue:36817", "right": "issue:43671", "accept": false, - "reason": "EuroBert and Qwen3-TTS are unrelated model families with different support paths." + "reason": "Different model requests: EuroBert config support vs Qwen3-TTS." }, { "left": "issue:36817", "right": "issue:36879", "accept": false, - "reason": "Different requested models: EuroBert vs RF-DETR." + "reason": "Different model families: EuroBert vs RF-DETR." }, { "left": "issue:36879", "right": "issue:41211", "accept": false, - "reason": "RF-DETR and DEIMv2 are distinct model requests, not the same change." + "reason": "RF-DETR and DEIMv2 are unrelated model additions." }, { "left": "issue:36879", "right": "issue:43909", "accept": false, - "reason": "RF-DETR vs LFM2.5 Audio 1.5B are unrelated models and code paths." + "reason": "RF-DETR is not the same as LFM2.5 Audio 1.5B." }, { "left": "issue:36879", "right": "issue:44121", "accept": false, - "reason": "RF-DETR and OpenAI weight-sparse transformer are different model requests." + "reason": "Object detection model vs weight-sparse transformer request." }, { "left": "issue:36879", "right": "issue:43837", "accept": false, - "reason": "RF-DETR and Qwen3-ASR are different model families and features." + "reason": "RF-DETR is unrelated to Qwen3-ASR support." }, { "left": "issue:36879", "right": "issue:43671", "accept": false, - "reason": "RF-DETR and Qwen3-TTS are unrelated model support requests." + "reason": "RF-DETR and Qwen3-TTS are different concrete additions." }, { "left": "issue:37015", "right": "issue:43837", "accept": false, - "reason": "NeoBERT and Qwen3-ASR are distinct models with different implementation needs." + "reason": "NeoBERT and Qwen3-ASR are different model requests." }, { "left": "issue:37015", "right": "issue:44121", "accept": false, - "reason": "NeoBERT vs weight-sparse transformer are different requests, not one bug/change." + "reason": "NeoBERT vs OpenAI weight-sparse transformer are distinct changes." }, { "left": "issue:37015", "right": "issue:43909", "accept": false, - "reason": "NeoBERT and LFM2.5 Audio 1.5B are unrelated model additions." + "reason": "NeoBERT is unrelated to LFM2.5 Audio 1.5B." }, { "left": "issue:36879", "right": "issue:37015", "accept": false, - "reason": "RF-DETR and NeoBERT are different model support requests." + "reason": "RF-DETR and NeoBERT are different models and code paths." }, { "left": "issue:37015", "right": "issue:42503", "accept": false, - "reason": "NeoBERT and ModernVBERT are different models despite similar naming." + "reason": "NeoBERT and ModernVBERT are separate model additions." }, { "left": "issue:41115", "right": "issue:43837", "accept": false, - "reason": "MiniCPM3 architecture support and Qwen3-ASR support are different code changes." + "reason": "MiniCPM3 architecture request vs Qwen3-ASR support." }, { "left": "issue:41115", "right": "issue:44121", "accept": false, - "reason": "MiniCPM3 and weight-sparse transformer are unrelated model requests." + "reason": "MiniCPM3 and the weight-sparse transformer request are not the same change." }, { "left": "issue:41115", "right": "issue:42503", "accept": false, - "reason": "MiniCPM3 and ModernVBERT are distinct model additions." + "reason": "MiniCPM3 architecture and ModernVBERT models are distinct additions." }, { "left": "issue:41115", "right": "issue:45027", "accept": false, - "reason": "MiniCPM3 and Voxtral-4B-TTS-2603 are different model support requests." + "reason": "MiniCPM3 and Voxtral TTS are unrelated model support requests." }, { "left": "issue:41211", "right": "issue:43671", "accept": false, - "reason": "DEIMv2 and Qwen3-TTS are unrelated model families." + "reason": "DEIMv2 and Qwen3-TTS are different model targets." }, { "left": "issue:42503", "right": "issue:43837", "accept": false, - "reason": "ModernVBERT and Qwen3-ASR are distinct support requests." + "reason": "ModernVBERT models and Qwen3-ASR support are unrelated." }, { "left": "issue:42503", "right": "issue:43671", "accept": false, - "reason": "ModernVBERT and Qwen3-TTS are different model additions." + "reason": "ModernVBERT and Qwen3-TTS are different concrete requests." }, { "left": "issue:42503", "right": "issue:42738", "accept": false, - "reason": "ModernVBERT and BERT-like models with RoPE are overlapping only at a very broad category level, not the same concrete change." + "reason": "Specific ModernVBERT support vs broad BERT-like RoPE request." }, { "left": "issue:42738", "right": "issue:45027", "accept": false, - "reason": "BERT-like RoPE support and Voxtral TTS support are unrelated." + "reason": "BERT-like RoPE models and Voxtral TTS are not the same underlying change." }, { "left": "issue:42738", "right": "issue:46653", "accept": false, - "reason": "BERT-like RoPE support and a person-name issue are not the same artifact or change." + "reason": "Model architecture request vs an unrelated personal-name issue." }, { "left": "issue:43671", "right": "issue:43837", "accept": false, - "reason": "Qwen3-TTS and Qwen3-ASR are related only by family name; they are separate capabilities and not one mergeable fix." + "reason": "Related family, but TTS and ASR are different concrete support changes." }, { "left": "issue:43671", "right": "issue:45027", "accept": false, - "reason": "Qwen3-TTS and Voxtral TTS are different model support requests." + "reason": "Different models: Qwen3-TTS vs Voxtral-4B-TTS-2603." }, { "left": "issue:43671", "right": "issue:45177", "accept": false, - "reason": "Qwen3-TTS and DeepSeek-OCR2 are unrelated model additions." + "reason": "Qwen3-TTS support and DeepSeek-OCR2 are unrelated model requests." }, { "left": "issue:43671", "right": "issue:44121", "accept": false, - "reason": "Qwen3-TTS and weight-sparse transformer are different model requests." + "reason": "Qwen3-TTS support is not the same as adding the weight-sparse transformer." }, { "left": "issue:43671", "right": "issue:43909", "accept": false, - "reason": "Qwen3-TTS and LFM2.5 Audio 1.5B are separate audio model support requests, not the same underlying change." + "reason": "Qwen3-TTS and LFM2.5 Audio are different audio models." }, { "left": "issue:43837", "right": "issue:43909", "accept": false, - "reason": "Qwen3-ASR and LFM2.5 Audio 1.5B are different model integrations." + "reason": "Qwen3-ASR and LFM2.5 Audio 1.5B are different speech models." }, { "left": "issue:43837", "right": "issue:45322", "accept": false, - "reason": "Qwen3-ASR and EUPE are unrelated model requests." + "reason": "ASR support and EUPE are unrelated model additions." }, { "left": "issue:43837", "right": "issue:44121", "accept": false, - "reason": "Qwen3-ASR and weight-sparse transformer are different concrete changes." + "reason": "Qwen3-ASR and the sparse transformer request are different code paths." }, { "left": "issue:43909", "right": "issue:44205", "accept": false, - "reason": "LFM2.5 Audio 1.5B and SAM3-LiteText are unrelated model requests." + "reason": "Audio model request vs SAM3-LiteText are unrelated." }, { "left": "issue:43909", "right": "issue:46645", "accept": false, - "reason": "LFM2.5 Audio 1.5B and SWA+sink for Granite are different support paths." + "reason": "LFM2.5 Audio support and Granite SWA+sink option are different changes." }, { "left": "issue:44121", "right": "issue:45322", "accept": false, - "reason": "Weight-sparse transformer and EUPE are unrelated model requests." + "reason": "Weight-sparse transformer request vs EUPE model request." }, { "left": "issue:43837", "right": "issue:45027", "accept": false, - "reason": "Qwen3-ASR and Voxtral-4B-TTS-2603 are different speech model additions." + "reason": "Qwen3-ASR and Voxtral TTS are different speech tasks/models." }, { "left": "issue:37015", "right": "issue:45027", "accept": false, - "reason": "NeoBERT and Voxtral-4B-TTS-2603 are unrelated model requests." + "reason": "NeoBERT and Voxtral TTS are unrelated model requests." }, { "left": "issue:43837", "right": "issue:45177", "accept": false, - "reason": "Qwen3-ASR and DeepSeek-OCR2 are different model support requests." + "reason": "Qwen3-ASR support and DeepSeek-OCR2 are different model additions." }, { "left": "issue:45322", "right": "issue:46133", "accept": false, - "reason": "EUPE and TIPSv2 are both vision-language related, but they are distinct models and not the same concrete change." + "reason": "EUPE and TIPSv2 are different model/architecture requests." }, { "left": "issue:42503", "right": "issue:46653", "accept": false, - "reason": "ModernVBERT and a name-only issue are not duplicates or the same underlying request." + "reason": "ModernVBERT models and the name-only issue are unrelated." }, { "left": "issue:45027", "right": "issue:46653", "accept": false, - "reason": "Voxtral-4B-TTS-2603 and a name-only issue are unrelated." + "reason": "Voxtral TTS support and the name-only issue are unrelated." } ] }, @@ -18715,12 +18759,12 @@ "soft_pairs": [ "issue:38740|issue:46443" ], - "prepared_review_unit_hash": "86361f4906a1753b0326068df9e3d1c582fda0d7a361430344d4f51cb8fa75a5", + "prepared_review_unit_hash": "9c8fffc9c79ddd21269b01aa91967d7e553ba3f6fa507dd5d81b6a22fd1ec0a5", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1505, + "serialized_chars": 1506, "estimated_input_tokens": 377, "estimated_eval_tokens": 1010 }, @@ -18728,7 +18772,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1505, + "serialized_chars": 1506, "estimated_input_tokens": 377, "estimated_eval_tokens": 1010 }, @@ -18737,18 +18781,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are unrelated: one is a docs request to add Pruna as an optimization framework, while the other concerns a Vision RoPE refactor. No duplicate triage linkage is supported by the titles or metadata.", - "confidence": 0.98, - "canonical_issue_reason": "Neither issue is a duplicate candidate for the other; the Pruna docs request and the Vision RoPE refactor concern completely different areas and failure modes.", + "summary": "The two issues are not duplicates: one is a closed docs request about adding Pruna as an optimization framework, and the other is an open technical issue about a Vision RoPE refactor. They concern entirely different areas and change types.", + "confidence": 0.97, + "canonical_issue_reason": "Issue #46443 is the better canonical issue for this cluster because it is the substantive open technical problem; #38740 is a closed documentation/proposal request and not the same topic.", "canonical_pr_reason": null, - "best_issue_reason": "issue:46443 is the better canonical issue only because it is a concrete technical refactor topic, but it is not related to issue:38740.", + "best_issue_reason": "#46443 is the strongest issue artifact here because it describes a concrete code-path/refactor area, whereas #38740 is only a docs addition request.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:38740", "right": "issue:46443", "accept": false, - "reason": "Different subjects and domains (docs/optimization framework vs vision RoPE refactor); no evidence they describe the same underlying bug or change." + "reason": "Different subjects and scopes: docs proposal for Pruna vs Vision RoPE refactor. No shared code path, filenames, or explicit linkage indicating the same underlying bug/change." } ] }, @@ -18778,12 +18822,12 @@ "soft_pairs": [ "issue:41084|issue:43037" ], - "prepared_review_unit_hash": "5c4466c349f02b7893dc6e5bc2f9fc986d9c9400660f6a16cd5c1bd1058d94bc", + "prepared_review_unit_hash": "4c3e365f69b4ec0737373e39d782cc4bd99c0bf9b34c823a605b25697501c7ce", "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 }, @@ -18791,7 +18835,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1480, + "serialized_chars": 1479, "estimated_input_tokens": 370, "estimated_eval_tokens": 996 }, @@ -18800,8 +18844,8 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues do not describe the same underlying problem: one is a specific decoding/control request, the other is a model-support request for DeepSeek v3.2. The soft similarity is too weak to treat them as duplicates.", - "confidence": 0.96, + "summary": "The two issues are unrelated: one concerns block decoding, while the other requests DeepSeek v3.2 support. They should not be clustered as duplicates.", + "confidence": 0.97, "canonical_issue_reason": null, "canonical_pr_reason": null, "best_issue_reason": null, @@ -18811,7 +18855,7 @@ "left": "issue:41084", "right": "issue:43037", "accept": false, - "reason": "Different subjects and scopes: block decoding vs DeepSeek v3.2 support. No shared code-path or concrete duplicate evidence." + "reason": "Different topics with no shared code-path or explicit linkage: \"Set Block Decoding\" vs \"DeepSeek v3.2 support\"." } ] }, @@ -18841,7 +18885,7 @@ "soft_pairs": [ "issue:43598|issue:43600" ], - "prepared_review_unit_hash": "24fbf4543fce3fc673795cf61e72a1e6a671fbaaab2868629ce6bbd94d7a317e", + "prepared_review_unit_hash": "fcb5bd8c129d8a56bbda2046be157b1ca5f102812e9fad984006c319dbc163c6", "component_budget": { "node_count": 2, "item_count": 2, @@ -18863,18 +18907,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are not duplicates: one asks to make `remove_unused_column` more customizable in `Trainer`, while the other is about accounting for custom trainers in FLOPS estimation. They share a broad Trainer/custom-trainer theme but describe different bugs and different fixes.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "These are related only at a very high level (Trainer customization), but they describe different problems: one is about `remove_unused_column` behavior, the other about FLOPS estimation for custom trainers. They should not be clustered as duplicates.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 43598 is the more concrete, narrowly scoped request: it names a specific Trainer behavior (`remove_unused_column`) and asks for better customizability.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43598 is the better standalone representative because it is more concrete and narrowly scoped around a specific Trainer behavior (`remove_unused_column`).", + "best_issue_reason": "Issue 43598 is the best representative of the cluster because it is the clearest actionable code-path change, whereas 43600 is a separate metric/estimation concern.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43598", "right": "issue:43600", "accept": false, - "reason": "Different underlying problems: customizability of `remove_unused_column` versus FLOPS estimation for custom trainers. Shared Trainer context is too broad to treat as the same bug." + "reason": "They share only the broad Trainer/customization theme. One concerns input column pruning, the other FLOPS estimation for custom trainers; these are different underlying bugs and would not be fixed by one change." } ] }, @@ -18904,7 +18948,7 @@ "soft_pairs": [ "issue:43684|issue:44405" ], - "prepared_review_unit_hash": "c1ce56f13c6a676049e0c6391a254eeff99c019186fa23a88ad709a559996e68", + "prepared_review_unit_hash": "9e869e2e04e7ee202cf17780abcf8e3184f23555916f4f823c946bd45ecada1e", "component_budget": { "node_count": 2, "item_count": 2, @@ -18926,18 +18970,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related only at a very broad theme (Qwen auto-model support), but they target different model families and different API surfaces, so they should not be merged as duplicates.", - "confidence": 0.92, - "canonical_issue_reason": null, + "summary": "These two issues are only superficially similar: both ask for new Qwen model support, but one is about AutoModel/AutoModelForConditionalGeneration registration for Qwen3-Omni, while the other is a separate AutoModelForSequenceClassification feature request for Qwen3.5. They should not be treated as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": "43684 is the broader, more general registration request and is the better cluster anchor if one must be chosen; 44405 targets a different model family and a different task head.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43684 is the broader, more general registration request and is the better representative if one must be chosen, but it is not a duplicate of 44405.", + "best_issue_reason": "43684 is the most representative of the only common theme here (model registration/support), but it does not subsume 44405's distinct sequence-classification request.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43684", "right": "issue:44405", "accept": false, - "reason": "Different underlying requests: 43684 asks for Qwen3-Omni registration in AutoModel/AutoModelForConditionalGeneration, while 44405 asks for AutoModelForSequenceClassification support for Qwen3.5. They are separate model/API changes, not the same bug or change." + "reason": "Different model variants and different capabilities: Qwen3-Omni registration for AutoModel/AutoModelForConditionalGeneration vs Qwen3.5 AutoModelForSequenceClassification support. Shared wording is not enough to indicate the same underlying change." } ] }, @@ -18974,12 +19018,12 @@ "issue:43845|issue:43859", "issue:43846|issue:43859" ], - "prepared_review_unit_hash": "ea4ee50cf4597c14fa82275e744ad8e573591d1d7bedf74dd0980de4020309ef", + "prepared_review_unit_hash": "c4c23fde8091e907088ff85b527fa386f7149d7f6516ceaff981461b65715fb9", "component_budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 4267, + "serialized_chars": 4266, "estimated_input_tokens": 1067, "estimated_eval_tokens": 2390 }, @@ -18987,7 +19031,7 @@ "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 4267, + "serialized_chars": 4266, "estimated_input_tokens": 1067, "estimated_eval_tokens": 2390 }, @@ -18996,58 +19040,55 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The three issues titled \"huggingface \" are strong duplicate candidates because they have an exact title match and were opened close together; \"Fett\" looks unrelated.", - "confidence": 0.86, - "canonical_issue_reason": "issue:43845 is the best canonical issue because it is the earliest of the exact-title \"huggingface \" reports and best represents the repeated cluster.", + "summary": "43845/43846/43859 look like near-identical vague issues, but there isn\u2019t enough concrete detail to safely collapse them; 43835 appears unrelated. No PRs are present.", + "confidence": 0.42, + "canonical_issue_reason": "issue:43845 is the earliest of the three 'huggingface' issues and the best representative if one must be chosen, but the evidence is too thin to treat the others as confirmed duplicates.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43845 is the strongest representative issue for the cluster: earliest among the repeated \"huggingface \" reports, with exact title match to the others.", + "best_issue_reason": "issue:43845 is the strongest cluster representative because it is one of the repeated 'huggingface' reports and has the earliest timestamp among the closest matches.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43835", "right": "issue:43845", "accept": false, - "reason": "Different title and no shared concrete bug details; this does not look like the same issue as the repeated \"huggingface \" reports." + "reason": "Different titles ('Fett' vs 'huggingface'); no substantive evidence of the same underlying issue." }, { "left": "issue:43835", "right": "issue:43846", "accept": false, - "reason": "Different title and no evidence of the same underlying report; \"Fett\" appears unrelated." + "reason": "Different titles ('Fett' vs 'huggingface'); no shared concrete details to justify a duplicate merge." }, { "left": "issue:43835", "right": "issue:43859", "accept": false, - "reason": "Different title with no shared target or concrete overlap, so this is not the same issue." + "reason": "Different titles ('Fett' vs 'huggingface'); too little information to link them as the same issue." }, { "left": "issue:43845", "right": "issue:43846", - "accept": true, - "reason": "Exact same title and opened seconds apart, which strongly suggests duplicate reports of the same issue." + "accept": false, + "reason": "The titles match, but they are too generic and there is no concrete description or target to confirm the same bug." }, { "left": "issue:43845", "right": "issue:43859", - "accept": true, - "reason": "Exact same title and close creation time; this looks like a duplicate of the same report." + "accept": false, + "reason": "Near-identical vague titles are not enough on their own to prove the same underlying issue." }, { "left": "issue:43846", "right": "issue:43859", - "accept": true, - "reason": "Exact same title and close timing support these being duplicate reports of the same underlying issue." + "accept": false, + "reason": "Same generic 'huggingface' title, but no substantive evidence that these are the same report." } ] }, - "evaluator_result": { - "accept": false, - "feedback": "The cluster-level conclusion is mostly grounded, but several reasons overstate the packet. We do not have body text, so claims like \"very similar short bodies\" and \"same title/body shape\" are unsupported. It is fine to say the three \"huggingface \" issues share an exact title and were opened close together, making them likely duplicates, but the canonical/best reasons should stay at that level of evidence." - }, - "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 }, @@ -19081,88 +19122,88 @@ "issue:44572|issue:44937", "issue:44573|issue:44937" ], - "prepared_review_unit_hash": "5a48e663967266fa2281b7cf253455dc3a12f5810aa89fe06a49df7df6f2e5c1", + "prepared_review_unit_hash": "83f066781ba4ad7db8e4d71290d81bc7cac7f7aaa99c67be40453caae4cdadb1", "component_budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 9, - "serialized_chars": 5794, - "estimated_input_tokens": 1449, - "estimated_eval_tokens": 3154 + "serialized_chars": 5791, + "estimated_input_tokens": 1448, + "estimated_eval_tokens": 3152 }, "budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 9, - "serialized_chars": 5794, - "estimated_input_tokens": 1449, - "estimated_eval_tokens": 3154 + "serialized_chars": 5791, + "estimated_input_tokens": 1448, + "estimated_eval_tokens": 3152 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The cluster looks like unrelated spam/placeholder issues rather than duplicates of one concrete bug or request. None of the soft pairs have enough technical detail to justify merging.", - "confidence": 0.96, - "canonical_issue_reason": "No reliable canonical issue: the items are generic spam-like placeholders with no shared concrete problem statement.", + "summary": "This cluster is dominated by low-information/spam-like issues with no actionable bug or change description. The two non-identical titles also look unrelated, so there is no safe duplicate consolidation here.", + "confidence": 0.93, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "If one must be picked, issue 44371 is the most generic placeholder and least misleading as a representative, but it is still not a real canonical duplicate.", + "best_issue_reason": "44371 is the earliest of the identical '' issues and is the least arbitrary representative, but none of these issues describe a real underlying bug/change well enough to serve as a meaningful canonical target.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44371", "right": "issue:44572", "accept": false, - "reason": "Both are placeholders, but there is no evidence they describe the same underlying issue." + "reason": "Same spam-like placeholder title, but no substantive bug/change is described, so this is not a safe duplicate match." }, { "left": "issue:44371", "right": "issue:44573", "accept": false, - "reason": "Identical spam-style titles are not enough to establish the same concrete bug or request." + "reason": "Same spam-like placeholder title, but the content is too thin to confirm a shared underlying issue." }, { "left": "issue:44371", "right": "issue:44418", "accept": false, - "reason": "Generic spam placeholder and a documentation suggestion are not the same issue." + "reason": "One is a spam placeholder and the other is a documentation enhancement request; they do not look like the same issue." }, { "left": "issue:44371", "right": "issue:44937", "accept": false, - "reason": "\"Check out \\\"Google LLC\\\"\" is unrelated to the spam placeholder report." + "reason": "Spam placeholder versus a different spammy-sounding title; no concrete shared bug or change is visible." }, { "left": "issue:44418", "right": "issue:44572", "accept": false, - "reason": "A documentation enhancement request is not the same as a spam placeholder issue." + "reason": "Documentation enhancement and spam placeholder are not the same underlying issue." }, { "left": "issue:44418", "right": "issue:44573", "accept": false, - "reason": "No shared concrete bug or change; these are unrelated generic issues." + "reason": "Documentation enhancement and spam placeholder are not the same underlying issue." }, { "left": "issue:44572", "right": "issue:44573", "accept": false, - "reason": "Same generic spam title, but no concrete evidence they are duplicates of one underlying problem." + "reason": "Both are identical spam placeholders, but there is no real issue description to justify treating them as the same bug/change." }, { "left": "issue:44572", "right": "issue:44937", "accept": false, - "reason": "Unrelated titles and no technical detail linking them to the same issue." + "reason": "Spam placeholder versus a distinct titled issue; no evidence of one concrete shared problem." }, { "left": "issue:44573", "right": "issue:44937", "accept": false, - "reason": "Unrelated titles and no shared concrete problem to merge." + "reason": "Spam placeholder versus a distinct titled issue; no evidence of one concrete shared problem." } ] }, @@ -19192,46 +19233,46 @@ "soft_pairs": [ "issue:45145|issue:45151" ], - "prepared_review_unit_hash": "03ceb88ca9b226f4cf4aa16720cb24c62f033d8a5e1dfe022431edb7c3624d44", + "prepared_review_unit_hash": "baed8f40dca467624ca5e2ce5bc2bbf2fb5009c8af83fae05b901a949c2cae9e", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1646, - "estimated_input_tokens": 412, - "estimated_eval_tokens": 1080 + "serialized_chars": 1644, + "estimated_input_tokens": 411, + "estimated_eval_tokens": 1078 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1646, - "estimated_input_tokens": 412, - "estimated_eval_tokens": 1080 + "serialized_chars": 1644, + "estimated_input_tokens": 411, + "estimated_eval_tokens": 1078 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "#45145 and #45151 are highly similar Energy/N6 Arithmetic issue reports and likely duplicates, but the available evidence is not enough to prove they are the exact same report. #45145 remains the best canonical issue because it is earlier and already has inbound reference activity.", - "confidence": 0.88, - "canonical_issue_reason": "#45145 is the earlier report and has the stronger thread footprint, so it is the best canonical representative for this near-duplicate pair.", + "summary": "These two closed issues look very similar and may be near-duplicates, but the evidence is only soft similarity (same title, close body length, Jaccard overlap) rather than direct content confirmation. #45145 is the better representative because it is earlier and already has inbound references.", + "confidence": 0.79, + "canonical_issue_reason": "Issue #45145 is the better canonical candidate because it was created first and already has inbound references, while #45151 appears to be a later near-duplicate report.", "canonical_pr_reason": null, - "best_issue_reason": "#45145 is the strongest issue representative because it was opened first and has slightly more graph activity, while #45151 adds no distinguishing signal.", + "best_issue_reason": "#45145 is the best cluster representative because it is the earliest report and has the most established reference activity.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45145", "right": "issue:45151", "accept": false, - "reason": "The titles match and the body lengths are close, which strongly suggests near-duplication, but the packet only provides soft-similarity evidence and no concrete shared targets, filenames, or explicit cross-links. That is enough to flag them as likely duplicates, not enough to confidently merge them as the same underlying issue." + "reason": "Same title and similar body metrics suggest a likely duplicate, but the evidence provided is only soft_similarity, so I can't confirm they are the same underlying issue with enough confidence." } ] }, "evaluator_result": { "accept": false, - "feedback": "The canonical-choice reasoning is grounded, but the soft-edge verdict is too strong for the evidence available. The packet only shows identical titles, similar body lengths, and a soft-similarity signal; it does not establish that the bodies are the same or that there is no separate bug report scope. A conservative assessment would treat these as likely duplicates/near-duplicates rather than definitively the same issue." + "feedback": "The packet only supports a soft similarity match, not a definitive duplicate. Same title and similar body lengths/Jaccard justify a likely-duplicate assessment, but the analyst states the issues \"are duplicates\" and that the body content is \"highly similar\" without direct body evidence. The canonical choice of the earlier issue is fine, but the soft-edge verdict and summary are a bit too strong for the available evidence." }, "evaluator_used": true, "retried": true, @@ -19312,40 +19353,40 @@ "soft_pairs": [ "issue:45672|issue:45674" ], - "prepared_review_unit_hash": "c23c079ecaf71ff5b059efe2082364640302d2fb7d6e444cafae430b30ee2a07", + "prepared_review_unit_hash": "0c7f5dfa0831aa9f63e91589c8f1b44c3f2b0054472445c04077e0af40d1a155", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1672, - "estimated_input_tokens": 418, - "estimated_eval_tokens": 1092 + "serialized_chars": 1673, + "estimated_input_tokens": 419, + "estimated_eval_tokens": 1094 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1672, - "estimated_input_tokens": 418, - "estimated_eval_tokens": 1092 + "serialized_chars": 1673, + "estimated_input_tokens": 419, + "estimated_eval_tokens": 1094 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are related only at a broad 4-bit inference level, but they describe different bugs: one is a Gemma4/audio uint8 torch.finfo TypeError, the other is a BitsAndBytesConfig skip-modules regression that drops the default lm_head exclusion and triggers an AssertionError.", - "confidence": 0.97, - "canonical_issue_reason": "No canonical duplicate: the reports point to different concrete failure modes and code paths, so they should not be merged into one issue.", + "summary": "These are two different 4-bit inference bugs: one is a Gemma4/audio-module uint8 torch.finfo TypeError, the other is a BitsAndBytesConfig lm_head exclusion regression caused by llm_int8_skip_modules. They share the same general quantized-inference context but not the same underlying failure path.", + "confidence": 0.91, + "canonical_issue_reason": "No single canonical issue: the titles and described failures point to distinct root causes in different code paths, so they should not be deduplicated.", "canonical_pr_reason": null, - "best_issue_reason": "If a single representative is needed, issue 45674 is the broader and more generally framed report, but it does not subsume 45672.", + "best_issue_reason": "issue:45674 is the better representative issue because it describes a broader, library-level regression with a clearer config-driven failure mode; issue:45672 is a narrower model-specific symptom.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45672", "right": "issue:45674", "accept": false, - "reason": "Different bugs: Gemma4 audio uint8 torch.finfo TypeError versus BitsAndBytesConfig lm_head exclusion AssertionError; same subsystem, but not the same underlying issue." + "reason": "Both involve 4-bit inference, but the failures are different: Gemma4/audio-module uint8 torch.finfo TypeError versus BitsAndBytesConfig lm_head exclusion being cleared by llm_int8_skip_modules. Different bug, not a duplicate." } ] }, @@ -19375,7 +19416,7 @@ "soft_pairs": [ "issue:45762|issue:45763" ], - "prepared_review_unit_hash": "bc54f98bb465ade5122e8b7171b28fd7df53fdea8546edcb5cc6a28f9febf93a", + "prepared_review_unit_hash": "3daf4129d24e51b0865c6246b2b9fe3b3f23bc99b8001942c5b67afbaac149f6", "component_budget": { "node_count": 2, "item_count": 2, @@ -19397,18 +19438,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues look unrelated: they are just different person-name/date entries with no shared problem statement or evidence of a duplicate bug/change.", - "confidence": 0.96, - "canonical_issue_reason": null, + "summary": "These two closed issues are unrelated personal-name/date entries and do not appear to describe the same underlying bug or request. Do not deduplicate or cluster them together.", + "confidence": 0.98, + "canonical_issue_reason": "No canonical issue: the titles reference different people and dates, with no body content, links, or other evidence of the same problem.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Neither issue is a good cluster representative because the items appear unrelated rather than duplicate reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45762", "right": "issue:45763", "accept": false, - "reason": "Only superficial title similarity; the names and dates differ and there is no body or linked evidence showing the same underlying issue." + "reason": "Different names and dates; no shared description, links, or evidence of the same underlying issue." } ] }, @@ -19438,40 +19479,40 @@ "soft_pairs": [ "issue:45764|issue:45765" ], - "prepared_review_unit_hash": "40b16eddc86b6994232fdd72ec69b381edc6dc18cf14b3b0c8aa48e702fe7026", + "prepared_review_unit_hash": "499e3e37b286d7d1aeaa2c55149b7d3ca6b08189a5530afbc4e0bfbdac6704f5", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1481, - "estimated_input_tokens": 371, - "estimated_eval_tokens": 998 + "serialized_chars": 1480, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1481, - "estimated_input_tokens": 371, - "estimated_eval_tokens": 998 + "serialized_chars": 1480, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two closed issues look like unrelated name/date entries rather than duplicate bug reports or the same underlying problem. The soft similarity is too weak to justify clustering them as duplicates.", - "confidence": 0.95, + "summary": "These two closed issues are not a duplicate pair; they have different person-name titles and only weak surface similarity (shared surname/date formatting). No code, targets, or discussion evidence suggests a common underlying bug or change.", + "confidence": 0.96, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:45764 is slightly cleaner and more self-contained as a representative item, but neither issue appears to be a true duplicate target.", + "best_issue_reason": "issue:45764 is the slightly better representative only because it was created first; there is no substantive evidence that either issue should be canonical for the other.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45764", "right": "issue:45765", "accept": false, - "reason": "The titles are different personal names with dates and provide no evidence of the same underlying bug, feature, or code-path change." + "reason": "Reject: the titles refer to different individuals and the only overlap is surname/date formatting, which is not enough to indicate the same underlying issue." } ] }, @@ -19501,12 +19542,12 @@ "soft_pairs": [ "issue:45810|issue:46295" ], - "prepared_review_unit_hash": "d0fc289b69a70f2b6d3f3b070793882f817c71f522eccba2ed98be270ad3892b", + "prepared_review_unit_hash": "ae8e21904cd3f9c7dc67750c5660bd6aaf9c4bc58ac234346d6f6bd4b8160cf7", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1542, + "serialized_chars": 1541, "estimated_input_tokens": 386, "estimated_eval_tokens": 1028 }, @@ -19514,7 +19555,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1542, + "serialized_chars": 1541, "estimated_input_tokens": 386, "estimated_eval_tokens": 1028 }, @@ -19523,18 +19564,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These issues are not duplicates: one requests adding a specific model class for token classification/value modeling, while the other asks for broad FP8 training support. They concern different features and code paths.", - "confidence": 0.96, - "canonical_issue_reason": "issue:45810 is the more concrete, narrowly scoped request with a clear implementation target (Qwen3_5ForTokenClassification).", + "summary": "These two issues are not duplicates: one requests a new Qwen3_5 token-classification/value-model class, the other requests FP8 training support. They concern different features and different code paths.", + "confidence": 0.97, + "canonical_issue_reason": "No canonical issue: issue 45810 is a model-class feature request, while issue 46295 is a training/runtime precision request; they are unrelated.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45810 is the best representative issue because it names a specific missing model capability, making it easier to triage and track.", + "best_issue_reason": "No single best issue for deduplication here because the cluster does not represent one underlying bug or change.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45810", "right": "issue:46295", "accept": false, - "reason": "Different underlying requests: model-class support for token classification/value modeling versus FP8 training support. They do not describe the same bug or change." + "reason": "Reject: the requests are about different capabilities (new token classification model class vs FP8 training support) and do not describe the same underlying problem or change." } ] }, @@ -19574,88 +19615,88 @@ "issue:45871|issue:45873", "issue:45872|issue:45873" ], - "prepared_review_unit_hash": "bd13fbddc25f27063dcfced16218205ead926ba9cbdb49c4276a7b0d43f42e14", + "prepared_review_unit_hash": "772e727a8a37ed70dd1876aad0f4ee720d2ccada5f00305f14dde55afe6557ae", "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 most likely contains two duplicate issue pairs: 45870/45871 and 45872/45873, based on identical titles and matching body lengths. 45869 appears related by theme but is a distinct issue, not part of those duplicate pairs.", - "confidence": 0.84, - "canonical_issue_reason": "45870 is the earliest member of one of the strongest likely-duplicate pairs, and it is a reasonable representative for that subthread. The available metadata supports duplication, but not a stronger claim than that.", + "summary": "Two likely duplicate pairs stand out: 45870/45871 and 45872/45873. 45869 is related by topic but appears to be a different request, so it should stay separate.", + "confidence": 0.93, + "canonical_issue_reason": "45872 is a reasonable canonical issue for the 45872/45873 pair because the titles match exactly and the body lengths match, so it can stand in for that same apparent request.", "canonical_pr_reason": null, - "best_issue_reason": "45870 is the best single issue representative in this set because it is part of the earliest clearly matching pair and has the clearest standalone title. That said, the cluster appears to contain two separate likely-duplicate subthreads rather than one unified exact duplicate group.", + "best_issue_reason": "45872 is the best overall issue candidate in this cluster because it anchors an exact-duplicate pair (45873) and has a specific title that distinguishes it from the more generic example issue.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45869", "right": "issue:45870", "accept": false, - "reason": "Same broad financial/DataForge theme, but the titles point to different concrete topics: a dataset listing versus an example. Not the same underlying issue." + "reason": "Related financial/DataForge topic, but the titles point to different requests: dataset/sample offer versus an example prediction issue." }, { "left": "issue:45869", "right": "issue:45871", "accept": false, - "reason": "Same broad financial/DataForge theme, but the titles point to different concrete topics: a dataset listing versus an example. Not the same underlying issue." + "reason": "Same as 45869 vs 45870: similar theme, but not the same apparent request." }, { "left": "issue:45870", "right": "issue:45871", "accept": true, - "reason": "Identical titles and matching body lengths make these look like likely duplicate issues, though the metadata alone does not prove they are exact duplicates." + "reason": "Exact same title and matching body length; they look like the same issue reported twice." }, { "left": "issue:45870", "right": "issue:45872", "accept": false, - "reason": "They share a broad DataForge/financial theme, but the titles describe different concrete requests: an example versus a scraping-service oriented issue." + "reason": "Different apparent requests: an example for financial time series prediction versus a financial time series + scraping service request." }, { "left": "issue:45870", "right": "issue:45873", "accept": false, - "reason": "They share a broad DataForge/financial theme, but the titles describe different concrete requests: an example versus a scraping-service oriented issue." + "reason": "Different apparent requests; only the broad DataForge/financial theme overlaps." }, { "left": "issue:45871", "right": "issue:45872", "accept": false, - "reason": "Same broad theme, but different concrete issue types and titles; not enough evidence that they are the same underlying issue." + "reason": "Different titles and different apparent request types, so not a likely duplicate pair." }, { "left": "issue:45871", "right": "issue:45873", "accept": false, - "reason": "Same broad theme, but different concrete issue types and titles; not enough evidence that they are the same underlying issue." + "reason": "Different apparent requests; shared topic is too broad to treat as a duplicate." }, { "left": "issue:45872", "right": "issue:45873", "accept": true, - "reason": "Identical titles and matching body lengths make these look like likely duplicate issues, though the metadata alone does not prove they are exact duplicates." + "reason": "Exact same title and matching body length; they appear to be duplicate copies of the same issue." } ] }, "evaluator_result": { "accept": false, - "feedback": "The cluster-level structure is plausible, but the writeup overstates the evidence. The packet only shows same titles and same body lengths for 45870/45871 and 45872/45873, which supports likely duplicates, not confirmed exact duplicates. Please soften those claims and keep the summary grounded in the available metadata." + "feedback": "The duplicate-pair calls are mostly well-grounded: 45870/45871 and 45872/45873 have identical titles and body lengths, so treating them as likely duplicates is conservative. However, the canonical/best-issue reasoning overstates the evidence: the packet does not contain bodies, so phrases like \"duplicated verbatim\" and \"more substantive\" are not supported, and \"strongest global issue candidate\" is not evidenced beyond title/body-length similarity. Please keep the language to \"likely duplicate\" / \"same apparent request\" and avoid claiming content-level distinctions that aren't in the packet." }, "evaluator_used": true, "retried": true, @@ -19682,47 +19723,50 @@ "soft_pairs": [ "issue:46439|issue:46441" ], - "prepared_review_unit_hash": "561e78631a690f21b7a9f7a14bb52f5bdd6ee0e6667174c4acd58f19ddaef2e9", + "prepared_review_unit_hash": "543df39bec430c1d13a1ac7c4d276c5cc715cc9564d07efa4662236cccfa665c", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1636, - "estimated_input_tokens": 409, - "estimated_eval_tokens": 1074 + "serialized_chars": 1637, + "estimated_input_tokens": 410, + "estimated_eval_tokens": 1076 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1636, - "estimated_input_tokens": 409, - "estimated_eval_tokens": 1074 + "serialized_chars": 1637, + "estimated_input_tokens": 410, + "estimated_eval_tokens": 1076 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both issues concern `cache_implementation=\"static\"` and `linear_attention`, but they describe different failure modes: one reports shape corruption during `early_initialization`, the other a `KeyError` for `linear_attention` in Qwen3.5. They may be related in subsystem, but not clearly the same underlying bug.", - "confidence": 0.66, - "canonical_issue_reason": "Issue 46439 is the more specific root-cause report: it names `early_initialization` and describes a concrete shape corruption path affecting linear-attention layers.", + "summary": "The two issues are plausibly related because both point to the static-cache path and `linear_attention`, but the evidence here is only title-level similarity. That supports a tentative duplicate/cluster link, not a strong claim that they share the exact same root cause.", + "confidence": 0.58, + "canonical_issue_reason": "Issue 46439 is the better canonical issue because it describes the broader failure mode in the static-cache/`early_initialization` path affecting linear-attention layers, whereas 46441 reads as a more model-specific crash symptom.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 46439 is the stronger canonical candidate because it points to a precise mechanism, while 46441 is a more product-specific crash symptom.", + "best_issue_reason": "Issue 46439 is the strongest standalone report since it names the mechanism and the general corruption/runtime-error behavior, making it the best representative for this cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46439", "right": "issue:46441", - "accept": false, - "reason": "Shared static-cache/linear-attention theme, but the reported errors differ (`RuntimeError` from wrong shapes vs `KeyError: 'linear_attention'`). Not enough evidence they are the same bug." + "accept": true, + "reason": "Tentatively accept: both titles involve static cache and `linear_attention`, so they are a plausible same-bug cluster candidate. However, the packet only provides soft similarity and titles, so this should be treated as provisional rather than a proven shared root cause." } ] }, - "evaluator_result": null, - "evaluator_used": false, - "retried": false, - "accepted_nontrivial_soft_edge": false, + "evaluator_result": { + "accept": false, + "feedback": "The linkage is plausible, but the analyst overstates the evidence available in the packet. We only have titles plus soft similarity, which support a conservative note that both issues involve static-cache and linear_attention, not a confident claim that they share the same root cause or that Qwen3.5/early_initialization applies to both. The soft-edge acceptance should be more tentative." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -19745,12 +19789,12 @@ "soft_pairs": [ "issue:46551|issue:46552" ], - "prepared_review_unit_hash": "b2cc4f9ce73122e92eeeec6387e7266a816e2cd4020a5ac0e3d4a49b2f9e5a70", + "prepared_review_unit_hash": "dac62016dca354642595b0144a96235e40e9f4d012b0fbb95079319fb05155b3", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1706, + "serialized_chars": 1707, "estimated_input_tokens": 427, "estimated_eval_tokens": 1110 }, @@ -19758,7 +19802,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1706, + "serialized_chars": 1707, "estimated_input_tokens": 427, "estimated_eval_tokens": 1110 }, @@ -19767,27 +19811,27 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are very likely duplicates: they have identical titles, the same body length, and maximal soft similarity. The evidence supports the same underlying bug report, though not literal verbatim body identity. The open issue is the better canonical thread.", - "confidence": 0.98, - "canonical_issue_reason": "issue:46552 is the better canonical issue because it is still open and already has inbound references, making it the active tracker for this bug.", + "summary": "These two issues are the same report with identical title/body about Wav2Vec2CTCTokenizer v5 mishandling the word delimiter as a special token. They should be clustered together, with the open issue as the better canonical record.", + "confidence": 0.99, + "canonical_issue_reason": "Issue 46552 is open and already has inbound references, so it is the better live tracker for the duplicated bug report.", "canonical_pr_reason": null, - "best_issue_reason": "issue:46552 is the best issue to keep since it is open and appears to be the living discussion for the same tokenizer bug.", + "best_issue_reason": "Issue 46552 is the strongest issue representative because it is the active duplicate target and carries the same exact report text as 46551.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46551", "right": "issue:46552", "accept": true, - "reason": "Identical titles plus equal body length and maximal soft similarity strongly indicate the same underlying issue report." + "reason": "Same title, same body, same underlying tokenizer bug report; this is a direct duplicate." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The duplicate conclusion is plausible, but the summary overstates the evidence by saying the issues are the same report verbatim. The packet only supports identical titles, equal body length, and maximal soft similarity (jaccard 1.0), not verbatim body identity. Please keep the wording more conservative." + "accept": true, + "feedback": "The clustering is well grounded in the packet: both issues have identical titles, identical body lengths, and the pair has Jaccard 1.0 with only soft_similarity evidence. Choosing the open issue (46552) as the canonical tracker is also conservative and supported by its inbound references. No overstatement beyond the available evidence." }, "evaluator_used": true, - "retried": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -19811,7 +19855,7 @@ "soft_pairs": [ "issue:46657|issue:46692" ], - "prepared_review_unit_hash": "f89861338899819128eea8124e4daf6001275297eca841db75974fcb8a763b30", + "prepared_review_unit_hash": "0050c78ae0f5b864f7eb05c083f52ffdc1b4dfe7bee60f2cb689ad7a086e2883", "component_budget": { "node_count": 2, "item_count": 2, @@ -19833,18 +19877,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are not duplicates: one reports a tensor size mismatch during object-detection training, the other reports OOM after several epochs while training RT-DETR v2. They describe different failure modes and likely different causes.", - "confidence": 0.96, + "summary": "These two issues do not look like the same bug: one reports a tensor shape mismatch during object-detection training, while the other reports GPU memory exhaustion after several epochs in RT-DETR v2 training.", + "confidence": 0.94, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue #46657 is the better representative if one must be picked, because it has the more concrete and specific error class (tensor shape mismatch) rather than a broad resource exhaustion symptom.", + "best_issue_reason": "If a single issue must represent the cluster, 46657 is the better anchor because it names a concrete, diagnosable failure mode; 46692 is a separate OOM problem with different remediation.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46657", "right": "issue:46692", "accept": false, - "reason": "Reject: shape-mismatch training failure and later OOM are different underlying problems, not the same bug or change." + "reason": "Different underlying failures: tensor size mismatch vs out-of-memory. They share a training context, but not the same code-path bug." } ] }, @@ -19904,154 +19948,160 @@ "pull_request:43823|pull_request:45213", "pull_request:43251|pull_request:43254" ], - "prepared_review_unit_hash": "bb30c6ff15f1f3a96c7fe8b2ff2e9132a1a6b2d9b3cdb9ecb960c14c6cff08ad", + "prepared_review_unit_hash": "e2f1e58824852d3668674cbc54b28681493307842d3a338cde4eb1706727f361", "component_budget": { "node_count": 14, "item_count": 14, "soft_pair_count": 19, - "serialized_chars": 38151, - "estimated_input_tokens": 9538, - "estimated_eval_tokens": 19332 + "serialized_chars": 38155, + "estimated_input_tokens": 9539, + "estimated_eval_tokens": 19334 }, "budget": { "node_count": 14, "item_count": 14, "soft_pair_count": 19, - "serialized_chars": 38151, - "estimated_input_tokens": 9538, - "estimated_eval_tokens": 19332 + "serialized_chars": 38155, + "estimated_input_tokens": 9539, + "estimated_eval_tokens": 19334 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This is mostly a loose similarity cluster of unrelated PRs across different models/features, with one genuine duplicate pair in `fixed_cross_entropy` kwargs handling.", - "confidence": 0.93, + "summary": "This set is mostly unrelated PRs from different models/features. The only true duplicate-like pair is the two `fixed_cross_entropy` patches for issue #43240; the rest are only loosely related by subsystem or generic \u201csoft similarity\u201d and should stay separate.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "`pull_request:43254` is the cleanest representative of the only true duplicate pair: it directly adds supported kwargs to `fixed_cross_entropy`, matching the same code-path change as `43251`.", + "canonical_pr_reason": "PR 43251 is the best canonical PR for the actual duplicate pair: it explicitly targets issue #43240 and directly fixes `fixed_cross_entropy` by forwarding supported `cross_entropy` kwargs. PR 43254 is a near-identical alternate patch to the same code path.", "best_issue_reason": null, - "best_pr_reason": "`pull_request:43254` is the best global representative because it captures the concrete loss-function fix most directly; the rest of the cluster is too heterogeneous to serve as a duplicate set.", + "best_pr_reason": "PR 43251 is the strongest representative of the only real duplicate cluster because it has the explicit issue link, touches the same file, and implements the same concrete fix in the cleanest form.", "soft_edge_verdicts": [ { "left": "pull_request:39895", "right": "pull_request:43424", "accept": false, - "reason": "Unrelated changes: new VideoPrism model/docs vs Executorch dynamic-shape export test." + "reason": "Unrelated model-addition PR vs. an executorch test change; no shared bug or mergeable code-path overlap." }, { "left": "pull_request:41855", "right": "pull_request:43823", "accept": false, - "reason": "Mistral tokenizer methods vs adding a new MobileLLM model; different code paths and goals." + "reason": "Tokenizer utility patch vs. adding a new model; same broad area of Transformers, but not the same underlying change." }, { "left": "pull_request:41855", "right": "pull_request:43424", "accept": false, - "reason": "Tokenizer API addition is unrelated to Executorch export testing." + "reason": "Mistral tokenizer methods and executorch export tests are different concerns." }, { "left": "pull_request:42668", "right": "pull_request:43823", "accept": false, - "reason": "Processor/pretrained robustness work is unrelated to introducing MobileLLM." + "reason": "Processor auto-registration / robustness changes are unrelated to adding MobileLLM." }, { "left": "pull_request:43251", "right": "pull_request:43823", "accept": false, - "reason": "Loss-function kwargs fix vs new MobileLLM model addition; no same underlying change." + "reason": "Cross-entropy kwargs fix is unrelated to introducing MobileLLM." }, { "left": "pull_request:43251", "right": "pull_request:44827", "accept": false, - "reason": "Cross-entropy kwargs fix and Mistral4 test fixes are different problems." + "reason": "Different code paths: loss utility kwargs handling vs. Mistral4 test/model changes." }, { "left": "pull_request:43424", "right": "pull_request:44827", "accept": false, - "reason": "Executorch dynamic-shape testing is unrelated to Mistral4 model/test fixes." + "reason": "Executorch dynamic-shape tests and Mistral4 fixes do not describe the same bug." }, { "left": "pull_request:43424", "right": "pull_request:43823", "accept": false, - "reason": "Executorch export tests and MobileLLM model code are unrelated." + "reason": "Executorch export tests are not the same change as MobileLLM model addition." }, { "left": "pull_request:43823", "right": "pull_request:45655", "accept": false, - "reason": "New model implementation vs class config-resolution fix; not the same bug/change." + "reason": "Model addition vs. class-resolution fix; no common underlying defect." }, { "left": "pull_request:43823", "right": "pull_request:43995", "accept": false, - "reason": "MobileLLM addition and Falcon output-interface refactor are different model changes." + "reason": "MobileLLM model scaffolding and Falcon output-interface refactoring are distinct changes." }, { - "left": "pull_request:43995", + "left": "pull_request:43823", "right": "pull_request:45213", "accept": false, - "reason": "Falcon refactor is unrelated to the model-creation skill scaffolding." + "reason": "Model addition and a model-creation skill draft are not the same artifact or fix." }, { "left": "pull_request:43995", "right": "pull_request:44827", "accept": false, - "reason": "Falcon output-collection refactor and Mistral4 test fixes do not share a concrete code-path problem." + "reason": "Falcon output collection refactor and Mistral4 test fixes target different model internals." + }, + { + "left": "pull_request:43995", + "right": "pull_request:45213", + "accept": false, + "reason": "Falcon model refactor is unrelated to the model-creation skill draft." }, { "left": "pull_request:44827", "right": "pull_request:45170", "accept": false, - "reason": "Mistral4 test fixes vs a LayerNorm typo rename across CLIP-like models are unrelated." + "reason": "Mistral4 test/model changes vs. `layrnorm` renaming across CLIP-like models; not the same bug." }, { "left": "pull_request:44827", "right": "pull_request:46267", "accept": false, - "reason": "Mistral4 test fixes and chat parsing/serve updates are different subsystems." + "reason": "Mistral4 test fixes and chat parsing updates are unrelated." }, { "left": "pull_request:44827", "right": "pull_request:46048", "accept": false, - "reason": "Mistral4 test fixes vs adding a TDT loss kernel are unrelated." + "reason": "Mistral4 fixes and TDT loss kernel integration address different subsystems and failure modes." }, { "left": "pull_request:44827", "right": "pull_request:45655", "accept": false, - "reason": "Mistral4 test fixes and config-resolution order fix are separate issues." + "reason": "Mistral4 test fixes are not the same issue as config-resolution ordering in `modeling_utils`." }, { "left": "pull_request:44827", "right": "pull_request:45213", "accept": false, - "reason": "A model test-fix PR is not the same change as the model-creation skill draft." + "reason": "Mistral4 PR and model-creation skill draft are unrelated." }, { "left": "pull_request:43823", "right": "pull_request:45213", "accept": false, - "reason": "MobileLLM model addition is unrelated to the model-creation skill PR." + "reason": "MobileLLM model addition and the model-creation skill draft are not the same underlying change." }, { "left": "pull_request:43251", "right": "pull_request:43254", "accept": true, - "reason": "Both PRs fix the same `fixed_cross_entropy` issue by exposing/pass-through kwargs to `nn.functional.cross_entropy`; they could plausibly be merged into one PR." + "reason": "Both patches fix the same concrete `fixed_cross_entropy` bug in `src/transformers/loss/loss_utils.py` by forwarding supported kwargs to `nn.functional.cross_entropy`, and one can plausibly subsume the other." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet, and the conservative rejects look appropriate. The only accepted soft edge is well-supported by the shared issue target and the same `fixed_cross_entropy` code path. No obvious overstatement or unsafe mergeability claim." + "feedback": "The summary stays grounded in the packet and is appropriately conservative. The only accepted pair, PR 43251/43254, is a plausible single-bugfix duplicate because both change `src/transformers/loss/loss_utils.py` for the same issue target (#43240) and implement the same `fixed_cross_entropy` kwargs forwarding. The remaining pairs are correctly left as unrelated soft similarities without overstating evidence." }, "evaluator_used": true, "retried": false, @@ -20078,7 +20128,7 @@ "soft_pairs": [ "pull_request:40546|pull_request:42781" ], - "prepared_review_unit_hash": "8ceb671284ff35d5cbce957263bb3783c7405f0ee5bb04d7ec03e287e9d515af", + "prepared_review_unit_hash": "9f6aca24fcf6c672600960b36b1abceaedba3adbab3987124966948a93974620", "component_budget": { "node_count": 2, "item_count": 2, @@ -20100,18 +20150,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two related-but-distinct PRs: one adds the core VibeVoice model, the other adds VibeVoice Realtime plus a separate acoustic tokenizer and generation changes. They share filenames because they extend the same model family, but they are not the same underlying fix/change.", + "summary": "Both PRs are related to VibeVoice, but they are not duplicates: #40546 is the core VibeVoice implementation, while #42781 adds a separate realtime/acoustic-tokenizer extension and related generation changes.", "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #40546 is the better canonical PR because it appears to introduce the base VibeVoice implementation, while #42781 builds a separate realtime extension on top of that model family.", + "canonical_pr_reason": "#40546 is the more central artifact: it introduces the base VibeVoice model, docs, auto-class wiring, and core conversion/modeling pieces. #42781 builds on the same model family but targets a different realtime variant with additional tokenizer/feature-extraction work.", "best_issue_reason": null, - "best_pr_reason": "PR #40546 is the strongest representative of the cluster: it is the foundational VibeVoice addition and has broader scope than the realtime follow-up.", + "best_pr_reason": "#40546 is the best representative for the cluster because it is the primary end-to-end VibeVoice implementation. #42781 is a distinct follow-on change for realtime support, not a duplicate of the base model PR.", "soft_edge_verdicts": [ { "left": "pull_request:40546", "right": "pull_request:42781", "accept": false, - "reason": "Do not merge: #40546 implements the base VibeVoice model, while #42781 adds VibeVoice Realtime plus acoustic-tokenizer and generation-path changes. They share subsystem files, but they are different concrete features and would not naturally collapse into one PR." + "reason": "Shared VibeVoice filenames are not enough to call these duplicates. #40546 implements the base model, while #42781 adds realtime support plus acoustic tokenizer and generation updates; they address different concrete changes and would not plausibly be merged as one PR." } ] }, @@ -20141,7 +20191,7 @@ "soft_pairs": [ "pull_request:40755|pull_request:41224" ], - "prepared_review_unit_hash": "71947da0564007df7698aa8bbd109b0eb2c544fd5e16b696c23c33dd8de36e07", + "prepared_review_unit_hash": "f9e3ccdffbfd980ef252b29e8ba5cbbeb7e4e5f1c6787b7b10088350f2cb90e4", "component_budget": { "node_count": 2, "item_count": 2, @@ -20163,18 +20213,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two PRs are unrelated feature additions in different subsystems: TimesFM covariate forecasting vs. DINOv3 image classification support. They should not be deduplicated.", - "confidence": 0.99, + "summary": "The two PRs are unrelated feature additions in different model areas: TimesFM covariate forecasting vs. DINOv3 image classification support. The similarity is only superficial (model/docs/test support), so they should not be clustered as duplicates.", + "confidence": 0.98, "canonical_issue_reason": null, "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "No single PR is a clear canonical representative because the cluster is not a true duplicate set; the PRs solve different problems and touch different code paths.", + "best_pr_reason": "No single best PR for duplicate triage: the PRs target different models and different concrete capabilities, so neither is a canonical representative of the other.", "soft_edge_verdicts": [ { "left": "pull_request:40755", "right": "pull_request:41224", "accept": false, - "reason": "Different models, different tasks, and different files/code paths. One adds TimesFM covariate forecasting; the other adds DINOv3 image classification support." + "reason": "Different underlying changes and code paths: TimesFM forecasting with covariates vs DINOv3 image classification support. Too broad/superficial to treat as the same PR." } ] }, @@ -20267,12 +20317,12 @@ "soft_pairs": [ "pull_request:42975|pull_request:42976" ], - "prepared_review_unit_hash": "66c0e7a1f821d3d7be79363dffe118c00d2ead029acfa376adcfbde98e08acc5", + "prepared_review_unit_hash": "f59cdad395855e3446ca2a8e1cf08467b45872d6ac0935efe1cc11a05659f922", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5934, + "serialized_chars": 5935, "estimated_input_tokens": 1484, "estimated_eval_tokens": 3224 }, @@ -20280,7 +20330,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5934, + "serialized_chars": 5935, "estimated_input_tokens": 1484, "estimated_eval_tokens": 3224 }, @@ -20289,18 +20339,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two separate workflow-dependency upgrade PRs with overlapping files, but they change different action families and target different upgrade goals, so they should not be clustered as the same change.", - "confidence": 0.94, + "summary": "These are overlapping workflow-maintenance PRs, but they do not fix the same underlying problem. One is a Node 24 compatibility sweep for core GitHub Actions, while the other is a broader latest-version bump centered on docker/build-push-action and related workflow actions.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 42975 is the broader, more central GitHub Actions maintenance change: it upgrades core actions like checkout/cache/upload-artifact across many workflows specifically for Node 24 compatibility.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "PR 42975 is the best representative because it captures the main cross-cutting maintenance theme and affects a larger set of workflow actions.", + "best_pr_reason": "No single PR is a clear canonical representative because the upgrades target different action families and different maintenance goals.", "soft_edge_verdicts": [ { "left": "pull_request:42975", "right": "pull_request:42976", "accept": false, - "reason": "Both are workflow action upgrades, but 42975 updates core GitHub Actions for Node 24 compatibility while 42976 only bumps docker/build-push-action to v6 and related workflow refs. They touch overlapping CI files, but not the same underlying bug or fixable code path." + "reason": "Different upgrade tracks: 42975 is specifically for Node 24 compatibility of actions like checkout/cache/upload-artifact, while 42976 updates docker/build-push-action and other actions. They touch similar workflows but do not appear to be the same concrete fix." } ] }, @@ -20322,101 +20372,130 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-42978-6", + "cluster_id": "cluster-42978-7", "nodes": [ "pull_request:42978", "pull_request:43448", "pull_request:43451", "pull_request:45490", "pull_request:46562", - "pull_request:46658" + "pull_request:46658", + "pull_request:46705" ], "soft_pairs": [ "pull_request:42978|pull_request:43448", "pull_request:42978|pull_request:43451", "pull_request:42978|pull_request:45490", + "pull_request:42978|pull_request:46705", "pull_request:43448|pull_request:43451", "pull_request:43448|pull_request:45490", + "pull_request:43448|pull_request:46705", "pull_request:43451|pull_request:45490", + "pull_request:43451|pull_request:46705", "pull_request:45490|pull_request:46562", + "pull_request:45490|pull_request:46705", "pull_request:46562|pull_request:46658" ], - "prepared_review_unit_hash": "45bec4ace93997e4dce84904078007e33ca0db8baa9c7721b143310faa91fbd9", + "prepared_review_unit_hash": "87524226e9a0ba46d9cf2283eba982a9c62905bf51b01eff62b1b96bcf835409", "component_budget": { - "node_count": 6, - "item_count": 6, - "soft_pair_count": 8, - "serialized_chars": 20829, - "estimated_input_tokens": 5208, - "estimated_eval_tokens": 10672 + "node_count": 7, + "item_count": 7, + "soft_pair_count": 12, + "serialized_chars": 24366, + "estimated_input_tokens": 6092, + "estimated_eval_tokens": 12440 }, "budget": { - "node_count": 6, - "item_count": 6, - "soft_pair_count": 8, - "serialized_chars": 20829, - "estimated_input_tokens": 5208, - "estimated_eval_tokens": 10672 + "node_count": 7, + "item_count": 7, + "soft_pair_count": 12, + "serialized_chars": 24366, + "estimated_input_tokens": 6092, + "estimated_eval_tokens": 12440 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is a set of separate model-addition PRs, not a duplicate set. The only potentially close pair is the two Step 3.7 Flash PRs, but one adds the core `step3p7` model package while the other adds a separate `step_3_7_flash` WIP/mini-model implementation and scripts, so they should stay separate.", - "confidence": 0.95, + "summary": "This cluster is mostly heterogeneous model-addition and infrastructure PRs. The only clear duplicate pair is the Step 3.7 Flash work: #46562 and #46658 target the same model/codepath, while the other PRs are separate model introductions or an unrelated distributed-config change.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #43451 (Molmo2) is the most complete and self-contained implementation in the set, with broad model, processor, video/image, auto-mapping, docs, and test coverage.", + "canonical_pr_reason": "#46562 is the best canonical PR for the only true duplicate pair: it is the more complete, model-facing Step 3.7 Flash implementation, while #46658 is a WIP/mini-model companion rather than a separate change.", "best_issue_reason": null, - "best_pr_reason": "PR #43451 is the best representative artifact because it is the most substantial, clearly scoped model-addition PR with the broadest implementation coverage and review activity.", + "best_pr_reason": "#46562 is the strongest representative PR in this cluster because it corresponds to the actual Step 3.7 Flash model addition and is the cleanest merge target among the duplicate pair.", "soft_edge_verdicts": [ { "left": "pull_request:42978", "right": "pull_request:43448", "accept": false, - "reason": "Different models entirely: ViT NEPA vs Molmo. Shared docs/auto-registry edits are just normal model-registration overlap, not the same underlying change." + "reason": "Both add different vision-language models (ViT NEPA vs Molmo); shared docs/auto-registration files are incidental, not the same change." }, { "left": "pull_request:42978", "right": "pull_request:43451", "accept": false, - "reason": "Different model families and code paths: ViT NEPA vs Molmo2." + "reason": "Different model additions with distinct architectures and files; overlap in auto docs is too broad to indicate duplication." }, { "left": "pull_request:42978", "right": "pull_request:45490", "accept": false, - "reason": "Different domains and implementations: vision model addition vs CTSM time-series model." + "reason": "ViT NEPA and CTSM are unrelated model additions; same repo plumbing does not make them the same underlying bug or feature." + }, + { + "left": "pull_request:42978", + "right": "pull_request:46705", + "accept": false, + "reason": "Model addition PR versus distributed-config infrastructure PR; no shared concrete code-path problem." }, { "left": "pull_request:43448", "right": "pull_request:43451", "accept": false, - "reason": "Molmo and Molmo2 are related by naming, but the files show distinct model packages and features; they are not the same concrete bug or change." + "reason": "Molmo and Molmo2 are separate model families/releases with different files and behavior; not mergeable as one PR." }, { "left": "pull_request:43448", "right": "pull_request:45490", "accept": false, - "reason": "Different model additions with unrelated architectures and code paths." + "reason": "Molmo and CTSM are distinct model additions; shared framework registration is not sufficient for duplicate triage." + }, + { + "left": "pull_request:43448", + "right": "pull_request:46705", + "accept": false, + "reason": "A model-addition PR and a distributed configuration PR are unrelated changes." }, { "left": "pull_request:43451", "right": "pull_request:45490", "accept": false, - "reason": "Molmo2 and CTSM are unrelated model additions; overlap is only in shared integration files." + "reason": "Molmo2 and CTSM are unrelated model additions; overlap is only in generic auto/model registry files." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46705", + "accept": false, + "reason": "Molmo2 model work is unrelated to distributed config support." }, { "left": "pull_request:45490", "right": "pull_request:46562", "accept": false, - "reason": "CTSM time-series integration and Step 3.7 Flash are unrelated model changes." + "reason": "CTSM and Step 3.7 Flash are different models; similar auto-registration edits do not imply the same bug or feature." + }, + { + "left": "pull_request:45490", + "right": "pull_request:46705", + "accept": false, + "reason": "CTSM model addition versus distributed config infrastructure; no duplicate relationship." }, { "left": "pull_request:46562", "right": "pull_request:46658", - "accept": false, - "reason": "Both mention Step 3.7 Flash, but they touch different packages and purposes: one adds the core `step3p7` model implementation, the other adds a separate WIP `step_3_7_flash` mini-model/demo/scripts bundle. They are not clearly the same mergeable code change." + "accept": true, + "reason": "Both PRs implement Step 3.7 Flash, share the same model registration path, and appear to be two versions of the same underlying feature/change." } ] }, @@ -20465,118 +20544,118 @@ "pull_request:46380|pull_request:46539", "pull_request:46380|pull_request:46579" ], - "prepared_review_unit_hash": "06dc4ffae5883af94e5354f95d29884045f25308befdb4870106b63fba49fd27", + "prepared_review_unit_hash": "2c57b19c10c70688314220c5fb5ab2940eb3aa54ed57980b6b2a676d765eb72d", "component_budget": { "node_count": 8, "item_count": 8, "soft_pair_count": 14, - "serialized_chars": 23163, - "estimated_input_tokens": 5791, - "estimated_eval_tokens": 11838 + "serialized_chars": 23165, + "estimated_input_tokens": 5792, + "estimated_eval_tokens": 11840 }, "budget": { "node_count": 8, "item_count": 8, "soft_pair_count": 14, - "serialized_chars": 23163, - "estimated_input_tokens": 5791, - "estimated_eval_tokens": 11838 + "serialized_chars": 23165, + "estimated_input_tokens": 5792, + "estimated_eval_tokens": 11840 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are mostly unrelated; the only apparent commonality is superficial similarity in titles or broad subsystem overlap. I found no convincing duplicate pairs.", - "confidence": 0.29, + "summary": "This cluster is heterogeneous: it mixes unrelated test fixes, a Whisper tokenizer change, a deprecated API alias, a generation-validation fix, CI infrastructure changes, and a typo cleanup. None of the soft edges look like the same underlying bug or change, so no merges should be made.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46380 is the most coherent functional bug fix in the set: it addresses a concrete generation-config validation issue and includes a targeted code change plus tests.", + "canonical_pr_reason": "PR 46579 is the broadest, most representative code-cleanup PR in the set, but the cluster is not actually a duplicate group; the other PRs are unrelated fixes in different subsystems.", "best_issue_reason": null, - "best_pr_reason": "PR 46380 is the strongest standalone candidate because it fixes a specific runtime bug in a clear code path, unlike the others which are mostly tests, docs, CI, or broad cleanup.", + "best_pr_reason": "PR 46579 is the best single representative only in a nominal sense because it spans many code files and is broader than the others, but it should not be treated as a duplicate target for the rest of the cluster.", "soft_edge_verdicts": [ { "left": "pull_request:43291", "right": "pull_request:43665", "accept": false, - "reason": "Whisper tokenizer/test fix vs unrelated CLIP/ViT test edits; no shared underlying bug." + "reason": "Both touch tests, but they target different models and different problems; not the same underlying bug." }, { "left": "pull_request:43291", "right": "pull_request:44815", "accept": false, - "reason": "Different code paths: Whisper decoding/tests versus FP8 dequantization." + "reason": "Whisper tokenizer/test fix is unrelated to the FP8 dequantization bug fix." }, { "left": "pull_request:43291", "right": "pull_request:46380", "accept": false, - "reason": "Whisper decode behavior and generation-config watermark validation are unrelated issues." + "reason": "Whisper test changes and generation watermarking validation are unrelated code paths." }, { "left": "pull_request:43498", "right": "pull_request:46579", "accept": false, - "reason": "Backward-compat alias for tie_weights is unrelated to a broad typo-cleanup PR." + "reason": "A backward-compatibility alias for tie_weights is unrelated to a broad typo-cleanup PR." }, { "left": "pull_request:43665", "right": "pull_request:44815", "accept": false, - "reason": "Test-only changes in vision models do not match the FP8 dequant fix." + "reason": "Test assertion changes for CLIP/ViT do not match the FP8 dequantization fix." }, { "left": "pull_request:43665", "right": "pull_request:46380", "accept": false, - "reason": "These touch unrelated test files versus generation validation logic." + "reason": "Different subsystems and different failure modes; only superficial test/bug-fix similarity." }, { "left": "pull_request:43665", "right": "pull_request:46539", "accept": false, - "reason": "Nonsense test edits and CI integration fixes are not the same change." + "reason": "Model test edits are unrelated to the Phimoe CI integration fix." }, { "left": "pull_request:43665", "right": "pull_request:46579", "accept": false, - "reason": "Test perturbations in CLIP/ViT are unrelated to code typo cleanup." + "reason": "Model test tweaks are unrelated to code typo cleanup across many files." }, { "left": "pull_request:44815", "right": "pull_request:46380", "accept": false, - "reason": "Both are functional fixes but for different problems and code paths; not mergeable as one PR." + "reason": "FP8 quantization/loading changes and generation config validation fixes are distinct bugs." }, { "left": "pull_request:44815", "right": "pull_request:46539", "accept": false, - "reason": "FP8 dequantization and Phimoe CI/test infrastructure are unrelated." + "reason": "Different problems: dequantization logic versus Phimoe integration CI behavior." }, { "left": "pull_request:44815", "right": "pull_request:46579", "accept": false, - "reason": "Shared files are not enough here; one is an FP8 bug fix, the other is a typo cleanup across many files." + "reason": "FP8 kernel/loading fixes are unrelated to typo-only cleanup." }, { "left": "pull_request:45114", "right": "pull_request:46579", "accept": false, - "reason": "Doctest documentation edits are unrelated to code typo fixes." + "reason": "Doctest formatting cleanup is not the same underlying change as code typo fixes." }, { "left": "pull_request:46380", "right": "pull_request:46539", "accept": false, - "reason": "Generation-config validation bug fix is unrelated to CI memory-limit/test harness work." + "reason": "Generation watermarking validation and Phimoe integration test infrastructure are unrelated." }, { "left": "pull_request:46380", "right": "pull_request:46579", "accept": false, - "reason": "A specific runtime bug fix does not match a broad typo-only cleanup PR." + "reason": "A generation config bug fix is unrelated to broad typo cleanup." } ] }, @@ -20606,7 +20685,7 @@ "soft_pairs": [ "pull_request:43488|pull_request:43532" ], - "prepared_review_unit_hash": "b56ba568e914e0773a8d16ef9d7f17beeb0bef4c38d639895db445bd98a77092", + "prepared_review_unit_hash": "e3099d4305c8fd845343735220549c70f26434e55f80ce349d0247e434c116a0", "component_budget": { "node_count": 2, "item_count": 2, @@ -20628,18 +20707,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two pull requests are not duplicates: one is a setup/version/dependency test change, the other is a workflow/testing-utils show-diff change. Their overlap is incidental (shared Vit test file) rather than the same underlying fix or feature.", - "confidence": 0.97, + "summary": "Both items are unrelated PRs that only overlap on a shared Vit test filename; they do not look like the same underlying change or bug fix.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "No clear canonical PR: the PRs target different code paths and purposes. If one must be chosen, #43532 is the broader change set, but it is not a substitute for #43488.", + "canonical_pr_reason": "43532 is the more representative PR in this cluster: it makes the broader CI/workflow and test-harness updates, while 43488 is a small repo-bot/check-style change with a version/dependency tweak.", "best_issue_reason": null, - "best_pr_reason": "#43532 is the more representative PR of the cluster only because it spans multiple workflow/test-related files; however, it still does not cover the separate setup/version-bump change in #43488.", + "best_pr_reason": "43532 is the best overall PR candidate because it targets the main CI/test execution path and is more substantive than 43488, which looks like a throwaway validation PR.", "soft_edge_verdicts": [ { "left": "pull_request:43488", "right": "pull_request:43532", "accept": false, - "reason": "Reject: they do not fix the same concrete bug/change. #43488 changes setup/version metadata and test deps, while #43532 rewires CI/workflow and test utilities. The shared Vit test file is not enough to treat them as one PR." + "reason": "Reject: the PRs have different purposes and code paths. 43488 is a small setup/version/test tweak, while 43532 changes workflow automation and testing utilities; the shared Vit test file is not enough to indicate the same underlying bug/change." } ] }, @@ -20669,12 +20748,12 @@ "soft_pairs": [ "pull_request:43612|pull_request:43613" ], - "prepared_review_unit_hash": "a20e42178fec5a24da598ae47491a61a6d5a4d0215230be2b960f51820764e4a", + "prepared_review_unit_hash": "9e434d983d855ba8a2a67d8eff95524cc9b8164ca9304b918d27b4a60b82ef69", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6082, + "serialized_chars": 6081, "estimated_input_tokens": 1521, "estimated_eval_tokens": 3298 }, @@ -20682,7 +20761,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6082, + "serialized_chars": 6081, "estimated_input_tokens": 1521, "estimated_eval_tokens": 3298 }, @@ -20691,18 +20770,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two open PRs add different pipelines: one for promptable concept segmentation and one for promptable visual segmentation. They share documentation/registration scaffolding, but they target different features, model integrations, and test suites.", - "confidence": 0.98, + "summary": "Two distinct PRs adding separate segmentation pipelines: one for promptable concept segmentation and one for promptable visual segmentation. They share documentation and pipeline scaffolding, but the underlying task/model support differs enough that they should not be clustered as duplicates.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: these are separate feature additions with different pipeline names, code paths, and model/test integrations.", + "canonical_pr_reason": "PR #43613 is the broader and more representative pipeline addition in this cluster: it adds promptable visual segmentation with more model/test coverage and a fuller integration surface.", "best_issue_reason": null, - "best_pr_reason": "Neither PR is a duplicate of the other; each is the best representative for its own distinct segmentation feature.", + "best_pr_reason": "PR #43613 is the strongest standalone candidate because it appears to implement the more complete, broader pipeline feature and has the richer set of supporting changes.", "soft_edge_verdicts": [ { "left": "pull_request:43612", "right": "pull_request:43613", "accept": false, - "reason": "Different underlying features and code paths: promptable concept segmentation vs promptable visual segmentation. The overlap is mostly shared docs/metadata/pipeline scaffolding, which is not enough to treat them as the same PR." + "reason": "Reject: these are different features targeting different tasks/code paths (concept segmentation vs visual segmentation). Shared docs and pipeline plumbing are not enough to treat them as the same change." } ] }, @@ -20732,7 +20811,7 @@ "soft_pairs": [ "pull_request:43651|pull_request:43663" ], - "prepared_review_unit_hash": "64509f87e2d8723bfdd85262863470d4521c0e514c2269c6d81ba746727f1628", + "prepared_review_unit_hash": "6d2af6a1f09caec203bdd5806e103583150e314a05e3d728f9a97c953c97dd9e", "component_budget": { "node_count": 2, "item_count": 2, @@ -20754,18 +20833,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests touch `src/transformers/trainer.py` and improve trainer extensibility, but they address different code paths: gradient-accumulation loss scaling vs. signature-column filtering.", - "confidence": 0.98, + "summary": "Both pull requests modify `src/transformers/trainer.py`, but they introduce different trainer extensibility hooks in different code paths: one changes gradient-accumulation loss scaling, the other extracts signature-column filtering into an overridable helper. They are related to Trainer customization, but not the same underlying change.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 43651 is the stronger canonical representative of the cluster because it makes a concrete trainer behavior change by introducing `_loss_is_scaled_for_ga`; PR 43663 is a separate refactor around input column filtering.", + "canonical_pr_reason": "PR #43663 is the better representative of the cluster because it refactors a core Trainer helper into an overridable method and is the broader API-style change.", "best_issue_reason": null, - "best_pr_reason": "PR 43651 is the best standalone PR in this cluster: it has a clear functional goal, a focused code change, and a direct behavioral impact. PR 43663 is related only at the broad trainer-extensibility level.", + "best_pr_reason": "PR #43663 is the stronger standalone candidate: it cleanly exposes a reusable hook for custom trainers and is more generally applicable than the loss-scaling tweak.", "soft_edge_verdicts": [ { "left": "pull_request:43651", "right": "pull_request:43663", "accept": false, - "reason": "Reject: these are different trainer changes. 43651 alters loss scaling during training, while 43663 refactors signature-column filtering for custom trainers. They do not fix the same underlying bug and would not naturally merge into one PR." + "reason": "Different trainer hooks and different behaviors: gradient-accumulation loss scaling vs. signature-column filtering. Same file/subsystem, but not the same bug or mergeable change set." } ] }, @@ -20795,12 +20874,12 @@ "soft_pairs": [ "pull_request:43656|pull_request:43842" ], - "prepared_review_unit_hash": "a8d34918839c6d3d3cf4ed7794c5f9cec8f31963014bd678b12bd04fa5b75033", + "prepared_review_unit_hash": "beda9940a0dad8634dced4c5402b2ed57e6ae56ddcb122871f6db6bb6eba3a75", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 3663, + "serialized_chars": 3664, "estimated_input_tokens": 916, "estimated_eval_tokens": 2088 }, @@ -20808,7 +20887,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 3663, + "serialized_chars": 3664, "estimated_input_tokens": 916, "estimated_eval_tokens": 2088 }, @@ -20817,24 +20896,27 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs touch `src/transformers/cli/serve.py` and mention `TypeAdapter` NameErrors, but they appear to address different annotation/runtime issues. PR 43842 is the more targeted fix; PR 43656 is broader typing cleanup with `from __future__ import annotations` and related annotation changes.", - "confidence": 0.72, + "summary": "The two PRs are related because they both touch `src/transformers/cli/serve.py` and address annotation-related `NameError`s, but they appear to fix different runtime paths: one broadly enables postponed annotation evaluation and updates several signatures, while the other specifically guards `TypeAdapter` behind `TYPE_CHECKING` for the pydantic-optional case. Related, yes; same fix cluster, not clearly.", + "confidence": 0.58, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 43842 is the clearest representative of this cluster: it directly fixes the CLI\u2019s optional-pydantic `TypeAdapter` annotation problem with a minimal, specific change.", + "canonical_pr_reason": "PR #43656 is the broader serve-CLI annotation fix and is the better representative of the shared file-level problem; #43842 is a narrower follow-up for a specific `TypeAdapter` import/evaluation case.", "best_issue_reason": null, - "best_pr_reason": "PR 43842 is the best overall fit because it is narrowly scoped to the concrete `TypeAdapter` NameError path and has an explicit issue link.", + "best_pr_reason": "PR #43656 is the best single representative because it addresses the more general annotation-evaluation issue in `serve.py`, while #43842 is a targeted variant tied to `TypeAdapter` and optional pydantic handling.", "soft_edge_verdicts": [ { "left": "pull_request:43656", "right": "pull_request:43842", "accept": false, - "reason": "Related in file and symptom, but the fixes are not the same concrete change: 43656 is broader annotation-evaluation cleanup, while 43842 specifically handles optional `pydantic`/`TypeAdapter` import and quoting. They should not be merged as duplicates." + "reason": "Both PRs are annotation-related and in the same file, but they solve different concrete issues: one uses `from __future__ import annotations` and signature rewrites, the other adds a `TYPE_CHECKING` import for `TypeAdapter`. That is related, but not enough to say they are the same underlying bugfix or merge into one PR." } ] }, - "evaluator_result": null, - "evaluator_used": false, - "retried": false, + "evaluator_result": { + "accept": false, + "feedback": "The relationship is plausible, but the evidence only clearly supports that both PRs touch `src/transformers/cli/serve.py` and both mitigate annotation-related NameErrors. That is weaker than saying they target the same underlying code path or the same bugfix. One PR adds `from __future__ import annotations` and adjusts several model/processor annotations; the other specifically handles `TypeAdapter` under `TYPE_CHECKING` for the pydantic-optional case. I\u2019d keep these as related, but not assert they are the same fix cluster with this level of confidence." + }, + "evaluator_used": true, + "retried": true, "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null @@ -20960,8 +21042,8 @@ "pull_request:44066|pull_request:44068", "pull_request:44018|pull_request:44066", "pull_request:44019|pull_request:44722", - "pull_request:44019|pull_request:44085", "pull_request:44086|pull_request:44722", + "pull_request:44019|pull_request:44085", "pull_request:44018|pull_request:44019", "pull_request:44019|pull_request:44071", "pull_request:44068|pull_request:44116", @@ -21020,397 +21102,373 @@ "pull_request:44044|pull_request:44129", "pull_request:44044|pull_request:44154" ], - "prepared_review_unit_hash": "6f51cf0b4a81ae30b50deaa46a36e0cfc34d3cc8b22c0f4a7ff9a0ac24630658", + "prepared_review_unit_hash": "adf3ac6614a6cd9cffe15d9d9ef338d1a23e3ed997df208aae45b29032a331df", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 78, - "serialized_chars": 84896, - "estimated_input_tokens": 21224, - "estimated_eval_tokens": 42704 + "serialized_chars": 84904, + "estimated_input_tokens": 21226, + "estimated_eval_tokens": 42708 }, "budget": { "node_count": 27, "item_count": 27, "soft_pair_count": 72, - "serialized_chars": 75092, - "estimated_input_tokens": 18773, - "estimated_eval_tokens": 37802 + "serialized_chars": 75099, + "estimated_input_tokens": 18775, + "estimated_eval_tokens": 37806 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The packet splits into two real duplicate pockets: GPT-Neo (44018/44068) and GPT-J/CodeGen centered on 44722 with variants 44066 and 44085. The rest are broader output-tracing refactors across different model files and should stay rejected.", - "confidence": 0.89, + "summary": "This cluster is mostly a collection of separate model-specific output-tracing refactors, not one broad duplicate set. The only clear duplicate-like subcluster is the GPT-J/CodeGen thread, plus a smaller GPT-Neo pair.", + "confidence": 0.88, "canonical_issue_reason": null, - "canonical_pr_reason": "No single canonical PR spans the whole packet; if one representative is needed, 44722 is the best anchor for the GPT-J/CodeGen pocket because both 44066 and 44085 converge on it, while 44018 is the anchor for the separate GPT-Neo pocket.", + "canonical_pr_reason": "44722 is the best representative of the GPT-J/CodeGen subcluster: it directly targets #43979, touches the same output-tracing code path as the closest neighbors, and is the broadest/latest variant in that thread.", "best_issue_reason": null, - "best_pr_reason": "44722 is the strongest single PR for global triage if forced to choose one, but only for the GPT-J/CodeGen pocket; the packet still contains a separate GPT-Neo duplicate pocket centered on 44018/44068.", + "best_pr_reason": "44722 is the strongest single PR anchor here because it is central to the only real overlap cluster and has the clearest direct relationship to the shared output-tracing refactor work.", "soft_edge_verdicts": [ { "left": "pull_request:43996", "right": "pull_request:44085", "accept": false, - "reason": "Different models/files; same output-tracing refactor theme only." + "reason": "Different model families and no shared concrete code path; both are only loosely about output tracing." }, { "left": "pull_request:43996", "right": "pull_request:44044", "accept": false, - "reason": "Different models/files; shared refactor style, not the same change." + "reason": "Different models and different refactors; shared issue target is not enough." }, { "left": "pull_request:44066", "right": "pull_request:44085", - "accept": false, - "reason": "Both are GPT-J-related, but the overlap is not the same concrete fix; 44722 is the actual duplicate hub." + "accept": true, + "reason": "Both are GPT-J output-tracing refactors in the same code path; 44066 just bundles an extra CodeGen change." }, { "left": "pull_request:44007", "right": "pull_request:44072", "accept": false, - "reason": "Different model implementations; only the standardized tracing pattern overlaps." + "reason": "Different model implementations; same theme, but not the same bug/change." }, { "left": "pull_request:44072", "right": "pull_request:44722", "accept": false, - "reason": "Different models/files; same migration theme only." + "reason": "Different models; output-tracing refactors are separate per architecture." }, { "left": "pull_request:44066", "right": "pull_request:44072", "accept": false, - "reason": "Different model families and code paths; not the same underlying change." + "reason": "CodeGen/GPT-J work versus EfficientNet; unrelated concrete code paths." }, { "left": "pull_request:44013", "right": "pull_request:44044", "accept": false, - "reason": "Different models/files; same refactor family only." + "reason": "MobileNetV2 versus DeBERTa-v2; only a broad standardization theme is shared." }, { "left": "pull_request:44066", "right": "pull_request:44086", "accept": false, - "reason": "Different model targets; similar decorator migration, not a duplicate fix." + "reason": "GPT-J/CodeGen versus MGP-STR; different model code paths." }, { "left": "pull_request:44066", "right": "pull_request:44071", "accept": false, - "reason": "Different models and code paths; only the output-tracing abstraction is shared." + "reason": "GPT-J/CodeGen versus MPT; same infrastructure theme, different change." }, { "left": "pull_request:44018", "right": "pull_request:44068", "accept": true, - "reason": "Same GPT-Neo model/file change; these look like the same underlying refactor." + "reason": "Same GPT-Neo files and same output-tracing refactor; these look like the same underlying change." }, { "left": "pull_request:44066", "right": "pull_request:44068", "accept": false, - "reason": "GPT-J/CodeGen versus GPT-Neo; related theme, different concrete code path." + "reason": "GPT-J/CodeGen versus GPT-Neo; not the same model or code path." }, { "left": "pull_request:44018", "right": "pull_request:44066", "accept": false, - "reason": "GPT-Neo versus GPT-J/CodeGen; same theme, different target change." + "reason": "GPT-Neo versus GPT-J/CodeGen; only the output-tracing pattern matches." }, { "left": "pull_request:44019", "right": "pull_request:44722", "accept": false, - "reason": "Different model families; both are output-tracing refactors, not the same fix." + "reason": "ResNet versus GPT-J/CodeGen; unrelated model refactors." }, { - "left": "pull_request:44019", - "right": "pull_request:44085", + "left": "pull_request:44086", + "right": "pull_request:44722", "accept": false, - "reason": "ResNet versus GPT-J; no shared concrete bug/change." + "reason": "MGP-STR versus GPT-J/CodeGen; different concrete code paths." }, { - "left": "pull_request:44086", - "right": "pull_request:44722", + "left": "pull_request:44019", + "right": "pull_request:44085", "accept": false, - "reason": "Different model/files; only the refactor pattern matches." + "reason": "ResNet versus GPT-J; different model families." }, { "left": "pull_request:44018", "right": "pull_request:44019", "accept": false, - "reason": "GPT-Neo and ResNet are unrelated code paths; same refactor class only." + "reason": "GPT-Neo and ResNet are unrelated implementations." }, { "left": "pull_request:44019", "right": "pull_request:44071", "accept": false, - "reason": "Different model implementations; not the same change." + "reason": "ResNet versus MPT; same broad refactor style, different model code." }, { "left": "pull_request:44068", "right": "pull_request:44116", "accept": false, - "reason": "Different models/files; similar output-handling cleanup only." + "reason": "GPT-Neo versus Flaubert/XLM; not the same underlying change." }, { "left": "pull_request:44068", "right": "pull_request:44722", "accept": false, - "reason": "GPT-Neo versus GPT-J/CodeGen; same modernization theme, not the same PR." + "reason": "GPT-Neo versus GPT-J/CodeGen; distinct model-specific refactors." }, { "left": "pull_request:44018", "right": "pull_request:44116", "accept": false, - "reason": "Different models/files; no shared concrete code-path problem." + "reason": "GPT-Neo versus Flaubert/XLM; no shared code path." }, { "left": "pull_request:44071", "right": "pull_request:44722", "accept": false, - "reason": "MPT versus GPT-J/CodeGen; same decorator pattern only." + "reason": "MPT versus GPT-J/CodeGen; different architectures." }, { "left": "pull_request:44018", "right": "pull_request:44085", "accept": false, - "reason": "GPT-Neo versus GPT-J; related refactor theme, different underlying change." + "reason": "GPT-Neo versus GPT-J; not the same model or change." }, { "left": "pull_request:44018", "right": "pull_request:44086", "accept": false, - "reason": "Different model families and files; not a duplicate." + "reason": "GPT-Neo versus MGP-STR; unrelated code paths." }, { "left": "pull_request:44068", - "right": "pull_request:44071", - "accept": false, - "reason": "GPT-Neo and MPT are different model implementations; no duplicate fix." - }, - { - "left": "pull_request:44059", - "right": "pull_request:44086", - "accept": false, - "reason": "GPT2 versus MGP-STR; same output-tracing idea, different code path." - }, - { - "left": "pull_request:44085", "right": "pull_request:44086", "accept": false, - "reason": "GPT-J/CodeGen versus MGP-STR; shared theme only." + "reason": "GPT-Neo versus MGP-STR; only the refactor style is similar." }, { "left": "pull_request:44098", "right": "pull_request:44101", "accept": false, - "reason": "ViLT and Flaubert are unrelated model implementations; not the same change." + "reason": "ViLT versus Flaubert/XLM; different model-specific changes." }, { "left": "pull_request:44059", - "right": "pull_request:44722", - "accept": false, - "reason": "GPT2 versus GPT-J/CodeGen; same refactor family, different target." - }, - { - "left": "pull_request:44018", "right": "pull_request:44071", "accept": false, - "reason": "GPT-Neo and MPT are different code paths; not duplicates." - }, - { - "left": "pull_request:44017", - "right": "pull_request:44161", - "accept": false, - "reason": "Segformer and LongT5 are different models; only the tracing pattern matches." + "reason": "GPT-2 versus MPT; not the same underlying bug/change." }, { - "left": "pull_request:44017", - "right": "pull_request:44722", + "left": "pull_request:44071", + "right": "pull_request:44086", "accept": false, - "reason": "Different architectures; same output-tracing migration, not the same fix." + "reason": "MPT versus MGP-STR; different code paths." }, { "left": "pull_request:44017", - "right": "pull_request:44073", + "right": "pull_request:44161", "accept": false, - "reason": "Different models/files; no shared concrete bug/change." + "reason": "SegFormer versus LongT5; different refactors in different models." }, { "left": "pull_request:43997", - "right": "pull_request:44073", + "right": "pull_request:44017", "accept": false, - "reason": "RegNet versus VisualBert; only the tracing refactor theme is shared." + "reason": "RegNet versus SegFormer; same output-tracing theme, different model code." }, { - "left": "pull_request:44071", - "right": "pull_request:44076", + "left": "pull_request:43997", + "right": "pull_request:44056", "accept": false, - "reason": "MPT and ImageGPT are different model paths; not the same change." + "reason": "RegNet versus MPNet; unrelated concrete changes." }, { "left": "pull_request:44074", "right": "pull_request:44101", "accept": false, - "reason": "TextNet and Flaubert are different implementations; not duplicates." + "reason": "TextNet versus Flaubert/XLM; not the same change." }, { "left": "pull_request:44056", "right": "pull_request:44161", "accept": false, - "reason": "MPNet versus LongT5; same broad refactor idea only." + "reason": "MPNet versus LongT5; different model implementations." }, { "left": "pull_request:43997", "right": "pull_request:44018", "accept": false, - "reason": "RegNet and GPT-Neo are different models; no duplicate code-path change." + "reason": "RegNet versus GPT-Neo; different code paths." }, { "left": "pull_request:43997", "right": "pull_request:44161", "accept": false, - "reason": "RegNet and LongT5 are unrelated implementations; same theme only." + "reason": "RegNet versus LongT5; unrelated model refactors." }, { "left": "pull_request:44074", "right": "pull_request:44161", "accept": false, - "reason": "TextNet and LongT5 are different model files; not the same fix." + "reason": "TextNet versus LongT5; different concrete changes." }, { "left": "pull_request:44018", - "right": "pull_request:44161", - "accept": false, - "reason": "GPT-Neo and LongT5 are unrelated targets; only the refactor style overlaps." - }, - { - "left": "pull_request:43997", "right": "pull_request:44071", "accept": false, - "reason": "RegNet and MPT are different models; same refactor effort, not the same bug." + "reason": "GPT-Neo versus MPT; different model code." }, { "left": "pull_request:44010", "right": "pull_request:44071", "accept": false, - "reason": "SqueezeBERT and MPT are different code paths; no duplicate change." + "reason": "SqueezeBERT versus MPT; same refactor pattern, different implementation." }, { "left": "pull_request:44010", "right": "pull_request:44073", "accept": false, - "reason": "SqueezeBERT and VisualBert are different models; same pattern only." + "reason": "SqueezeBERT versus VisualBERT; unrelated models." }, { "left": "pull_request:44018", "right": "pull_request:44129", "accept": false, - "reason": "GPT-Neo and SpeechT5 are unrelated model changes; not the same bug/change." + "reason": "GPT-Neo versus SpeechT5; not the same code path." }, { - "left": "pull_request:43997", - "right": "pull_request:44074", + "left": "pull_request:44073", + "right": "pull_request:44101", "accept": false, - "reason": "RegNet and TextNet are different models; only the output-tracing migration is common." + "reason": "VisualBERT versus Flaubert/XLM; different model changes." }, { - "left": "pull_request:44010", - "right": "pull_request:44018", + "left": "pull_request:44059", + "right": "pull_request:44068", "accept": false, - "reason": "SqueezeBERT and GPT-Neo are unrelated model changes; same refactor class only." + "reason": "GPT-2 versus GPT-Neo; only the output-tracing theme overlaps." }, { - "left": "pull_request:43997", - "right": "pull_request:44129", + "left": "pull_request:44056", + "right": "pull_request:44101", "accept": false, - "reason": "RegNet and SpeechT5 are different code paths; not a duplicate." + "reason": "MPNet versus Flaubert/XLM; different implementations." }, { - "left": "pull_request:44071", - "right": "pull_request:44129", + "left": "pull_request:44018", + "right": "pull_request:44059", "accept": false, - "reason": "MPT and SpeechT5 are different models; share only the refactor direction." + "reason": "GPT-Neo versus GPT-2; same broad subsystem, but not the same change." }, { - "left": "pull_request:43996", - "right": "pull_request:44161", + "left": "pull_request:44059", + "right": "pull_request:44086", "accept": false, - "reason": "FNet/CVT and LongT5 are different models; same tracing theme, not the same change." + "reason": "GPT-2 versus MGP-STR; unrelated models." }, { - "left": "pull_request:43996", - "right": "pull_request:44017", + "left": "pull_request:44071", + "right": "pull_request:44076", + "accept": false, + "reason": "MPT versus ImageGPT; separate refactors." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44061", "accept": false, - "reason": "FNet/CVT and Segformer are different model implementations; not duplicates." + "reason": "No matching concrete duplicate change is evident." }, { "left": "pull_request:44010", "right": "pull_request:44044", "accept": false, - "reason": "SqueezeBERT and DeBERTa V2 are different code paths; only the output-tracing pattern overlaps." + "reason": "SqueezeBERT versus DeBERTa-v2; different model-specific work." }, { "left": "pull_request:44044", "right": "pull_request:44074", "accept": false, - "reason": "DeBERTa V2 and TextNet are unrelated model changes; not the same fix." + "reason": "DeBERTa-v2 versus TextNet; unrelated changes." }, { "left": "pull_request:44066", "right": "pull_request:44722", "accept": true, - "reason": "Same GPT-J/CodeGen refactor family; 44722 is the duplicate hub for this pocket." + "reason": "Same GPT-J/CodeGen output-tracing refactor thread, with nearly identical code path and issue target." }, { "left": "pull_request:44085", "right": "pull_request:44722", "accept": true, - "reason": "Both target the same GPT-J output-tracing refactor and overlap on the same file." + "reason": "Both are GPT-J output-tracing refactors in the same file family; 44722 is the broader version." }, { "left": "pull_request:44085", "right": "pull_request:44129", "accept": false, - "reason": "GPT-J and SpeechT5 are different models; no shared concrete bug/change." + "reason": "GPT-J versus SpeechT5; different models and code paths." }, { "left": "pull_request:43996", "right": "pull_request:44007", "accept": false, - "reason": "FNet/CVT and RegNet are different model implementations; not the same change." + "reason": "CVT/FNet versus RegNet/ResNet/RT-DETR-ResNet; different model families." }, { "left": "pull_request:43996", "right": "pull_request:44129", "accept": false, - "reason": "FNet/CVT and SpeechT5 are unrelated code paths; only the refactor family matches." + "reason": "CVT/FNet versus SpeechT5; unrelated changes." }, { "left": "pull_request:44044", "right": "pull_request:44129", "accept": false, - "reason": "DeBERTa V2 and SpeechT5 are different models; same tracing refactor theme only." + "reason": "DeBERTa-v2 versus SpeechT5; different concrete implementations." }, { "left": "pull_request:44044", "right": "pull_request:44154", "accept": false, - "reason": "DeBERTa V2 and VITS are different implementations; not a duplicate fix." + "reason": "DeBERTa-v2 versus ViTS; only the output-tracing pattern matches, not the change." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The packet supports two separate duplicate pockets, not one combined GPT-J/GPT-Neo subcluster: GPT-Neo (44018/44068) and GPT-J/CodeGen (44066/44722, 44085/44722). The summary conflates these and overstates the evidence a bit, especially since 44066/44085 is explicitly rejected in the packet. The rest of the soft-edge rejections are conservative." + "accept": true, + "feedback": "The summary is well grounded in the packet: it correctly identifies the GPT-J/CodeGen thread as the only strong duplicate-like cluster, the GPT-Neo pair as a smaller subcluster, and treats the rest as separate model-specific refactors. The accepted soft-edge verdicts are conservative and look mergeable as same-bugfix PRs within the same code path." }, "evaluator_used": true, - "retried": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -21445,20 +21503,20 @@ "pull_request:43997|pull_request:44010", "pull_request:44010|pull_request:44129" ], - "prepared_review_unit_hash": "efb1278a9ef3958e58ea91076466d77e0b633e45a017e471a8d7edc555fb062c", + "prepared_review_unit_hash": "0d925e397e58ecbb477ba55dfa52b11beb35cc38545565ce26370a216ae3bede", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 78, - "serialized_chars": 84896, - "estimated_input_tokens": 21224, - "estimated_eval_tokens": 42704 + "serialized_chars": 84904, + "estimated_input_tokens": 21226, + "estimated_eval_tokens": 42708 }, "budget": { "node_count": 8, "item_count": 8, "soft_pair_count": 6, - "serialized_chars": 17125, + "serialized_chars": 17126, "estimated_input_tokens": 4282, "estimated_eval_tokens": 8820 }, @@ -21467,48 +21525,48 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a set of model-specific PRs that all refactor output tracing/standardized output capture, but they are not duplicates of one another. Each targets a different model implementation and code path, so they should stay separate.", - "confidence": 0.97, + "summary": "These PRs are all model-specific refactors to the new standardized output-capture/tracing machinery, but they touch different model implementations and do not look mergeable as duplicates of the same concrete fix.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "44129 is the strongest anchor: it has an explicit issue target, a concrete SpeechT5 code-path refactor, and clearly fits the standardized output capture theme without being as ambiguous as the title-only PRs.", + "canonical_pr_reason": "PR 44073 is the cleanest representative of the cluster: it clearly introduces the standardized output-capture pattern in a single model, with explicit _can_record_outputs wiring and the new decorator imports.", "best_issue_reason": null, - "best_pr_reason": "44129 is the best representative PR for the cluster because it is the most concretely scoped and explicitly linked to an issue, while still matching the shared output-capture refactor pattern.", + "best_pr_reason": "PR 44073 is the best overall representative because its title and diff most directly express the shared refactor theme without being unusually narrow or noisy.", "soft_edge_verdicts": [ { "left": "pull_request:44024", "right": "pull_request:44076", "accept": false, - "reason": "Different model families and code paths (FocalNet vs ImageGPT). Same broad refactor theme, but not the same underlying bug/change and not mergeable into one PR." + "reason": "Both are similar output-tracing refactors, but they apply to different models (FocalNet vs ImageGPT) and distinct code paths, so they are not the same underlying change." }, { "left": "pull_request:44073", "right": "pull_request:44074", "accept": false, - "reason": "Different models (VisualBert vs TextNet). Both adjust standardized output tracing, but they are separate model-specific refactors; one also changes tests, so this is not a duplicate." + "reason": "VisualBert and TextNet are separate model implementations; these are parallel refactors, not one shared bugfix or one PR that could plausibly subsume the other." }, { "left": "pull_request:44056", "right": "pull_request:44073", "accept": false, - "reason": "MPNet and VisualBert are unrelated implementations. Shared decorator/output-tracing work is too broad to treat as one underlying change." + "reason": "MPNet and VisualBert each add output-capture plumbing in different model files; same broad subsystem, but not the same concrete change." }, { "left": "pull_request:44056", "right": "pull_request:44074", "accept": false, - "reason": "Different models and forward implementations (MPNet vs TextNet). They touch the same modernization pattern, not the same concrete fix." + "reason": "MPNet and TextNet are unrelated model-specific refactors, even though both migrate to the same decorator-based output collection style." }, { "left": "pull_request:43997", "right": "pull_request:44010", "accept": false, - "reason": "RegNet vs SqueezeBert are separate model code paths. The similarity is only at the refactor pattern level, so they are not duplicates." + "reason": "RegNet and SqueezeBert changes are separate model migrations to the new output tracing API, not a duplicate fix on the same code path." }, { "left": "pull_request:44010", "right": "pull_request:44129", "accept": false, - "reason": "SqueezeBert and SpeechT5 are different model implementations with different output contracts. Too broad to merge as one PR or consider the same change." + "reason": "SqueezeBert and SpeechT5 affect different model stacks and different output-tracing implementations; they should stay separate." } ] }, @@ -21588,274 +21646,274 @@ "pull_request:44028|pull_request:44030", "pull_request:44029|pull_request:44030" ], - "prepared_review_unit_hash": "28050623566f6787f8c9b4d45982bbf47aa11d85c5406b13077a350d687fc325", + "prepared_review_unit_hash": "9c14ec8944788469923035968933b472224a9ec24a55c8a09d8b9be3835908cb", "component_budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 40, - "serialized_chars": 40167, - "estimated_input_tokens": 10042, - "estimated_eval_tokens": 20340 + "serialized_chars": 40181, + "estimated_input_tokens": 10046, + "estimated_eval_tokens": 20348 }, "budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 40, - "serialized_chars": 40167, - "estimated_input_tokens": 10042, - "estimated_eval_tokens": 20340 + "serialized_chars": 40181, + "estimated_input_tokens": 10046, + "estimated_eval_tokens": 20348 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are mostly model-specific refactors that share a common theme (output tracing / can_return_tuple), but they affect different code paths and should not be deduped as the same change.", - "confidence": 0.93, + "summary": "This cluster is a set of separate per-model output-tracing refactors, not duplicate fixes. The similar titles and diff patterns come from a shared refactor template, but each PR targets a different architecture and code path.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 44003 is the best representative of the cluster because it is a concrete, broader output-tracing refactor spanning two related model files and the same utility pattern.", + "canonical_pr_reason": "No true canonical duplicate exists here; the PRs are model-specific refactors rather than the same change. If one representative is needed, PR 44003 is the broadest example because it spans both mamba and falcon_mamba.", "best_issue_reason": null, - "best_pr_reason": "PR 44003 is the strongest cluster representative: it is a substantive model refactor, not just a narrow one-file tweak, and most clearly matches the shared theme across the set.", + "best_pr_reason": "PR 44003 is the best cluster representative because it has the broadest scope and the most substantial output-tracing rewrite among the candidates, but it is still not a duplicate of the others.", "soft_edge_verdicts": [ { "left": "pull_request:43998", "right": "pull_request:44002", "accept": false, - "reason": "Same refactor theme, but different model families and different forward paths; not the same underlying bug/change." + "reason": "Different model families and forward paths; same refactor motif only." }, { "left": "pull_request:43998", "right": "pull_request:44000", "accept": false, - "reason": "Both update output tracing, but they target unrelated models and code paths." + "reason": "Timm backbone vs vision-text dual encoder; unrelated code paths." }, { "left": "pull_request:43998", "right": "pull_request:44001", "accept": false, - "reason": "Shared helper usage only; timm_backbone and univnet are separate implementations." + "reason": "Timm backbone vs UnivNet; shared output-tracing theme is not enough." }, { "left": "pull_request:43998", "right": "pull_request:44003", "accept": false, - "reason": "Different architectures and distinct model files; too broad to be one mergeable PR." + "reason": "Different architectures (timm_backbone vs mamba/falcon_mamba); not the same bug/change." }, { "left": "pull_request:43998", "right": "pull_request:44029", "accept": false, - "reason": "Same general refactor style, but RWKV and timm_backbone are unrelated code paths." + "reason": "Timm backbone and RWKV are distinct models with separate implementation paths." }, { "left": "pull_request:43999", "right": "pull_request:44001", "accept": false, - "reason": "Different models and different output plumbing; not a duplicate change." + "reason": "MobileNetV1 and UnivNet are unrelated; only the refactor pattern matches." }, { "left": "pull_request:43999", "right": "pull_request:44002", "accept": false, - "reason": "Only the tracing refactor pattern matches; the model implementations are unrelated." + "reason": "MobileNetV1 vs UperNet; different model code paths." }, { "left": "pull_request:43999", "right": "pull_request:44025", "accept": false, - "reason": "Different model families and files; no shared concrete bug to merge together." + "reason": "MobileNetV1 and Depth Anything are separate architectures; not mergeable as one PR." }, { "left": "pull_request:43999", "right": "pull_request:44026", "accept": false, - "reason": "Both are output-tracing refactors, but for distinct wrapper/model code paths." + "reason": "MobileNetV1 vs VisionEncoderDecoder; same template, different change." }, { "left": "pull_request:43999", "right": "pull_request:44027", "accept": false, - "reason": "Same theme, different encoder-decoder wrapper and model behavior." + "reason": "MobileNetV1 and SpeechEncoderDecoder are different code paths." }, { "left": "pull_request:44000", "right": "pull_request:44001", "accept": false, - "reason": "Shared refactor vocabulary, but vision-text dual encoder and univnet are unrelated implementations." + "reason": "Vision-text dual encoder and UnivNet are unrelated models." }, { "left": "pull_request:44000", "right": "pull_request:44002", "accept": false, - "reason": "Different model code paths; no evidence of the same concrete fix." + "reason": "Different architectures and call stacks; no shared underlying defect." }, { "left": "pull_request:44000", "right": "pull_request:44003", "accept": false, - "reason": "These touch separate model families and cannot plausibly be one PR." + "reason": "Vision-text dual encoder vs mamba/falcon_mamba; not the same concrete change." }, { "left": "pull_request:44000", "right": "pull_request:44025", "accept": false, - "reason": "Both mention output tracing, but they apply to different models and wrappers." + "reason": "Different model families; only broadly similar output-handling refactors." }, { "left": "pull_request:44001", "right": "pull_request:44002", "accept": false, - "reason": "Distinct model-specific refactors, not the same bug or change." + "reason": "UnivNet and UperNet are separate code paths with different outputs." }, { "left": "pull_request:44001", "right": "pull_request:44003", "accept": false, - "reason": "UnivNet/Falcon-Mamba and UperNet are unrelated; only the refactor motif is shared." + "reason": "UnivNet and mamba/falcon_mamba do not share the same bug or fix." }, { "left": "pull_request:44001", "right": "pull_request:44025", "accept": false, - "reason": "Different model code paths and filenames; no duplicate relationship." + "reason": "Different architectures; same refactor wording is not enough for duplicate status." }, { "left": "pull_request:44002", "right": "pull_request:44029", "accept": false, - "reason": "Different architecture and tracing implementation; not mergeable as one PR." + "reason": "UperNet and RWKV are distinct model implementations." }, { "left": "pull_request:44002", "right": "pull_request:44003", "accept": false, - "reason": "Separate model families and changes; shared theme is too generic." + "reason": "UperNet vs mamba/falcon_mamba; different concrete changes." }, { "left": "pull_request:44003", "right": "pull_request:44028", "accept": false, - "reason": "Mamba/Falcon-Mamba refactor is unrelated to SuperPoint's tracing rewrite." + "reason": "Mamba/FalconMamba and SuperPoint are unrelated models." }, { "left": "pull_request:44003", "right": "pull_request:44004", "accept": false, - "reason": "Different model code and different change shape; not the same fix." + "reason": "Different subsystems (mamba vs codegen) and different forward semantics." }, { "left": "pull_request:43998", "right": "pull_request:44004", "accept": false, - "reason": "They both alter output handling, but in different models and different call signatures." + "reason": "Timm backbone and CodeGen are unrelated code paths." }, { "left": "pull_request:44001", "right": "pull_request:44004", "accept": false, - "reason": "CodeGen and UnivNet are unrelated; same broad theme is insufficient for deduping." + "reason": "UnivNet and CodeGen are different model families." }, { "left": "pull_request:44002", "right": "pull_request:44004", "accept": false, - "reason": "Different model implementations and different tracing mechanics." + "reason": "UperNet and CodeGen do not share a single underlying bug/change." }, { "left": "pull_request:44000", "right": "pull_request:44004", "accept": false, - "reason": "These are separate model refactors with no shared concrete code-path bug." + "reason": "Vision-text dual encoder vs CodeGen; same refactor theme only." }, { "left": "pull_request:44025", "right": "pull_request:44026", "accept": false, - "reason": "Both are output-tracing refactors, but depth-anything and vision-encoder-decoder are unrelated." + "reason": "Depth Anything and VisionEncoderDecoder are separate architectures." }, { "left": "pull_request:44025", "right": "pull_request:44027", "accept": false, - "reason": "Different wrapper/model code paths; not the same underlying change." + "reason": "Depth Anything vs SpeechEncoderDecoder; not the same code-path problem." }, { "left": "pull_request:44025", "right": "pull_request:44028", "accept": false, - "reason": "Depth Anything and SuperPoint refactors are distinct model-specific changes." + "reason": "Depth Anything and SuperPoint are unrelated implementations." }, { "left": "pull_request:44025", "right": "pull_request:44029", "accept": false, - "reason": "Different model families; only a generic refactor pattern is shared." + "reason": "Depth Anything and RWKV are distinct models with different refactors." }, { "left": "pull_request:44025", "right": "pull_request:44030", "accept": false, - "reason": "Separate model code paths and different output contracts." + "reason": "Depth Anything vs DPR; different code paths and outputs." }, { "left": "pull_request:44026", "right": "pull_request:44029", "accept": false, - "reason": "Vision-encoder-decoder and RWKV are unrelated implementations." + "reason": "VisionEncoderDecoder and RWKV are unrelated models." }, { "left": "pull_request:44026", "right": "pull_request:44027", "accept": false, - "reason": "These are distinct encoder-decoder wrappers with different internals." + "reason": "VisionEncoderDecoder and SpeechEncoderDecoder are different architectures." }, { "left": "pull_request:44026", "right": "pull_request:44028", "accept": false, - "reason": "Different model families and different refactor targets." + "reason": "VisionEncoderDecoder vs SuperPoint; same refactor motif only." }, { "left": "pull_request:44026", "right": "pull_request:44030", "accept": false, - "reason": "Shared output-tracing theme only; not the same concrete bug or change." + "reason": "Different model families and code paths." }, { "left": "pull_request:44027", "right": "pull_request:44028", "accept": false, - "reason": "Speech encoder-decoder and SuperPoint are unrelated model paths." + "reason": "SpeechEncoderDecoder and SuperPoint are unrelated implementations." }, { "left": "pull_request:44027", "right": "pull_request:44029", "accept": false, - "reason": "Different architecture and forward logic; not mergeable into one PR." + "reason": "SpeechEncoderDecoder and RWKV do not share a concrete fix." }, { "left": "pull_request:44027", "right": "pull_request:44030", "accept": false, - "reason": "Separate model wrappers and tracing changes." + "reason": "SpeechEncoderDecoder vs DPR; different underlying changes." }, { "left": "pull_request:44028", "right": "pull_request:44029", "accept": false, - "reason": "SuperPoint and RWKV are unrelated; the similarity is only thematic." + "reason": "SuperPoint and RWKV are different model code paths." }, { "left": "pull_request:44028", "right": "pull_request:44030", "accept": false, - "reason": "Different model code paths and distinct refactor scopes." + "reason": "SuperPoint vs DPR; only the output-tracing pattern overlaps." }, { "left": "pull_request:44029", "right": "pull_request:44030", "accept": false, - "reason": "RWKV and DPR implement different forward flows; not the same underlying fix." + "reason": "RWKV and DPR are separate architectures with different fixes." } ] }, @@ -21885,12 +21943,12 @@ "soft_pairs": [ "pull_request:44662|pull_request:45401" ], - "prepared_review_unit_hash": "57ddea3716b4ca7f8d225d6611c9ed5dcc81e2903ec1e7eabffaff315c250e4c", + "prepared_review_unit_hash": "471ad25dd9aaa744d9cc6a244e8bfb453617b7bfb512c0295465ddc4fb64b990", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6552, + "serialized_chars": 6551, "estimated_input_tokens": 1638, "estimated_eval_tokens": 3532 }, @@ -21898,7 +21956,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6552, + "serialized_chars": 6551, "estimated_input_tokens": 1638, "estimated_eval_tokens": 3532 }, @@ -21907,18 +21965,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two unrelated model-addition PRs: one adds PenguinVL and the other adds Voxtral TTS. They overlap on Transformers boilerplate, but they target different models, configs, processors, and tests, so they are not duplicates.", - "confidence": 0.98, + "summary": "These are two separate model-implementation PRs: one adds PenguinVL, the other adds Voxtral TTS. They share standard transformers scaffolding, but not the same underlying bug or change.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45401 is the cleaner representative because it has an explicit issue target and a focused, model-specific scope for Voxtral TTS.", + "canonical_pr_reason": "If a representative PR is needed, 45401 is the cleaner one to anchor on because it has an explicit issue target and a narrower, self-contained TTS feature scope.", "best_issue_reason": null, - "best_pr_reason": "PR 45401 is the best single representative of the cluster because it is tied to a concrete issue and describes one coherent implementation. PR 44662 is a separate implementation for a different model.", + "best_pr_reason": "45401 is the better global representative among the two because it is explicitly tied to an issue and is more narrowly scoped; 44662 is a larger standalone model addition with no issue linkage.", "soft_edge_verdicts": [ { "left": "pull_request:44662", "right": "pull_request:45401", "accept": false, - "reason": "Different underlying changes: PenguinVL vs Voxtral TTS. Shared Transformers registration/doc boilerplate is expected for separate model integrations and does not indicate the same bug or feature." + "reason": "Different model additions (PenguinVL vs Voxtral TTS) with different code paths and goals; shared registration/docs boilerplate is not enough to treat them as the same change." } ] }, @@ -21948,12 +22006,12 @@ "soft_pairs": [ "pull_request:44979|pull_request:45363" ], - "prepared_review_unit_hash": "65dcec8c41598137caef0d18d7ea0e234015e36d1429f90844d57600d4095f02", + "prepared_review_unit_hash": "4f813547662fe17b524bd147e5b827ffffa8e3377f87073f58e15b4bb03b4057", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4408, + "serialized_chars": 4407, "estimated_input_tokens": 1102, "estimated_eval_tokens": 2460 }, @@ -21961,7 +22019,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4408, + "serialized_chars": 4407, "estimated_input_tokens": 1102, "estimated_eval_tokens": 2460 }, @@ -21970,18 +22028,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are related in the broad area of module/kernel fusion, but they are not the same change: one introduces a generic module-fusion API scaffold, while the other wires up n-to-1 kernel fusion through KernelConfig across model loading/integration code.", - "confidence": 0.87, + "summary": "These are related but distinct pull requests: one introduces a generic Module Fusion API, while the other wires up n-to-1 kernel fusion through KernelConfig and adds an example. They do not look like the same mergeable change.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45363 is the better canonical representative because it implements the more complete, user-facing fusion feature and touches the actual loading/integration path; PR 44979 is a narrower API foundation.", + "canonical_pr_reason": "PR #44979 is the more canonical representative because it introduces the foundational Module Fusion API; PR #45363 is an application/integration on top of that API rather than the same change.", "best_issue_reason": null, - "best_pr_reason": "PR 45363 is the stronger overall exemplar of the cluster since it contains the concrete end-to-end kernel-fusion implementation and example usage, not just the lower-level module-fusion API layer.", + "best_pr_reason": "PR #44979 is the best cluster representative since it defines the core abstraction and new module implementation, whereas PR #45363 mainly consumes that work to enable a specific kernel-fusion path.", "soft_edge_verdicts": [ { "left": "pull_request:44979", "right": "pull_request:45363", "accept": false, - "reason": "Reject: they are conceptually adjacent but not the same underlying change. 44979 adds a standalone Module Fusion API and tests, while 45363 adds KernelConfig-based n-to-1 kernel fusion integration across model loading/utils plus an example. They could coexist rather than be merged into one PR." + "reason": "Different underlying changes: #44979 adds the generic fusion API/module, while #45363 integrates n-to-1 kernel fusion via KernelConfig and example code. They are related but not the same concrete fix/change." } ] }, @@ -22011,7 +22069,7 @@ "soft_pairs": [ "pull_request:45086|pull_request:45317" ], - "prepared_review_unit_hash": "453f084135124805800d6664969b94ed4200fcfdc567bb0b10368cf4fb46c81e", + "prepared_review_unit_hash": "8114a26b465bc41c06f67a8b949f8a8c89fb41322276beb6af5243829c38e5f4", "component_budget": { "node_count": 2, "item_count": 2, @@ -22033,18 +22091,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests address the same Mistral regex patch bug in `_patch_mistral_regex` and make the same code change; 45317 is the stronger canonical PR because it includes the implementation plus tests.", - "confidence": 0.98, + "summary": "Both pull requests address the same Mistral tokenizer bug in `_patch_mistral_regex`: using `tokenizer.backend_tokenizer.pre_tokenizer` causes an `AttributeError`, and both switch to `tokenizer.pre_tokenizer`. PR 45317 is the more complete fix because it includes test coverage.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "45317 is the most complete representative: it fixes the AttributeError in the Mistral regex patch and adds test coverage, while matching the same underlying code change as 45086.", + "canonical_pr_reason": "PR 45317 is the better canonical PR because it implements the same fix as 45086 and adds tests, making it the more complete and reviewable change.", "best_issue_reason": null, - "best_pr_reason": "45317 is best suited as the canonical PR because it is the more polished, test-backed version of the same fix and subsumes the smaller patch in 45086.", + "best_pr_reason": "PR 45317 is the strongest representative of the cluster: same concrete bug fix, same file, same tracked issue, plus tests that validate the regression.", "soft_edge_verdicts": [ { "left": "pull_request:45086", "right": "pull_request:45317", "accept": true, - "reason": "Same concrete bug fix in the same function and file; 45317 is effectively the same patch plus tests, so they could plausibly be merged into one PR." + "reason": "Same underlying AttributeError fix in `_patch_mistral_regex`; the second PR is essentially a superset with tests and the same code-path change." } ] }, @@ -22078,7 +22136,7 @@ "pull_request:45569|pull_request:46565", "pull_request:46332|pull_request:46565" ], - "prepared_review_unit_hash": "8a03b8689a0e5aa5555dffbb9a00ece5d89d75797ad7607cf1d609751a2f8d50", + "prepared_review_unit_hash": "632cc621e59a96cb262169c87d55237dabc795c8d8b0c07b3a55bb6ceebd9fd3", "component_budget": { "node_count": 4, "item_count": 4, @@ -22100,30 +22158,30 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is heterogeneous: two PRs are about NemotronH model refactors/variants, and two are about Nemotron ASR streaming. The overlaps are mostly superficial (shared auto-registry/docs scaffolding and Nemotron naming), not the same underlying change.", + "summary": "This cluster is heterogeneous: two NemotronH refactor/architecture PRs and two unrelated Nemotron ASR streaming PRs. None of the soft pairs look like the same underlying bug or change, so there is no true duplicate/canonical merge candidate across the cluster.", "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46565 is the broadest, most self-contained change set in the cluster and the clearest anchor for the ASR subtopic.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "PR 46565 has the strongest standalone scope and completeness (new Nemotron3_5 ASR model, conversion, auto-mapping, docs, and tests), making it the best representative PR here.", + "best_pr_reason": "46565 is the strongest standalone PR in the cluster because it is the broadest/latest ASR-streaming feature change, but it is still not a duplicate of the other PRs and only weakly represents the cluster.", "soft_edge_verdicts": [ { "left": "pull_request:45115", "right": "pull_request:45569", "accept": false, - "reason": "Related NemotronH work, but not the same change: 45115 is a conversion/refactor mapping tweak, while 45569 adds the broader NemotronH/NemotronH Dense/Sparse model split and docs." + "reason": "Both are NemotronH-related, but one is a GranitemoeHybrid inheritance/refactor and the other adds a dense/sparse split plus docs/auto wiring. Similar subsystem, different change." }, { "left": "pull_request:45569", "right": "pull_request:46565", "accept": false, - "reason": "Different model families and tasks: NemotronH architecture refactor versus Nemotron 3.5 ASR streaming. Shared auto/doc files do not indicate the same underlying bug or feature." + "reason": "These touch overlapping auto/docs files, but they target different models and problems: NemotronH refactor vs Nemotron ASR streaming. Not the same code-path fix." }, { "left": "pull_request:46332", "right": "pull_request:46565", "accept": false, - "reason": "Both are ASR-related, but they target different concrete models (Nemotron ASR vs Nemotron 3.5 ASR) with different codepaths and files; not mergeable as one duplicate PR." + "reason": "Both are ASR streaming PRs, but they cover different model families and different implementation scopes. They are related features, not the same underlying change." } ] }, @@ -22153,12 +22211,12 @@ "soft_pairs": [ "pull_request:45332|pull_request:45333" ], - "prepared_review_unit_hash": "fddef5f05dbdf79f10244675c3946f08d420e821395e53284ec4cc0f8f83aa3e", + "prepared_review_unit_hash": "d3a7fbe432f48fcd8ee23438fe62d74a0f3d6926c48daf640f1f1d25561aaf9b", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5466, + "serialized_chars": 5465, "estimated_input_tokens": 1367, "estimated_eval_tokens": 2990 }, @@ -22166,7 +22224,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5466, + "serialized_chars": 5465, "estimated_input_tokens": 1367, "estimated_eval_tokens": 2990 }, @@ -22175,18 +22233,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two related open PRs in the heterogeneous-model feature area, but they are not duplicates: one adds per-layer config support, while the other is a broader modeling/cache integration built on top of it.", - "confidence": 0.87, + "summary": "These are related feature PRs for heterogeneous model support, but they split the work into a config-layer foundation and a broader modeling/integration change. They share files and scope, yet they do not look like the same concrete change set or a single fix that should be deduplicated.", + "confidence": 0.79, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45332 is the broader umbrella change, covering per-layer config usage across modeling/cache/masking paths, so it is the best representative for the cluster.", + "canonical_pr_reason": "PR #45332 is the better representative because it is the broader end-to-end feature PR: it includes the per-layer config integration plus cache/modeling changes needed to actually use heterogeneous configs across the stack.", "best_issue_reason": null, - "best_pr_reason": "PR #45332 is the better global representative because it spans the full heterogeneous-model implementation rather than only the configuration layer.", + "best_pr_reason": "PR #45332 is the strongest single PR for the cluster because it covers the full heterogeneous model support path, while #45333 is the narrower config-only foundation.", "soft_edge_verdicts": [ { "left": "pull_request:45332", "right": "pull_request:45333", "accept": false, - "reason": "They overlap on heterogeneous config plumbing, but #45333 is only the per-layer configuration work while #45332 is the larger downstream modeling/cache integration. That is a dependency relationship, not the same concrete change, and they would not plausibly be merged as one PR under strict duplicate criteria." + "reason": "Related but not duplicates: #45333 adds per-layer configuration infrastructure, while #45332 builds on that with cache/modeling support and other integrations. They touch overlapping files, but they do not appear to fix the same concrete code-path problem and would not naturally collapse into one PR without losing separation of concerns." } ] }, @@ -22216,7 +22274,7 @@ "soft_pairs": [ "pull_request:45415|pull_request:45425" ], - "prepared_review_unit_hash": "75523f5d906521db4b027f7865f8cf38d681e464942c2077b5b59c27321f682d", + "prepared_review_unit_hash": "003cf4a47dc8f8c69f173d6a66f6fb55c3ed2ccca7eb20e10e87fdeb9eaf8e42", "component_budget": { "node_count": 2, "item_count": 2, @@ -22238,18 +22296,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related typing-focused PRs, but they are not the same change: #45415 is a broad type-checking sweep across many Transformers modules and type-ignore config, while #45425 adds/adjusts typing protocols and helper aliases for modeling_utils and related internals. They overlap in some touched files, but the underlying work differs.", - "confidence": 0.9, + "summary": "These are both typing-related PRs in transformers, but they are not the same change: one is a broad type-checking sweep across many files, while the other is a focused typing-support patch for modeling_utils and related integrations.", + "confidence": 0.94, "canonical_issue_reason": null, - "canonical_pr_reason": "#45415 is the broader umbrella change: it introduces type-checking updates across many `src/transformers/*.py` files and configures type-ignore handling, making it the better representative of this typing cleanup cluster.", + "canonical_pr_reason": "PR 45415 is the broader umbrella typing pass across many files, while PR 45425 is a narrower follow-up touching a subset of shared typing helpers.", "best_issue_reason": null, - "best_pr_reason": "#45415 is the best canonical PR candidate because it covers the widest portion of the typing effort and looks like the main sweep rather than a narrow helper-typing follow-up.", + "best_pr_reason": "PR 45415 is the better cluster representative because it is the more general, cross-cutting typing cleanup, but it still does not match 45425 as a duplicate.", "soft_edge_verdicts": [ { "left": "pull_request:45415", "right": "pull_request:45425", "accept": false, - "reason": "They are both typing-related, but not the same underlying code change. #45415 is a broad type-checking pass across many modules; #45425 adds specific typing protocols/aliases for modeling_utils and friends. Shared files alone do not make them duplicates." + "reason": "Reject: the overlap is only in general typing infrastructure. 45415 is a broad repo-wide type-checking sweep; 45425 adds specific protocols/aliases for modeling_utils and PEFT-related code, so they do not fix the same concrete change." } ] }, @@ -22279,7 +22337,7 @@ "soft_pairs": [ "pull_request:45513|pull_request:46084" ], - "prepared_review_unit_hash": "9fe714b97585bc7189cc93f23308bb6827653ff03cccf3185e7e74644d11cba1", + "prepared_review_unit_hash": "9786ea98f9d976b14f711abcd84ac77996fbeab8c7668d46cb316f9bb68348fd", "component_budget": { "node_count": 2, "item_count": 2, @@ -22301,18 +22359,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two pull requests are both about cached forward paths with seq_len > 1, but they fix different model families and different code paths: Qwen/OLMo linear attention vs. Mamba2 use_cache handling. They are related in theme, not duplicates.", - "confidence": 0.93, + "summary": "These are two unrelated PRs that both adjust cached forward behavior for sequence-length > 1, but in different model families and code paths. They should not be deduplicated.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #46084 is the clearest representative of a concrete cache bug fix because it has an explicit issue target and a focused seq_len > 1 use_cache correction in Mamba2.", + "canonical_pr_reason": "PR #46084 is the better canonical PR for this cluster because it is narrowly scoped, has an explicit issue target, and directly fixes a concrete use_cache bug for seq_len > 1. PR #45513 is broader and spans multiple Qwen/Olmo model files.", "best_issue_reason": null, - "best_pr_reason": "PR #46084 is the stronger standalone representative: it is tightly scoped, linked to a specific issue, and addresses a precise cached-state bug path. PR #45513 is broader and spans multiple model implementations, but it is a different fix.", + "best_pr_reason": "#46084 is the best representative PR: it has the clearest bug statement, an explicit linked issue, and a focused patch. #45513 is more sprawling and appears to cover a different model-specific caching defect.", "soft_edge_verdicts": [ { "left": "pull_request:45513", "right": "pull_request:46084", "accept": false, - "reason": "Same broad theme (cached multi-token forward) but different models and different mechanisms; no shared files/targets and the diffs address distinct bugs, so they should not be merged as duplicates." + "reason": "Different underlying bugs in different model implementations: Qwen3.5/Olmo hybrid linear attention cached forward vs Mamba2 use_cache with seq_len > 1. Similar theme, but not the same fixable code-path." } ] }, @@ -22342,7 +22400,7 @@ "soft_pairs": [ "pull_request:45548|pull_request:45662" ], - "prepared_review_unit_hash": "c74b0568565cc5298dfc9a8f0f4494f5a54da0f8e53662e9419c1c4132911bdc", + "prepared_review_unit_hash": "ce2384b1d5570fac25ad2d811949d841a3b4fa0c67e5d6c31617ca2343434cb0", "component_budget": { "node_count": 2, "item_count": 2, @@ -22364,18 +22422,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two open PRs both touch expert parallelism, but they address different failure modes: one preserves distributed config during loading/serialization, the other prevents expert parameters from being overwritten during sharded loading/broadcast.", + "summary": "Both PRs are about expert parallelism, but they fix different bugs on different loading paths: one targets DeepSpeed ZeRO-3/accelerate launch config handling, the other fixes expert parameters being overwritten during FSDP2/distributed loading. They should not be merged as duplicates.", "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45662 is the stronger representative of the EP-loading bug class because it fixes a concrete overwrite path with broader runtime impact across loading and MoE code paths.", + "canonical_pr_reason": "PR 45662 is the better canonical PR because it fixes a concrete expert-parameter overwrite bug in the distributed loading path and has the broader, more direct code-path impact.", "best_issue_reason": null, - "best_pr_reason": "PR #45662 is the best standalone candidate: it targets a specific, reproducible EP/FSDP2 loading bug rather than a narrower configuration-preservation fix.", + "best_pr_reason": "PR 45662 is the strongest representative of the cluster: it addresses a specific, user-visible EP failure mode with a focused fix across loading and tensor-parallel integration.", "soft_edge_verdicts": [ { "left": "pull_request:45548", "right": "pull_request:45662", "accept": false, - "reason": "Reject: both are EP-related, but #45548 fixes distributed_config serialization/accelerate+ZeRO-3 loading while #45662 fixes expert params being overwritten after post-shard/broadcast. Different concrete bugs and not one mergeable PR." + "reason": "They share the broad EP theme, but 45548 is about DeepSpeed ZeRO-3 loading/config serialization via accelerate launch, while 45662 fixes experts being silently overwritten by rank-0 broadcast during FSDP2/distributed loading. Different concrete bugs and code paths." } ] }, @@ -22405,40 +22463,40 @@ "soft_pairs": [ "pull_request:45639|pull_request:45645" ], - "prepared_review_unit_hash": "88f8296de48f7f18171b6ef5aae53147837dcc764c2e10412fbdb5ed7f165e5b", + "prepared_review_unit_hash": "9330c25b42182b406fdfb91d6274229093a873a82a1f78f85cefc2edbc65e0b8", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4588, - "estimated_input_tokens": 1147, - "estimated_eval_tokens": 2550 + "serialized_chars": 4589, + "estimated_input_tokens": 1148, + "estimated_eval_tokens": 2552 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4588, - "estimated_input_tokens": 1147, - "estimated_eval_tokens": 2550 + "serialized_chars": 4589, + "estimated_input_tokens": 1148, + "estimated_eval_tokens": 2552 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs target the same xdist collision around `captured_info` output in `testing_utils.py`; PR 45645 appears to be the broader, more complete fix, while 45639 is a narrower subset of that change.", + "summary": "Both PRs address the same xdist-related captured_info logging collision in testing_utils; PR 45645 is the broader, more complete fix that also updates CI to read multiple log files.", "confidence": 0.84, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45645 is the broader canonical fix: it addresses the xdist-safe output handling in `testing_utils.py` and also updates CI/workflow handling for multiple captured log files, making it the more complete implementation of the shared issue.", + "canonical_pr_reason": "PR 45645 is the better canonical PR because it contains the core xdist-safe logging fix plus the CI/workflow adjustments needed to consume the new per-worker artifacts.", "best_issue_reason": null, - "best_pr_reason": "PR 45645 is the best representative because it covers the core bug and the downstream CI artifact handling needed to preserve debug logs; 45639 only covers the library-side portion.", + "best_pr_reason": "PR 45645 is the strongest representative of the change set: same underlying bug, broader coverage, and the supporting workflow updates make it the more complete fix.", "soft_edge_verdicts": [ { "left": "pull_request:45639", "right": "pull_request:45645", "accept": true, - "reason": "Same underlying bug: xdist collisions for `captured_info` artifacts in the patched testing debug path. 45645 expands the same fix with CI log aggregation, so the two are close enough to treat as one change cluster." + "reason": "Both PRs fix the same concrete problem: xdist collisions around captured_info output in testing_utils. 45645 appears to extend 45639 with the same core fix plus CI handling for multiple output files, so they are plausibly mergeable into one PR." } ] }, @@ -22468,7 +22526,7 @@ "soft_pairs": [ "pull_request:45670|pull_request:46320" ], - "prepared_review_unit_hash": "643f0d07450be6f89ebc461e351e597ad7399e32ac5e83b4384bd1143c70ae0b", + "prepared_review_unit_hash": "da4ca3f9439594ae1425fd31d61b760794e161976890f6ba44b4b533c6866dd8", "component_budget": { "node_count": 2, "item_count": 2, @@ -22490,18 +22548,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two PRs are unrelated: one is a small AutoModel registry mapping for GLM-ASR, the other adds attn_implementation dispatch support for RecurrentGemma with code/docs/tests. They should not be clustered as duplicates.", - "confidence": 0.97, + "summary": "The two PRs are unrelated: one updates AutoModel registration for glmasr/dia, while the other adds RecurrentGemma attention-dispatch support plus docs/tests. They do not appear to be the same underlying change or a single mergeable fix.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #46320 is the better canonical representative because it is a concrete feature/change set with code, docs, and tests centered on one behavior change, whereas #45670 is a narrow registry tweak.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "PR #46320 is the strongest standalone artifact in the set: it describes a specific code-path enhancement and includes validation coverage. #45670 is too small and different in scope to serve as a good cluster anchor.", + "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "pull_request:45670", "right": "pull_request:46320", "accept": false, - "reason": "These are different changes in different model areas: #45670 adds a model class to auto-generation mappings, while #46320 implements RecurrentGemma attention-implementation dispatch. They do not fix the same underlying bug or change the same code path." + "reason": "Different features and code paths: PR 45670 only changes auto-model class mapping, while PR 46320 modifies RecurrentGemma attention implementation dispatch and related docs/tests. They are not the same bug/change and would not be merged into one PR." } ] }, @@ -22531,12 +22589,12 @@ "soft_pairs": [ "pull_request:45720|pull_request:46259" ], - "prepared_review_unit_hash": "4d9f032dc8b736dedb5c225c9e45faaa1e9c418226d847fba1f9cbca627e5bda", + "prepared_review_unit_hash": "73224546847048c491f4fc84f40209fdd3c8effd059623808a285d6ce8ab5fd0", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 7247, + "serialized_chars": 7248, "estimated_input_tokens": 1812, "estimated_eval_tokens": 3880 }, @@ -22544,7 +22602,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 7247, + "serialized_chars": 7248, "estimated_input_tokens": 1812, "estimated_eval_tokens": 3880 }, @@ -22553,18 +22611,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two unrelated linter/chore pull requests that happen to overlap in several model files. PR 45720 applies TRF003\u2019s return_dict/can_return_tuple refactor, while PR 46259 implements TRF018 weight-initialization modeling changes. The overlap is incidental, not the same underlying fix.", - "confidence": 0.96, + "summary": "These PRs are only superficially similar: both are linter-driven, but one adds tuple-return handling for model forwards (TRF003) while the other expands weight-initialization modeling rules (TRF018). They address different code paths and should not be merged as duplicates.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45720 is the better representative because it is a focused, concrete implementation of TRF003 across model forwards and is more coherent as a standalone change set.", + "canonical_pr_reason": "No canonical PR: 45720 and 46259 implement different lint rules and fix different behaviors, so they are separate changes.", "best_issue_reason": null, - "best_pr_reason": "PR 45720 is the best PR for the cluster because it is the clearer, more self-contained code change; PR 46259 addresses a different linter rule and a different bug class.", + "best_pr_reason": "Neither PR is a good global best for the cluster because they solve unrelated problems; if forced to pick a representative, 46259 is broader, but the cluster should remain split.", "soft_edge_verdicts": [ { "left": "pull_request:45720", "right": "pull_request:46259", "accept": false, - "reason": "Reject: these PRs fix different problems. 45720 changes model forward signatures/decorators for TRF003, while 46259 changes module weight initialization for TRF018. Shared filenames reflect broad model coverage, not the same concrete code-path bug." + "reason": "Different rules and fixes: 45720 adds can_return_tuple/return_dict handling across model forwards, while 46259 refactors _init_weights behavior for many modules. Shared files are incidental, not the same underlying change." } ] }, @@ -22660,40 +22718,40 @@ "soft_pairs": [ "pull_request:45932|pull_request:46100" ], - "prepared_review_unit_hash": "5e0174b8b776b2a8f38d66b98e962b3051186ecb6e6a59780848e10e0e911cc2", + "prepared_review_unit_hash": "57ceb34b13313f7cd456e992b4bd2930d7132d5e070f4d3441193e98aded017f", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5161, - "estimated_input_tokens": 1291, - "estimated_eval_tokens": 2838 + "serialized_chars": 5160, + "estimated_input_tokens": 1290, + "estimated_eval_tokens": 2836 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5161, - "estimated_input_tokens": 1291, - "estimated_eval_tokens": 2838 + "serialized_chars": 5160, + "estimated_input_tokens": 1290, + "estimated_eval_tokens": 2836 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are unrelated open PRs: #45932 is a CI/telemetry plus model-config patch, while #46100 is a test migration adding AltCLIP/ViT kernel tests and DDP config. The shared similarity is superficial and not a duplicate change.", + "summary": "The two PRs are not duplicates: one is a model/workflow change with telemetry and architecture updates, while the other is a test-migration PR adding DDP/kernel test infrastructure and model tests. They touch different code paths and should not be clustered together as the same change.", "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR for this cluster: the PRs address different code paths and goals, so neither should be treated as a duplicate of the other.", + "canonical_pr_reason": "No canonical PR: PR #45932 and PR #46100 have different goals, touched files, and code paths, so they are not the same underlying change.", "best_issue_reason": null, - "best_pr_reason": "Neither PR is a good global representative of the pair because they are not the same change; #45932 is broader infra/model churn, while #46100 is focused on test migration and new kernel coverage.", + "best_pr_reason": "PR #45932 is the better representative PR only in the sense that it is a more substantive code-change PR; PR #46100 is primarily test migration/infrastructure.", "soft_edge_verdicts": [ { "left": "pull_request:45932", "right": "pull_request:46100", "accept": false, - "reason": "Different changes: #45932 touches CI/OTel wiring and model configuration, whereas #46100 adds new tests and DDP setup for AltCLIP/ViT. They do not fix the same concrete bug or implement the same feature." + "reason": "Reject: these are unrelated \"do not merge\" PRs. #45932 changes model code and CI secrets/workflow; #46100 adds distributed test config and test migrations. They do not fix the same concrete bug or implement the same change." } ] }, @@ -22723,7 +22781,7 @@ "soft_pairs": [ "pull_request:46126|pull_request:46269" ], - "prepared_review_unit_hash": "5839b669cf48062fde89b8c7d4f31706eae7372e30ada6e5b596594963e00bf6", + "prepared_review_unit_hash": "93b79d783fd5a92cce4a5c58b817695381ebceeaf7b83f34cb0af2445ffcfcd9", "component_budget": { "node_count": 2, "item_count": 2, @@ -22745,18 +22803,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two PRs overlap in distributed-module filenames, but they are not the same change: one is a focused EP/Trainer integration, while the other is a broad distributed branch base spanning FSDP, integrations, modeling, and CI. They should not be deduplicated.", - "confidence": 0.94, + "summary": "These PRs overlap in distributed/EP-related files, but they do not look like the same change: PR 46126 is a focused EP + Trainer integration feature, while PR 46269 is a broad distributed branch base touching many unrelated areas. The shared filenames are not enough to treat them as duplicates.", + "confidence": 0.92, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #46126 is the more focused, concrete change set centered on EP routing and Trainer integration; it is a better canonical representative than the broad branch-base PR.", + "canonical_pr_reason": "PR 46126 is the more focused, concrete change: it targets EP + Trainer integration and has a narrow set of distributed-model updates. PR 46269 is a broad branch-base PR with many unrelated edits, so it is not a good canonical representative for this cluster.", "best_issue_reason": null, - "best_pr_reason": "PR #46126 is the best single PR to keep as the representative artifact because it has a narrow, coherent feature scope. PR #46269 is too broad and mixed to serve as a clean canonical duplicate target.", + "best_pr_reason": "PR 46126 is the best representative because it describes a specific code-path change and has a coherent scope. PR 46269 is too broad and mixed to serve as the cluster\u2019s main PR.", "soft_edge_verdicts": [ { "left": "pull_request:46126", "right": "pull_request:46269", "accept": false, - "reason": "Shared distributed filenames do not indicate the same underlying change. #46126 targets EP routing/Trainer integration, while #46269 is a much larger distributed branch base touching many unrelated areas; they would not plausibly merge as one PR." + "reason": "Shared distributed filenames do not indicate the same underlying bug or change. 46126 is a targeted EP/Trainer integration PR, while 46269 is a large base branch with many additional distributed and unrelated modifications; they would not plausibly be merged as one PR." } ] }, @@ -22818,190 +22876,202 @@ "pull_request:46504|pull_request:46563", "pull_request:46432|pull_request:46563" ], - "prepared_review_unit_hash": "73990c20da32ca61dea1b77f2294243b92b303e75dfb2ba892ee56d2ac647ea6", + "prepared_review_unit_hash": "c739833276fa42339bfc7099f48d8947cffdea11b4227a3205d04dc0733edb8c", "component_budget": { "node_count": 9, "item_count": 9, "soft_pair_count": 26, - "serialized_chars": 23362, - "estimated_input_tokens": 5841, - "estimated_eval_tokens": 11938 + "serialized_chars": 23358, + "estimated_input_tokens": 5840, + "estimated_eval_tokens": 11936 }, "budget": { "node_count": 9, "item_count": 9, "soft_pair_count": 26, - "serialized_chars": 23362, - "estimated_input_tokens": 5841, - "estimated_eval_tokens": 11938 + "serialized_chars": 23358, + "estimated_input_tokens": 5840, + "estimated_eval_tokens": 11936 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are not duplicate PRs; they are a loose cluster of unrelated workflow/config updates that only share generic 'chore: update ...yml' similarity. The only close subgroup is the three pr-ci-security-gate edits, but they change different aspects of that workflow and would not plausibly merge into one PR as the same fix.", - "confidence": 0.98, + "summary": "Mostly unrelated workflow maintenance PRs. The only tight textual cluster is three changes to `pr-ci-security-gate.yml`, but they alter different behaviors and should not be merged as duplicates.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "No true canonical PR exists for a duplicate set here. If one representative is needed, pull_request:46554 is the closest because it makes the most substantive security-gate logic change, but it is still not a duplicate of the others.", + "canonical_pr_reason": "No true duplicate PR canonical emerges: the set spans several distinct workflow edits, and the `pr-ci-security-gate.yml` PRs are separate behavior changes rather than the same fix. If a representative is needed, `#46554` is the most substantive security-gate change, but it is not a duplicate of the others.", "best_issue_reason": null, - "best_pr_reason": "pull_request:46554 is the strongest representative of the cluster because it introduces concrete security-gate behavior changes; the others are separate workflow tweaks in different files or different concerns.", + "best_pr_reason": "`#46554` is the strongest representative of the cluster because it makes the most concrete security-gate/code-path change, but it still stands alone and should not absorb the other PRs as duplicates.", "soft_edge_verdicts": [ { "left": "pull_request:46138", "right": "pull_request:46554", "accept": false, - "reason": "Different workflow files and different changes: self-comment CI command-input handling vs pr-ci security gate logic." + "reason": "Different workflow files and different fixes: stdin handling for self-comment CI vs new security-gate logic." }, { "left": "pull_request:46138", "right": "pull_request:46558", "accept": false, - "reason": "Different workflow files and unrelated changes; only the generic 'workflow update' pattern matches." + "reason": "Unrelated changes in different workflows; one updates self-comment CI, the other tightens PR CI security-gate permissions/approval flow." }, { "left": "pull_request:46138", "right": "pull_request:46563", "accept": false, - "reason": "Different workflow files and different concerns; no shared underlying bug or fix." + "reason": "Different workflow and behavior: self-comment CI input handling vs switching the security gate to a reusable workflow." }, { "left": "pull_request:46138", "right": "pull_request:46504", "accept": false, - "reason": "Different workflow files and unrelated behavior changes in CI image-tag handling vs self-comment CI input handling." + "reason": "Same broad area of CI workflows, but different scripts and fixes; not the same underlying bug or change." }, { "left": "pull_request:46138", "right": "pull_request:46432", "accept": false, - "reason": "Different workflow files and unrelated fixes; not the same code-path problem." + "reason": "Different workflow files and purposes: comment parsing in one PR, SSH runner hardening and env cleanup in the other." }, { "left": "pull_request:46432", "right": "pull_request:46503", "accept": false, - "reason": "Different workflows and unrelated CI logic; ssh-runner hardening is not the same as benchmark triggering." + "reason": "Distinct workflow changes: SSH runner hardening vs allowing benchmark_v2 to run on workflow_dispatch." }, { "left": "pull_request:46432", "right": "pull_request:46501", "accept": false, - "reason": "Different workflow files and unrelated changes; same broad CI area is not enough for duplication." + "reason": "Different workflow files and unrelated updates; no shared concrete code-path problem." }, { "left": "pull_request:46432", "right": "pull_request:46502", "accept": false, - "reason": "Different workflow files and unrelated changes; no single underlying bug." + "reason": "SSH runner security/hardening changes are unrelated to assign-reviewers checkout behavior." }, { "left": "pull_request:46432", "right": "pull_request:46554", "accept": false, - "reason": "Different workflow files and unrelated fixes; ssh-runner env cleanup is not the security-gate change." + "reason": "Different workflows and different fixes; one hardens SSH runner download/env handling, the other changes PR CI security gating logic." }, { "left": "pull_request:46432", "right": "pull_request:46558", "accept": false, - "reason": "Different workflow files and unrelated changes." + "reason": "Not the same change: SSH runner workflow hardening vs PR CI gate permission/approval tweaks." }, { "left": "pull_request:46432", "right": "pull_request:46563", "accept": false, - "reason": "Different workflow files and unrelated concerns." + "reason": "Separate workflow concerns; no common concrete bug to merge as one PR." }, { "left": "pull_request:46501", "right": "pull_request:46503", "accept": false, - "reason": "Both touch CI workflows, but they modify different files for different behaviors; not the same change." + "reason": "Both are workflow chores, but they change different files and different behaviors." }, { "left": "pull_request:46501", "right": "pull_request:46502", "accept": false, - "reason": "Different workflow files and different update intents." + "reason": "Unrelated workflow edits: anti-slop permissions vs reviewer assignment checkout ref." }, { "left": "pull_request:46501", "right": "pull_request:46504", "accept": false, - "reason": "Different workflow files and different CI mechanics; no shared underlying fix." + "reason": "Different workflow files and separate fixes; not duplicate change requests." }, { "left": "pull_request:46501", "right": "pull_request:46554", "accept": false, - "reason": "Unrelated workflows and changes; generic similarity only." + "reason": "No shared underlying bug: anti-slop permission cleanup vs security-gate file scanning and checkout changes." }, { "left": "pull_request:46502", "right": "pull_request:46503", "accept": false, - "reason": "Different workflow files and separate functionality changes." + "reason": "Different workflows and distinct behavior changes." }, { "left": "pull_request:46502", "right": "pull_request:46504", "accept": false, - "reason": "Different workflow files and unrelated edits." + "reason": "Assign-reviewers checkout ref fix is unrelated to build-ci-docker-images shell input handling." }, { "left": "pull_request:46502", "right": "pull_request:46554", "accept": false, - "reason": "Different workflow files and different concerns." + "reason": "Different workflow code paths; not the same underlying issue." }, { "left": "pull_request:46503", "right": "pull_request:46504", "accept": false, - "reason": "Different workflow files and different CI behaviors; not mergeable as one fix." + "reason": "Benchmark dispatch support and build-ci docker trigger handling are separate changes." }, { "left": "pull_request:46503", "right": "pull_request:46554", "accept": false, - "reason": "Different workflow files and unrelated changes." + "reason": "Unrelated workflows and fixes; no plausible single PR merge target." }, { "left": "pull_request:46504", "right": "pull_request:46554", "accept": false, - "reason": "Different workflow files and distinct CI changes; same broad area only." + "reason": "Both touch CI workflow scripts, but they address different jobs and different problems." }, { "left": "pull_request:46504", "right": "pull_request:46558", "accept": false, - "reason": "Different workflow files and unrelated updates." - }, - { - "left": "pull_request:46504", - "right": "pull_request:46563", - "accept": false, - "reason": "Different workflow files and unrelated changes." + "reason": "Different workflow files and unrelated logic changes." }, { "left": "pull_request:46554", "right": "pull_request:46558", "accept": false, - "reason": "Same file, but different parts of the workflow and different intended fixes; not the same concrete bug." + "reason": "Same workflow file, but different behavior changes: 46554 adds file-diff security checks, while 46558 adjusts permissions and approval gating." }, { "left": "pull_request:46554", "right": "pull_request:46563", "accept": false, - "reason": "Same file, but one changes security-gate logic while the other changes the reusable workflow call/comment; different issues." + "reason": "Same workflow file, but one adds local security checks and the other replaces the job with a reusable workflow; different change sets." }, { "left": "pull_request:46558", "right": "pull_request:46563", "accept": false, - "reason": "Same file, but different workflow concerns and no shared underlying bug." + "reason": "Both edit `pr-ci-security-gate.yml`, but they implement different mechanisms and are not the same fix." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46563", + "accept": false, + "reason": "Different workflow files and different maintenance changes." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46563", + "accept": false, + "reason": "SSH runner hardening and PR CI gate refactor are unrelated." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflows and distinct code paths; not duplicate." } ] }, @@ -23031,12 +23101,12 @@ "soft_pairs": [ "pull_request:46152|pull_request:46207" ], - "prepared_review_unit_hash": "5f3c7ab867d8421fef26e5ce64adeed5d5e3fd9a05faeef0914a4f49d0020f4f", + "prepared_review_unit_hash": "32fc0c5e924c7544bf607006f0a20fb0a4c290a5986a7b0998429d2efad86bcf", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4275, + "serialized_chars": 4276, "estimated_input_tokens": 1069, "estimated_eval_tokens": 2394 }, @@ -23044,7 +23114,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4275, + "serialized_chars": 4276, "estimated_input_tokens": 1069, "estimated_eval_tokens": 2394 }, @@ -23053,18 +23123,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs replace file-open based source inspection with safer introspection, but they target different code paths in different modules and are not the same underlying change.", - "confidence": 0.92, + "summary": "Both items are PRs about replacing raw file-open source reads with a safer source-reading mechanism, but they target different code paths and different bugs, so they should not be deduplicated.", + "confidence": 0.94, "canonical_issue_reason": null, - "canonical_pr_reason": "No PR here is a true canonical duplicate of the other; if forced to pick one cluster representative, #46207 is the stronger standalone change because it has a clear, focused behavior update plus tests in a single module.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "#46207 is the best representative of the cluster because it addresses one concrete implementation path with an explicit source-reading fix and accompanying test coverage, while #46152 is a separate fix in a different utility path.", + "best_pr_reason": "PR #46207 is the stronger standalone change: it targets a specific helper in modeling_utils, includes tests, and addresses a concrete source-reading path rather than a broader import-utils refactor.", "soft_edge_verdicts": [ { "left": "pull_request:46152", "right": "pull_request:46207", "accept": false, - "reason": "Both avoid open() for reading source, but #46152 changes import_utils path resolution and #46207 changes _can_set_attn_implementation in modeling_utils; they are different bugs/changes and would not plausibly merge into one PR." + "reason": "They are related only at a high level (avoiding open() for source reading). One changes import_utils module path resolution; the other changes _can_set_attn_implementation in modeling_utils. Different code paths, different failure modes, not one mergeable fix." } ] }, @@ -23094,12 +23164,12 @@ "soft_pairs": [ "pull_request:46154|pull_request:46196" ], - "prepared_review_unit_hash": "39d8ce927117a5b03751919878588607328cb1a0e6873af3fc3a4185f925d51b", + "prepared_review_unit_hash": "29bb69f5e8fda1d7d42d40980e6c8e13e4d5471b6b453c1d5394a3df7dc7a3d5", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5699, + "serialized_chars": 5698, "estimated_input_tokens": 1425, "estimated_eval_tokens": 3106 }, @@ -23107,7 +23177,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5699, + "serialized_chars": 5698, "estimated_input_tokens": 1425, "estimated_eval_tokens": 3106 }, @@ -23116,18 +23186,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests are about SDPA attention dispatch support, but they target different model families and code paths: T5 versus wav2vec2/audio models. They are related in theme, not duplicates.", - "confidence": 0.97, + "summary": "Both PRs add SDPA dispatch support, but they target different model families and separate code paths (T5 vs wav2vec2-family speech models). They are related in theme, not duplicates.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #46196 is the broader and more representative change set, touching multiple audio model implementations and docs for SDPA dispatch support.", + "canonical_pr_reason": "PR #46196 is the better representative because it covers a broader, more central SDPA dispatch update across multiple speech-model implementations and has the larger change set/activity.", "best_issue_reason": null, - "best_pr_reason": "PR #46196 is the strongest representative PR because it covers a larger, more central SDPA dispatch effort across several models.", + "best_pr_reason": "PR #46196 is the strongest standalone PR in this cluster due to wider scope and clearer cross-model impact; PR #46154 is a narrower T5-specific addition.", "soft_edge_verdicts": [ { "left": "pull_request:46154", "right": "pull_request:46196", "accept": false, - "reason": "Same general feature area, but different concrete model implementations and code paths; they are not the same fix and would not plausibly be merged into one PR." + "reason": "Both concern SDPA dispatch, but they fix different model stacks and likely different attention implementations; they do not look like the same concrete bug/change and would not plausibly be merged into one PR." } ] }, @@ -23146,8 +23216,8 @@ "review_unit_count": 1, "status": "reviewed", "reason": null, - "source": "llm", - "cache_hit": false, + "source": "cache", + "cache_hit": true, "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-46173-2", "nodes": [ @@ -23223,7 +23293,7 @@ "soft_pairs": [ "pull_request:46310|pull_request:46546" ], - "prepared_review_unit_hash": "2c79cc59d0a5d30c88c58a5d2eb16a75839277699bb48d80dad3d4831d23f39e", + "prepared_review_unit_hash": "04fad04f796d7c11018c653a5b6eb2fb038bc8882eb45143664ca7a43a528af9", "component_budget": { "node_count": 2, "item_count": 2, @@ -23245,18 +23315,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests fix the same sliding-window boundary bug: the right-side mask used a strict `<` instead of `<=`, excluding the boundary token. They target the same issue and make essentially the same code change in the same code path.", - "confidence": 0.98, - "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46546 is the stronger canonical PR because it includes the Moonshine streaming fix plus a test update and broader coverage, while still addressing the same concrete boundary condition.", + "summary": "Both PRs fix the same Moonshine streaming sliding-window right-boundary off-by-one bug (shared issue target 46305, identical moonshine_streaming mask change). PR 46546 is broader because it also carries the Gemma4 counterpart and tests.", + "confidence": 0.92, + "canonical_issue_reason": "No issue artifact is present in the cluster; the shared target 46305 appears to be the underlying bug report for the right-window boundary off-by-one.", + "canonical_pr_reason": "PR 46546 is the better canonical PR because it includes the Moonshine fix plus the corresponding Gemma4 fix and test coverage, making it the more complete change set.", "best_issue_reason": null, - "best_pr_reason": "PR 46546 is the best representative of the underlying change because it is the more complete patch for the same bug and is more likely to be the final intended fix.", + "best_pr_reason": "PR 46546 is the strongest representative of the fix: it contains the same Moonshine boundary correction as 46310 and adds the related Gemma4 patch and tests.", "soft_edge_verdicts": [ { "left": "pull_request:46310", "right": "pull_request:46546", "accept": true, - "reason": "Same bug, same issue target, and the diff changes the same right-window boundary condition from `<` to `<=` in the same mask function. The second PR is a near-duplicate expansion of the first rather than a different fix." + "reason": "Same underlying Moonshine right-window boundary bug; identical fix in the shared Moonshine files and same explicit issue target 46305, with 46546 adding a related companion fix/test coverage." } ] }, @@ -23286,7 +23356,7 @@ "soft_pairs": [ "pull_request:46376|pull_request:46417" ], - "prepared_review_unit_hash": "edee4cdb537049586b09151e4a3409465339d0ffb1608df2cae46f25530c5b0d", + "prepared_review_unit_hash": "6a21810767d7019b34ed13cfd87d3e564896e7f1e692760de4a5ff3c4c12d3f5", "component_budget": { "node_count": 2, "item_count": 2, @@ -23308,18 +23378,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two pull requests are unrelated: one adds multimodal support to continuous batching, while the other adds the HunYuan VL model and its docs/tests. They should not be clustered as duplicates.", + "summary": "The cluster contains two unrelated pull requests: one adds multimodal support to continuous batching, and the other adds the HunYuan VL model and its docs/tests. They do not appear to fix the same code-path problem.", "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46417 is the better cluster representative because it is a self-contained, non-draft model addition with extensive docs and tests; PR 46376 is a separate draft infrastructure change.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "PR 46417 is the stronger standalone PR overall: it has a clear feature scope, broad implementation coverage, and supporting documentation/tests. PR 46376 is still a draft and covers a different subsystem.", + "best_pr_reason": "pull_request:46417 is the stronger standalone PR: it is a self-contained model addition with docs, code, and tests, whereas pull_request:46376 is a draft infrastructure change and not a duplicate of the other PR.", "soft_edge_verdicts": [ { "left": "pull_request:46376", "right": "pull_request:46417", "accept": false, - "reason": "Different underlying changes and code paths: continuous batching multimodal support vs. adding a HunYuan VL model. No shared files and no evidence they could be merged into one PR." + "reason": "Different goals and code paths: continuous batching/multimodal cache work versus adding a new HunYuan VL model stack. They are not mergeable as one PR and do not represent the same bug or feature." } ] }, @@ -23349,7 +23419,7 @@ "soft_pairs": [ "pull_request:46574|pull_request:46580" ], - "prepared_review_unit_hash": "1e3d7d7cb0fdc827bad75e326faabad2150922bb2cfe74b7bb11f21600bad108", + "prepared_review_unit_hash": "573e85a73aa5aea08f6fd6e66d29cd314131cb266ac77a55dca53423f11c33a8", "component_budget": { "node_count": 2, "item_count": 2, @@ -23371,18 +23441,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are both tokenizer-related PRs, but they fix different model-specific bugs: one changes CodeLlama decode stripping behavior, the other fixes DeepSeek checkpoints loading the wrong tokenizer class. They should not be clustered as duplicates.", - "confidence": 0.97, + "summary": "These two PRs are unrelated tokenizer fixes: one changes CodeLlama decode behavior to preserve leading spaces, while the other adjusts AutoTokenizer routing for specific DeepSeek checkpoints. They should not be clustered as duplicates.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: the PRs address different underlying bugs in different code paths, so neither is a duplicate representative of the other.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "No best PR for deduping this cluster; each PR targets a distinct model-specific tokenizer problem.", + "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "pull_request:46574", "right": "pull_request:46580", "accept": false, - "reason": "Different bugs and different fixes: CodeLlama decode-space stripping vs DeepSeek checkpoint tokenizer-class selection. They do not look mergeable into one PR." + "reason": "Different models and different code paths: CodeLlama decoder stripping vs DeepSeek checkpoint-to-tokenizer-class selection. They do not appear mergeable into one PR." } ] }, @@ -23412,7 +23482,7 @@ "soft_pairs": [ "pull_request:46603|pull_request:46604" ], - "prepared_review_unit_hash": "751d667c746306541951f0c4906196a471e91f27e27afe711427612b36a7b340", + "prepared_review_unit_hash": "1b822f7769f8a5da1dec56c6db8cba99639bf0091398d137539f617a6dfffb35", "component_budget": { "node_count": 2, "item_count": 2, @@ -23434,18 +23504,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two related Mistral tokenizer PRs, but they address different changes: one refactors/moves MistralConverter into the integrations package, the other adds tekken tokenizer detection, conversion, and save utilities. They share files but are not the same underlying change.", - "confidence": 0.95, + "summary": "Both PRs touch the Mistral tokenizer area and even share several files, but they are different changes: one moves `MistralConverter` into a dedicated `integrations/mistral` package, while the other adds tekken tokenizer detection, conversion, and save utilities. They are related follow-up/refactor work, not the same underlying bug or feature change.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46604 is the more substantive and feature-complete Mistral tokenizer change; PR 46603 is a narrower refactor/move of MistralConverter.", + "canonical_pr_reason": "PR #46604 is the more substantive feature change, adding tekken tokenizer detection/conversion/save support; #46603 is a narrower packaging refactor that extracts `MistralConverter`.", "best_issue_reason": null, - "best_pr_reason": "PR 46604 is the best representative because it contains the main functional addition (tekken tokenizer detection/conversion/save support) rather than just relocating existing code.", + "best_pr_reason": "PR #46604 best represents the cluster because it contains the concrete new tokenizer functionality, whereas #46603 is primarily an internal move of existing Mistral conversion code.", "soft_edge_verdicts": [ { "left": "pull_request:46603", "right": "pull_request:46604", "accept": false, - "reason": "Reject: the PRs have different goals and code paths. 46603 moves MistralConverter into a new package; 46604 adds tekken tokenizer detection, conversion, and save utilities. Shared Mistral files alone do not make them the same change." + "reason": "Shared Mistral tokenizer files do not make these the same change: #46603 relocates `MistralConverter`, while #46604 implements tekken tokenizer support and utilities. They would not plausibly merge into one single PR as duplicate code-path fixes." } ] }, diff --git a/analysis/current/manifest.json b/analysis/current/manifest.json index 522a2dd142f8373e6238f23f16f588808ba02120..57a15ab436bd8ad241bca4e2a0c123ffca1c20f8 100644 --- a/analysis/current/manifest.json +++ b/analysis/current/manifest.json @@ -1,8 +1,8 @@ { - "analysis_id": "hybrid-model-20260617t000107z", + "analysis_id": "hybrid-model-20260617t060052z", "archived_artifacts": { - "hybrid": "snapshots/20260617T000107Z/analysis-runs/hybrid-model-20260617t000107z/analysis-report-hybrid.json", - "hybrid_reviews": "snapshots/20260617T000107Z/analysis-runs/hybrid-model-20260617t000107z/analysis-report-hybrid.llm-reviews.json" + "hybrid": "snapshots/20260617T060052Z/analysis-runs/hybrid-model-20260617t060052z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260617T060052Z/analysis-runs/hybrid-model-20260617t060052z/analysis-report-hybrid.llm-reviews.json" }, "artifacts": { "hybrid": "analysis/current/analysis-report-hybrid.json", @@ -10,9 +10,9 @@ }, "channel": "canonical", "model": null, - "published_at": "2026-06-17T00:10:37Z", + "published_at": "2026-06-17T06:10:53Z", "repo": "huggingface/transformers", "schema_version": 1, - "snapshot_id": "20260617T000107Z", + "snapshot_id": "20260617T060052Z", "variant": "hybrid" } diff --git a/snapshots/20260617T060052Z/analysis-runs/hybrid-model-20260617t060052z/analysis-report-hybrid.json b/snapshots/20260617T060052Z/analysis-runs/hybrid-model-20260617t060052z/analysis-report-hybrid.json new file mode 100644 index 0000000000000000000000000000000000000000..e24bf39e424f0edb3003d58de7491d750f6adfc8 --- /dev/null +++ b/snapshots/20260617T060052Z/analysis-runs/hybrid-model-20260617t060052z/analysis-report-hybrid.json @@ -0,0 +1,2220 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260617T060052Z", + "generated_at": "2026-06-17T06:10:51Z", + "evidence_quality": "full", + "llm_enrichment": true, + "meta_bugs": [ + { + "cluster_id": "cluster-43979-11", + "summary": "Cluster of 1 issues and 10 PRs centered on issue #43979.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 43979, + "canonical_pr_number": 44007, + "issue_numbers": [ + 43979 + ], + "pr_numbers": [ + 43996, + 44007, + 44013, + 44044, + 44066, + 44072, + 44085, + 44129, + 44154, + 44722 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 43996, + "right_pr_number": 44007, + "code_similarity": 0.179, + "size_similarity": 0.576, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.429, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44013, + "code_similarity": 0.122, + "size_similarity": 0.318, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.392, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44044, + "code_similarity": 0.245, + "size_similarity": 0.864, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.479, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44066, + "code_similarity": 0.225, + "size_similarity": 0.818, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.408, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44072, + "code_similarity": 0.14, + "size_similarity": 0.303, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.528, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44085, + "code_similarity": 0.216, + "size_similarity": 0.783, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.398, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44129, + "code_similarity": 0.163, + "size_similarity": 0.643, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.229, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44154, + "code_similarity": 0.153, + "size_similarity": 0.535, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.31, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44722, + "code_similarity": 0.225, + "size_similarity": 0.848, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.368, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44013, + "code_similarity": 0.19, + "size_similarity": 0.553, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.531, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44044, + "code_similarity": 0.186, + "size_similarity": 0.667, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.354, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44066, + "code_similarity": 0.188, + "size_similarity": 0.704, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.315, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44072, + "code_similarity": 0.212, + "size_similarity": 0.526, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.708, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44085, + "code_similarity": 0.195, + "size_similarity": 0.735, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.318, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44129, + "code_similarity": 0.103, + "size_similarity": 0.37, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.191, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44154, + "code_similarity": 0.238, + "size_similarity": 0.93, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.344, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44722, + "code_similarity": 0.178, + "size_similarity": 0.679, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.28, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44044, + "code_similarity": 0.126, + "size_similarity": 0.368, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.351, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44066, + "code_similarity": 0.127, + "size_similarity": 0.389, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.325, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44072, + "code_similarity": 0.29, + "size_similarity": 0.952, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.667, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44085, + "code_similarity": 0.131, + "size_similarity": 0.406, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.329, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44129, + "code_similarity": 0.07, + "size_similarity": 0.205, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.192, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44154, + "code_similarity": 0.177, + "size_similarity": 0.594, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.389, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44722, + "code_similarity": 0.118, + "size_similarity": 0.375, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.287, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44066, + "code_similarity": 0.25, + "size_similarity": 0.947, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.404, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44072, + "code_similarity": 0.136, + "size_similarity": 0.351, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.442, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44085, + "code_similarity": 0.24, + "size_similarity": 0.906, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.394, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44129, + "code_similarity": 0.147, + "size_similarity": 0.555, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.243, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44154, + "code_similarity": 0.17, + "size_similarity": 0.62, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.306, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44722, + "code_similarity": 0.257, + "size_similarity": 0.982, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.402, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44066, + "right_pr_number": 44072, + "code_similarity": 0.133, + "size_similarity": 0.37, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.393, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44066, + "right_pr_number": 44085, + "code_similarity": 0.763, + "size_similarity": 0.957, + "file_overlap": 0.5, + "area_overlap": 0.825, + "patch_similarity": 0.887, + "shared_filenames": [ + "src/transformers/models/gptj/modeling_gptj.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/gptj/modeling_gptj.py", + "left_ranges": [ + [ + 33, + 41 + ], + [ + 174, + 181 + ], + [ + 250, + 257 + ], + [ + 398, + 419 + ], + [ + 425, + 434 + ], + [ + 459, + 466 + ], + [ + 471, + 488 + ], + [ + 518, + 540 + ], + [ + 554, + 560 + ], + [ + 566, + 575 + ], + [ + 580, + 598 + ], + [ + 601, + 606 + ], + [ + 634, + 640 + ], + [ + 646, + 653 + ], + [ + 658, + 674 + ], + [ + 717, + 722 + ], + [ + 738, + 744 + ], + [ + 749, + 772 + ], + [ + 790, + 795 + ] + ], + "right_ranges": [ + [ + 33, + 40 + ], + [ + 173, + 180 + ], + [ + 249, + 256 + ], + [ + 397, + 405 + ], + [ + 408, + 419 + ], + [ + 425, + 434 + ], + [ + 459, + 465 + ], + [ + 470, + 489 + ], + [ + 517, + 539 + ], + [ + 553, + 559 + ], + [ + 565, + 574 + ], + [ + 579, + 597 + ], + [ + 600, + 611 + ], + [ + 633, + 639 + ], + [ + 645, + 652 + ], + [ + 657, + 673 + ], + [ + 716, + 728 + ], + [ + 737, + 743 + ], + [ + 748, + 771 + ], + [ + 789, + 794 + ] + ] + } + ] + }, + { + "left_pr_number": 44066, + "right_pr_number": 44129, + "code_similarity": 0.145, + "size_similarity": 0.526, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.263, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44066, + "right_pr_number": 44154, + "code_similarity": 0.174, + "size_similarity": 0.654, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.286, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44066, + "right_pr_number": 44722, + "code_similarity": 0.907, + "size_similarity": 0.964, + "file_overlap": 1.0, + "area_overlap": 0.808, + "patch_similarity": 0.874, + "shared_filenames": [ + "src/transformers/models/codegen/modeling_codegen.py", + "src/transformers/models/gptj/modeling_gptj.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/codegen/modeling_codegen.py", + "left_ranges": [ + [ + 245, + 266 + ] + ], + "right_ranges": [ + [ + 228, + 234 + ] + ] + }, + { + "filename": "src/transformers/models/gptj/modeling_gptj.py", + "left_ranges": [ + [ + 33, + 41 + ], + [ + 174, + 181 + ], + [ + 250, + 257 + ], + [ + 398, + 419 + ], + [ + 425, + 434 + ], + [ + 459, + 466 + ], + [ + 471, + 488 + ], + [ + 518, + 540 + ], + [ + 554, + 560 + ], + [ + 566, + 575 + ], + [ + 580, + 598 + ], + [ + 601, + 606 + ], + [ + 634, + 640 + ], + [ + 646, + 653 + ], + [ + 658, + 674 + ], + [ + 717, + 722 + ], + [ + 738, + 744 + ], + [ + 749, + 772 + ], + [ + 790, + 795 + ] + ], + "right_ranges": [ + [ + 33, + 42 + ], + [ + 175, + 182 + ], + [ + 251, + 258 + ], + [ + 399, + 420 + ], + [ + 426, + 435 + ], + [ + 460, + 467 + ], + [ + 472, + 489 + ], + [ + 519, + 540 + ], + [ + 554, + 560 + ], + [ + 566, + 575 + ], + [ + 580, + 598 + ], + [ + 601, + 612 + ], + [ + 634, + 640 + ], + [ + 646, + 653 + ], + [ + 658, + 663 + ], + [ + 666, + 674 + ], + [ + 717, + 722 + ], + [ + 738, + 744 + ], + [ + 749, + 772 + ], + [ + 790, + 795 + ] + ] + } + ] + }, + { + "left_pr_number": 44072, + "right_pr_number": 44085, + "code_similarity": 0.137, + "size_similarity": 0.387, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.398, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44072, + "right_pr_number": 44129, + "code_similarity": 0.074, + "size_similarity": 0.195, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.231, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44072, + "right_pr_number": 44154, + "code_similarity": 0.175, + "size_similarity": 0.566, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.414, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44072, + "right_pr_number": 44722, + "code_similarity": 0.124, + "size_similarity": 0.357, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.347, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44085, + "right_pr_number": 44129, + "code_similarity": 0.141, + "size_similarity": 0.503, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.272, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44085, + "right_pr_number": 44154, + "code_similarity": 0.18, + "size_similarity": 0.684, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.289, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44085, + "right_pr_number": 44722, + "code_similarity": 0.728, + "size_similarity": 0.923, + "file_overlap": 0.5, + "area_overlap": 0.791, + "patch_similarity": 0.78, + "shared_filenames": [ + "src/transformers/models/gptj/modeling_gptj.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/gptj/modeling_gptj.py", + "left_ranges": [ + [ + 33, + 40 + ], + [ + 173, + 180 + ], + [ + 249, + 256 + ], + [ + 397, + 405 + ], + [ + 408, + 419 + ], + [ + 425, + 434 + ], + [ + 459, + 465 + ], + [ + 470, + 489 + ], + [ + 517, + 539 + ], + [ + 553, + 559 + ], + [ + 565, + 574 + ], + [ + 579, + 597 + ], + [ + 600, + 611 + ], + [ + 633, + 639 + ], + [ + 645, + 652 + ], + [ + 657, + 673 + ], + [ + 716, + 728 + ], + [ + 737, + 743 + ], + [ + 748, + 771 + ], + [ + 789, + 794 + ] + ], + "right_ranges": [ + [ + 33, + 42 + ], + [ + 175, + 182 + ], + [ + 251, + 258 + ], + [ + 399, + 420 + ], + [ + 426, + 435 + ], + [ + 460, + 467 + ], + [ + 472, + 489 + ], + [ + 519, + 540 + ], + [ + 554, + 560 + ], + [ + 566, + 575 + ], + [ + 580, + 598 + ], + [ + 601, + 612 + ], + [ + 634, + 640 + ], + [ + 646, + 653 + ], + [ + 658, + 663 + ], + [ + 666, + 674 + ], + [ + 717, + 722 + ], + [ + 738, + 744 + ], + [ + 749, + 772 + ], + [ + 790, + 795 + ] + ] + } + ] + }, + { + "left_pr_number": 44129, + "right_pr_number": 44154, + "code_similarity": 0.099, + "size_similarity": 0.344, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.199, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44129, + "right_pr_number": 44722, + "code_similarity": 0.146, + "size_similarity": 0.545, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.247, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44154, + "right_pr_number": 44722, + "code_similarity": 0.164, + "size_similarity": 0.631, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.255, + "shared_filenames": [], + "shared_file_areas": [] + } + ] + }, + { + "cluster_id": "cluster-41211-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #41211.", + "status": "open", + "confidence": 0.55, + "canonical_issue_number": 41211, + "canonical_pr_number": 44339, + "issue_numbers": [ + 41211 + ], + "pr_numbers": [ + 41356, + 44339 + ], + "evidence_types": [ + "closing_reference" + ], + "pr_comparisons": [ + { + "left_pr_number": 41356, + "right_pr_number": 44339, + "code_similarity": 0.155, + "size_similarity": 0.094, + "file_overlap": 0.273, + "area_overlap": 0.078, + "patch_similarity": 0.183, + "shared_filenames": [ + "docs/source/en/model_doc/deimv2.md", + "src/transformers/models/deimv2/__init__.py", + "src/transformers/models/deimv2/configuration_deimv2.py", + "src/transformers/models/deimv2/modeling_deimv2.py", + "tests/models/deimv2/__init__.py", + "tests/models/deimv2/test_modeling_deimv2.py" + ], + "shared_file_areas": [ + { + "filename": "docs/source/en/model_doc/deimv2.md", + "left_ranges": [ + [ + 1, + 132 + ] + ], + "right_ranges": [ + [ + 1, + 65 + ] + ] + }, + { + "filename": "src/transformers/models/deimv2/__init__.py", + "left_ranges": [ + [ + 1, + 15 + ] + ], + "right_ranges": [ + [ + 1, + 29 + ] + ] + }, + { + "filename": "src/transformers/models/deimv2/configuration_deimv2.py", + "left_ranges": [ + [ + 1, + 74 + ] + ], + "right_ranges": [ + [ + 1, + 266 + ] + ] + }, + { + "filename": "tests/models/deimv2/test_modeling_deimv2.py", + "left_ranges": [ + [ + 1, + 15 + ] + ], + "right_ranges": [ + [ + 1, + 1734 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-41115-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #41115.", + "status": "open", + "confidence": 0.55, + "canonical_issue_number": 41115, + "canonical_pr_number": 41116, + "issue_numbers": [ + 41115 + ], + "pr_numbers": [ + 41116, + 45613 + ], + "evidence_types": [ + "closing_reference" + ], + "pr_comparisons": [ + { + "left_pr_number": 41116, + "right_pr_number": 45613, + "code_similarity": 0.808, + "size_similarity": 0.893, + "file_overlap": 0.75, + "area_overlap": 0.869, + "patch_similarity": 0.667, + "shared_filenames": [ + "docs/source/en/_toctree.yml", + "docs/source/en/model_doc/minicpm3.md", + "src/transformers/models/auto/modeling_auto.py", + "src/transformers/models/minicpm3/__init__.py", + "src/transformers/models/minicpm3/configuration_minicpm3.py", + "src/transformers/models/minicpm3/modeling_minicpm3.py", + "src/transformers/models/minicpm3/modular_minicpm3.py", + "tests/models/minicpm3/__init__.py", + "tests/models/minicpm3/test_modeling_minicpm3.py" + ], + "shared_file_areas": [ + { + "filename": "docs/source/en/_toctree.yml", + "left_ranges": [ + [ + 713, + 720 + ] + ], + "right_ranges": [ + [ + 730, + 737 + ] + ] + }, + { + "filename": "docs/source/en/model_doc/minicpm3.md", + "left_ranges": [ + [ + 1, + 67 + ] + ], + "right_ranges": [ + [ + 1, + 45 + ] + ] + }, + { + "filename": "src/transformers/models/auto/modeling_auto.py", + "left_ranges": [ + [ + 285, + 291 + ], + [ + 700, + 706 + ], + [ + 1298, + 1304 + ] + ], + "right_ranges": [ + [ + 720, + 726 + ], + [ + 1331, + 1337 + ] + ] + }, + { + "filename": "src/transformers/models/minicpm3/__init__.py", + "left_ranges": [ + [ + 1, + 29 + ] + ], + "right_ranges": [ + [ + 1, + 29 + ] + ] + }, + { + "filename": "src/transformers/models/minicpm3/configuration_minicpm3.py", + "left_ranges": [ + [ + 1, + 149 + ] + ], + "right_ranges": [ + [ + 1, + 126 + ] + ] + }, + { + "filename": "src/transformers/models/minicpm3/modeling_minicpm3.py", + "left_ranges": [ + [ + 1, + 550 + ] + ], + "right_ranges": [ + [ + 1, + 522 + ] + ] + }, + { + "filename": "src/transformers/models/minicpm3/modular_minicpm3.py", + "left_ranges": [ + [ + 1, + 394 + ] + ], + "right_ranges": [ + [ + 1, + 342 + ] + ] + }, + { + "filename": "tests/models/minicpm3/test_modeling_minicpm3.py", + "left_ranges": [ + [ + 1, + 132 + ] + ], + "right_ranges": [ + [ + 1, + 136 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-43240-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #43240.", + "status": "open", + "confidence": 0.75, + "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-45987-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45987.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 45987, + "canonical_pr_number": 46173, + "issue_numbers": [ + 45987 + ], + "pr_numbers": [ + 46173, + 46474 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 46173, + "right_pr_number": 46474, + "code_similarity": 0.65, + "size_similarity": 1.0, + "file_overlap": 1.0, + "area_overlap": 0.0, + "patch_similarity": 1.0, + "shared_filenames": [ + "src/transformers/cache_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/cache_utils.py", + "left_ranges": [ + [ + 69, + 75 + ], + [ + 375, + 381 + ] + ], + "right_ranges": [ + [ + 79, + 85 + ], + [ + 461, + 467 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-45561-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45561.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 45561, + "canonical_pr_number": 45645, + "issue_numbers": [ + 45561 + ], + "pr_numbers": [ + 45639, + 45645 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 45639, + "right_pr_number": 45645, + "code_similarity": 0.494, + "size_similarity": 0.619, + "file_overlap": 0.5, + "area_overlap": 0.434, + "patch_similarity": 0.451, + "shared_filenames": [ + "src/transformers/testing_utils.py", + "tests/utils/test_testing_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/testing_utils.py", + "left_ranges": [ + [ + 3595, + 3602 + ], + [ + 3819, + 3853 + ] + ], + "right_ranges": [ + [ + 3525, + 3558 + ], + [ + 3782, + 3788 + ] + ] + }, + { + "filename": "tests/utils/test_testing_utils.py", + "left_ranges": [ + [ + 1, + 86 + ] + ], + "right_ranges": [ + [ + 1, + 114 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-46305-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #46305.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 46305, + "canonical_pr_number": 46546, + "issue_numbers": [ + 46305 + ], + "pr_numbers": [ + 46310, + 46546 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 46310, + "right_pr_number": 46546, + "code_similarity": 0.566, + "size_similarity": 0.444, + "file_overlap": 0.5, + "area_overlap": 0.867, + "patch_similarity": 0.16, + "shared_filenames": [ + "src/transformers/models/moonshine_streaming/modeling_moonshine_streaming.py", + "src/transformers/models/moonshine_streaming/modular_moonshine_streaming.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/moonshine_streaming/modeling_moonshine_streaming.py", + "left_ranges": [ + [ + 356, + 362 + ] + ], + "right_ranges": [ + [ + 355, + 361 + ] + ] + }, + { + "filename": "src/transformers/models/moonshine_streaming/modular_moonshine_streaming.py", + "left_ranges": [ + [ + 269, + 275 + ] + ], + "right_ranges": [ + [ + 269, + 275 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-45081-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45081.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 45081, + "canonical_pr_number": 45317, + "issue_numbers": [ + 45081 + ], + "pr_numbers": [ + 45086, + 45317 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 45086, + "right_pr_number": 45317, + "code_similarity": 0.589, + "size_similarity": 0.34, + "file_overlap": 0.5, + "area_overlap": 1.0, + "patch_similarity": 0.136, + "shared_filenames": [ + "src/transformers/tokenization_utils_tokenizers.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/tokenization_utils_tokenizers.py", + "left_ranges": [ + [ + 1360, + 1370 + ], + [ + 1374, + 1380 + ] + ], + "right_ranges": [ + [ + 1360, + 1370 + ], + [ + 1374, + 1380 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-43698-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #43698.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 43698, + "canonical_pr_number": 43779, + "issue_numbers": [ + 43698 + ], + "pr_numbers": [ + 43779, + 43816 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 43779, + "right_pr_number": 43816, + "code_similarity": 0.418, + "size_similarity": 0.538, + "file_overlap": 1.0, + "area_overlap": 0.02, + "patch_similarity": 0.02, + "shared_filenames": [ + "src/transformers/integrations/integration_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/integrations/integration_utils.py", + "left_ranges": [ + [ + 2227, + 2241 + ], + [ + 2303, + 2309 + ] + ], + "right_ranges": [ + [ + 2278, + 2291 + ], + [ + 2309, + 2322 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-46562-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 46562, + "issue_numbers": [], + "pr_numbers": [ + 46562, + 46658 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 46562, + "right_pr_number": 46658, + "code_similarity": 0.17, + "size_similarity": 0.397, + "file_overlap": 0.032, + "area_overlap": 0.0, + "patch_similarity": 0.537, + "shared_filenames": [ + "src/transformers/models/auto/auto_mappings.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/auto/auto_mappings.py", + "left_ranges": [ + [ + 441, + 447 + ], + [ + 591, + 598 + ], + [ + 823, + 829 + ], + [ + 883, + 889 + ], + [ + 1082, + 1088 + ], + [ + 1205, + 1211 + ] + ], + "right_ranges": [ + [ + 362, + 368 + ], + [ + 795, + 801 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-44018-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 44068, + "issue_numbers": [], + "pr_numbers": [ + 44018, + 44068 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 44018, + "right_pr_number": 44068, + "code_similarity": 0.766, + "size_similarity": 0.939, + "file_overlap": 1.0, + "area_overlap": 0.425, + "patch_similarity": 0.866, + "shared_filenames": [ + "src/transformers/models/gpt_neo/modeling_gpt_neo.py", + "tests/models/gpt_neo/test_modeling_gpt_neo.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/gpt_neo/modeling_gpt_neo.py", + "left_ranges": [ + [ + 26, + 46 + ], + [ + 138, + 143 + ], + [ + 182, + 187 + ], + [ + 283, + 295 + ], + [ + 327, + 341 + ], + [ + 348, + 354 + ], + [ + 360, + 369 + ], + [ + 399, + 406 + ], + [ + 411, + 419 + ], + [ + 428, + 433 + ], + [ + 472, + 492 + ], + [ + 507, + 513 + ], + [ + 519, + 528 + ], + [ + 541, + 559 + ], + [ + 562, + 567 + ], + [ + 595, + 601 + ], + [ + 607, + 614 + ], + [ + 627, + 643 + ], + [ + 685, + 690 + ], + [ + 708, + 714 + ], + [ + 720, + 727 + ], + [ + 740, + 757 + ], + [ + 761, + 766 + ], + [ + 780, + 786 + ], + [ + 791, + 798 + ], + [ + 807, + 822 + ], + [ + 840, + 845 + ] + ], + "right_ranges": [ + [ + 26, + 31 + ], + [ + 34, + 44 + ], + [ + 136, + 141 + ], + [ + 180, + 185 + ], + [ + 281, + 293 + ], + [ + 325, + 339 + ], + [ + 346, + 352 + ], + [ + 358, + 367 + ], + [ + 397, + 404 + ], + [ + 409, + 417 + ], + [ + 426, + 434 + ], + [ + 463, + 483 + ], + [ + 498, + 504 + ], + [ + 510, + 519 + ], + [ + 532, + 550 + ], + [ + 553, + 564 + ], + [ + 586, + 592 + ], + [ + 598, + 605 + ], + [ + 618, + 634 + ], + [ + 676, + 688 + ], + [ + 699, + 705 + ], + [ + 711, + 718 + ], + [ + 731, + 748 + ], + [ + 752, + 762 + ], + [ + 771, + 777 + ], + [ + 782, + 789 + ], + [ + 798, + 813 + ], + [ + 831, + 836 + ] + ] + }, + { + "filename": "tests/models/gpt_neo/test_modeling_gpt_neo.py", + "left_ranges": [ + [ + 458, + 464 + ] + ], + "right_ranges": [ + [ + 458, + 464 + ] + ] + } + ] + } + ] + } + ], + "duplicate_issues": [], + "duplicate_prs": [ + { + "cluster_id": "cluster-41115-3", + "canonical_pr_number": 41116, + "duplicate_pr_numbers": [ + 45613 + ], + "target_issue_number": 41115, + "reason": "PRs in cluster-41115-3 are treated as duplicates because they converge on issue #41115 with closing_reference evidence." + }, + { + "cluster_id": "cluster-41211-3", + "canonical_pr_number": 44339, + "duplicate_pr_numbers": [ + 41356 + ], + "target_issue_number": 41211, + "reason": "PRs in cluster-41211-3 are treated as duplicates because they converge on issue #41211 with closing_reference evidence." + }, + { + "cluster_id": "cluster-43240-3", + "canonical_pr_number": 43251, + "duplicate_pr_numbers": [ + 43254 + ], + "target_issue_number": 43240, + "reason": "PRs in cluster-43240-3 are treated as duplicates because they converge on issue #43240 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-43698-3", + "canonical_pr_number": 43779, + "duplicate_pr_numbers": [ + 43816 + ], + "target_issue_number": 43698, + "reason": "PRs in cluster-43698-3 are treated as duplicates because they converge on issue #43698 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-43979-11", + "canonical_pr_number": 44007, + "duplicate_pr_numbers": [ + 43996, + 44013, + 44044, + 44066, + 44072, + 44085, + 44129, + 44154, + 44722 + ], + "target_issue_number": 43979, + "reason": "PRs in cluster-43979-11 are treated as duplicates because they converge on issue #43979 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-44018-2", + "canonical_pr_number": 44068, + "duplicate_pr_numbers": [ + 44018 + ], + "target_issue_number": null, + "reason": "PRs in cluster-44018-2 are treated as duplicates because they share soft_similarity evidence." + }, + { + "cluster_id": "cluster-45081-3", + "canonical_pr_number": 45317, + "duplicate_pr_numbers": [ + 45086 + ], + "target_issue_number": 45081, + "reason": "PRs in cluster-45081-3 are treated as duplicates because they converge on issue #45081 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-45561-3", + "canonical_pr_number": 45645, + "duplicate_pr_numbers": [ + 45639 + ], + "target_issue_number": 45561, + "reason": "PRs in cluster-45561-3 are treated as duplicates because they converge on issue #45561 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-45647-3", + "canonical_pr_number": 45738, + "duplicate_pr_numbers": [ + 45737 + ], + "target_issue_number": 45647, + "reason": "PRs in cluster-45647-3 are treated as duplicates because they converge on issue #45647 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-45987-3", + "canonical_pr_number": 46173, + "duplicate_pr_numbers": [ + 46474 + ], + "target_issue_number": 45987, + "reason": "PRs in cluster-45987-3 are treated as duplicates because they converge on issue #45987 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-46305-3", + "canonical_pr_number": 46546, + "duplicate_pr_numbers": [ + 46310 + ], + "target_issue_number": 46305, + "reason": "PRs in cluster-46305-3 are treated as duplicates because they converge on issue #46305 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-46562-2", + "canonical_pr_number": 46562, + "duplicate_pr_numbers": [ + 46658 + ], + "target_issue_number": null, + "reason": "PRs in cluster-46562-2 are treated as duplicates because they share soft_similarity evidence." + } + ], + "best_issue": { + "cluster_id": "cluster-41211-3", + "issue_number": 41211, + "reason": "Issue #41211 is the strongest global issue candidate because it is open, belongs to a cluster with 3 artifacts, and carries 6 discussion comments plus 6 inbound references.", + "score": 148.0 + }, + "best_pr": { + "cluster_id": "cluster-43979-11", + "pr_number": 44007, + "reason": "PR #44007 is the strongest global PR candidate because it is open, belongs to a cluster with 11 artifacts, links to 1 issues, and carries 2 review/discussion events.", + "score": 238.886 + } +} diff --git a/snapshots/20260617T060052Z/analysis-runs/hybrid-model-20260617t060052z/analysis-report-hybrid.llm-reviews.json b/snapshots/20260617T060052Z/analysis-runs/hybrid-model-20260617t060052z/analysis-report-hybrid.llm-reviews.json new file mode 100644 index 0000000000000000000000000000000000000000..0521a665c9b0c54de0a624b95386e59db230ef94 --- /dev/null +++ b/snapshots/20260617T060052Z/analysis-runs/hybrid-model-20260617t060052z/analysis-report-hybrid.llm-reviews.json @@ -0,0 +1,23530 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260617T060052Z", + "generated_at": "2026-06-17T06:10:51Z", + "model": "gpt-5.4-mini?service_tier=flex", + "reviews": [ + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 1, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:24643", + "issue:30277", + "issue:34634", + "issue:34928", + "issue:35545", + "issue:36010", + "issue:36331", + "issue:39290", + "issue:41720", + "issue:41762", + "issue:42915", + "issue:43039", + "issue:43425", + "issue:43541", + "issue:43596", + "issue:43638", + "issue:43716", + "issue:43827", + "issue:43828", + "issue:43975", + "issue:44292", + "issue:44322", + "issue:44457", + "issue:44509", + "issue:44512", + "issue:44560", + "issue:44805", + "issue:44841", + "issue:44918", + "issue:45081", + "issue:45161", + "issue:45237", + "issue:45464", + "issue:45507", + "issue:45685", + "issue:45800", + "issue:45823", + "issue:45859", + "issue:45874", + "issue:45901", + "issue:45902", + "issue:46015", + "issue:46165", + "issue:46294", + "issue:46468", + "issue:46515", + "issue:46661" + ], + "soft_pairs": [ + "issue:44322|issue:45464", + "issue:45237|issue:45685", + "issue:43716|issue:45237", + "issue:43828|issue:45237", + "issue:45237|issue:46468", + "issue:45237|issue:45507", + "issue:30277|issue:35545", + "issue:35545|issue:42915", + "issue:35545|issue:41720", + "issue:45901|issue:46015", + "issue:44509|issue:44512", + "issue:35545|issue:36010", + "issue:44918|issue:46165", + "issue:44805|issue:45823", + "issue:43638|issue:44805", + "issue:43638|issue:46165", + "issue:43596|issue:44805", + "issue:44805|issue:46294", + "issue:44805|issue:45161", + "issue:46165|issue:46468", + "issue:44560|issue:46165", + "issue:43596|issue:46165", + "issue:34634|issue:35545", + "issue:30277|issue:43638", + "issue:43596|issue:43638", + "issue:43039|issue:45161", + "issue:43975|issue:44457", + "issue:44292|issue:44457", + "issue:44841|issue:46661", + "issue:34928|issue:44805", + "issue:24643|issue:30277", + "issue:45859|issue:45902", + "issue:43638|issue:45161", + "issue:34928|issue:36331", + "issue:43596|issue:45161", + "issue:43716|issue:45507", + "issue:45874|issue:46294", + "issue:44292|issue:45237", + "issue:39290|issue:45081", + "issue:45823|issue:45874", + "issue:45800|issue:46515", + "issue:43425|issue:44292", + "issue:34928|issue:41720", + "issue:43828|issue:45161", + "issue:43828|issue:44292", + "issue:43541|issue:43716", + "issue:43716|issue:46468", + "issue:43716|issue:45685", + "issue:44292|issue:44918", + "issue:45507|issue:45685", + "issue:43716|issue:44292", + "issue:44918|issue:45161", + "issue:43039|issue:44918", + "issue:41762|issue:43638", + "issue:43827|issue:44512", + "issue:43828|issue:45685", + "issue:44918|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": "fc7b34c7eb574940dcda7453b066bc512ee649506bfc65572dc67fd1a8227a9c", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 62, + "serialized_chars": 45731, + "estimated_input_tokens": 11433, + "estimated_eval_tokens": 23122 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly a heterogeneous issue cluster with a few true duplicate pairs. The clearest duplicate subclusters are the DeepSpeed ZeRO-3 + Bert index-out-of-bounds reports and the torch 2.4.1 compatibility reports; most other pairs only share broad subsystem similarity and should stay separate.", + "confidence": 0.84, + "canonical_issue_reason": "Issue 43596 is the best canonical issue: it is a precise DeepSpeed ZeRO-3 + Bert reproduction of the same index-out-of-bounds failure seen in 43638, with the cleanest specific bug description.", + "canonical_pr_reason": null, + "best_issue_reason": "If one issue must anchor the cluster, 43596 is the strongest representative because it sits in a real duplicate pair and has the most concrete, narrowly scoped failure mode.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44322", + "right": "issue:45464", + "accept": false, + "reason": "Different failures: Qwen3.5 config/model-shape error vs chat/completions streaming inference." + }, + { + "left": "issue:45237", + "right": "issue:45685", + "accept": false, + "reason": "Different backends and bugs: AMD GPU runtime issue vs MPS histogram op not implemented." + }, + { + "left": "issue:43716", + "right": "issue:45237", + "accept": false, + "reason": "Mistral image dtype mismatch is unrelated to GPT-OSS failing on AMD GPUs." + }, + { + "left": "issue:43828", + "right": "issue:45237", + "accept": false, + "reason": "Phi autocast dtype mismatch and GPT-OSS AMD GPU failure are different code paths." + }, + { + "left": "issue:45237", + "right": "issue:46468", + "accept": false, + "reason": "AMD GPU failure for GPT-OSS is not the same as the Nemotron example breakage." + }, + { + "left": "issue:45237", + "right": "issue:45507", + "accept": false, + "reason": "Different model-specific bugs: GPT-OSS AMD runtime issue vs GraniteMoEHybrid invalid method call." + }, + { + "left": "issue:30277", + "right": "issue:35545", + "accept": false, + "reason": "Deepspeed collective mismatch on Jamba is unrelated to ModernBERT ONNX export." + }, + { + "left": "issue:35545", + "right": "issue:42915", + "accept": false, + "reason": "ModernBERT export error and Qwen3Moe FP8 failure are different problems." + }, + { + "left": "issue:35545", + "right": "issue:41720", + "accept": false, + "reason": "ONNX export bug and Qwen3 device-mapping CUDA assert are unrelated." + }, + { + "left": "issue:45901", + "right": "issue:46015", + "accept": false, + "reason": "Table QA crash and outdated RAG docs are different issues." + }, + { + "left": "issue:44509", + "right": "issue:44512", + "accept": false, + "reason": "Both are docs cleanups for v5 removals, but they target different removed material and are not the same fix." + }, + { + "left": "issue:35545", + "right": "issue:36010", + "accept": false, + "reason": "ONNX export failure is unrelated to the GenerationMixin import error." + }, + { + "left": "issue:44918", + "right": "issue:46165", + "accept": false, + "reason": "Qwen3.5 embedding unpacking with TRL is unrelated to resume-training with trust_remote_code." + }, + { + "left": "issue:44805", + "right": "issue:45823", + "accept": false, + "reason": "Mask shape mismatch during indexing is unrelated to Gemma4 device-map mismatch." + }, + { + "left": "issue:43638", + "right": "issue:44805", + "accept": false, + "reason": "Both mention shape/index errors, but the Bert ZeRO-3 load crash and the mask/index mismatch are different bugs." + }, + { + "left": "issue:43638", + "right": "issue:46165", + "accept": false, + "reason": "Bert ZeRO-3 loading failure is unrelated to resuming local-model training with remote code." + }, + { + "left": "issue:43596", + "right": "issue:44805", + "accept": false, + "reason": "Different failures and repros: Bert ZeRO-3 init index error vs generic mask/tensor shape mismatch." + }, + { + "left": "issue:44805", + "right": "issue:46294", + "accept": false, + "reason": "Generic mask/index mismatch is unrelated to Qwen3.5 video-token template mismatch." + }, + { + "left": "issue:44805", + "right": "issue:45161", + "accept": false, + "reason": "Mask/index mismatch and GPT-OSS tensor-parallel failure are not the same bug." + }, + { + "left": "issue:46165", + "right": "issue:46468", + "accept": false, + "reason": "Remote-code resume-training issue is unrelated to the Nemotron example failure." + }, + { + "left": "issue:44560", + "right": "issue:46165", + "accept": false, + "reason": "Qwen3-VL video StopIteration and remote-code resume-training are different problems." + }, + { + "left": "issue:43596", + "right": "issue:46165", + "accept": false, + "reason": "Bert ZeRO-3 load crash is unrelated to the remote-code resume-training issue." + }, + { + "left": "issue:34634", + "right": "issue:35545", + "accept": false, + "reason": "BarkProcessor voice preset bug is unrelated to ModernBERT export." + }, + { + "left": "issue:30277", + "right": "issue:43638", + "accept": false, + "reason": "Both involve DeepSpeed, but the Jamba collective mismatch and Bert ZeRO-3 index error are not the same underlying bug." + }, + { + "left": "issue:43596", + "right": "issue:43638", + "accept": true, + "reason": "Same DeepSpeed ZeRO-3/Bert index-out-of-bounds failure, just described with pretrained vs non-pretrained wording." + }, + { + "left": "issue:43039", + "right": "issue:45161", + "accept": false, + "reason": "Liger Kernel cross-entropy dispatch and GPT-OSS tensor-parallel failure are unrelated." + }, + { + "left": "issue:43975", + "right": "issue:44457", + "accept": false, + "reason": "DeepSeek detokenization and LoRA merge/save-load output mismatch are different issues." + }, + { + "left": "issue:44292", + "right": "issue:44457", + "accept": false, + "reason": "Qwen NVFP4 runtime failure is unrelated to the LoRA save/reload discrepancy." + }, + { + "left": "issue:44841", + "right": "issue:46661", + "accept": false, + "reason": "Processor failure for Voxtral is unrelated to remote-code config_class import errors." + }, + { + "left": "issue:34928", + "right": "issue:44805", + "accept": false, + "reason": "Different training-time mismatch bugs: FSDP checkpointing vs generic mask/index shape error." + }, + { + "left": "issue:24643", + "right": "issue:30277", + "accept": false, + "reason": "DeepSpeed training weight-dimension error is unrelated to the Jamba collective mismatch." + }, + { + "left": "issue:45859", + "right": "issue:45902", + "accept": false, + "reason": "Same model family, but CPU-offload incompatibility and uninitialized-memory materialization are different bugs." + }, + { + "left": "issue:43638", + "right": "issue:45161", + "accept": false, + "reason": "Bert ZeRO-3 index error and GPT-OSS tensor-parallel failure are unrelated." + }, + { + "left": "issue:34928", + "right": "issue:36331", + "accept": false, + "reason": "Activation-checkpointing/FSDP mismatch is unrelated to the compute_loss signature change." + }, + { + "left": "issue:43596", + "right": "issue:45161", + "accept": false, + "reason": "Bert ZeRO-3 loading failure is unrelated to GPT-OSS tensor-parallel issues." + }, + { + "left": "issue:43716", + "right": "issue:44292", + "accept": false, + "reason": "Mistral image dtype mismatch and Qwen NVFP4 runtime error are different model-specific bugs." + }, + { + "left": "issue:45874", + "right": "issue:46294", + "accept": false, + "reason": "Gemma4 reverse-embedding lookup cost is unrelated to Qwen video feature/token mismatch." + }, + { + "left": "issue:43425", + "right": "issue:44292", + "accept": false, + "reason": "Torch version compatibility issue is unrelated to Qwen NVFP4 runtime failure." + }, + { + "left": "issue:45800", + "right": "issue:46515", + "accept": true, + "reason": "Same torch 2.4.1 compatibility problem; one report is just the more specific transformers 5.8.0 manifestation." + }, + { + "left": "issue:43827", + "right": "issue:44512", + "accept": false, + "reason": "Both are v5 docs fixes, but they cover different removed docs content and should not be merged as one bug." + }, + { + "left": "issue:43828", + "right": "issue:45161", + "accept": false, + "reason": "Phi autocast dtype mismatch is unrelated to GPT-OSS tensor-parallel failure." + }, + { + "left": "issue:30277", + "right": "issue:36331", + "accept": false, + "reason": "DeepSpeed collective mismatch and CustomTrainer API breakage are different layers and different bugs." + }, + { + "left": "issue:34928", + "right": "issue:36010", + "accept": false, + "reason": "FSDP activation-checkpointing mismatch is unrelated to the GenerationMixin import error." + }, + { + "left": "issue:45237", + "right": "issue:46015", + "accept": false, + "reason": "GPT-OSS AMD GPU runtime failure is unrelated to outdated RAG examples." + }, + { + "left": "issue:43039", + "right": "issue:44805", + "accept": false, + "reason": "Liger Kernel cross-entropy dispatch is unrelated to the mask shape/index error." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The cluster summary is grounded in the packet and stays conservative. The two accepted duplicate pairs are well supported by the issue titles/body patterns, and the rejected soft edges mostly separate clearly different model/backend failures or distinct docs/API issues. I don\u2019t see evidence of overstatement in the canonical issue choice or the summary." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 2, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:36010", + "issue:39290", + "issue:41093", + "issue:41720", + "issue:41762", + "issue:43039", + "issue:43296", + "issue:43366", + "issue:43404", + "issue:43425", + "issue:43531", + "issue:43541", + "issue:43572", + "issue:43596", + "issue:43638", + "issue:43792", + "issue:43828", + "issue:43854", + "issue:43866", + "issue:43872", + "issue:43901", + "issue:43975", + "issue:44292", + "issue:44512", + "issue:44560", + "issue:44589", + "issue:44805", + "issue:44841", + "issue:44863", + "issue:44918", + "issue:45020", + "issue:45070", + "issue:45084", + "issue:45237", + "issue:45362", + "issue:45399", + "issue:45464", + "issue:45685", + "issue:45800", + "issue:45812", + "issue:45823", + "issue:45901", + "issue:45920", + "issue:46015", + "issue:46294", + "issue:46304", + "issue:46402", + "issue:46661" + ], + "soft_pairs": [ + "issue:43866|issue:46402", + "issue:45800|issue:46304", + "issue:43296|issue:46304", + "issue:41762|issue:43596", + "issue:39290|issue:41720", + "issue:43901|issue:44512", + "issue:44589|issue:46304", + "issue:45237|issue:45901", + "issue:43039|issue:44560", + "issue:45823|issue:46294", + "issue:43975|issue:46015", + "issue:44841|issue:45823", + "issue:44863|issue:46015", + "issue:45020|issue:46304", + "issue:45812|issue:45920", + "issue:43866|issue:44863", + "issue:44292|issue:46015", + "issue:43872|issue:46304", + "issue:39290|issue:43531", + "issue:41093|issue:43638", + "issue:43854|issue:43866", + "issue:43792|issue:45901", + "issue:43425|issue:45070", + "issue:43975|issue:45901", + "issue:43596|issue:45800", + "issue:43792|issue:45070", + "issue:45362|issue:45464", + "issue:44292|issue:45901", + "issue:39290|issue:45823", + "issue:45464|issue:45685", + "issue:43541|issue:45685", + "issue:43541|issue:43828", + "issue:36010|issue:39290", + "issue:43404|issue:43866", + "issue:44841|issue:45084", + "issue:41093|issue:41762", + "issue:45399|issue:46661", + "issue:43572|issue:45070", + "issue:45800|issue:45823", + "issue:43296|issue:43366", + "issue:44805|issue:44918" + ], + "prepared_review_unit_hash": "241f43f6d1ea3b7631a6ec558e9195595977355ab64cb1552eb98a4b40d78b5f", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 41, + "serialized_chars": 39815, + "estimated_input_tokens": 9954, + "estimated_eval_tokens": 20164 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose mix of transformers/vLLM/DeepSpeed/model-loading regressions, not a single tight duplicate family. The only supported merge-worthy pairs are the ZeRO-3 index-out-of-bounds reports and the AutoTokenizer wrong-token-ID regression reports. Most other items share only broad subsystem or version-regression themes and should remain separate.", + "confidence": 0.92, + "canonical_issue_reason": "Issue 45020 is the broadest umbrella for the recurring recent-version regression reports, but the cluster remains heterogeneous and only a couple of pairs are close enough to combine.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45020 is the strongest single representative because it captures the recurring recent-version breakage theme without requiring the cluster to be much tighter than it is.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43866", + "right": "issue:46402", + "accept": false, + "reason": "Different models and failure modes: corrupted checkpoint vs missing text encoder weights." + }, + { + "left": "issue:45800", + "right": "issue:46304", + "accept": false, + "reason": "Both are version-compatibility complaints, but they describe different breakages and code paths." + }, + { + "left": "issue:43296", + "right": "issue:46304", + "accept": false, + "reason": "PaddleOCR-VL loading in vLLM is unrelated to USE_DISTRIBUTED=0 support in PyTorch." + }, + { + "left": "issue:41762", + "right": "issue:43596", + "accept": true, + "reason": "Same ZeRO-3 load-time signature: index 0 out of bounds for dimension 0 with size 0 during model initialization." + }, + { + "left": "issue:39290", + "right": "issue:41720", + "accept": false, + "reason": "Different symptoms and subsystems: Gemma3 config attribute error vs Qwen3 auto device-map CUDA assert." + }, + { + "left": "issue:43901", + "right": "issue:44512", + "accept": false, + "reason": "Both are docs issues, but they cover different documentation mistakes." + }, + { + "left": "issue:44589", + "right": "issue:46304", + "accept": false, + "reason": "Storage-type lookup failure is unrelated to the PyTorch distributed support regression." + }, + { + "left": "issue:45237", + "right": "issue:45901", + "accept": false, + "reason": "AMD GPU support for GPT-OSS and table-question-answering crashes are different issues." + }, + { + "left": "issue:43039", + "right": "issue:44560", + "accept": false, + "reason": "Liger Kernel cross-entropy dispatch and Qwen3.5-VL video StopIteration are different bugs." + }, + { + "left": "issue:45823", + "right": "issue:46294", + "accept": false, + "reason": "Device-map mismatch in Gemma4 PLE is unrelated to Qwen3.5 video token/feature alignment." + }, + { + "left": "issue:43975", + "right": "issue:46015", + "accept": false, + "reason": "Wrong detokenization and outdated RAG examples are unrelated." + }, + { + "left": "issue:44841", + "right": "issue:45823", + "accept": false, + "reason": "Voxtral processor failure and Gemma4 device mismatch are different code paths." + }, + { + "left": "issue:44863", + "right": "issue:46015", + "accept": false, + "reason": "NemotronH checkpoint loading is unrelated to RAG documentation drift." + }, + { + "left": "issue:45020", + "right": "issue:46304", + "accept": false, + "reason": "Both are broad regression reports, but one is about remote_code models and the other about USE_DISTRIBUTED=0 support." + }, + { + "left": "issue:45812", + "right": "issue:45920", + "accept": true, + "reason": "Same AutoTokenizer wrong-token-ID v5 regression; 45920 is the broader follow-up covering more model families." + }, + { + "left": "issue:43866", + "right": "issue:44863", + "accept": false, + "reason": "Corrupted checkpoint and NemotronH load failure are not the same bug." + }, + { + "left": "issue:44292", + "right": "issue:46015", + "accept": false, + "reason": "Qwen-3-8B-NVFP4 runtime error is unrelated to outdated RAG examples." + }, + { + "left": "issue:43872", + "right": "issue:46304", + "accept": false, + "reason": "bitsandbytes Int8Params incompatibility is distinct from the PyTorch distributed build issue." + }, + { + "left": "issue:39290", + "right": "issue:43531", + "accept": false, + "reason": "Both mention sliding window, but the configs and failures are different model-specific problems." + }, + { + "left": "issue:41093", + "right": "issue:43638", + "accept": false, + "reason": "Different errors: mask/tensor shape mismatch vs ZeRO-3 index-out-of-bounds during BertModel loading." + }, + { + "left": "issue:43854", + "right": "issue:43866", + "accept": false, + "reason": "Model load failure in tests and corrupted checkpoint are different problems." + }, + { + "left": "issue:43792", + "right": "issue:45901", + "accept": false, + "reason": "Whisper load failure and table-question-answering crash are unrelated." + }, + { + "left": "issue:43425", + "right": "issue:45070", + "accept": false, + "reason": "Torch 2.10 incompatibility and the PretrainedConfig pydantic-field regression are different issues." + }, + { + "left": "issue:43975", + "right": "issue:45901", + "accept": false, + "reason": "Detokenization regression and table-question-answering crash are unrelated." + }, + { + "left": "issue:43596", + "right": "issue:45800", + "accept": false, + "reason": "Both involve versioned infrastructure, but one is a ZeRO-3 index error and the other is torch/transformers incompatibility." + }, + { + "left": "issue:43792", + "right": "issue:45070", + "accept": false, + "reason": "Whisper loading and pydantic config-field breakage do not share the same bug." + }, + { + "left": "issue:45362", + "right": "issue:45464", + "accept": false, + "reason": "Different Qwen3.5 failures and different surfaces (chat vs streaming API)." + }, + { + "left": "issue:45685", + "right": "issue:45464", + "accept": false, + "reason": "MPS MoE histogram failure is unrelated to Qwen3.5 streaming inference." + }, + { + "left": "issue:43541", + "right": "issue:45685", + "accept": false, + "reason": "Different MoE-related runtime errors with different ops and failure modes." + }, + { + "left": "issue:43541", + "right": "issue:43828", + "accept": false, + "reason": "Grouped_mm tracing error and autocast dtype mismatch are not the same underlying bug." + }, + { + "left": "issue:36010", + "right": "issue:39290", + "accept": false, + "reason": "ImportError for GenerationMixin is unrelated to Gemma3 sliding_window_pattern." + }, + { + "left": "issue:43404", + "right": "issue:43866", + "accept": false, + "reason": "Untied lm_head weights in Mistral3 are unrelated to a corrupted Ovis2 checkpoint." + }, + { + "left": "issue:44841", + "right": "issue:45084", + "accept": false, + "reason": "Processor failure and template-node compilation error are different issues." + }, + { + "left": "issue:41093", + "right": "issue:41762", + "accept": false, + "reason": "Both are IndexErrors, but they are different failure signatures and code paths." + }, + { + "left": "issue:45399", + "right": "issue:46661", + "accept": false, + "reason": "Flash-attn fallback logic and remote-code AutoModel config_class handling are unrelated." + }, + { + "left": "issue:43572", + "right": "issue:45070", + "accept": false, + "reason": "Missing pad_token_idx and pydantic PretrainedConfig breakage are different config regressions." + }, + { + "left": "issue:45800", + "right": "issue:45823", + "accept": false, + "reason": "Torch/transformers incompatibility and Gemma4 device mismatch are unrelated." + }, + { + "left": "issue:43296", + "right": "issue:43366", + "accept": false, + "reason": "PaddleOCR-VL loading in vLLM and GGUF gpt-oss support are different model-support requests." + }, + { + "left": "issue:44805", + "right": "issue:44918", + "accept": false, + "reason": "Mask shape mismatch and Qwen3.5 embedding unpacking with TRL are different bugs." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The cluster-level takeaway is mostly grounded, but the summary overstates the evidence by claiming acceptance of \u201cthe mask/tensor shape mismatch reports\u201d even though no soft-edge verdict in the packet accepts such a pair. The supported accepted pairs are the ZeRO-3 index-out-of-bounds pair and the AutoTokenizer wrong-token-ID pair. Also, calling issue 45020 \u201ccentral\u201d is a bit stronger than the packet justifies; it is broad, but the cluster is still quite heterogeneous. Please tighten the summary to match only the explicitly supported pairings and keep the umbrella wording more conservative." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 3, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:39186", + "issue:40990", + "issue:41093", + "issue:42915", + "issue:43054", + "issue:43366", + "issue:43404", + "issue:43531", + "issue:43541", + "issue:43596", + "issue:43645", + "issue:43716", + "issue:43828", + "issue:43873", + "issue:43975", + "issue:43994", + "issue:44291", + "issue:44292", + "issue:44360", + "issue:44387", + "issue:44479", + "issue:44492", + "issue:44512", + "issue:44530", + "issue:44589", + "issue:44863", + "issue:44918", + "issue:44960", + "issue:45005", + "issue:45020", + "issue:45070", + "issue:45161", + "issue:45183", + "issue:45399", + "issue:45464", + "issue:45507", + "issue:45632", + "issue:45698", + "issue:45759", + "issue:45800", + "issue:45902", + "issue:46294", + "issue:46463", + "issue:46489", + "issue:46491", + "issue:46640", + "issue:46661" + ], + "soft_pairs": [ + "issue:44387|issue:45005", + "issue:43541|issue:45161", + "issue:43828|issue:45464", + "issue:43366|issue:45464", + "issue:44291|issue:45070", + "issue:43531|issue:45070", + "issue:43541|issue:44589", + "issue:43716|issue:46294", + "issue:45020|issue:45800", + "issue:41093|issue:43596", + "issue:44479|issue:46294", + "issue:43054|issue:43994", + "issue:45759|issue:45902", + "issue:46640|issue:46661", + "issue:39186|issue:44918", + "issue:43716|issue:44918", + "issue:40990|issue:44960", + "issue:45632|issue:45698", + "issue:43645|issue:45399", + "issue:43404|issue:45005", + "issue:42915|issue:45399", + "issue:44530|issue:45005", + "issue:43873|issue:45005", + "issue:44360|issue:44512", + "issue:44863|issue:45005", + "issue:44492|issue:44512", + "issue:45183|issue:46463", + "issue:46489|issue:46491", + "issue:44918|issue:45507", + "issue:43975|issue:44292", + "issue:44292|issue:45507" + ], + "prepared_review_unit_hash": "c25c597c39fc324ddb959da42755f1286f07a7c0523fbb362a6500da02fa7eaf", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36233, + "estimated_input_tokens": 9059, + "estimated_eval_tokens": 18374 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly heterogeneous; the soft pairs are largely not true duplicates. The only plausible duplicate-like pair is the T5 input-validation error messaging issue, while the rest differ by model, symptom, or code path.", + "confidence": 0.73, + "canonical_issue_reason": "issue:45005 is the broadest issue in the set and the best anchor for the tied-weights regression subgroup, with multiple nearby soft links, whereas most other items are narrower model-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "If one issue must represent the cluster, issue:45005 is the best fit because it covers the widest related regression surface and sits at the center of the closest soft ties.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44387", + "right": "issue:45005", + "accept": false, + "reason": "Both involve v5 regressions, but one is an int4 memory/OOM problem and the other is tied-weights behavior; different failure modes and likely different fixes." + }, + { + "left": "issue:43541", + "right": "issue:45161", + "accept": false, + "reason": "Both mention MoE models, but one is a grouped_mm torch-dynamo tracing error and the other is a tensor-parallel support issue for GPT-OSS; not the same bug." + }, + { + "left": "issue:43828", + "right": "issue:45464", + "accept": false, + "reason": "One is an autocast dtype mismatch in Phi-tiny-MoE; the other is a streaming chat/completions failure in Qwen3.5-0.8B. Different code paths." + }, + { + "left": "issue:43366", + "right": "issue:45464", + "accept": false, + "reason": "GGUF/gpt-oss support and Qwen3.5 streaming inference are unrelated issues." + }, + { + "left": "issue:44291", + "right": "issue:45070", + "accept": false, + "reason": "init_empty_weights/_is_hf_initialized is a constructor-signature bug; the pydantic issue is about PretrainedConfig fields. Different regressions." + }, + { + "left": "issue:43531", + "right": "issue:45070", + "accept": false, + "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to the PretrainedConfig/pydantic breakage." + }, + { + "left": "issue:43541", + "right": "issue:44589", + "accept": false, + "reason": "Grouped_mm tracing failure and missing Float8 storage are distinct runtime errors." + }, + { + "left": "issue:43716", + "right": "issue:46294", + "accept": false, + "reason": "Both are multimodal mismatches, but one is image preprocessor dtype handling for Mistral-3 and the other is Qwen3.5 video token/feature alignment; not the same underlying bug." + }, + { + "left": "issue:45020", + "right": "issue:45800", + "accept": false, + "reason": "Remote_code breakage across versions is not the same as a torch/transformers version incompatibility report." + }, + { + "left": "issue:41093", + "right": "issue:43596", + "accept": false, + "reason": "Mask-length mismatch in one issue and an index-out-of-bounds error during deepspeed zero3 init in the other are different failures." + }, + { + "left": "issue:44479", + "right": "issue:46294", + "accept": false, + "reason": "Both concern video inputs, but 44479 is a broad v5.3.0 regression across several model families while 46294 is a specific apply_chat_template token/feature mismatch." + }, + { + "left": "issue:43054", + "right": "issue:43994", + "accept": false, + "reason": "Both are about siglip2, but one reports poor embedding quality and the other incorrect AutoModel/pipeline outputs; the symptoms and likely fixes differ." + }, + { + "left": "issue:45759", + "right": "issue:45902", + "accept": false, + "reason": "One is about unwrapping text_config in composite multimodal configs; the other is about uninitialized memory in rotary embedding after to_empty. Different code paths." + }, + { + "left": "issue:46640", + "right": "issue:46661", + "accept": false, + "reason": "Flash Attention 2 setup/troubleshooting and remote-code config_class lookup are unrelated." + }, + { + "left": "issue:39186", + "right": "issue:44918", + "accept": false, + "reason": "FSDP 2-D weight errors and TRL SFT unpacking failures are unrelated model/runtime problems." + }, + { + "left": "issue:43716", + "right": "issue:44918", + "accept": false, + "reason": "Image dtype mismatch and Qwen3.5 input-embedding unpacking are distinct multimodal issues." + }, + { + "left": "issue:40990", + "right": "issue:44960", + "accept": false, + "reason": "High perplexity on gpt-oss and a GLM5 issue are unrelated model-specific reports." + }, + { + "left": "issue:45632", + "right": "issue:45698", + "accept": false, + "reason": "Both involve custom-module loading, but one is a cache-path collision for local models while the other is wrong module selection after save_pretrained; too different to merge confidently." + }, + { + "left": "issue:43645", + "right": "issue:45399", + "accept": false, + "reason": "Custom model initialization in notebooks and flash-attn2 fallback gating are unrelated." + }, + { + "left": "issue:43404", + "right": "issue:45005", + "accept": false, + "reason": "Both mention tied weights, but 43404 is a specific Mistral3 multimodal lm_head tying bug and 45005 is a broader translation-model regression; not clearly the same concrete fix." + }, + { + "left": "issue:42915", + "right": "issue:45399", + "accept": false, + "reason": "Qwen3Moe FineGrainedFP8Config failure and flash-attn2 fallback checks are different issues." + }, + { + "left": "issue:44530", + "right": "issue:45005", + "accept": false, + "reason": "PagedAttentionCache/linear_attention crash is unrelated to tied-weights behavior." + }, + { + "left": "issue:43873", + "right": "issue:45005", + "accept": false, + "reason": "Quantization/offloading behavior and tied weights are different problem areas." + }, + { + "left": "issue:44360", + "right": "issue:44512", + "accept": false, + "reason": "A DSA indexer implementation comment and a docs typo are unrelated." + }, + { + "left": "issue:44863", + "right": "issue:45005", + "accept": false, + "reason": "NemotronH checkpoint loading and translation-model tied weights are not the same bug." + }, + { + "left": "issue:44492", + "right": "issue:44512", + "accept": false, + "reason": "Cache-strategy typo and removed run-command documentation are both docs-level issues, but not the same one." + }, + { + "left": "issue:45183", + "right": "issue:46463", + "accept": true, + "reason": "Both are about input validation for input_ids/inputs_embeds and misleading/incorrect error messages; 46463 looks like a specific T5-family instance of the broader validation bug." + }, + { + "left": "issue:46489", + "right": "issue:46491", + "accept": false, + "reason": "Both are tokenizer regressions, but one is DeepSeek-Coder v1 incorrect detokenization and the other is CodeLlama leading-space round-trip loss; different model behaviors." + }, + { + "left": "issue:44918", + "right": "issue:45507", + "accept": false, + "reason": "Qwen3.5 input-embedding unpacking in TRL and GraniteMoEHybrid calling an invalid method are unrelated failures." + }, + { + "left": "issue:43975", + "right": "issue:44292", + "accept": false, + "reason": "DeepSeek-Coder tokenization and Qwen-3-8B-NVFP4 runtime errors are unrelated." + }, + { + "left": "issue:44292", + "right": "issue:45507", + "accept": false, + "reason": "NVFP4 runtime failure and GraniteMoEHybrid invalid-method calls are different issues." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is well grounded in the packet: it stays conservative about the soft links, correctly treats most pairs as non-duplicates, and the only accepted soft pair (issue:45183 with issue:46463) matches the validation/error-message family in the titles. The choice of issue:45005 as the anchor is also supported by its multiple nearby soft ties." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 4, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:24643", + "issue:28530", + "issue:36331", + "issue:38175", + "issue:39692", + "issue:41553", + "issue:41762", + "issue:42491", + "issue:42915", + "issue:42947", + "issue:43054", + "issue:43493", + "issue:43643", + "issue:43644", + "issue:43782", + "issue:43824", + "issue:43828", + "issue:43856", + "issue:43866", + "issue:43872", + "issue:43881", + "issue:43883", + "issue:43950", + "issue:43975", + "issue:44368", + "issue:44387", + "issue:44403", + "issue:44451", + "issue:44488", + "issue:44534", + "issue:44589", + "issue:44661", + "issue:44910", + "issue:44928", + "issue:44960", + "issue:44991", + "issue:45020", + "issue:45081", + "issue:45399", + "issue:45520", + "issue:45663", + "issue:45685", + "issue:45701", + "issue:45823", + "issue:45859", + "issue:45902", + "issue:45920", + "issue:46489" + ], + "soft_pairs": [ + "issue:43950|issue:44534", + "issue:39692|issue:43054", + "issue:43644|issue:43950", + "issue:36331|issue:41762", + "issue:43828|issue:44928", + "issue:43872|issue:44589", + "issue:45920|issue:46489", + "issue:43054|issue:45701", + "issue:43866|issue:44403", + "issue:38175|issue:43054", + "issue:43950|issue:45902", + "issue:41553|issue:43883", + "issue:44910|issue:45859", + "issue:42947|issue:44387", + "issue:43054|issue:43493", + "issue:45663|issue:45902", + "issue:41553|issue:43881", + "issue:44403|issue:44451", + "issue:43975|issue:45520", + "issue:42947|issue:43856", + "issue:44403|issue:44488", + "issue:24643|issue:28530", + "issue:43643|issue:45020", + "issue:41762|issue:42915", + "issue:44991|issue:45399", + "issue:43782|issue:44960", + "issue:45685|issue:45823", + "issue:43824|issue:43872", + "issue:43881|issue:43883", + "issue:45020|issue:45823", + "issue:44661|issue:45020", + "issue:45081|issue:45399", + "issue:42491|issue:44368", + "issue:43828|issue:45823" + ], + "prepared_review_unit_hash": "ca7aefed57df90a3e9363516078f6423f35ca42112e7674798e8547b37d1749e", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 38113, + "estimated_input_tokens": 9529, + "estimated_eval_tokens": 19314 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The packet suggests a conservative thematic cluster around Transformers v5 non-persistent-buffer corruption, but the evidence is mostly title-level similarity. I would treat the closely related reports as related-but-not-proven duplicates, with issue 43950 as the best anchor only because it states the broad regression most explicitly.", + "confidence": 0.72, + "canonical_issue_reason": "issue:43950 is the broadest and most concrete description of the v5 non-persistent-buffer corruption regression, so it is the best anchor if this cluster is being tracked conservatively.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43950 is the strongest representative of the buffer-corruption theme, but the packet does not firmly prove the nearby reports are true duplicates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43950", + "right": "issue:44534", + "accept": false, + "reason": "Very similar wording, but the packet only gives title-level similarity; that is not enough to confirm a duplicate merge." + }, + { + "left": "issue:39692", + "right": "issue:43054", + "accept": false, + "reason": "Both mention SigLIP2, but one is a docs/example error and the other is a model-quality regression." + }, + { + "left": "issue:43644", + "right": "issue:43950", + "accept": false, + "reason": "Missing fields with trust_remote_code is a different failure mode from non-persistent-buffer corruption." + }, + { + "left": "issue:36331", + "right": "issue:41762", + "accept": false, + "reason": "Unexpected compute_loss kwarg and DeepSpeed ZeRO-3 load index errors are different bugs." + }, + { + "left": "issue:43828", + "right": "issue:44928", + "accept": false, + "reason": "Autocast dtype mismatch and Qwen3.5 position_ids/SDPA issues are not the same code-path problem." + }, + { + "left": "issue:43872", + "right": "issue:44589", + "accept": false, + "reason": "bitsandbytes constructor incompatibility and missing Float8 storage are unrelated compatibility errors." + }, + { + "left": "issue:45920", + "right": "issue:46489", + "accept": false, + "reason": "Both are tokenizer regressions, but 46489 is a specific DeepSeek-Coder gap in an earlier fix, not the same broad multi-model bug." + }, + { + "left": "issue:43054", + "right": "issue:45701", + "accept": false, + "reason": "Text-embedding degradation and tokenization changes are only loosely related, not duplicates." + }, + { + "left": "issue:43866", + "right": "issue:44403", + "accept": false, + "reason": "Checkpoint corruption and noisy loading messages are different symptoms with no shared concrete bug shown." + }, + { + "left": "issue:38175", + "right": "issue:43054", + "accept": false, + "reason": "Unexpected zero probabilities and worse text embeddings are different SigLIP2 problems." + }, + { + "left": "issue:43950", + "right": "issue:45902", + "accept": false, + "reason": "Highly related buffer-corruption reports, but the packet does not provide enough direct evidence to confirm a duplicate pair." + }, + { + "left": "issue:41553", + "right": "issue:43883", + "accept": false, + "reason": "Voxtral tokenizer loading and Molmo tied-weights attribute errors are different failures." + }, + { + "left": "issue:44910", + "right": "issue:45859", + "accept": false, + "reason": "FlashAttention crash from 3D position_ids and CPU-offload incompatibility are distinct Qwen3.5 issues." + }, + { + "left": "issue:42947", + "right": "issue:44387", + "accept": false, + "reason": "Gradient checkpointing ineffectiveness and int4 memory OOM are different performance bugs." + }, + { + "left": "issue:43054", + "right": "issue:43493", + "accept": false, + "reason": "SigLIP2 embedding quality and HF-vs-JAX implementation discrepancy are not the same concrete defect." + }, + { + "left": "issue:45663", + "right": "issue:45902", + "accept": false, + "reason": "Gemma-4 FSDP2 shared_kv_states and Qwen3_5Moe buffer initialization are unrelated model-path bugs." + }, + { + "left": "issue:41553", + "right": "issue:43881", + "accept": false, + "reason": "Bad AutoTokenizer error for Voxtral and glm-4v-9b load failure are different issues." + }, + { + "left": "issue:44403", + "right": "issue:44451", + "accept": false, + "reason": "Generic loading noise and a specific ScandiBERT load failure are not the same bug." + }, + { + "left": "issue:43975", + "right": "issue:45520", + "accept": false, + "reason": "DeepSeek detokenization regression and flash_attn import-key failure are unrelated." + }, + { + "left": "issue:42947", + "right": "issue:43856", + "accept": false, + "reason": "Both involve training/memory, but checkpointing behavior and MoE memory inefficiency are different root causes." + }, + { + "left": "issue:44403", + "right": "issue:44488", + "accept": false, + "reason": "Same broad loading area, but the reports target different models and failure modes." + }, + { + "left": "issue:24643", + "right": "issue:28530", + "accept": false, + "reason": "DeepSpeed weight-shape training error and early-stopping metric mismatch are separate trainer issues." + }, + { + "left": "issue:43643", + "right": "issue:45020", + "accept": false, + "reason": "Generic remote_code breakage and a specific AutoConfig missing-fields bug are not duplicates." + }, + { + "left": "issue:41762", + "right": "issue:42915", + "accept": false, + "reason": "ZeRO-3 index-out-of-bounds during loading and FineGrainedFP8Config failure are different load/config problems." + }, + { + "left": "issue:44991", + "right": "issue:45399", + "accept": false, + "reason": "Tokenizer loading failure and flash-attn fallback gating are unrelated." + }, + { + "left": "issue:43782", + "right": "issue:44960", + "accept": false, + "reason": "Qwen3VL weight_only load error and GLM5 are not the same bug." + }, + { + "left": "issue:45685", + "right": "issue:45823", + "accept": false, + "reason": "MPS histogram int error and Gemma4 device mismatch are different device/runtime issues." + }, + { + "left": "issue:43824", + "right": "issue:43872", + "accept": false, + "reason": "Qwen2.5-VL import failure and bitsandbytes Int8Params kwarg mismatch are unrelated." + }, + { + "left": "issue:43881", + "right": "issue:43883", + "accept": false, + "reason": "glm-4v-9b loading failure and Molmo missing all_tied_weights_keys are different code paths." + }, + { + "left": "issue:45020", + "right": "issue:45823", + "accept": false, + "reason": "Remote_code regression and Gemma4 device_map forward mismatch are not the same underlying defect." + }, + { + "left": "issue:44661", + "right": "issue:45020", + "accept": false, + "reason": "Tokenizer mapping edge case and remote_code loading breakage are only superficially related." + }, + { + "left": "issue:45081", + "right": "issue:45399", + "accept": false, + "reason": "Tokenizer regex crash and flash-attn fallback blocking are different bugs." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The summary is a bit too strong for the evidence in the packet. While issue:43950, issue:44534, and issue:45902 are all thematically related to non-persistent-buffer corruption / uninitialized memory, the packet\u2019s soft-pair signal marks those pairs as deterministic_accept=false, and the analyst\u2019s hard `accept: true` verdicts go beyond what is firmly grounded from the titles alone. The claim of a \"clear duplicate cluster\" is therefore overstated; a more conservative read would keep those as related-but-not-proven duplicates." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 5, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:24643", + "issue:30064", + "issue:36331", + "issue:41093", + "issue:41553", + "issue:42915", + "issue:42947", + "issue:43257", + "issue:43366", + "issue:43404", + "issue:43531", + "issue:43596", + "issue:43792", + "issue:43827", + "issue:43828", + "issue:43854", + "issue:43872", + "issue:43940", + "issue:43950", + "issue:44220", + "issue:44291", + "issue:44479", + "issue:44488", + "issue:44509", + "issue:44530", + "issue:44560", + "issue:44805", + "issue:44863", + "issue:44928", + "issue:44991", + "issue:45020", + "issue:45305", + "issue:45313", + "issue:45325", + "issue:45356", + "issue:45488", + "issue:45588", + "issue:45663", + "issue:45715", + "issue:45759", + "issue:45823", + "issue:45902", + "issue:45923", + "issue:46315", + "issue:46402", + "issue:46489", + "issue:46515", + "issue:46661" + ], + "soft_pairs": [ + "issue:45715|issue:46315", + "issue:45020|issue:45356", + "issue:45488|issue:46489", + "issue:43827|issue:44509", + "issue:43404|issue:43950", + "issue:45759|issue:46661", + "issue:43872|issue:44291", + "issue:42915|issue:43872", + "issue:45020|issue:46661", + "issue:43792|issue:44220", + "issue:42947|issue:45923", + "issue:43257|issue:45305", + "issue:41553|issue:44991", + "issue:36331|issue:41093", + "issue:43940|issue:45313", + "issue:43854|issue:44863", + "issue:44928|issue:45588", + "issue:41553|issue:44488", + "issue:44479|issue:44560", + "issue:41093|issue:44805", + "issue:44863|issue:46402", + "issue:24643|issue:30064", + "issue:44530|issue:44863", + "issue:43366|issue:43828", + "issue:43596|issue:46515", + "issue:45325|issue:45902", + "issue:43366|issue:43531", + "issue:45663|issue:45823" + ], + "prepared_review_unit_hash": "dae4ded0153c02b476524e6fe52f2cba865704773679e00863dd782c8a5ab2f7", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36623, + "estimated_input_tokens": 9156, + "estimated_eval_tokens": 18568 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly heterogeneous Transformer regression reports, not one tight duplicate cluster. Only a few soft pairs look like true same-bug duplicates: the removed-pipeline docs issue, the `_is_hf_initialized` loading regression, the identical mask-shape IndexError, and likely the DeepSeek tokenizer regression pair.", + "confidence": 0.64, + "canonical_issue_reason": "There is no single clean canonical across the whole batch; if one must be chosen, issue 43827 is the best anchor because it is part of the clearest duplicate pair and describes the docs regression most concretely.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43827 is the strongest overall issue to anchor on because it is concrete, reproducible, and has the clearest one-to-one duplicate relationship in this set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45715", + "right": "issue:46315", + "accept": false, + "reason": "Related tokenizer performance regressions, but one is about `convert_ids_to_tokens` and the other about `convert_tokens_to_ids`; different code paths." + }, + { + "left": "issue:45020", + "right": "issue:45356", + "accept": false, + "reason": "Both are broad tokenizer/model regressions, but the symptoms and likely fixes are different (`remote_code` breakage vs Kimi tokenizer codec handling)." + }, + { + "left": "issue:45488", + "right": "issue:46489", + "accept": true, + "reason": "Both point to the same DeepSeek tokenizer regression in v5: incorrect handling of tokenizer pre-tokenization/output for the DeepSeek family." + }, + { + "left": "issue:43827", + "right": "issue:44509", + "accept": true, + "reason": "Same docs bug: v5 removed pipeline tasks, but the documentation still references summarization/translation/text2text-generation pipelines." + }, + { + "left": "issue:43404", + "right": "issue:43950", + "accept": false, + "reason": "Both involve model loading/state handling, but one is a tied-weight bug in Mistral3 and the other is a non-persistent buffer corruption regression." + }, + { + "left": "issue:45759", + "right": "issue:46661", + "accept": false, + "reason": "Different auto-model failures: composite multimodal config unwrapping vs remote-code `config_class` lookup." + }, + { + "left": "issue:43872", + "right": "issue:44291", + "accept": true, + "reason": "Same root cause: the new `_is_hf_initialized` keyword breaks model materialization/loading paths." + }, + { + "left": "issue:42915", + "right": "issue:43872", + "accept": false, + "reason": "Both mention loading/quantization incompatibilities, but the model/config and failure mode are different." + }, + { + "left": "issue:45020", + "right": "issue:46661", + "accept": false, + "reason": "`remote_code` regressions are too broad here; these are different concrete failures and not the same bug." + }, + { + "left": "issue:43792", + "right": "issue:44220", + "accept": false, + "reason": "Whisper failing to run and `_torch_extract_fbank_features()` are related, but the reported problems are not clearly the same underlying bug." + }, + { + "left": "issue:42947", + "right": "issue:45923", + "accept": false, + "reason": "Gradient checkpointing issues, but one is ineffective LoRA checkpointing and the other is a missing support flag on a remote-code variant." + }, + { + "left": "issue:43257", + "right": "issue:45305", + "accept": false, + "reason": "Both involve DeepSpeed, but one is a weight-conversion failure and the other is a gradient-averaging bug." + }, + { + "left": "issue:41553", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-loading problems, but they affect different models and different failure causes." + }, + { + "left": "issue:36331", + "right": "issue:41093", + "accept": false, + "reason": "Different training-time errors: custom `compute_loss` signature mismatch vs tensor/mask shape mismatch." + }, + { + "left": "issue:43940", + "right": "issue:45313", + "accept": false, + "reason": "Both are ZeRO-3 weight-loading complaints for Qwen models, but the model family and missing-weights pattern differ." + }, + { + "left": "issue:43854", + "right": "issue:44863", + "accept": false, + "reason": "Different model-loading failures for unrelated architectures; no clear shared bug." + }, + { + "left": "issue:44928", + "right": "issue:45588", + "accept": false, + "reason": "Different runtime paths and symptoms: RLHF gradient explosion vs flash-attention auxiliary-state crash." + }, + { + "left": "issue:41553", + "right": "issue:44488", + "accept": false, + "reason": "Both are tokenizer/model loading issues, but they are different models and different errors." + }, + { + "left": "issue:44479", + "right": "issue:44560", + "accept": false, + "reason": "Different multimodal regressions: video input handling vs Qwen3-vl-embedding video StopIteration." + }, + { + "left": "issue:41093", + "right": "issue:44805", + "accept": true, + "reason": "Same concrete mask/tensor shape mismatch error; the later report is essentially a rephrased duplicate." + }, + { + "left": "issue:44863", + "right": "issue:46402", + "accept": false, + "reason": "Different checkpoint-loading failures for different model families and different missing components." + }, + { + "left": "issue:24643", + "right": "issue:30064", + "accept": false, + "reason": "Unrelated issues: DeepSpeed training weight-shape error vs image processor handling of void segmentation maps." + }, + { + "left": "issue:44530", + "right": "issue:44863", + "accept": false, + "reason": "Different model-specific failures; one is a PagedAttentionCache crash, the other is a checkpoint loading problem." + }, + { + "left": "issue:43366", + "right": "issue:43828", + "accept": false, + "reason": "Unrelated: GGUF architecture support vs autocast dtype mismatch in a different model." + }, + { + "left": "issue:43596", + "right": "issue:46515", + "accept": false, + "reason": "A ZeRO-3/BERT initialization failure is not the same as a general torch compatibility report." + }, + { + "left": "issue:45325", + "right": "issue:45902", + "accept": false, + "reason": "Both touch rotary/position handling, but they describe different bugs and different model families." + }, + { + "left": "issue:43366", + "right": "issue:43531", + "accept": false, + "reason": "Both mention Qwen/Qwen-MoE-adjacent models, but the actual bugs are unrelated." + }, + { + "left": "issue:45663", + "right": "issue:45823", + "accept": false, + "reason": "Same family (Gemma4) but different failures: shared KV state construction vs device placement under `device_map='auto'`." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays appropriately cautious about the batch being heterogeneous. The accepted soft pairs are supported by the issue titles, especially the duplicate-looking docs regression, the `_is_hf_initialized` loading bug, the identical mask-shape error, and the likely DeepSeek tokenizer regression. The negative verdicts also read conservative rather than overclaimed." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 6, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:36010", + "issue:38175", + "issue:39692", + "issue:41720", + "issue:42491", + "issue:42617", + "issue:42915", + "issue:43317", + "issue:43366", + "issue:43550", + "issue:43596", + "issue:43632", + "issue:43638", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43716", + "issue:43720", + "issue:43824", + "issue:43950", + "issue:43957", + "issue:43994", + "issue:44155", + "issue:44451", + "issue:44488", + "issue:44534", + "issue:44560", + "issue:44589", + "issue:44661", + "issue:44683", + "issue:44756", + "issue:44805", + "issue:44877", + "issue:44910", + "issue:44918", + "issue:44960", + "issue:44977", + "issue:45030", + "issue:45276", + "issue:45584", + "issue:45685", + "issue:45812", + "issue:45823", + "issue:45834", + "issue:46459", + "issue:46489", + "issue:46512", + "issue:46515" + ], + "soft_pairs": [ + "issue:43638|issue:46515", + "issue:44155|issue:45584", + "issue:43716|issue:44560", + "issue:43950|issue:45834", + "issue:43366|issue:44589", + "issue:43824|issue:43957", + "issue:44560|issue:45685", + "issue:42915|issue:43957", + "issue:43645|issue:43646", + "issue:43550|issue:44910", + "issue:44560|issue:44918", + "issue:44683|issue:46515", + "issue:42617|issue:43366", + "issue:45812|issue:46489", + "issue:36010|issue:42915", + "issue:45823|issue:46515", + "issue:44560|issue:44805", + "issue:44661|issue:45823", + "issue:38175|issue:43994", + "issue:43632|issue:45834", + "issue:44877|issue:45030", + "issue:43638|issue:44661", + "issue:43596|issue:44661", + "issue:43317|issue:44756", + "issue:44661|issue:44805", + "issue:42491|issue:43720", + "issue:38175|issue:39692", + "issue:43644|issue:44534", + "issue:41720|issue:42915", + "issue:42491|issue:45276", + "issue:42915|issue:44756", + "issue:44451|issue:44488", + "issue:46459|issue:46512", + "issue:44960|issue:44977" + ], + "prepared_review_unit_hash": "d1c330c7fe78edcc382690d798508f9306e5869d8ccde5b2467032546af50630", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37775, + "estimated_input_tokens": 9444, + "estimated_eval_tokens": 19144 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is mostly heterogeneous, with only a few clear duplicate mini-clusters: the non-persistent-buffer regression, the Transformers 5 custom-model init regression, and the SigLIP2 inference regression. Most other pairs are related only at a broad subsystem/model-family level and should not be merged.", + "confidence": 0.82, + "canonical_issue_reason": "No single issue cleanly represents the whole cluster; the strongest duplicate-style issue is 43644 because it has an exact near-title duplicate and a precise regression description.", + "canonical_pr_reason": null, + "best_issue_reason": "43644 is the best representative issue in this set: it states the regression clearly and has the closest duplicate match (44534) with the same underlying buffer-corruption bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43638", + "right": "issue:46515", + "accept": false, + "reason": "Both mention failures after upgrades, but one is a Deepspeed/BERT zero3 init IndexError and the other is a generic torch 2.4.1 incompatibility; no shared concrete bug." + }, + { + "left": "issue:44155", + "right": "issue:45584", + "accept": false, + "reason": "Different models and failure modes: AudioFlamingo3 track leakage vs Whisper empty-transcription handling after token alignment." + }, + { + "left": "issue:43716", + "right": "issue:44560", + "accept": false, + "reason": "Both are multimodal issues, but they affect different models and code paths (SigLIP2 preprocessing mismatch vs Qwen3-VL video StopIteration)." + }, + { + "left": "issue:43950", + "right": "issue:45834", + "accept": false, + "reason": "Unrelated regressions: non-persistent buffer corruption during loading vs Kosmos2.5 long-OCR index error." + }, + { + "left": "issue:43366", + "right": "issue:44589", + "accept": false, + "reason": "GGUF/gpt-oss support and Float8 storage lookup are unrelated loader/runtime issues." + }, + { + "left": "issue:43824", + "right": "issue:43957", + "accept": false, + "reason": "An import/export symbol regression is not the same bug as meta-device loading failures across some models." + }, + { + "left": "issue:44560", + "right": "issue:45685", + "accept": false, + "reason": "Different architectures and errors: Qwen3-VL video inference StopIteration vs an MPS histogram op missing for Int." + }, + { + "left": "issue:42915", + "right": "issue:43957", + "accept": false, + "reason": "Qwen3Moe FP8 config failure and meta-device loading failures are separate bugs." + }, + { + "left": "issue:43645", + "right": "issue:43646", + "accept": true, + "reason": "Same underlying Transformers v5 custom-model initialization regression; 43645 is a notebook-specific manifestation of 43646." + }, + { + "left": "issue:43550", + "right": "issue:44910", + "accept": false, + "reason": "Different attention/compile failures in different models; no shared code-path bug is evident." + }, + { + "left": "issue:44560", + "right": "issue:44918", + "accept": false, + "reason": "Both involve Qwen-family multimodal behavior, but one is a video embedding StopIteration and the other is TRL SFT embedding unpacking." + }, + { + "left": "issue:44683", + "right": "issue:46515", + "accept": false, + "reason": "Torch-version incompatibility, but on different features and releases; not the same concrete problem." + }, + { + "left": "issue:42617", + "right": "issue:43366", + "accept": false, + "reason": "3D parallel script execution and GGUF architecture support are unrelated." + }, + { + "left": "issue:45812", + "right": "issue:46489", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different model families and likely different code paths." + }, + { + "left": "issue:36010", + "right": "issue:42915", + "accept": false, + "reason": "ImportError for GenerationMixin is unrelated to Qwen3Moe FP8 execution failure." + }, + { + "left": "issue:45823", + "right": "issue:46515", + "accept": false, + "reason": "Gemma4 device-map mismatch and a generic torch incompatibility are not the same bug." + }, + { + "left": "issue:44560", + "right": "issue:44805", + "accept": false, + "reason": "Video StopIteration and mask/tensor shape mismatch are different runtime failures." + }, + { + "left": "issue:44661", + "right": "issue:45823", + "accept": false, + "reason": "Tokenizer-mapping validation and Gemma4 device placement are unrelated." + }, + { + "left": "issue:38175", + "right": "issue:43994", + "accept": true, + "reason": "Same SigLIP2 base-patch16-224 inference regression: zero probabilities / nonsensical outputs on the same model indicate the same underlying bug." + }, + { + "left": "issue:43632", + "right": "issue:45834", + "accept": false, + "reason": "The `_is_hf_initialized` regression and Kosmos2.5 OCR index error are unrelated." + }, + { + "left": "issue:44877", + "right": "issue:45030", + "accept": false, + "reason": "Both are config-validation failures, but they concern different models and there is no evidence of a shared root cause." + }, + { + "left": "issue:43638", + "right": "issue:44661", + "accept": false, + "reason": "Deepspeed zero3/BERT initialization and TOKENIZER_MAPPING_NAMES handling are different issues." + }, + { + "left": "issue:43596", + "right": "issue:44661", + "accept": false, + "reason": "A zero3/BERT IndexError is not the same bug as add-new-model-like failing in tokenizer mapping names." + }, + { + "left": "issue:43317", + "right": "issue:44756", + "accept": false, + "reason": "Device-map offload loading failure and mmap/OOM on Strix Halo are different memory/loading problems." + }, + { + "left": "issue:42491", + "right": "issue:43720", + "accept": false, + "reason": "Qwen3_moe LoRA compatibility across versions and BitNet packed-weight loading are unrelated." + }, + { + "left": "issue:38175", + "right": "issue:39692", + "accept": false, + "reason": "Same model family, but one is an inference symptom and the other is a documentation/example issue with additional quantization failure; not clearly the same bug." + }, + { + "left": "issue:43644", + "right": "issue:44534", + "accept": true, + "reason": "Exact same Transformers v5 non-persistent-buffer regression, just reworded." + }, + { + "left": "issue:41720", + "right": "issue:42915", + "accept": false, + "reason": "Both are Qwen-related, but one is auto device mapping CUDA assert and the other is FP8 config failure." + }, + { + "left": "issue:42491", + "right": "issue:45276", + "accept": false, + "reason": "LoRA/Qwen3_moe version compatibility and Gemma4 resize_token_embeddings behavior are unrelated." + }, + { + "left": "issue:42915", + "right": "issue:44756", + "accept": false, + "reason": "Qwen3Moe FP8 failure and Strix Halo mmap OOM are unrelated." + }, + { + "left": "issue:44451", + "right": "issue:44488", + "accept": false, + "reason": "Both report loading failures for BERT-like models, but the titles do not show the same underlying regression or code path." + }, + { + "left": "issue:46459", + "right": "issue:46512", + "accept": false, + "reason": "Both involve dtype handling, but one is ignored dtype on composite checkpoint loading and the other is stale config dtype after from_config; different bugs." + }, + { + "left": "issue:44960", + "right": "issue:44977", + "accept": false, + "reason": "GLM5 and Qwen3.5 flash-attention generation failures are separate model-specific problems." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The cluster summary is grounded in the packet and stays conservative about soft matches. The accepted duplicates (43645/43646, 38175/43994, 43644/44534) are plausible from the titles and descriptions, and the rejected pairs mostly avoid overclaiming shared root cause where the evidence is only broad subsystem or model-family similarity." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 7, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:36010", + "issue:39404", + "issue:39692", + "issue:41720", + "issue:42491", + "issue:42947", + "issue:43257", + "issue:43404", + "issue:43425", + "issue:43454", + "issue:43644", + "issue:43645", + "issue:43708", + "issue:43854", + "issue:43856", + "issue:43873", + "issue:43931", + "issue:43957", + "issue:43994", + "issue:44075", + "issue:44155", + "issue:44164", + "issue:44291", + "issue:44387", + "issue:44402", + "issue:44488", + "issue:44756", + "issue:44910", + "issue:44960", + "issue:44991", + "issue:45276", + "issue:45292", + "issue:45313", + "issue:45335", + "issue:45357", + "issue:45362", + "issue:45406", + "issue:45488", + "issue:45507", + "issue:45588", + "issue:45663", + "issue:45704", + "issue:45759", + "issue:45812", + "issue:45859", + "issue:46050", + "issue:46468" + ], + "soft_pairs": [ + "issue:45313|issue:45357", + "issue:45507|issue:46468", + "issue:44488|issue:44991", + "issue:43856|issue:43873", + "issue:43644|issue:43645", + "issue:39692|issue:43994", + "issue:43708|issue:44075", + "issue:43856|issue:44387", + "issue:43257|issue:43854", + "issue:43854|issue:43931", + "issue:42491|issue:45292", + "issue:42491|issue:43931", + "issue:44960|issue:46050", + "issue:45488|issue:45812", + "issue:45276|issue:45335", + "issue:44164|issue:44291", + "issue:45362|issue:45406", + "issue:44910|issue:45588", + "issue:44402|issue:44488", + "issue:43404|issue:43454", + "issue:42947|issue:45704", + "issue:39692|issue:44960", + "issue:43873|issue:44387", + "issue:41720|issue:44155", + "issue:43425|issue:43957", + "issue:45663|issue:45759", + "issue:44291|issue:44756", + "issue:36010|issue:39404", + "issue:45663|issue:45859", + "issue:44387|issue:44756" + ], + "prepared_review_unit_hash": "fbfd6a95b5a37acdc81e5b1339b08b6ff354e4a05194820aa552eacfe4e840b2", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36207, + "estimated_input_tokens": 9052, + "estimated_eval_tokens": 18360 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most candidates are thematically related but not true duplicates. The only clear duplicate pair is the resize_token_embeddings regression across Gemma4/T5Gemma; the rest differ in model family, code path, or failure mode.", + "confidence": 0.74, + "canonical_issue_reason": "issue:45276 is the best canonical issue: it describes the broader resize_token_embeddings propagation bug and cleanly covers the t5gemma variant.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45276 is the strongest representative issue overall because it is the most general statement of the shared embedding-resize bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45313", + "right": "issue:45357", + "accept": false, + "reason": "Same model family, but one is a ZeRO-3 weight-loading failure and the other is a save_pretrained key regression; different code paths and symptoms." + }, + { + "left": "issue:45507", + "right": "issue:46468", + "accept": false, + "reason": "Different models and failures; too broad to be the same underlying bug." + }, + { + "left": "issue:44488", + "right": "issue:44991", + "accept": false, + "reason": "Both concern tokenizer/model loading in v5, but they affect different models and likely different tokenizer failure modes." + }, + { + "left": "issue:43856", + "right": "issue:43873", + "accept": false, + "reason": "Both are memory-related, but one is Qwen3 MoE training memory usage and the other is quantization/offloading behavior; not the same bug." + }, + { + "left": "issue:43644", + "right": "issue:43645", + "accept": false, + "reason": "Both are Transformers 5.0 regressions, but one is buffer initialization and the other is notebook custom model init; different issues." + }, + { + "left": "issue:39692", + "right": "issue:43994", + "accept": false, + "reason": "Both mention SigLIP2, but one is a documentation example with multiple errors and the other is incorrect outputs in AutoModel/pipeline usage." + }, + { + "left": "issue:43708", + "right": "issue:44075", + "accept": false, + "reason": "Trainer checkpoint math and SGD argument handling are unrelated." + }, + { + "left": "issue:43856", + "right": "issue:44387", + "accept": false, + "reason": "Both involve memory, but one is training-time inefficiency and the other is int4 reserved-memory/OOM regression; not the same bug." + }, + { + "left": "issue:43257", + "right": "issue:43854", + "accept": false, + "reason": "Different models and load paths; no concrete shared bug is evident." + }, + { + "left": "issue:43854", + "right": "issue:43931", + "accept": false, + "reason": "Both are model-loading problems, but for different Qwen/GLM-VL models and different failure signatures." + }, + { + "left": "issue:42491", + "right": "issue:45292", + "accept": false, + "reason": "LoRA/Qwen3 MoE compatibility is unrelated to resize_token_embeddings behavior." + }, + { + "left": "issue:42491", + "right": "issue:43931", + "accept": false, + "reason": "Different models and failures; not the same bug." + }, + { + "left": "issue:44960", + "right": "issue:46050", + "accept": false, + "reason": "GLM5 loading/model issue vs sinq quantization persistence bug; unrelated." + }, + { + "left": "issue:45488", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions, but they involve different tokenizer classes, models, and failure modes." + }, + { + "left": "issue:45276", + "right": "issue:45335", + "accept": true, + "reason": "Near-paraphrase of the same bug: resize_token_embeddings does not propagate to model-specific embedding/output modules in both Gemma4 and T5Gemma." + }, + { + "left": "issue:44164", + "right": "issue:44291", + "accept": false, + "reason": "save/from_pretrained extra_state handling and init_empty_weights argument compatibility are separate loading bugs." + }, + { + "left": "issue:45362", + "right": "issue:45406", + "accept": false, + "reason": "Chat crash on Qwen3.5-35B and Gemma4Processor missing _tokenizer in serve are different subsystems and errors." + }, + { + "left": "issue:44910", + "right": "issue:45588", + "accept": false, + "reason": "Both touch flash_attention, but one is a position_ids packing crash and the other is an s_aux=None AttributeError; different code paths." + }, + { + "left": "issue:44402", + "right": "issue:44488", + "accept": false, + "reason": "Different models and different symptoms; not enough evidence of the same root cause." + }, + { + "left": "issue:43404", + "right": "issue:43454", + "accept": false, + "reason": "Both mention untied lm_head weights, but they target different model classes and lack evidence of a shared concrete code-path bug." + }, + { + "left": "issue:42947", + "right": "issue:45704", + "accept": false, + "reason": "Gradient checkpointing/LoRA behavior and an apex RMSNorm memory leak are unrelated." + }, + { + "left": "issue:39692", + "right": "issue:44960", + "accept": false, + "reason": "Unrelated model families and failure modes." + }, + { + "left": "issue:43873", + "right": "issue:44387", + "accept": false, + "reason": "Both are quantization/memory themed, but one is offloading not working and the other is increased reserved memory/OOM; not the same bug." + }, + { + "left": "issue:41720", + "right": "issue:44155", + "accept": false, + "reason": "Qwen3 auto device mapping CUDA assert and AudioFlamingo3 batched inference leakage are unrelated." + }, + { + "left": "issue:43425", + "right": "issue:43957", + "accept": false, + "reason": "Torch 2.10 incompatibility and meta-device loading regressions are distinct compatibility issues." + }, + { + "left": "issue:45663", + "right": "issue:45759", + "accept": false, + "reason": "Different models and different failure mechanisms; not the same underlying bug." + }, + { + "left": "issue:44291", + "right": "issue:44756", + "accept": false, + "reason": "init_empty_weights loading error and Strix Halo mmap OOM are unrelated." + }, + { + "left": "issue:36010", + "right": "issue:39404", + "accept": false, + "reason": "ImportError for GenerationMixin and Whisper return_language pipeline regression are different issues." + }, + { + "left": "issue:45663", + "right": "issue:45859", + "accept": false, + "reason": "Different Qwen/Gemma model code paths and different offload/attention issues." + }, + { + "left": "issue:44387", + "right": "issue:44756", + "accept": false, + "reason": "Both are memory problems, but one is int4 quantization reserved-memory growth and the other is mmap on Strix Halo; not the same bug." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The main duplicate call is well grounded by the titles: both 45276 and 45335 describe resize_token_embeddings not propagating to model-specific embedding/output modules. The remaining pairwise rejections are appropriately conservative from the packet evidence, with no obvious overstatement of shared root cause. " + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 8, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:39404", + "issue:42796", + "issue:42947", + "issue:43335", + "issue:43381", + "issue:43425", + "issue:43611", + "issue:43645", + "issue:43646", + "issue:43708", + "issue:43716", + "issue:43828", + "issue:43854", + "issue:43906", + "issue:43957", + "issue:44155", + "issue:44164", + "issue:44292", + "issue:44402", + "issue:44451", + "issue:44479", + "issue:44530", + "issue:44589", + "issue:44683", + "issue:44756", + "issue:44805", + "issue:44863", + "issue:44877", + "issue:44912", + "issue:44928", + "issue:45084", + "issue:45216", + "issue:45276", + "issue:45292", + "issue:45335", + "issue:45588", + "issue:45663", + "issue:45684", + "issue:45685", + "issue:45698", + "issue:45701", + "issue:45812", + "issue:45823", + "issue:46315", + "issue:46326", + "issue:46468", + "issue:46491", + "issue:46665" + ], + "soft_pairs": [ + "issue:44292|issue:44912", + "issue:43335|issue:45663", + "issue:44155|issue:45823", + "issue:42947|issue:43381", + "issue:43957|issue:44756", + "issue:45276|issue:45292", + "issue:44863|issue:46468", + "issue:45685|issue:46468", + "issue:44479|issue:44877", + "issue:43381|issue:44928", + "issue:44164|issue:44756", + "issue:45684|issue:45698", + "issue:43854|issue:45216", + "issue:45588|issue:45663", + "issue:42796|issue:46326", + "issue:45292|issue:45335", + "issue:43646|issue:43708", + "issue:43828|issue:43957", + "issue:44402|issue:44451", + "issue:43645|issue:44164", + "issue:44589|issue:44756", + "issue:43425|issue:43828", + "issue:45812|issue:46315", + "issue:46315|issue:46491", + "issue:39404|issue:45701", + "issue:43611|issue:43645", + "issue:39404|issue:46665", + "issue:44530|issue:45084", + "issue:44683|issue:45823", + "issue:43425|issue:45823", + "issue:43425|issue:43716", + "issue:44479|issue:44805", + "issue:43906|issue:44530" + ], + "prepared_review_unit_hash": "c1d63970f65c4989610e2fa1b8d0abe1e1e2e779da60fa19ff839300eb23eb2d", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37582, + "estimated_input_tokens": 9396, + "estimated_eval_tokens": 19048 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most items are unrelated. The only clear duplicate subcluster is the `resize_token_embeddings` regression, where a generic report and two model-specific reports point to the same embedding-update bug.", + "confidence": 0.73, + "canonical_issue_reason": "issue:45292 is the broadest formulation of the `resize_token_embeddings` bug: output embeddings are not updated. The related Gemma4 and T5Gemma reports look like concrete manifestations of the same missing propagation.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45292 is the best representative because it states the underlying bug at the right level of generality and can subsume the more specific model variants.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44292", + "right": "issue:44912", + "accept": false, + "reason": "Both are quantization/load failures, but they involve different models and different quantization setups; no clear same code-path bug." + }, + { + "left": "issue:43335", + "right": "issue:45663", + "accept": false, + "reason": "Unrelated symptoms: SwitchTransformers layer creation vs Gemma4 FSDP2 shared_kv_states handling." + }, + { + "left": "issue:44155", + "right": "issue:45823", + "accept": false, + "reason": "Different models and failures: track leakage in batched audio inference vs Gemma4 device placement mismatch." + }, + { + "left": "issue:42947", + "right": "issue:43381", + "accept": false, + "reason": "Both mention gradient checkpointing, but one is about ineffectiveness with LoRA and the other about eval-mode restrictions." + }, + { + "left": "issue:43957", + "right": "issue:44756", + "accept": false, + "reason": "Meta-device loading bug vs Strix Halo mmap/OOM issue; different root causes." + }, + { + "left": "issue:45276", + "right": "issue:45292", + "accept": true, + "reason": "Same underlying `resize_token_embeddings` propagation bug; 45276 is a Gemma4-specific manifestation of the generic output-embeddings issue." + }, + { + "left": "issue:44863", + "right": "issue:46468", + "accept": false, + "reason": "Both are Nemotron-related, but one is checkpoint loading and the other is an example breakage; too little evidence of the same bug." + }, + { + "left": "issue:45685", + "right": "issue:46468", + "accept": false, + "reason": "No common failure mode beyond model family name." + }, + { + "left": "issue:44479", + "right": "issue:44877", + "accept": false, + "reason": "Different subsystems: video-input regression vs strict config loading." + }, + { + "left": "issue:43381", + "right": "issue:44928", + "accept": false, + "reason": "Gradient-checkpointing eval-mode behavior is unrelated to the RLHF NaN / SDPA fallback regression." + }, + { + "left": "issue:44164", + "right": "issue:44756", + "accept": false, + "reason": "Serialization extra_state handling and mmap/OOM are unrelated." + }, + { + "left": "issue:45684", + "right": "issue:45698", + "accept": false, + "reason": "Both involve save/from_pretrained around custom modules, but one is a permissions issue and the other is loading the wrong module." + }, + { + "left": "issue:43854", + "right": "issue:45216", + "accept": false, + "reason": "Different models and different load/save failures; no shared underlying bug is evident." + }, + { + "left": "issue:45588", + "right": "issue:45663", + "accept": false, + "reason": "Flash-attention sink-less-model crash vs Gemma4 FSDP2 key error are different code paths." + }, + { + "left": "issue:42796", + "right": "issue:46326", + "accept": false, + "reason": "Tokenizer-related, but one is BPE marker handling and the other is `continue_final_message` behavior in mistral_common tokenizers." + }, + { + "left": "issue:45292", + "right": "issue:45335", + "accept": true, + "reason": "Same core regression: resizing token embeddings does not propagate to model embedding/output layers; these are architecture-specific instances." + }, + { + "left": "issue:43646", + "right": "issue:43708", + "accept": false, + "reason": "Custom model initialization in notebooks is unrelated to Trainer checkpoint resume max_steps calculation." + }, + { + "left": "issue:43828", + "right": "issue:43957", + "accept": false, + "reason": "Autocast dtype mismatch for a specific model is not the same as meta-device loading failures." + }, + { + "left": "issue:44402", + "right": "issue:44451", + "accept": false, + "reason": "One is a tokenizer vocab-size inconsistency, the other is a model-loading failure for a specific checkpoint." + }, + { + "left": "issue:43645", + "right": "issue:44164", + "accept": false, + "reason": "Notebook-defined custom model initialization and extra_state save/load handling are different bugs." + }, + { + "left": "issue:44589", + "right": "issue:44756", + "accept": false, + "reason": "Float8 storage lookup failure is unrelated to mmap/OOM." + }, + { + "left": "issue:43425", + "right": "issue:43828", + "accept": false, + "reason": "Generic Torch 2.10 incompatibility is too broad and not the same concrete dtype-mismatch bug." + }, + { + "left": "issue:45812", + "right": "issue:46315", + "accept": false, + "reason": "Both are tokenizer regressions, but one is wrong token IDs and the other is a performance regression in `convert_tokens_to_ids`." + }, + { + "left": "issue:46315", + "right": "issue:46491", + "accept": false, + "reason": "Performance regression vs encode/decode correctness regression; not the same underlying change." + }, + { + "left": "issue:39404", + "right": "issue:45701", + "accept": false, + "reason": "Whisper pipeline `return_language` regression and generic tokenization changes are not the same bug." + }, + { + "left": "issue:43611", + "right": "issue:43645", + "accept": false, + "reason": "Both touch custom-model loading in Transformers 5, but the failure modes are different enough that they should stay separate." + }, + { + "left": "issue:39404", + "right": "issue:46665", + "accept": false, + "reason": "Whisper pipeline behavior and `from_pretrained` inside Python classes do not point to the same code path." + }, + { + "left": "issue:44530", + "right": "issue:45084", + "accept": false, + "reason": "Qwen3.5 PagedAttentionCache crash and compilation/template-node type error are unrelated." + }, + { + "left": "issue:44683", + "right": "issue:45823", + "accept": false, + "reason": "Compiled flex_attention on torch>=2.9 is unrelated to Gemma4 device-map mismatch." + }, + { + "left": "issue:43425", + "right": "issue:45823", + "accept": false, + "reason": "Broad Torch 2.10 incompatibility does not match the specific Gemma4 PLE device placement bug." + }, + { + "left": "issue:43425", + "right": "issue:43716", + "accept": false, + "reason": "Torch compatibility and Mistral-3 image preprocessor dtype mismatch are distinct issues." + }, + { + "left": "issue:43906", + "right": "issue:44530", + "accept": false, + "reason": "The reproduction issue does not clearly match the Qwen3.5 linear_attention group-type crash; insufficient evidence to merge." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The clustering is conservative and grounded in the packet. The only clearly supported duplicate subcluster is the `resize_token_embeddings` regression around output/decoder embeddings, and the analyst\u2019s canonical choice of issue:45292 is reasonable. The remaining pairwise rejects avoid overclaiming shared root causes and stay within the evidence from titles/issue descriptions." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 9, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:36032", + "issue:39404", + "issue:42796", + "issue:42915", + "issue:43299", + "issue:43404", + "issue:43425", + "issue:43452", + "issue:43582", + "issue:43596", + "issue:43611", + "issue:43632", + "issue:43643", + "issue:43644", + "issue:43716", + "issue:43856", + "issue:43901", + "issue:43906", + "issue:43931", + "issue:43994", + "issue:44079", + "issue:44164", + "issue:44265", + "issue:44462", + "issue:44479", + "issue:44488", + "issue:44509", + "issue:44530", + "issue:44589", + "issue:44661", + "issue:44671", + "issue:44683", + "issue:44737", + "issue:44756", + "issue:44898", + "issue:44912", + "issue:44918", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45759", + "issue:45859", + "issue:46050", + "issue:46285", + "issue:46326", + "issue:46402", + "issue:46489" + ], + "soft_pairs": [ + "issue:43901|issue:44509", + "issue:44589|issue:45084", + "issue:44265|issue:44898", + "issue:43994|issue:44079", + "issue:44671|issue:44991", + "issue:43425|issue:44918", + "issue:39404|issue:44912", + "issue:36032|issue:43452", + "issue:46050|issue:46326", + "issue:44164|issue:45759", + "issue:43299|issue:43931", + "issue:43906|issue:46326", + "issue:44991|issue:45081", + "issue:43856|issue:44756", + "issue:43596|issue:44683", + "issue:44164|issue:44479", + "issue:44462|issue:46489", + "issue:44488|issue:45081", + "issue:44488|issue:44987", + "issue:43611|issue:43644", + "issue:42796|issue:43404", + "issue:39404|issue:44671", + "issue:44737|issue:45859", + "issue:46285|issue:46402", + "issue:36032|issue:42915", + "issue:43716|issue:45084", + "issue:44661|issue:44683", + "issue:43643|issue:44530", + "issue:43582|issue:43632" + ], + "prepared_review_unit_hash": "88d39897d6778fa7dc11d279943763a8fca88842f0f1cc43f3d2408413b10063", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36445, + "estimated_input_tokens": 9112, + "estimated_eval_tokens": 18480 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a mix of unrelated Transformers v5 regressions: tokenizer loading, model/config loading, quantization, and a few docs issues. The soft-similarity pairs mostly share broad subsystem vocabulary but not the same concrete bug or fix path.", + "confidence": 0.88, + "canonical_issue_reason": "issue:43611 is the broadest, most central report in the set: a general Transformers v5 model-loading regression. It best represents the cluster\u2019s common theme of v5 breakages, even though many other issues are narrower and model/tokenizer-specific.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43611 is the best overall representative because it captures the overarching v5 loading regression pattern that many of the other issues are variants of, whereas the others are more specialized failure modes.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43901", + "right": "issue:44509", + "accept": false, + "reason": "Both are docs-related, but one is about return_all_scores and the other about removed pipeline tasks; different documentation bugs." + }, + { + "left": "issue:44589", + "right": "issue:45084", + "accept": false, + "reason": "Different failures: Float8 storage lookup vs template compilation error. No shared code-path bug." + }, + { + "left": "issue:44265", + "right": "issue:44898", + "accept": false, + "reason": "torch.export/export hook failure and Perceiver interpolation failure are unrelated model paths." + }, + { + "left": "issue:43994", + "right": "issue:44079", + "accept": false, + "reason": "One is a SigLIP2 bad-output regression, the other is ModelOutput key assignment; different underlying bugs." + }, + { + "left": "issue:44671", + "right": "issue:44991", + "accept": false, + "reason": "Different models and symptoms: CamemBERT MLM predictions vs tokenizer loading for est-roberta." + }, + { + "left": "issue:43425", + "right": "issue:44918", + "accept": false, + "reason": "Torch version incompatibility is unrelated to Qwen3.5 TRL embedding unpacking." + }, + { + "left": "issue:39404", + "right": "issue:44912", + "accept": false, + "reason": "Whisper pipeline return_language regression and MXFP4 fallback are separate issues." + }, + { + "left": "issue:36032", + "right": "issue:43452", + "accept": false, + "reason": "T5Tokenizer method conflict and gguf_file loading failures are different tokenizer/model loading bugs." + }, + { + "left": "issue:46050", + "right": "issue:46326", + "accept": false, + "reason": "Quantized save/load breakage and mistral_common continue_final_message handling are unrelated." + }, + { + "left": "issue:44164", + "right": "issue:45759", + "accept": false, + "reason": "extra_state save/load handling is different from unwrapping text_config in composite configs." + }, + { + "left": "issue:43299", + "right": "issue:43931", + "accept": false, + "reason": "Both involve Qwen3-VL loading, but one is a dev0 regression and the other is a specific weight-shape mismatch; not the same concrete bug." + }, + { + "left": "issue:43906", + "right": "issue:46326", + "accept": false, + "reason": "Both mention tokenizer/chat behavior, but they are different reproductions and code paths." + }, + { + "left": "issue:44991", + "right": "issue:45081", + "accept": false, + "reason": "Different tokenizer bugs: est-roberta loading vs Mistral regex patch crash." + }, + { + "left": "issue:43856", + "right": "issue:44756", + "accept": false, + "reason": "Memory inefficiency during Qwen3 MoE training is unrelated to disabling mmap on Strix Halo." + }, + { + "left": "issue:43596", + "right": "issue:44683", + "accept": false, + "reason": "DeepSpeed ZeRO3/BertModel indexing error and compiled flex_attention on new torch are unrelated." + }, + { + "left": "issue:44164", + "right": "issue:44479", + "accept": false, + "reason": "extra_state persistence and Qwen video-input regression are separate failures." + }, + { + "left": "issue:44462", + "right": "issue:46489", + "accept": false, + "reason": "Tokenizer.json being ignored is not the same as DeepSeek-Coder output corruption after a prior fix." + }, + { + "left": "issue:44488", + "right": "issue:45081", + "accept": false, + "reason": "Both are tokenizer-loading issues, but for different models and different failure points." + }, + { + "left": "issue:44488", + "right": "issue:44987", + "accept": false, + "reason": "Different model-specific load failures with no shared root cause evident." + }, + { + "left": "issue:43611", + "right": "issue:43644", + "accept": false, + "reason": "Both are v5 regressions, but one is model loading/base_model_prefix and the other is non-persistent buffer initialization." + }, + { + "left": "issue:42796", + "right": "issue:43404", + "accept": false, + "reason": "Tokenizer BPE marker handling and Mistral lm_head tying are unrelated bugs." + }, + { + "left": "issue:39404", + "right": "issue:44671", + "accept": false, + "reason": "Whisper pipeline language return and CamemBERT MLM predictions are different behaviors in different models." + }, + { + "left": "issue:44737", + "right": "issue:45859", + "accept": false, + "reason": "CPU arange placement bug in XLNet and CPU-offload incompatibility in Qwen3_5Moe are separate issues." + }, + { + "left": "issue:46285", + "right": "issue:46402", + "accept": false, + "reason": "CLIP structure inconsistency and Sam3 text encoder weight loading are unrelated model-specific regressions." + }, + { + "left": "issue:36032", + "right": "issue:42915", + "accept": false, + "reason": "T5 tokenizer method conflict and Qwen3 MoE FineGrainedFP8 failure are different code paths." + }, + { + "left": "issue:43716", + "right": "issue:45084", + "accept": false, + "reason": "Mistral image-preprocessor dtype mismatch is unrelated to template-node compilation failures." + }, + { + "left": "issue:44661", + "right": "issue:44683", + "accept": false, + "reason": "Tokenizer mapping-name insertion bug and compiled flex_attention regression are distinct problems." + }, + { + "left": "issue:43643", + "right": "issue:44530", + "accept": false, + "reason": "AutoConfig trust_remote_code field loss and Qwen3.5 paged attention group-type crash are unrelated." + }, + { + "left": "issue:43582", + "right": "issue:43632", + "accept": false, + "reason": "AppleSilicon warmup TypeError and _is_hf_initialized breakage are different regressions." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 10, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:28530", + "issue:34567", + "issue:36032", + "issue:40444", + "issue:40990", + "issue:42175", + "issue:42915", + "issue:43089", + "issue:43262", + "issue:43317", + "issue:43377", + "issue:43404", + "issue:43421", + "issue:43452", + "issue:43482", + "issue:43541", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43742", + "issue:43746", + "issue:43819", + "issue:43883", + "issue:43994", + "issue:44206", + "issue:44220", + "issue:44265", + "issue:44291", + "issue:44403", + "issue:44451", + "issue:44464", + "issue:44466", + "issue:44488", + "issue:44530", + "issue:44534", + "issue:44610", + "issue:44811", + "issue:44849", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45325", + "issue:45507", + "issue:45584", + "issue:45987", + "issue:46178", + "issue:46326" + ], + "soft_pairs": [ + "issue:40990|issue:43421", + "issue:42915|issue:44530", + "issue:44451|issue:44987", + "issue:43541|issue:45084", + "issue:34567|issue:40444", + "issue:43644|issue:43646", + "issue:42175|issue:46178", + "issue:44811|issue:45584", + "issue:44206|issue:44220", + "issue:43883|issue:44488", + "issue:40444|issue:46326", + "issue:36032|issue:43645", + "issue:43452|issue:43482", + "issue:43317|issue:43482", + "issue:45084|issue:45507", + "issue:44987|issue:44991", + "issue:43262|issue:43377", + "issue:43746|issue:46326", + "issue:43883|issue:44991", + "issue:44464|issue:45987", + "issue:43646|issue:44534", + "issue:44403|issue:45081", + "issue:42915|issue:45084", + "issue:43645|issue:44534", + "issue:43404|issue:44466", + "issue:44849|issue:45325", + "issue:43377|issue:43819", + "issue:43994|issue:44610", + "issue:28530|issue:34567", + "issue:44451|issue:45081", + "issue:44987|issue:45081", + "issue:42915|issue:44291", + "issue:43089|issue:45987", + "issue:44265|issue:44610", + "issue:36032|issue:43742" + ], + "prepared_review_unit_hash": "f541d63ac5e6efa2b0ab2e75169a6a49ce75eb46442fc8cbe11a5c9fd6b25f6e", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 38403, + "estimated_input_tokens": 9601, + "estimated_eval_tokens": 19458 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a collection of unrelated closed issues that only look superficially similar because many involve loading or model-regression symptoms. The only soft pair that looks like the same underlying bug is the GGUF loading regression pair.", + "confidence": 0.82, + "canonical_issue_reason": "issue:43452 is the best canonical issue because it describes the broadest, most concrete failure mode in the cluster: GGUF loading breaks across both tokenizer and causal LM from_pretrained paths. It is more general than the specific Qwen2.5-GGUF report and best represents that duplicate subgroup.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43452 is the strongest issue candidate overall: it is the most general report among the only clearly related pair, and its wording matches a reusable code-path regression rather than a model-specific symptom.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:40990", + "right": "issue:43421", + "accept": false, + "reason": "Different symptoms and code paths: perplexity/eval behavior versus runtime special-token post-processing updates." + }, + { + "left": "issue:42915", + "right": "issue:44530", + "accept": false, + "reason": "Both mention Qwen models, but one is an FP8 loading/config failure and the other is a paged-attention cache/group-type crash; not the same bug." + }, + { + "left": "issue:44451", + "right": "issue:44987", + "accept": false, + "reason": "Both are model-loading failures, but they concern different models and likely different regressions." + }, + { + "left": "issue:43541", + "right": "issue:45084", + "accept": false, + "reason": "A torch dynamo grouped_mm tracing error is unrelated to a template compilation TypeError." + }, + { + "left": "issue:34567", + "right": "issue:40444", + "accept": false, + "reason": "Trainer token-count tracking and multimodal iterable-dataset finetuning are unrelated." + }, + { + "left": "issue:43644", + "right": "issue:43646", + "accept": false, + "reason": "Both are v5 regressions, but one is about non-persistent buffers being corrupted while the other is about custom model initialization; too broad and not the same concrete bug." + }, + { + "left": "issue:42175", + "right": "issue:46178", + "accept": false, + "reason": "Both concern backend/package handling, but one is missing tensorflow in a torch extra and the other is a missing tensorflow_text mapping entry." + }, + { + "left": "issue:44811", + "right": "issue:45584", + "accept": false, + "reason": "Both involve Whisper, but one is batch_decode skip_special_tokens handling and the other is empty-transcription generation after align_special_tokens." + }, + { + "left": "issue:44206", + "right": "issue:44220", + "accept": false, + "reason": "Both are audio-feature-related, but the titles point to different failures and there is not enough evidence they share the same exact code-path bug." + }, + { + "left": "issue:43883", + "right": "issue:44488", + "accept": false, + "reason": "One is a missing attribute on MolmoForCausalLM and the other is a specific tokenizer/model loading failure; not the same issue." + }, + { + "left": "issue:40444", + "right": "issue:46326", + "accept": false, + "reason": "Different domains: multimodal finetuning with iterable datasets versus mistral_common continue_final_message behavior." + }, + { + "left": "issue:36032", + "right": "issue:43645", + "accept": false, + "reason": "Tokenizer method-name conflict and Jupyter custom-model initialization are unrelated." + }, + { + "left": "issue:43452", + "right": "issue:43482", + "accept": true, + "reason": "Both describe the same GGUF loading regression in transformers v5, with 43482 appearing to be a model-specific instance of the broader from_pretrained GGUF failure in 43452." + }, + { + "left": "issue:43317", + "right": "issue:43482", + "accept": false, + "reason": "Device-map/offload loading failure and GGUF loading failure are different code paths." + }, + { + "left": "issue:45084", + "right": "issue:45507", + "accept": false, + "reason": "Template-node compilation error and GraniteMoEHybrid invalid-method call are different failures." + }, + { + "left": "issue:44987", + "right": "issue:44991", + "accept": false, + "reason": "Both are loading regressions, but they affect different models/tokenizers and do not look like one concrete bug." + }, + { + "left": "issue:43262", + "right": "issue:43377", + "accept": false, + "reason": "Audio chat-template sampling-rate defaults and MIMI padding-mask batching are distinct issues." + }, + { + "left": "issue:43746", + "right": "issue:46326", + "accept": false, + "reason": "PEFT/local-checkpoint loading for GraniteSpeech is unrelated to mistral_common continue_final_message behavior." + }, + { + "left": "issue:43883", + "right": "issue:44991", + "accept": false, + "reason": "Missing all_tied_weights_keys and tokenizer loading for a different model are not the same problem." + }, + { + "left": "issue:44464", + "right": "issue:45987", + "accept": false, + "reason": "Chunked generation inconsistency under compiled forward and StaticCache return-type contract are related only at a very high level." + }, + { + "left": "issue:43646", + "right": "issue:44534", + "accept": false, + "reason": "Both mention v5/custom model behavior, but the specific symptoms differ enough that they should not be merged." + }, + { + "left": "issue:44403", + "right": "issue:45081", + "accept": false, + "reason": "Noise on load and a Mistral regex patch crash are different loading-related issues." + }, + { + "left": "issue:42915", + "right": "issue:45084", + "accept": false, + "reason": "Qwen3Moe FP8 failure and template compilation TypeError are unrelated." + }, + { + "left": "issue:43645", + "right": "issue:44534", + "accept": false, + "reason": "Custom model initialization in notebooks and non-persistent-buffer corruption are not the same bug." + }, + { + "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 a device-dependent serialization issue." + }, + { + "left": "issue:44849", + "right": "issue:45325", + "accept": false, + "reason": "Different Qwen model families and different failure modes." + }, + { + "left": "issue:43377", + "right": "issue:43819", + "accept": false, + "reason": "Batched-vs-single MIMI encoder outputs and DAC.from_latents/forward mismatch are unrelated approximation bugs." + }, + { + "left": "issue:43994", + "right": "issue:44610", + "accept": false, + "reason": "Nonsensical model outputs and wrong processor input size are different problems." + }, + { + "left": "issue:28530", + "right": "issue:34567", + "accept": false, + "reason": "Early-stopping metric selection and token-count tracking are unrelated Trainer issues." + }, + { + "left": "issue:44451", + "right": "issue:45081", + "accept": false, + "reason": "Different models and different tokenizer/loading failures." + }, + { + "left": "issue:44987", + "right": "issue:45081", + "accept": false, + "reason": "Model-loading regression for one model versus a Mistral tokenizer regex crash for another." + }, + { + "left": "issue:42915", + "right": "issue:44291", + "accept": false, + "reason": "FineGrainedFP8Config failure and init_empty_weights unexpected-argument failure are different loading paths." + }, + { + "left": "issue:43089", + "right": "issue:45987", + "accept": false, + "reason": "Generation overhead/per-token syncs is a performance issue, not a cache-length contract bug." + }, + { + "left": "issue:44265", + "right": "issue:44610", + "accept": false, + "reason": "torch.export/torch_compilable_check and a processor size mismatch are unrelated." + }, + { + "left": "issue:36032", + "right": "issue:43742", + "accept": false, + "reason": "Tokenizer method conflict and MobileLLM key error are not the same failure." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays appropriately conservative. The only accepted soft pair, issue:43452/issue:43482, is a plausible same-bug relation based on the shared GGUF loading regression, and the rejections avoid overclaiming beyond what the titles support." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 11, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:33666", + "issue:35707", + "issue:36032", + "issue:40444", + "issue:40990", + "issue:41553", + "issue:41720", + "issue:42796", + "issue:42915", + "issue:43296", + "issue:43322", + "issue:43381", + "issue:43421", + "issue:43425", + "issue:43482", + "issue:43525", + "issue:43531", + "issue:43540", + "issue:43582", + "issue:43611", + "issue:43632", + "issue:43746", + "issue:43792", + "issue:44062", + "issue:44079", + "issue:44220", + "issue:44291", + "issue:44403", + "issue:44479", + "issue:44534", + "issue:44568", + "issue:44589", + "issue:44683", + "issue:44756", + "issue:44811", + "issue:44877", + "issue:44987", + "issue:44991", + "issue:45216", + "issue:45325", + "issue:45357", + "issue:45538", + "issue:45663", + "issue:45812", + "issue:45923", + "issue:46315", + "issue:46491" + ], + "soft_pairs": [ + "issue:43322|issue:43540", + "issue:44062|issue:44987", + "issue:42915|issue:44589", + "issue:44479|issue:45325", + "issue:45216|issue:45357", + "issue:44991|issue:46315", + "issue:36032|issue:41720", + "issue:43531|issue:44877", + "issue:43611|issue:44534", + "issue:43425|issue:44403", + "issue:43296|issue:43531", + "issue:44220|issue:44683", + "issue:40444|issue:45325", + "issue:45812|issue:46491", + "issue:43632|issue:44756", + "issue:42796|issue:43746", + "issue:44403|issue:44991", + "issue:43381|issue:45923", + "issue:33666|issue:40444", + "issue:43792|issue:44589", + "issue:40990|issue:44811", + "issue:44079|issue:45663", + "issue:44291|issue:44589", + "issue:43421|issue:44568", + "issue:43632|issue:44534", + "issue:41553|issue:43582", + "issue:35707|issue:40444", + "issue:43482|issue:43525", + "issue:43482|issue:45538" + ], + "prepared_review_unit_hash": "39d0bedffdd40dd28aae295250ed72830d46afd68254a030171be55fd99ff64f", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35621, + "estimated_input_tokens": 8906, + "estimated_eval_tokens": 18068 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most candidate edges are false positives across unrelated bugs, models, or failure modes. The only clear duplicate is the Qwen3.5 `save_pretrained` checkpoint-regression pair.", + "confidence": 0.77, + "canonical_issue_reason": "Issue 45216 is the broader, earlier report of the Qwen3.5 checkpoint serialization regression; 45357 reads like a later, narrower manifestation focused on the visual encoder keys.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45216 is the best representative because it captures the full `save_pretrained` regression at the model level, while still matching the later visual-key corruption report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43322", + "right": "issue:43540", + "accept": false, + "reason": "Different multimodal models and different failure types; both are video/image-related but not the same code path." + }, + { + "left": "issue:44062", + "right": "issue:44987", + "accept": false, + "reason": "One is an AddedToken argument conflict, the other is a model-specific tokenizer load failure; no clear shared bug." + }, + { + "left": "issue:42915", + "right": "issue:44589", + "accept": false, + "reason": "Both involve FP8-related errors, but they point to different underlying failures and code paths." + }, + { + "left": "issue:44479", + "right": "issue:45325", + "accept": false, + "reason": "`44479` is a broad video-input regression across several models; `45325` is a specific still-image RoPE scaling bug in Qwen2.5-VL." + }, + { + "left": "issue:45216", + "right": "issue:45357", + "accept": true, + "reason": "Same model family and same `save_pretrained` checkpoint corruption regression; `45357` is a narrower follow-up about the visual encoder keys." + }, + { + "left": "issue:44991", + "right": "issue:46315", + "accept": false, + "reason": "Both are tokenizer issues in v5, but one is a load failure and the other is a performance regression in `convert_tokens_to_ids`." + }, + { + "left": "issue:36032", + "right": "issue:41720", + "accept": false, + "reason": "Unrelated failures: a T5 tokenizer attribute conflict versus a Qwen3 auto device-mapping CUDA assert." + }, + { + "left": "issue:43531", + "right": "issue:44877", + "accept": false, + "reason": "Different models and unrelated bugs; one is sliding-window handling, the other is config strictness." + }, + { + "left": "issue:43611", + "right": "issue:44534", + "accept": false, + "reason": "Both are v5 regressions, but they affect different loading/init paths and are not the same concrete bug." + }, + { + "left": "issue:43425", + "right": "issue:44403", + "accept": false, + "reason": "Torch-version incompatibility and noisy load logs are unrelated." + }, + { + "left": "issue:43296", + "right": "issue:43531", + "accept": false, + "reason": "Different model families and different failure modes; no shared underlying bug." + }, + { + "left": "issue:44220", + "right": "issue:44683", + "accept": false, + "reason": "Audio feature extraction issue versus compiled flex_attention failure; unrelated." + }, + { + "left": "issue:40444", + "right": "issue:45325", + "accept": false, + "reason": "Same model family but different symptoms and paths: multi-image iterable training failure versus still-image temporal-position scaling." + }, + { + "left": "issue:45812", + "right": "issue:46491", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different tokenizers and different behaviors." + }, + { + "left": "issue:43632", + "right": "issue:44756", + "accept": false, + "reason": "Unrelated: HF initialization flag regression versus mmap/OOM on Strix Halo." + }, + { + "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 load noise is not the same bug as a tokenizer load failure." + }, + { + "left": "issue:43381", + "right": "issue:45923", + "accept": false, + "reason": "One is about eval-mode gradient checkpointing misuse, the other is a missing capability flag on a trust_remote_code model." + }, + { + "left": "issue:33666", + "right": "issue:40444", + "accept": false, + "reason": "Both involve Qwen-VL training, but they target different models and different failure scenarios." + }, + { + "left": "issue:43792", + "right": "issue:44589", + "accept": false, + "reason": "Whisper load failure and missing Float8 storage type are unrelated." + }, + { + "left": "issue:40990", + "right": "issue:44811", + "accept": false, + "reason": "Perplexity regression and Whisper batch_decode special-token handling are unrelated." + }, + { + "left": "issue:44079", + "right": "issue:45663", + "accept": false, + "reason": "ModelOutput key assignment and Gemma4 FSDP2 attention state rebuilding are different bugs." + }, + { + "left": "issue:44291", + "right": "issue:44589", + "accept": false, + "reason": "Both are load-time TypeErrors, but they stem from different causes and code paths." + }, + { + "left": "issue:43421", + "right": "issue:44568", + "accept": false, + "reason": "Both mention special tokens, but one is runtime post-processor updates and the other is a tokenizer-specific BOS/EOS addition regression." + }, + { + "left": "issue:43632", + "right": "issue:44534", + "accept": false, + "reason": "Different v5 regressions affecting different internal mechanisms; not the same issue." + }, + { + "left": "issue:41553", + "right": "issue:43582", + "accept": false, + "reason": "Bad tokenizer error messaging and Apple Silicon allocator TypeError are unrelated." + }, + { + "left": "issue:35707", + "right": "issue:40444", + "accept": false, + "reason": "Progressive generation with `inputs_embeds`/`past_key_values` is unrelated to Qwen2.5-VL multi-image finetuning." + }, + { + "left": "issue:43482", + "right": "issue:43525", + "accept": false, + "reason": "Different model load failures: GGUF/v5 compatibility versus missing `pad_token_id` on Llama4Config." + }, + { + "left": "issue:43482", + "right": "issue:45538", + "accept": false, + "reason": "Unrelated tokenizer/load bugs; no shared underlying defect." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is well grounded in the packet, and the lone accepted edge (45216/45357) is conservative given both titles reference the same Qwen3.5 `save_pretrained` regression. The rejected pairs are also framed cautiously, mostly distinguishing by model, symptom, or code path rather than overstating duplication." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 12, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:40444", + "issue:41720", + "issue:42175", + "issue:42796", + "issue:42915", + "issue:43296", + "issue:43317", + "issue:43377", + "issue:43454", + "issue:43482", + "issue:43493", + "issue:43531", + "issue:43606", + "issue:43611", + "issue:43632", + "issue:43644", + "issue:43646", + "issue:43701", + "issue:43749", + "issue:43792", + "issue:43818", + "issue:43856", + "issue:43873", + "issue:43957", + "issue:43994", + "issue:44079", + "issue:44265", + "issue:44291", + "issue:44462", + "issue:44466", + "issue:44479", + "issue:44534", + "issue:44829", + "issue:44877", + "issue:45072", + "issue:45081", + "issue:45325", + "issue:45357", + "issue:45381", + "issue:45479", + "issue:45491", + "issue:45542", + "issue:45701", + "issue:45800", + "issue:45812", + "issue:45823", + "issue:45834", + "issue:46402" + ], + "soft_pairs": [ + "issue:43317|issue:43856", + "issue:42175|issue:45542", + "issue:43531|issue:44291", + "issue:44829|issue:45479", + "issue:44079|issue:44534", + "issue:44462|issue:45812", + "issue:43818|issue:46402", + "issue:41720|issue:45081", + "issue:43818|issue:45357", + "issue:43377|issue:45491", + "issue:43611|issue:43646", + "issue:43644|issue:44079", + "issue:43749|issue:44079", + "issue:44265|issue:45072", + "issue:42175|issue:43296", + "issue:45325|issue:45381", + "issue:43493|issue:43994", + "issue:40444|issue:45381", + "issue:43454|issue:44466", + "issue:45325|issue:45834", + "issue:43792|issue:43957", + "issue:42915|issue:43482", + "issue:43632|issue:45072", + "issue:42796|issue:45381", + "issue:43701|issue:45325", + "issue:44291|issue:44877", + "issue:43873|issue:45701", + "issue:43296|issue:45800", + "issue:43606|issue:44265", + "issue:44479|issue:45823" + ], + "prepared_review_unit_hash": "516d94f624fd8758911b145575a806e2ff0db83e5f30d47b0ce3fbf7752807d5", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36835, + "estimated_input_tokens": 9209, + "estimated_eval_tokens": 18674 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous: many unrelated Transformers bugs were grouped by loose title similarity. I only see two likely duplicate pairs: the Qwen2.5-VL position-id regressions and the SigLIP2 correctness regressions. No PRs are present.", + "confidence": 0.71, + "canonical_issue_reason": "Issue 43646 is the broadest representative of the set: it describes a general Transformers v5 model-initialization regression, whereas most other items are either unrelated or narrower model-specific bugs.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43646 is the best available issue representative because it is the most generic loading/initialization regression among the related v5-model issues; it is still not a true umbrella for the whole cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43317", + "right": "issue:43856", + "accept": false, + "reason": "Both mention offload/memory, but one is a device-map dequantized loading failure and the other is a Qwen3 MoE training memory issue." + }, + { + "left": "issue:42175", + "right": "issue:45542", + "accept": false, + "reason": "Both touch TensorFlow packaging, but one is a missing backend dependency in the torch extra and the other is a runtime undefined-tf-backend error with only tensorboard installed." + }, + { + "left": "issue:43531", + "right": "issue:44291", + "accept": false, + "reason": "Qwen3-MoE sliding-window handling and init_empty_weights/_is_hf_initialized are different code paths and different failure modes." + }, + { + "left": "issue:44829", + "right": "issue:45479", + "accept": false, + "reason": "Both are training regressions, but one is flash_attention_3 degenerate training and the other is single-label classification with num_labels=1 causing zero loss." + }, + { + "left": "issue:44079", + "right": "issue:44534", + "accept": false, + "reason": "ModelOutput key assignment and non-persistent buffer initialization are unrelated bugs." + }, + { + "left": "issue:44462", + "right": "issue:45812", + "accept": false, + "reason": "Both are AutoTokenizer regressions, but one is ignoring tokenizer.json and the other is wrong Granite token IDs; not enough to treat as the same bug." + }, + { + "left": "issue:43818", + "right": "issue:46402", + "accept": false, + "reason": "Video-LLaVA vision-tower/temporal-attention issues are unrelated to Sam3Model text-encoder weight loading." + }, + { + "left": "issue:41720", + "right": "issue:45081", + "accept": false, + "reason": "CUDA assert in Qwen3 auto device mapping is unrelated to a tokenizer regex patch AttributeError." + }, + { + "left": "issue:43818", + "right": "issue:45357", + "accept": false, + "reason": "Video-LLaVA architecture problems and Qwen3.5 save_pretrained key serialization are different bugs." + }, + { + "left": "issue:43377", + "right": "issue:45491", + "accept": false, + "reason": "Both involve batching/masks, but one is MIMI encoder padding-mask mismatch and the other is Gemma3 sliding-window NaNs on all-padding windows." + }, + { + "left": "issue:43611", + "right": "issue:43646", + "accept": false, + "reason": "Both are Transformers v5 loading/init regressions, but they are distinct concrete failures: base_model_prefix loading versus custom model initialization." + }, + { + "left": "issue:43644", + "right": "issue:44079", + "accept": false, + "reason": "Non-persistent buffer junk initialization is unrelated to ModelOutput key bookkeeping." + }, + { + "left": "issue:43749", + "right": "issue:44079", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and ModelOutput key assignment are different subsystems and bugs." + }, + { + "left": "issue:44265", + "right": "issue:45072", + "accept": false, + "reason": "torch.export with torch_compilable_check is unrelated to bfloat16 dtype mismatches in inference wrappers." + }, + { + "left": "issue:42175", + "right": "issue:43296", + "accept": false, + "reason": "TensorFlow backend packaging and PaddleOCR-VL load failures in vLLM are different problems." + }, + { + "left": "issue:45325", + "right": "issue:45381", + "accept": true, + "reason": "Both are Qwen2.5-VL 5.3.0 position-id regressions in the same RoPE/vision-position-id code path, one for still images and one for video input." + }, + { + "left": "issue:43493", + "right": "issue:43994", + "accept": true, + "reason": "Both report SigLIP2 producing incorrect outputs; one compares HF to JAX and the other shows nonsensical AutoModel/pipeline results, pointing to the same model implementation bug." + }, + { + "left": "issue:40444", + "right": "issue:45381", + "accept": false, + "reason": "IterableDataset with multiple images per prompt is a finetuning/input-packing bug, not the same as Qwen2.5-VL video position-id handling." + }, + { + "left": "issue:43454", + "right": "issue:44466", + "accept": false, + "reason": "Tied lm_head weight initialization and save_pretrained serialization inconsistency are related only at a broad theme level, not the same concrete bug." + }, + { + "left": "issue:45325", + "right": "issue:45834", + "accept": false, + "reason": "Qwen2.5-VL position-id scaling and Kosmos2.5 OCR index errors are unrelated." + }, + { + "left": "issue:43792", + "right": "issue:43957", + "accept": false, + "reason": "Whisper inference failure and meta-device model loading regressions are different model-loading issues." + }, + { + "left": "issue:42915", + "right": "issue:43482", + "accept": false, + "reason": "Qwen3MoE FP8 config failure and Qwen2.5 GGUF loading failure are unrelated." + }, + { + "left": "issue:43632", + "right": "issue:45072", + "accept": false, + "reason": "_is_hf_initialized flag handling and bfloat16 dtype mismatches are separate regressions." + }, + { + "left": "issue:42796", + "right": "issue:45381", + "accept": false, + "reason": "Tokenizer BPE-marker handling and Qwen2.5-VL video position ids are unrelated." + }, + { + "left": "issue:43701", + "right": "issue:45325", + "accept": false, + "reason": "Checkpoint key mismatch is unrelated to Qwen2.5-VL position-id scaling." + }, + { + "left": "issue:44291", + "right": "issue:44877", + "accept": false, + "reason": "init_empty_weights argument handling and strict Granite speech config loading are different bugs." + }, + { + "left": "issue:43873", + "right": "issue:45701", + "accept": false, + "reason": "Quantization/offloading behavior and tokenization changing across versions are not the same issue." + }, + { + "left": "issue:43296", + "right": "issue:45800", + "accept": false, + "reason": "PaddleOCR-VL load failure in vLLM and torch/transformers version incompatibility are unrelated." + }, + { + "left": "issue:43606", + "right": "issue:44265", + "accept": false, + "reason": "CPU offload device mismatch and torch.export/torch_compilable_check are different failure modes." + }, + { + "left": "issue:44479", + "right": "issue:45823", + "accept": false, + "reason": "Qwen video-input regression and Gemma4 device-map forward mismatch are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative: it identifies the two strongest duplicate-like pairs supported by the issue titles/body themes, and the other soft-edge rejections are appropriately cautious. The note that issue 43646 is only a rough representative, not a true umbrella, is also consistent with the heterogeneous cluster. No PR pair analysis was needed because the packet contains no PRs." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 13, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:34634", + "issue:35707", + "issue:39692", + "issue:40444", + "issue:40990", + "issue:41720", + "issue:42617", + "issue:42796", + "issue:42915", + "issue:42947", + "issue:43257", + "issue:43262", + "issue:43317", + "issue:43322", + "issue:43377", + "issue:43454", + "issue:43526", + "issue:43550", + "issue:43572", + "issue:43606", + "issue:43720", + "issue:43746", + "issue:43749", + "issue:43819", + "issue:43873", + "issue:43881", + "issue:43883", + "issue:44186", + "issue:44265", + "issue:44387", + "issue:44488", + "issue:44534", + "issue:44568", + "issue:44610", + "issue:44898", + "issue:44977", + "issue:44991", + "issue:45072", + "issue:45584", + "issue:45704", + "issue:45759", + "issue:45823", + "issue:45920", + "issue:45923", + "issue:46050", + "issue:46209", + "issue:46305", + "issue:46315" + ], + "soft_pairs": [ + "issue:43262|issue:43746", + "issue:40990|issue:45584", + "issue:43881|issue:44488", + "issue:34634|issue:35707", + "issue:35707|issue:41720", + "issue:43454|issue:43550", + "issue:43606|issue:43819", + "issue:39692|issue:40444", + "issue:44186|issue:44265", + "issue:41720|issue:45823", + "issue:46305|issue:46315", + "issue:43377|issue:44610", + "issue:44568|issue:45920", + "issue:43257|issue:45923", + "issue:42796|issue:46209", + "issue:43454|issue:45072", + "issue:43317|issue:43873", + "issue:43377|issue:44898", + "issue:43454|issue:43720", + "issue:40990|issue:42947", + "issue:42796|issue:44977", + "issue:43749|issue:44534", + "issue:42617|issue:42915", + "issue:43572|issue:45759", + "issue:43322|issue:46050", + "issue:43526|issue:43883", + "issue:41720|issue:43322", + "issue:43881|issue:44991", + "issue:44387|issue:45704" + ], + "prepared_review_unit_hash": "1a4f9c0d0607a85e8d8d9eb9786662653541fcc999353f4c622fee8ead7fc973", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36488, + "estimated_input_tokens": 9122, + "estimated_eval_tokens": 18500 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous: nearly all pairs are superficially similar topic-wise but do not look like the same concrete bug. The only plausible duplicate is the quantized/offloading loading failure pair (#43317 vs #43873).", + "confidence": 0.33, + "canonical_issue_reason": "#43873 is the broadest umbrella issue in the set and is the only one that plausibly subsumes a nearby duplicate around offloading + quantization loading failures.", + "canonical_pr_reason": null, + "best_issue_reason": "#43873 is the most representative issue for the small offloading/quantization subgroup; the rest of the cluster is too mixed to treat as one duplicate set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43262", + "right": "issue:43746", + "accept": false, + "reason": "Audio chat-template sampling-rate default vs PEFT/local-checkpoint loading failure; different code paths and bugs." + }, + { + "left": "issue:40990", + "right": "issue:45584", + "accept": false, + "reason": "Perplexity evaluation issue on GPT-OSS is unrelated to Whisper empty-transcription generation failure." + }, + { + "left": "issue:43881", + "right": "issue:44488", + "accept": false, + "reason": "Both are model-loading failures, but they involve different models and likely different root causes." + }, + { + "left": "issue:34634", + "right": "issue:35707", + "accept": false, + "reason": "Bark voice_preset handling vs progressive generation with inputs_embeds/past_key_values are unrelated bugs." + }, + { + "left": "issue:35707", + "right": "issue:41720", + "accept": false, + "reason": "Generation-cache behavior and CUDA assert under auto device mapping are different failures." + }, + { + "left": "issue:43454", + "right": "issue:43550", + "accept": false, + "reason": "lm_head weight tying bug vs torch.compile/SDPA failure are not the same underlying issue." + }, + { + "left": "issue:43606", + "right": "issue:43819", + "accept": false, + "reason": "CPU offload device mismatch in Bark is unrelated to DAC latent reconstruction/STE mismatch." + }, + { + "left": "issue:39692", + "right": "issue:40444", + "accept": false, + "reason": "Different multimodal examples and failures: doc example mismatch/quantization vs iterable-dataset multi-image finetuning." + }, + { + "left": "issue:44186", + "right": "issue:44265", + "accept": false, + "reason": "Tokenizer crash on NER/padding is unrelated to torch.export failures in torch_compilable_check models." + }, + { + "left": "issue:41720", + "right": "issue:45823", + "accept": false, + "reason": "Both mention device mapping, but the models and failure modes differ (CUDA assert vs forward-time device mismatch)." + }, + { + "left": "issue:46305", + "right": "issue:46315", + "accept": false, + "reason": "Streaming mask off-by-one bug is unrelated to tokenizer conversion performance regression." + }, + { + "left": "issue:43377", + "right": "issue:44610", + "accept": false, + "reason": "MIMI padding-mask correctness and OmDet-Turbo processor image-size mismatch are different subsystems and bugs." + }, + { + "left": "issue:44568", + "right": "issue:45920", + "accept": false, + "reason": "Both are tokenizer regressions, but one is missing BOS/EOS insertion and the other is wrong token IDs across many models." + }, + { + "left": "issue:43257", + "right": "issue:45923", + "accept": false, + "reason": "MOE weight conversion during accelerate+deepspeed loading is unrelated to a missing gradient-checkpointing flag." + }, + { + "left": "issue:42796", + "right": "issue:46209", + "accept": false, + "reason": "Tokenizer BPE-marker handling and FP8 inference on B200 are unrelated." + }, + { + "left": "issue:43454", + "right": "issue:45072", + "accept": false, + "reason": "Weight-tying regression is unrelated to bfloat16 dtype mismatches in inference." + }, + { + "left": "issue:43317", + "right": "issue:43873", + "accept": true, + "reason": "Both describe offloading failing for quantized/dequantized models under device_map/accelerate-style loading, plausibly the same underlying bug." + }, + { + "left": "issue:43377", + "right": "issue:44898", + "accept": false, + "reason": "Audio encoder padding-mask correctness is unrelated to Perceiver image-resolution handling." + }, + { + "left": "issue:43454", + "right": "issue:43720", + "accept": false, + "reason": "lm_head tying bug and packed-weight unpacking during accelerate loading are distinct issues." + }, + { + "left": "issue:40990", + "right": "issue:42947", + "accept": false, + "reason": "High perplexity evaluation on a model is unrelated to gradient checkpointing behavior with PEFT LoRA." + }, + { + "left": "issue:42796", + "right": "issue:44977", + "accept": false, + "reason": "Different tokenizer/regression bugs: BPE-marker handling vs flash-attention generation failure." + }, + { + "left": "issue:43749", + "right": "issue:44534", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and non-persistent buffer junk are different v5 regressions." + }, + { + "left": "issue:42617", + "right": "issue:42915", + "accept": false, + "reason": "3d_parallel.py runtime failure and FineGrainedFP8Config failure are unrelated." + }, + { + "left": "issue:43572", + "right": "issue:45759", + "accept": false, + "reason": "Missing pad_token_idx in a config is unrelated to not unwrapping text_config for composite multimodal configs." + }, + { + "left": "issue:43322", + "right": "issue:46050", + "accept": false, + "reason": "Llava Next segmentation fault and sinq save/load breakage are different failure modes and code paths." + }, + { + "left": "issue:43526", + "right": "issue:43883", + "accept": false, + "reason": "BEiT label reduction bug is unrelated to Molmo weight-tying attribute errors." + }, + { + "left": "issue:41720", + "right": "issue:43322", + "accept": false, + "reason": "Both are load-time crashes, but the models and root causes are different." + }, + { + "left": "issue:43881", + "right": "issue:44991", + "accept": false, + "reason": "Different tokenizer/model-loading regressions for different models; no same concrete bug is evident." + }, + { + "left": "issue:44387", + "right": "issue:45704", + "accept": false, + "reason": "Int4 CUDA reserved-memory OOM and T5 apex RMSNorm memory leak are separate quantization/runtime problems." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is well grounded in the packet and stays conservative. It correctly treats the cluster as heterogeneous and only elevates the 43317 \u2194 43873 pair, which is the one soft pair whose titles and descriptions both point to offloading/quantization loading behavior. The canonical/best-issue choice is subjective but not overstated, and there are no unsupported duplicate claims." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 14, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:26413", + "issue:28530", + "issue:33666", + "issue:36010", + "issue:36331", + "issue:40990", + "issue:41720", + "issue:42617", + "issue:42915", + "issue:43065", + "issue:43295", + "issue:43317", + "issue:43404", + "issue:43450", + "issue:43526", + "issue:43531", + "issue:43550", + "issue:43572", + "issue:43577", + "issue:43606", + "issue:43646", + "issue:43708", + "issue:43720", + "issue:43873", + "issue:43881", + "issue:44008", + "issue:44186", + "issue:44423", + "issue:44466", + "issue:44610", + "issue:44671", + "issue:44743", + "issue:44756", + "issue:44877", + "issue:44898", + "issue:44977", + "issue:44991", + "issue:45072", + "issue:45161", + "issue:45357", + "issue:45406", + "issue:45632", + "issue:45684", + "issue:45701", + "issue:45823", + "issue:46291", + "issue:46402", + "issue:46665" + ], + "soft_pairs": [ + "issue:44466|issue:45357", + "issue:36331|issue:42617", + "issue:41720|issue:44977", + "issue:36331|issue:42915", + "issue:43450|issue:45701", + "issue:26413|issue:28530", + "issue:44877|issue:44991", + "issue:43065|issue:46402", + "issue:40990|issue:45701", + "issue:44186|issue:45072", + "issue:43873|issue:44756", + "issue:41720|issue:42617", + "issue:43550|issue:43720", + "issue:43577|issue:43708", + "issue:43404|issue:45357", + "issue:43550|issue:43606", + "issue:43295|issue:45701", + "issue:43646|issue:46665", + "issue:44186|issue:44610", + "issue:44008|issue:45823", + "issue:44610|issue:44898", + "issue:42915|issue:43317", + "issue:44423|issue:45406", + "issue:36010|issue:42617", + "issue:45632|issue:45684", + "issue:41720|issue:45161", + "issue:28530|issue:33666", + "issue:44671|issue:44743", + "issue:43531|issue:43572", + "issue:42915|issue:46291", + "issue:43526|issue:43881", + "issue:36331|issue:41720" + ], + "prepared_review_unit_hash": "bad94399beb68e6f091b1af9e503609cff03e1431501106e9f9ee075fc8dc939", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37363, + "estimated_input_tokens": 9341, + "estimated_eval_tokens": 18938 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: the items span unrelated trainer, tokenizer, multimodal, quantization, loading, and model-specific bugs. None of the soft pairs look like true duplicates, so there is no safe canonical duplicate hub.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue is safe to choose because the set is a broad semantic bucket rather than one underlying bug; even the closest trainer-related items have different failure modes.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43708 is the strongest standalone report: it is concrete, reproducible, and narrowly scoped around a specific resume_from_checkpoint/max_steps bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44466", + "right": "issue:45357", + "accept": false, + "reason": "Both involve save/tied-weight behavior, but one is about lm_head serialization and the other about incorrect visual encoder keys; different code paths and symptoms." + }, + { + "left": "issue:36331", + "right": "issue:42617", + "accept": false, + "reason": "Unexpected compute_loss kwargs and failure to run 3d_parallel.py are unrelated trainer/runtime problems." + }, + { + "left": "issue:41720", + "right": "issue:44977", + "accept": false, + "reason": "Both mention Qwen models, but one is a device-mapping crash and the other is flash-attention generation behavior; not the same bug." + }, + { + "left": "issue:36331", + "right": "issue:42915", + "accept": false, + "reason": "A compute_loss signature mismatch is unrelated to a Qwen3-MoE fine-grained FP8 failure." + }, + { + "left": "issue:43450", + "right": "issue:45701", + "accept": false, + "reason": "Batched video-shape handling and version-dependent tokenization are different subsystems and failure modes." + }, + { + "left": "issue:26413", + "right": "issue:28530", + "accept": false, + "reason": "Both are trainer-related, but one is checkpoint resume failing on epoch iteration and the other is early stopping being disabled due to a missing metric." + }, + { + "left": "issue:44877", + "right": "issue:44991", + "accept": false, + "reason": "Strict config loading for granite_speech and tokenizer loading for est-roberta are unrelated bugs." + }, + { + "left": "issue:43065", + "right": "issue:46402", + "accept": false, + "reason": "A dummy Conv2d in Sam3PixelDecoder is not the same as missing text-encoder weights in Sam3Model." + }, + { + "left": "issue:40990", + "right": "issue:45701", + "accept": false, + "reason": "High perplexity on gpt-oss-20b and tokenization changes across versions are not the same underlying issue." + }, + { + "left": "issue:44186", + "right": "issue:45072", + "accept": false, + "reason": "LayoutLMv2 tokenizer crashes and dtype mismatches in bfloat16 inference are unrelated." + }, + { + "left": "issue:43873", + "right": "issue:44756", + "accept": false, + "reason": "Quantization/offload malfunction and mmap OOM on Strix Halo are distinct problems." + }, + { + "left": "issue:41720", + "right": "issue:42617", + "accept": false, + "reason": "Multi-GPU training for Qwen2-VL and a 3d_parallel.py runtime failure are not the same bug." + }, + { + "left": "issue:43550", + "right": "issue:43720", + "accept": false, + "reason": "torch.compile with SDPA on Bamba-9B-v2 and packed-weight handling in BitNet loading are different code paths." + }, + { + "left": "issue:43577", + "right": "issue:43708", + "accept": false, + "reason": "Blip2 dtype loading and Trainer resume/max_steps calculation are unrelated." + }, + { + "left": "issue:43404", + "right": "issue:45357", + "accept": false, + "reason": "Tied lm_head weights in Mistral3ForConditionalGeneration and incorrect visual encoder keys on save_pretrained are different issues." + }, + { + "left": "issue:43550", + "right": "issue:43606", + "accept": false, + "reason": "A compile/SDPA problem and a CPU-offload device mismatch are not the same defect." + }, + { + "left": "issue:43295", + "right": "issue:45701", + "accept": false, + "reason": "A regression in processor.tokenizer access is not the same as version-dependent tokenization changes." + }, + { + "left": "issue:43646", + "right": "issue:46665", + "accept": false, + "reason": "Both mention custom loading behavior, but one is broad Transformers 5.0 init breakage and the other is from_pretrained behavior inside Python classes." + }, + { + "left": "issue:44186", + "right": "issue:44610", + "accept": false, + "reason": "Tokenizer crashes on NER/padding and an image processor output-size mismatch are unrelated." + }, + { + "left": "issue:44008", + "right": "issue:45823", + "accept": false, + "reason": "Gemma3n variable-name collision in forward and Gemma4 PLE device mismatch are different bugs." + }, + { + "left": "issue:44610", + "right": "issue:44898", + "accept": false, + "reason": "Both involve image resolution, but one is wrong processor output size and the other is a model failure despite interpolate_pos_encoding." + }, + { + "left": "issue:42915", + "right": "issue:43317", + "accept": false, + "reason": "FineGrainedFP8Config failure and dequantized model offload loading failure are different quantization/device-map problems." + }, + { + "left": "issue:44423", + "right": "issue:45406", + "accept": false, + "reason": "Both are serve multimodal crashes, but the failing attributes and processor stacks differ, so they are not clearly the same bug." + }, + { + "left": "issue:36010", + "right": "issue:42617", + "accept": false, + "reason": "GenerationMixin import failure and 3d_parallel.py execution failure are unrelated." + }, + { + "left": "issue:45632", + "right": "issue:45684", + "accept": false, + "reason": "Cache-path collisions for trust_remote_code and read-only permission propagation in save_pretrained are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:45161", + "accept": false, + "reason": "Both involve distributed/parallel use, but they target different models and different failures." + }, + { + "left": "issue:28530", + "right": "issue:33666", + "accept": false, + "reason": "Early stopping metric handling and Qwen2-VL multi-GPU training are not the same issue." + }, + { + "left": "issue:44671", + "right": "issue:44743", + "accept": false, + "reason": "CamemBERT masked-LM regression and Qwen3.5 recurrent-state reset are unrelated model-specific bugs." + }, + { + "left": "issue:43531", + "right": "issue:43572", + "accept": false, + "reason": "Sliding-window behavior in Qwen3-MoE and a missing pad_token_idx in StableLmConfig are different regressions." + }, + { + "left": "issue:42915", + "right": "issue:46291", + "accept": false, + "reason": "A Qwen3-MoE FP8 failure and a kernels import requirement change 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:36331", + "right": "issue:41720", + "accept": false, + "reason": "A trainer compute_loss signature issue is unrelated to Qwen2-VL multi-GPU training." + } + ] + }, + "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": 22, + "review_unit_index": 15, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "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:43531", + "issue:43540", + "issue:43550", + "issue:43606", + "issue:43644", + "issue:43653", + "issue:43720", + "issue:43746", + "issue:43782", + "issue:43873", + "issue:44077", + "issue:44186", + "issue:44361", + "issue:44387", + "issue:44451", + "issue:44492", + "issue:44509", + "issue:44534", + "issue:44589", + "issue:44625", + "issue:44913", + "issue:44991", + "issue:45059", + "issue:45357", + "issue:45406", + "issue:45464", + "issue:45479", + "issue:45701", + "issue:45704", + "issue:45759", + "issue:45859", + "issue:46402", + "issue:46665" + ], + "soft_pairs": [ + "issue:44186|issue:44361", + "issue:43316|issue:44913", + "issue:42915|issue:43322", + "issue:28530|issue:30064", + "issue:33357|issue:35707", + "issue:43540|issue:43782", + "issue:29159|issue:36032", + "issue:43550|issue:43653", + "issue:43873|issue:45704", + "issue:43606|issue:43720", + "issue:44387|issue:45701", + "issue:44492|issue:44509", + "issue:30064|issue:34634", + "issue:44534|issue:46665", + "issue:41720|issue:43746", + "issue:43388|issue:45059", + "issue:43531|issue:45759", + "issue:42490|issue:43329", + "issue:43606|issue:43653", + "issue:45406|issue:45464", + "issue:45357|issue:46402", + "issue:34567|issue:35707", + "issue:34634|issue:41720", + "issue:45759|issue:45859", + "issue:44451|issue:44991", + "issue:43531|issue:44589", + "issue:44625|issue:45479", + "issue:44077|issue:44625", + "issue:43644|issue:46665" + ], + "prepared_review_unit_hash": "68a43e621647e571cde589a1110ec32d26fbae6efcd9f77c4403be8bee592cbf", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36387, + "estimated_input_tokens": 9097, + "estimated_eval_tokens": 18450 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is mostly a heterogeneous mix of unrelated bug reports. I found no strong duplicate pair among the listed soft edges; the closest thematic overlap is the non-persistent-buffer regression, but the provided pairings do not line up tightly enough to merge.", + "confidence": 0.81, + "canonical_issue_reason": "No clear duplicate cluster emerges from the listed pairs. Issue 43644 is the best representative because it is a concrete, well-scoped regression report with a crisp failure mode and version context.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43644 is the strongest standalone issue in the set: specific, reproducible, and easy to action. It is the clearest representative even though the other items are not true duplicates of it.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44186", + "right": "issue:44361", + "accept": false, + "reason": "Both are tokenizer/model-input crashes, but they affect different models and different failure modes (LayoutLMv2 NER padding/truncation vs MLuke AttributeError)." + }, + { + "left": "issue:43316", + "right": "issue:44913", + "accept": false, + "reason": "Both mention config inconsistencies, but one is a Gemma3TextConfig API mismatch and the other is GPTNeoX rotary_pct not persisting on reload; different fields and code paths." + }, + { + "left": "issue:42915", + "right": "issue:43322", + "accept": false, + "reason": "Different subsystems and failures: Qwen3Moe + FineGrainedFP8Config versus Llava Next segmentation fault on model load." + }, + { + "left": "issue:28530", + "right": "issue:30064", + "accept": false, + "reason": "Early-stopping metric resolution and image processor segmentation-map handling are unrelated bugs." + }, + { + "left": "issue:33357", + "right": "issue:35707", + "accept": false, + "reason": "A MacOS bus error in CLIP loading and a generation bug with inputs_embeds/past_key_values are different runtime paths." + }, + { + "left": "issue:43540", + "right": "issue:43782", + "accept": false, + "reason": "Both are Qwen multimodal issues, but one is video-input processing in Qwen3OmniMoe and the other is a weight_only from_pretrained failure in Qwen3VL; not the same bug." + }, + { + "left": "issue:29159", + "right": "issue:36032", + "accept": false, + "reason": "Slow/fast tokenizer inconsistency is a broad tokenizer behavior issue, while T5Tokenizer load fails due to an add_special_tokens method conflict." + }, + { + "left": "issue:43550", + "right": "issue:43653", + "accept": false, + "reason": "Bamba SDPA/torch.compile failure and BigBirdTokenizer special-token registration are unrelated." + }, + { + "left": "issue:43873", + "right": "issue:45704", + "accept": false, + "reason": "Quantization/offloading behavior versus an Apex RMSNorm memory leak are different underlying problems." + }, + { + "left": "issue:43606", + "right": "issue:43720", + "accept": false, + "reason": "Both involve loading/offload machinery, but one is a Bark CPU-offload device mismatch and the other is BitNet packed-weight unpacking during accelerate loading." + }, + { + "left": "issue:44387", + "right": "issue:45701", + "accept": false, + "reason": "CUDA memory growth under int4 quantization is unrelated to tokenizer changes across versions." + }, + { + "left": "issue:44492", + "right": "issue:44509", + "accept": false, + "reason": "A typo in cache strategies is not the same as stale docs mentioning removed pipeline tasks." + }, + { + "left": "issue:30064", + "right": "issue:34634", + "accept": false, + "reason": "Void segmentation-map processing and BarkProcessor voice_preset handling are separate model/component bugs." + }, + { + "left": "issue:44534", + "right": "issue:46665", + "accept": false, + "reason": "Both touch loading behavior in v5, but one is non-persistent buffer initialization and the other is from_pretrained used inside Python classes; different code paths." + }, + { + "left": "issue:41720", + "right": "issue:43746", + "accept": false, + "reason": "Qwen3 auto device mapping cudaErrorAssert and GraniteSpeech PEFT local checkpoint loading are unrelated failures." + }, + { + "left": "issue:43388", + "right": "issue:45059", + "accept": false, + "reason": "A metrics-gathering batch truncation bug is unrelated to SAM3 text/bounding-box processing behavior." + }, + { + "left": "issue:43531", + "right": "issue:45759", + "accept": false, + "reason": "Sliding-window handling in Qwen3-MoE and from_config text_config unwrapping for composite Qwen multimodal configs are different bugs." + }, + { + "left": "issue:42490", + "right": "issue:43329", + "accept": false, + "reason": "Import warning for trailing slashes and an undefined video branch in multimodal token counting are unrelated." + }, + { + "left": "issue:43606", + "right": "issue:43653", + "accept": false, + "reason": "CPU offload device mismatch in Bark and BigBirdTokenizer special-token decode behavior are different issues." + }, + { + "left": "issue:45406", + "right": "issue:45464", + "accept": false, + "reason": "Gemma4Processor missing _tokenizer in serve and Qwen3.5 streaming chat/completions failure are separate serving bugs." + }, + { + "left": "issue:45357", + "right": "issue:46402", + "accept": false, + "reason": "Both concern multimodal weight loading, but one is incorrect visual encoder keys on save_pretrained and the other is Sam3Model text encoder weights not loading." + }, + { + "left": "issue:34567", + "right": "issue:35707", + "accept": false, + "reason": "TrainerState token-count tracking and progressive generation with inputs_embeds/past_key_values are unrelated." + }, + { + "left": "issue:34634", + "right": "issue:41720", + "accept": false, + "reason": "Bark voice_preset handling and Qwen3 auto device mapping are not the same failure mode." + }, + { + "left": "issue:45759", + "right": "issue:45859", + "accept": false, + "reason": "Composite config unwrapping and CPU-offload incompatibility in Qwen3_5MoeTextRotaryEmbedding are different problems." + }, + { + "left": "issue:44451", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-loading regressions, but they affect different models and are not clearly the same root cause." + }, + { + "left": "issue:43531", + "right": "issue:44589", + "accept": false, + "reason": "Sliding-window logic and missing Float8 storage object errors are unrelated." + }, + { + "left": "issue:44625", + "right": "issue:45479", + "accept": false, + "reason": "A config propagation bug for Qwen3.5 and a sequence-classification loss degeneration bug are different issues." + }, + { + "left": "issue:44077", + "right": "issue:44625", + "accept": false, + "reason": "Optional post_init on patchtsmixer and num_labels propagation in Qwen3.5 are unrelated." + }, + { + "left": "issue:43644", + "right": "issue:46665", + "accept": false, + "reason": "Non-persistent buffer junk on v5 and from_pretrained behavior inside Python classes do not describe the same defect." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 16, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:30064", + "issue:36010", + "issue:39692", + "issue:41720", + "issue:42886", + "issue:43381", + "issue:43421", + "issue:43531", + "issue:43550", + "issue:43645", + "issue:43653", + "issue:43720", + "issue:43746", + "issue:43782", + "issue:43819", + "issue:43844", + "issue:43873", + "issue:43927", + "issue:43994", + "issue:44062", + "issue:44186", + "issue:44368", + "issue:44451", + "issue:44514", + "issue:44545", + "issue:44610", + "issue:44671", + "issue:44843", + "issue:44898", + "issue:44918", + "issue:44936", + "issue:44987", + "issue:45042", + "issue:45072", + "issue:45081", + "issue:45356", + "issue:45357", + "issue:45381", + "issue:45704", + "issue:45759", + "issue:45907", + "issue:45923", + "issue:46032", + "issue:46050", + "issue:46463", + "issue:46655", + "issue:46693" + ], + "soft_pairs": [ + "issue:44898|issue:45072", + "issue:44062|issue:44451", + "issue:44368|issue:45357", + "issue:43381|issue:43844", + "issue:44987|issue:45042", + "issue:46655|issue:46693", + "issue:39692|issue:44610", + "issue:43421|issue:43927", + "issue:45704|issue:45923", + "issue:44918|issue:44936", + "issue:41720|issue:45704", + "issue:43381|issue:45704", + "issue:43746|issue:46050", + "issue:44514|issue:44545", + "issue:43531|issue:43645", + "issue:42886|issue:44843", + "issue:46032|issue:46463", + "issue:45759|issue:45923", + "issue:43421|issue:43653", + "issue:39692|issue:43873", + "issue:43994|issue:44671", + "issue:43746|issue:43873", + "issue:43782|issue:45381", + "issue:43421|issue:44186", + "issue:45081|issue:45356", + "issue:30064|issue:36010", + "issue:45356|issue:45907", + "issue:43720|issue:43819", + "issue:43653|issue:43720", + "issue:43550|issue:45072" + ], + "prepared_review_unit_hash": "9737738d72c5f31cc657b3eed3e28be62b279c2974782106192134c9c3d74595", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36489, + "estimated_input_tokens": 9123, + "estimated_eval_tokens": 18502 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most items are unrelated bug reports. Only two soft edges look like true duplicates: the Qwen2.5-VL batched chat-template crash, and the HF_HUB_OFFLINE tokenizer regression around _patch_mistral_regex/model_info.", + "confidence": 0.88, + "canonical_issue_reason": "Issue 44843 is the clearest root-cause report for the offline tokenizer/Mistral-regex regression; issue 42886 reads as the broader symptom report of the same failure.", + "canonical_pr_reason": null, + "best_issue_reason": "44843 is the strongest canonical issue in this set: it names a specific code path, a reproducible offline-mode regression, and cleanly subsumes the related cache-loading complaint.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44898", + "right": "issue:45072", + "accept": false, + "reason": "Different models and failures: image-resolution/interpolation bug vs bfloat16 dtype mismatch in another model family." + }, + { + "left": "issue:44062", + "right": "issue:44451", + "accept": false, + "reason": "AddedToken special-argument error is unrelated to the ScandiBERT loading failure." + }, + { + "left": "issue:44368", + "right": "issue:45357", + "accept": false, + "reason": "Different problems: tie_word_embeddings warning during LoRA fine-tuning vs incorrect visual-encoder keys on save_pretrained." + }, + { + "left": "issue:43381", + "right": "issue:43844", + "accept": false, + "reason": "Eval-mode gradient checkpointing restriction is not the same as the ZeRO-3 gradient explosion regression." + }, + { + "left": "issue:44987", + "right": "issue:45042", + "accept": false, + "reason": "Different subsystems and symptoms; one is a model-loading regression, the other an image-processor torchvision dependency issue." + }, + { + "left": "issue:46655", + "right": "issue:46693", + "accept": false, + "reason": "Video padding bug in SmolVLM is unrelated to the FA2 long-context performance regression." + }, + { + "left": "issue:39692", + "right": "issue:44610", + "accept": false, + "reason": "Both mention processor/model mismatch, but they are different models and different concrete bugs." + }, + { + "left": "issue:43421", + "right": "issue:43927", + "accept": false, + "reason": "Runtime special-token/post-processor update bug is distinct from DiaConfig losing custom token IDs on save/load." + }, + { + "left": "issue:45704", + "right": "issue:45923", + "accept": false, + "reason": "T5 RMSNorm memory leak and missing gradient-checkpointing flag are unrelated issues." + }, + { + "left": "issue:44918", + "right": "issue:44936", + "accept": false, + "reason": "TRL input-embedding unpacking failure is not the same as a generic evaluate-after-train failure." + }, + { + "left": "issue:41720", + "right": "issue:45704", + "accept": false, + "reason": "Qwen3 cudaErrorAssert and T5 apex memory leak are unrelated." + }, + { + "left": "issue:43381", + "right": "issue:45704", + "accept": false, + "reason": "Gradient checkpointing eval-mode restriction and apex memory leak do not share a code path." + }, + { + "left": "issue:43746", + "right": "issue:46050", + "accept": false, + "reason": "PEFT adapter local-checkpoint loading differs from sinq quantized save/load corruption." + }, + { + "left": "issue:44514", + "right": "issue:44545", + "accept": true, + "reason": "Same Qwen2_5_VLProcessor.apply_chat_template batched-input crash with padding=False; these are duplicate reports of the same bug." + }, + { + "left": "issue:43531", + "right": "issue:43645", + "accept": false, + "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to the Jupyter custom-model initialization regression." + }, + { + "left": "issue:42886", + "right": "issue:44843", + "accept": true, + "reason": "Both report tokenizer loading breaking in HF_HUB_OFFLINE mode from the same _patch_mistral_regex/model_info code path." + }, + { + "left": "issue:46032", + "right": "issue:46463", + "accept": false, + "reason": "Incorrect use_cache results in Mamba2Mixer is unrelated to a T5 error-message validation bug." + }, + { + "left": "issue:45759", + "right": "issue:45923", + "accept": false, + "reason": "Composite-config unwrapping for Qwen multimodal models is unrelated to a missing gradient-checkpointing flag." + }, + { + "left": "issue:43421", + "right": "issue:43653", + "accept": false, + "reason": "Both involve special tokens, but 43653 is a missing registration bug while 43421 is about post-processor refresh after runtime edits." + }, + { + "left": "issue:39692", + "right": "issue:43873", + "accept": false, + "reason": "SigLIP2 docs/model mismatch is unrelated to quantization offloading behavior." + }, + { + "left": "issue:43994", + "right": "issue:44671", + "accept": false, + "reason": "Different models and failure modes: SigLIP2 nonsensical output vs CamemBERT masked-LM regression." + }, + { + "left": "issue:43746", + "right": "issue:43873", + "accept": false, + "reason": "PEFT adapter loading and quantization offloading are separate bugs." + }, + { + "left": "issue:43782", + "right": "issue:45381", + "accept": false, + "reason": "Qwen3VL weight_only loading error is not the same as Qwen2.5-VL video position-id corruption." + }, + { + "left": "issue:43421", + "right": "issue:44186", + "accept": false, + "reason": "LayoutLMv2 tokenizer crashes on NER/padding are not the same bug as special-token post-processor refresh." + }, + { + "left": "issue:45081", + "right": "issue:45356", + "accept": false, + "reason": "Both touch fix_mistral_regex, but one is an AttributeError crash and the other is a codec-handling/warning regression; not clearly the same underlying bug." + }, + { + "left": "issue:30064", + "right": "issue:36010", + "accept": false, + "reason": "Void segmentation-map processing and a GenerationMixin import error are unrelated." + }, + { + "left": "issue:45356", + "right": "issue:45907", + "accept": false, + "reason": "Tokenizer codec/regression issues are unrelated to list_repo_files missing the revision argument." + }, + { + "left": "issue:43720", + "right": "issue:43819", + "accept": false, + "reason": "BitNet packed-weight loading and DAC from_latents/STE mismatch are different code paths." + }, + { + "left": "issue:43653", + "right": "issue:43720", + "accept": false, + "reason": "BigBird special-token decode issue is unrelated to BitNet packed-weight loading." + }, + { + "left": "issue:43550", + "right": "issue:45072", + "accept": false, + "reason": "Torch.compile+SDPA failure in Bamba-9B-v2 is unrelated to bfloat16 dtype mismatches in other models." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded overall and appropriately conservative. The two accepted duplicate pairs are well supported by the titles, and the other soft-edge rejections avoid overclaiming despite some superficial similarity." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 17, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:30064", + "issue:36331", + "issue:39692", + "issue:41720", + "issue:43295", + "issue:43322", + "issue:43335", + "issue:43388", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43596", + "issue:43611", + "issue:43618", + "issue:43643", + "issue:43653", + "issue:43697", + "issue:43756", + "issue:43819", + "issue:43927", + "issue:44008", + "issue:44062", + "issue:44186", + "issue:44220", + "issue:44442", + "issue:44448", + "issue:44485", + "issue:44488", + "issue:44509", + "issue:44661", + "issue:44671", + "issue:44821", + "issue:44843", + "issue:44877", + "issue:44898", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45183", + "issue:45381", + "issue:45479", + "issue:45685", + "issue:45834", + "issue:46032", + "issue:46294", + "issue:46665", + "issue:46693" + ], + "soft_pairs": [ + "issue:43295|issue:45834", + "issue:43540|issue:46294", + "issue:43550|issue:43927", + "issue:44062|issue:44488", + "issue:43596|issue:44220", + "issue:45381|issue:46693", + "issue:43618|issue:43697", + "issue:43611|issue:46665", + "issue:43653|issue:43819", + "issue:44186|issue:44898", + "issue:43531|issue:43643", + "issue:44220|issue:44661", + "issue:43335|issue:43756", + "issue:43550|issue:43819", + "issue:44843|issue:45081", + "issue:41720|issue:44991", + "issue:43643|issue:44877", + "issue:30064|issue:36331", + "issue:43540|issue:44008", + "issue:44821|issue:45084", + "issue:39692|issue:45834", + "issue:44186|issue:44442", + "issue:44485|issue:44509", + "issue:44448|issue:44671", + "issue:45183|issue:45479", + "issue:43388|issue:46032", + "issue:43322|issue:45685" + ], + "prepared_review_unit_hash": "ff40d8f9a3fde6f94cac4e4da52b927ac19bb1045aa21b822c58a1e4e1b3eb84", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35521, + "estimated_input_tokens": 8881, + "estimated_eval_tokens": 18018 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is heterogeneous: it contains many unrelated model, tokenizer, config, and runtime bugs across different code paths. The soft-similarity pairs are mostly broad thematic overlaps rather than the same underlying defect, so none should be merged as duplicates. No single issue is a clear canonical representative.", + "confidence": 0.93, + "canonical_issue_reason": "No clear canonical issue: the items span unrelated subsystems and failure modes, with only loose topical similarity.", + "canonical_pr_reason": null, + "best_issue_reason": "No strong global representative issue stands out because the cluster is too mixed to summarize with one artifact.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43295", + "right": "issue:45834", + "accept": false, + "reason": "Different bugs: processor/tokenizer regression vs long OCR index error in a different model." + }, + { + "left": "issue:43540", + "right": "issue:46294", + "accept": false, + "reason": "Both involve video inputs, but they fail in different Qwen models and different code paths (processing vs chat template token/feature mismatch)." + }, + { + "left": "issue:43550", + "right": "issue:43927", + "accept": false, + "reason": "Unrelated: torch.compile/SDPA failure in Bamba vs DiaConfig token-id persistence bug." + }, + { + "left": "issue:44062", + "right": "issue:44488", + "accept": false, + "reason": "Both tokenizer-related, but the errors and affected models are different; not the same defect." + }, + { + "left": "issue:43596", + "right": "issue:44220", + "accept": false, + "reason": "Completely different areas: Deepspeed/BertModel init error vs audio fbank feature extraction." + }, + { + "left": "issue:45381", + "right": "issue:46693", + "accept": false, + "reason": "Different regressions: Qwen2.5-VL video position IDs vs FA2 long-context performance regression." + }, + { + "left": "issue:43618", + "right": "issue:43697", + "accept": false, + "reason": "Both are v5 vision-model regressions, but they affect different outputs and code paths; not one bug." + }, + { + "left": "issue:43611", + "right": "issue:46665", + "accept": false, + "reason": "Both concern loading, but one is base_model_prefix behavior and the other is from_pretrained inside Python classes." + }, + { + "left": "issue:43653", + "right": "issue:43819", + "accept": false, + "reason": "Tokenizer special-token registration bug vs audio codec forward/from_latents mismatch; unrelated." + }, + { + "left": "issue:44186", + "right": "issue:44898", + "accept": false, + "reason": "Different models and failures: LayoutLMv2 tokenizer padding/truncation vs Perceiver image classification resolution handling." + }, + { + "left": "issue:43531", + "right": "issue:43643", + "accept": false, + "reason": "Sliding-window MoE behavior is unrelated to trust_remote_code config field loss." + }, + { + "left": "issue:44220", + "right": "issue:44661", + "accept": false, + "reason": "Audio feature extraction bug vs tooling failure in add-new-model-like; different problem domains." + }, + { + "left": "issue:43335", + "right": "issue:43756", + "accept": false, + "reason": "Both are architecture/config issues, but they concern different models and different layer-selection logic." + }, + { + "left": "issue:43550", + "right": "issue:43819", + "accept": false, + "reason": "Different bug types and code paths: compile/SDPA vs DAC latent decoding mismatch." + }, + { + "left": "issue:44843", + "right": "issue:45081", + "accept": false, + "reason": "Same helper function name, but distinct failures: offline hub call vs attribute error on tokenizer object." + }, + { + "left": "issue:41720", + "right": "issue:44991", + "accept": false, + "reason": "Unrelated: Qwen3 device-mapping assert vs tokenizer loading failure for a specific RoBERTa model." + }, + { + "left": "issue:43643", + "right": "issue:44877", + "accept": false, + "reason": "Different config-loading problems with different triggers; not the same underlying defect." + }, + { + "left": "issue:30064", + "right": "issue:36331", + "accept": false, + "reason": "Image processor segmentation-map failure vs CustomTrainer compute_loss signature regression are unrelated." + }, + { + "left": "issue:43540", + "right": "issue:44008", + "accept": false, + "reason": "Both mention multimodal inputs, but one is a video-processing ValueError and the other is a Gemma3n variable collision." + }, + { + "left": "issue:44821", + "right": "issue:45084", + "accept": false, + "reason": "Different issues: AutoImageProcessor URL loading vs template-node compilation error." + }, + { + "left": "issue:39692", + "right": "issue:45834", + "accept": false, + "reason": "Docs/example mistakes in SigLIP2 vs runtime OCR index error in Kosmos2.5; not the same bug." + }, + { + "left": "issue:44186", + "right": "issue:44442", + "accept": false, + "reason": "Tokenizer padding/truncation crash vs AutoTokenizer loading failure for another tokenizer; unrelated." + }, + { + "left": "issue:44485", + "right": "issue:44509", + "accept": false, + "reason": "RoPE implementation discussion vs stale docs mentioning removed pipeline tasks; different changes." + }, + { + "left": "issue:44448", + "right": "issue:44671", + "accept": false, + "reason": "Both are v5 output regressions, but they affect different models and behaviors, so not mergeable as one fix." + }, + { + "left": "issue:45183", + "right": "issue:45479", + "accept": false, + "reason": "Validation XOR error across models vs degenerate single-label classification loss; different bugs." + }, + { + "left": "issue:43388", + "right": "issue:46032", + "accept": false, + "reason": "Metric-gathering last-batch truncation vs Mamba2Mixer use_cache correctness issue; unrelated." + }, + { + "left": "issue:43322", + "right": "issue:45685", + "accept": false, + "reason": "Segfault loading Llava Next models vs MPS histogram implementation error in moe; no shared underlying bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 18, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:28218", + "issue:29159", + "issue:34634", + "issue:36010", + "issue:36331", + "issue:42175", + "issue:43421", + "issue:43540", + "issue:43577", + "issue:43582", + "issue:43606", + "issue:43644", + "issue:43645", + "issue:43716", + "issue:43720", + "issue:43749", + "issue:43782", + "issue:43818", + "issue:43825", + "issue:43827", + "issue:43950", + "issue:43994", + "issue:44008", + "issue:44291", + "issue:44479", + "issue:44561", + "issue:44610", + "issue:44671", + "issue:44805", + "issue:44811", + "issue:44936", + "issue:44977", + "issue:44991", + "issue:45020", + "issue:45072", + "issue:45081", + "issue:45381", + "issue:45584", + "issue:45701", + "issue:45834", + "issue:45987", + "issue:46192", + "issue:46291", + "issue:46294", + "issue:46463", + "issue:46512", + "issue:46665" + ], + "soft_pairs": [ + "issue:43644|issue:43749", + "issue:45584|issue:45834", + "issue:43716|issue:44008", + "issue:43606|issue:45072", + "issue:34634|issue:36010", + "issue:43950|issue:45987", + "issue:44479|issue:45834", + "issue:43582|issue:46512", + "issue:43825|issue:43994", + "issue:43577|issue:43818", + "issue:43421|issue:44811", + "issue:46294|issue:46463", + "issue:43782|issue:44977", + "issue:44991|issue:46291", + "issue:44561|issue:45020", + "issue:44805|issue:44936", + "issue:43540|issue:45381", + "issue:44610|issue:45072", + "issue:43720|issue:45072", + "issue:28218|issue:29159", + "issue:43825|issue:43827", + "issue:43749|issue:44291", + "issue:45987|issue:46192", + "issue:43645|issue:46665", + "issue:44671|issue:45701", + "issue:43577|issue:43582", + "issue:42175|issue:45081", + "issue:34634|issue:36331" + ], + "prepared_review_unit_hash": "7607169b9b999cf28a11fdd562af472cbadbd37474a8021c37c6538b11a4f1f2", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35837, + "estimated_input_tokens": 8960, + "estimated_eval_tokens": 18176 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a set of unrelated regression reports, but one tokenizer pair looks like the same underlying bug: added-token spacing / slow-vs-fast tokenizer inconsistency. All other soft edges appear to be different failures in different code paths.", + "confidence": 0.63, + "canonical_issue_reason": "issue:28218 is the most concrete tokenizer regression in the set: it describes a specific, reproducible added-token spacing bug and is a better anchor than the broader inconsistency report.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:28218 is the best representative issue for the only likely duplicate subgroup, because it states the core tokenizer symptom clearly and narrowly.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43644", + "right": "issue:43749", + "accept": false, + "reason": "Both are loading-related regressions, but one is about non-persistent buffers being corrupted and the other is FSDP CPU RAM efficient loading; too different to be the same bug." + }, + { + "left": "issue:45584", + "right": "issue:45834", + "accept": false, + "reason": "Whisper empty-transcription generation and Kosmos2.5 long OCR index errors are unrelated model-specific failures." + }, + { + "left": "issue:43716", + "right": "issue:44008", + "accept": false, + "reason": "Different models and different failure modes: dtype mismatch in preprocessing vs an AttributeError from variable collision." + }, + { + "left": "issue:43606", + "right": "issue:45072", + "accept": false, + "reason": "CPU offload device mismatch for Bark is unrelated to bfloat16 dtype mismatches in other models." + }, + { + "left": "issue:34634", + "right": "issue:36010", + "accept": false, + "reason": "BarkProcessor voice_preset behavior and a GenerationMixin import error are unrelated issues." + }, + { + "left": "issue:43950", + "right": "issue:45987", + "accept": false, + "reason": "Non-persistent buffer corruption during loading and StaticCache sequence-length return-type mismatch are different code paths." + }, + { + "left": "issue:44479", + "right": "issue:45834", + "accept": false, + "reason": "Both mention vision/video or indexing failures, but they are different model families and distinct regressions." + }, + { + "left": "issue:43582", + "right": "issue:46512", + "accept": false, + "reason": "Apple Silicon allocator TypeError is unrelated to from_config leaving config.dtype stale." + }, + { + "left": "issue:43825", + "right": "issue:43994", + "accept": false, + "reason": "An error-message/doc mismatch for pipeline support is not the same as SigLIP2 producing nonsensical outputs." + }, + { + "left": "issue:43577", + "right": "issue:43818", + "accept": false, + "reason": "Blip2 dtype loading and Video-LLaVA missing temporal attention are different model-specific bugs." + }, + { + "left": "issue:43421", + "right": "issue:44811", + "accept": false, + "reason": "One is runtime special-token post-processor syncing; the other is Whisper batch_decode ignoring skip_special_tokens. Similar theme, different bug." + }, + { + "left": "issue:46294", + "right": "issue:46463", + "accept": false, + "reason": "Qwen video/chat-template feature-token mismatch and a T5 decoder-input error message are unrelated." + }, + { + "left": "issue:43782", + "right": "issue:44977", + "accept": false, + "reason": "Qwen3VL weight_only loading and Qwen3.5 flash-attention generation failures are different problems." + }, + { + "left": "issue:44991", + "right": "issue:46291", + "accept": false, + "reason": "Tokenizer loading failure and kernels package import failure are unrelated dependency/runtime issues." + }, + { + "left": "issue:44561", + "right": "issue:45020", + "accept": false, + "reason": "Both involve remote_code breakage, but the specific removed API and the broader recent-version incompatibility are not clearly the same bug." + }, + { + "left": "issue:44805", + "right": "issue:44936", + "accept": false, + "reason": "A mask/index shape mismatch and trainer.evaluate() failing after train are not the same underlying defect." + }, + { + "left": "issue:43540", + "right": "issue:45381", + "accept": false, + "reason": "Both are video-input regressions, but they affect different Qwen variants and have different reported symptoms; not enough to merge as one bug." + }, + { + "left": "issue:44610", + "right": "issue:45072", + "accept": false, + "reason": "Processor image-size mismatch and dtype mismatch in inference are unrelated." + }, + { + "left": "issue:43720", + "right": "issue:45072", + "accept": false, + "reason": "BitNet packed-weight unpacking during accelerate loading is unrelated to generic dtype mismatch issues." + }, + { + "left": "issue:28218", + "right": "issue:29159", + "accept": true, + "reason": "These both point to tokenizer behavior differences, and the specific added-token spacing bug plausibly explains the slow-vs-fast inconsistency." + }, + { + "left": "issue:43825", + "right": "issue:43827", + "accept": false, + "reason": "One is a misleading error message about translation tasks; the other is outdated docs still referencing pipeline. Same area, different artifacts." + }, + { + "left": "issue:43749", + "right": "issue:44291", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and init_empty_weights/_is_hf_initialized are different loading-path failures." + }, + { + "left": "issue:45987", + "right": "issue:46192", + "accept": false, + "reason": "StaticCache return type and return_tensors being ignored are unrelated API bugs." + }, + { + "left": "issue:43645", + "right": "issue:46665", + "accept": false, + "reason": "Notebook custom-model initialization and from_pretrained inside Python classes are similar in spirit but not clearly the same code-path bug." + }, + { + "left": "issue:44671", + "right": "issue:45701", + "accept": false, + "reason": "Masked LM prediction regression and a broad tokenization-change report are not the same specific failure." + }, + { + "left": "issue:43577", + "right": "issue:43582", + "accept": false, + "reason": "Blip2 dtype mismatch and Apple Silicon allocator TypeError are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:45081", + "accept": false, + "reason": "Backend install extras and Mistral tokenizer regex patching are unrelated." + }, + { + "left": "issue:34634", + "right": "issue:36331", + "accept": false, + "reason": "BarkProcessor voice_preset and CustomTrainer.compute_loss signature mismatch are unrelated bugs." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays appropriately cautious. The only positive soft edge (28218/29159) is plausibly supported by the titles, and the rest of the verdicts are conservative given the limited evidence. If anything, keep the tokenizer linkage phrased as 'likely related' rather than asserting the same underlying bug too strongly." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 19, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:36010", + "issue:36331", + "issue:38175", + "issue:39692", + "issue:42175", + "issue:42490", + "issue:43257", + "issue:43262", + "issue:43329", + "issue:43452", + "issue:43526", + "issue:43531", + "issue:43550", + "issue:43582", + "issue:43638", + "issue:43645", + "issue:43653", + "issue:43697", + "issue:43720", + "issue:43742", + "issue:43782", + "issue:43819", + "issue:43906", + "issue:43927", + "issue:44060", + "issue:44368", + "issue:44554", + "issue:44589", + "issue:44610", + "issue:44857", + "issue:44912", + "issue:44936", + "issue:44991", + "issue:45003", + "issue:45020", + "issue:45042", + "issue:45081", + "issue:45084", + "issue:45325", + "issue:45372", + "issue:45406", + "issue:45538", + "issue:45600", + "issue:46209", + "issue:46285", + "issue:46291", + "issue:46326" + ], + "soft_pairs": [ + "issue:45003|issue:46285", + "issue:43582|issue:44857", + "issue:43653|issue:44912", + "issue:43262|issue:43819", + "issue:43782|issue:45325", + "issue:45081|issue:45406", + "issue:45372|issue:46326", + "issue:44060|issue:44368", + "issue:43329|issue:45600", + "issue:43742|issue:45020", + "issue:44589|issue:45042", + "issue:43526|issue:43697", + "issue:43550|issue:44912", + "issue:45084|issue:45538", + "issue:43526|issue:44554", + "issue:43452|issue:44991", + "issue:43906|issue:46209", + "issue:43257|issue:43329", + "issue:43819|issue:43927", + "issue:43638|issue:44936", + "issue:36010|issue:36331", + "issue:36010|issue:43645", + "issue:43531|issue:45042", + "issue:43720|issue:44610", + "issue:42175|issue:46291", + "issue:39692|issue:44912", + "issue:42490|issue:43257", + "issue:38175|issue:44610" + ], + "prepared_review_unit_hash": "677404128b274debad9ebf88332c8e08a2e9b6ed835aadc2827c4945097e0061", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35939, + "estimated_input_tokens": 8985, + "estimated_eval_tokens": 18226 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous: the soft-similarity pairs share only broad subsystem/model-family overlap, not the same underlying bug or change. I would not merge any of the candidate pairs as duplicates.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 45020 is the broadest representative of the set\u2019s recurring theme of recent version-related loading regressions, but the cluster is too diverse to support a true duplicate canonical.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45020 is the closest thing to a cluster anchor because it covers a wide class of loading/compatibility breakages, but it still does not subsume the other reports well enough to treat them as duplicates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45003", + "right": "issue:46285", + "accept": false, + "reason": "Different bug classes: unsafe sys.modules access vs a CLIP text-model structure inconsistency." + }, + { + "left": "issue:43582", + "right": "issue:44857", + "accept": false, + "reason": "Unrelated failure modes on different platforms: Apple Silicon TypeError vs CUDA AMP crash." + }, + { + "left": "issue:43653", + "right": "issue:44912", + "accept": false, + "reason": "Tokenizer special-token registration bug vs MXFP4 quantization loading/fallback issue." + }, + { + "left": "issue:43262", + "right": "issue:43819", + "accept": false, + "reason": "Audio chat-template sampling-rate default vs DAC latent/STE mismatch are separate code paths." + }, + { + "left": "issue:43782", + "right": "issue:45325", + "accept": false, + "reason": "Both involve VL models, but one is a from_pretrained weight-only loading error and the other is a rope-index scaling bug." + }, + { + "left": "issue:45081", + "right": "issue:45406", + "accept": false, + "reason": "Different tokenizer/processor crashes in different model integrations and different attributes." + }, + { + "left": "issue:45372", + "right": "issue:46326", + "accept": false, + "reason": "One is a missing import breaking Gemma 4 processor loading; the other is a continue_final_message bug for mistral_common tokenizers." + }, + { + "left": "issue:44060", + "right": "issue:44368", + "accept": false, + "reason": "Different Qwen model variants and different warning/content: tied-weights miswire vs tie_word_embeddings warning." + }, + { + "left": "issue:43329", + "right": "issue:45600", + "accept": false, + "reason": "Multimodal token counting bug vs stale auto_mappings references are unrelated." + }, + { + "left": "issue:43742", + "right": "issue:45020", + "accept": false, + "reason": "Both are loading regressions, but they are not the same concrete failure or code path." + }, + { + "left": "issue:44589", + "right": "issue:45042", + "accept": false, + "reason": "Storage-object lookup failure is unrelated to PIL backend torchvision dependency handling." + }, + { + "left": "issue:43526", + "right": "issue:43697", + "accept": false, + "reason": "BEiT label reduction bug vs RTDetr output-drift regression are different model behaviors." + }, + { + "left": "issue:43550", + "right": "issue:44912", + "accept": false, + "reason": "torch.compile/SDPA failure is unrelated to MXFP4 quantization fallback during loading." + }, + { + "left": "issue:45084", + "right": "issue:45538", + "accept": false, + "reason": "Template compilation error vs tokenizer model_max_length constant are unrelated." + }, + { + "left": "issue:43526", + "right": "issue:44554", + "accept": false, + "reason": "Different functionality entirely: image label reduction vs MPS attention correctness." + }, + { + "left": "issue:43452", + "right": "issue:44991", + "accept": false, + "reason": "GGUF file handling breakage is not the same as a specific tokenizer-loading regression for one model." + }, + { + "left": "issue:43906", + "right": "issue:46209", + "accept": false, + "reason": "An isolated reproduction of an old issue is not the same bug as FP8 inference failure on B200." + }, + { + "left": "issue:43257", + "right": "issue:43329", + "accept": false, + "reason": "Qwen3 MoE loading/accelerate conversion issue vs multimodal token counting undefined names are unrelated." + }, + { + "left": "issue:43819", + "right": "issue:43927", + "accept": false, + "reason": "DAC latent reconstruction bug and DiaConfig custom-token-ID persistence bug are different code paths." + }, + { + "left": "issue:43638", + "right": "issue:44936", + "accept": false, + "reason": "DeepSpeed zero3/Bert index error vs trainer.evaluate() after train are different training/runtime issues." + }, + { + "left": "issue:36010", + "right": "issue:36331", + "accept": false, + "reason": "Import path breakage for GenerationMixin is unrelated to CustomTrainer.compute_loss signature mismatch." + }, + { + "left": "issue:36010", + "right": "issue:43645", + "accept": false, + "reason": "Importing GenerationMixin and Jupyter custom-model initialization are different failures." + }, + { + "left": "issue:43531", + "right": "issue:45042", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior is unrelated to PIL backend torchvision dependency enforcement." + }, + { + "left": "issue:43720", + "right": "issue:44610", + "accept": false, + "reason": "BitNet packed-weight loading bug and OmDet-Turbo input-size mismatch are not the same bug." + }, + { + "left": "issue:42175", + "right": "issue:46291", + "accept": false, + "reason": "Package extra composition and kernels import requirements are unrelated dependency issues." + }, + { + "left": "issue:39692", + "right": "issue:44912", + "accept": false, + "reason": "SigLIP2 documentation mistakes do not match a quantization-loading fallback bug." + }, + { + "left": "issue:42490", + "right": "issue:43257", + "accept": false, + "reason": "Trailing-slash local model import warning is unrelated to Qwen3 MoE weight-conversion loading." + }, + { + "left": "issue:38175", + "right": "issue:44610", + "accept": false, + "reason": "Unexpected zero probabilities in SigLIP2 is not the same as OmDet-Turbo preprocessing size mismatch." + } + ] + }, + "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": 22, + "review_unit_index": 20, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:29159", + "issue:34567", + "issue:40990", + "issue:42915", + "issue:43257", + "issue:43262", + "issue:43329", + "issue:43404", + "issue:43450", + "issue:43454", + "issue:43525", + "issue:43540", + "issue:43577", + "issue:43606", + "issue:43643", + "issue:43650", + "issue:43696", + "issue:43742", + "issue:43746", + "issue:43749", + "issue:43854", + "issue:43906", + "issue:43927", + "issue:43931", + "issue:44291", + "issue:44387", + "issue:44479", + "issue:44568", + "issue:44617", + "issue:44756", + "issue:44811", + "issue:44821", + "issue:44912", + "issue:44977", + "issue:44987", + "issue:45042", + "issue:45127", + "issue:45356", + "issue:45459", + "issue:45538", + "issue:45759", + "issue:45803", + "issue:45859", + "issue:46291", + "issue:46421", + "issue:46424", + "issue:46701" + ], + "soft_pairs": [ + "issue:43746|issue:44811", + "issue:44387|issue:44617", + "issue:43854|issue:44977", + "issue:43643|issue:45759", + "issue:43257|issue:43696", + "issue:43262|issue:44811", + "issue:45459|issue:45803", + "issue:46421|issue:46424", + "issue:43329|issue:44479", + "issue:43257|issue:43931", + "issue:43329|issue:43450", + "issue:43606|issue:45859", + "issue:42915|issue:43742", + "issue:43749|issue:44756", + "issue:43650|issue:43906", + "issue:34567|issue:40990", + "issue:44291|issue:45042", + "issue:43540|issue:46701", + "issue:42915|issue:44821", + "issue:43454|issue:43577", + "issue:29159|issue:34567", + "issue:43927|issue:44912", + "issue:44568|issue:45356", + "issue:43525|issue:46291", + "issue:43404|issue:45127", + "issue:44987|issue:45538" + ], + "prepared_review_unit_hash": "5b1d95fb4c4ec5b7f48134dbf5b6b16f6405a6183e0db74facb958c6810b4103", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 26, + "serialized_chars": 35072, + "estimated_input_tokens": 8768, + "estimated_eval_tokens": 17792 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is a heterogeneous mix of unrelated issue reports. None of the soft pairs look like true duplicates; most share only a broad subsystem label (tokenizers, video, loading, offload, cache, etc.) rather than the same concrete bug.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 46421 is the best weak anchor for the cluster only because it is a clearly scoped, active cache-related bug with discussion and references; however, the overall set is too diverse to have a true canonical duplicate issue.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 46421 has the strongest triage value among the listed items: specific reproduction context, open status, and clear user impact. It is not a true representative of the whole cluster, just the most actionable single issue here.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43746", + "right": "issue:44811", + "accept": false, + "reason": "Different bugs and code paths: GraniteSpeech PEFT/local checkpoint loading vs Whisper batch_decode ignoring skip_special_tokens." + }, + { + "left": "issue:44387", + "right": "issue:44617", + "accept": false, + "reason": "One is a broad int4 quantization memory regression; the other is a model-specific Sam3Video OOM. Not the same underlying issue." + }, + { + "left": "issue:43854", + "right": "issue:44977", + "accept": false, + "reason": "Different model families and failures: GLM-4.7-Flash test loading vs Qwen3.5 flash-attention generation behavior." + }, + { + "left": "issue:43643", + "right": "issue:45759", + "accept": false, + "reason": "Config-loading with trust_remote_code missing fields is distinct from from_config failing to unwrap text_config in multimodal configs." + }, + { + "left": "issue:43257", + "right": "issue:43696", + "accept": false, + "reason": "Qwen3 MOE weight conversion with accelerate+deepspeed is unrelated to GPT-oss-20b CUDA OOM." + }, + { + "left": "issue:43262", + "right": "issue:44811", + "accept": false, + "reason": "Both are processor-related, but one is chat template sampling-rate defaulting and the other is Whisper decoding token filtering." + }, + { + "left": "issue:45459", + "right": "issue:45803", + "accept": false, + "reason": "Both involve exception handling, but they are different functions and different failure modes: protobuf decode import handling vs bare except in tensor conversion." + }, + { + "left": "issue:46421", + "right": "issue:46424", + "accept": false, + "reason": "Related cache area, but one is lazy init causing recompilation and the other is max_cache_length being ignored outside Quantized Cache. Same subsystem, different bug." + }, + { + "left": "issue:43329", + "right": "issue:44479", + "accept": false, + "reason": "Undefined video-token helper variables in multimodal token counting is not the same as the broader v5.3.0 video input regression across several Qwen variants." + }, + { + "left": "issue:43257", + "right": "issue:43931", + "accept": false, + "reason": "Different loading problems: Qwen3 MOE conversion failure vs Qwen3-VL weight shape mismatch." + }, + { + "left": "issue:43329", + "right": "issue:43450", + "accept": false, + "reason": "Token-counting helper bug with undefined video branch variables is different from batched video processor shape output." + }, + { + "left": "issue:43606", + "right": "issue:45859", + "accept": false, + "reason": "Both mention CPU offload, but they affect different models and code paths: Bark device mismatch vs Qwen3_5MoeTextRotaryEmbedding compatibility." + }, + { + "left": "issue:42915", + "right": "issue:43742", + "accept": false, + "reason": "FineGrainedFP8Config failure for Qwen3Moe is unrelated to a key error while loading MobileLLM-125M." + }, + { + "left": "issue:43749", + "right": "issue:44756", + "accept": false, + "reason": "Different memory/loading problems: FSDP CPU-RAM-efficient loading vs disabling mmap on Strix Halo to avoid OOM." + }, + { + "left": "issue:43650", + "right": "issue:43906", + "accept": false, + "reason": "No substantive overlap: one is a nonspecific placeholder title, the other is a reproduction of an external issue." + }, + { + "left": "issue:34567", + "right": "issue:40990", + "accept": false, + "reason": "TrainerState token-count tracking and high perplexity on GPT-oss-20b are unrelated bugs." + }, + { + "left": "issue:44291", + "right": "issue:45042", + "accept": false, + "reason": "Different areas: init_empty_weights / _is_hf_initialized argument handling vs PIL image processors incorrectly requiring torchvision." + }, + { + "left": "issue:43540", + "right": "issue:46701", + "accept": false, + "reason": "Different model/pipeline failures: Qwen3OmniMoe video input ValueError vs DiffusionGemmaGenerationOutput pipeline not working." + }, + { + "left": "issue:42915", + "right": "issue:44821", + "accept": false, + "reason": "Unrelated issues: FineGrainedFP8Config failure for Qwen3Moe vs AutoImageProcessor loading from URL." + }, + { + "left": "issue:43454", + "right": "issue:43577", + "accept": false, + "reason": "Weight-tying failure causing generation corruption is distinct from dtype propagation remaining float32 in BLIP2." + }, + { + "left": "issue:29159", + "right": "issue:34567", + "accept": false, + "reason": "Tokenizer slow/fast inconsistency and TrainerState token accounting are entirely different components." + }, + { + "left": "issue:43927", + "right": "issue:44912", + "accept": false, + "reason": "DiaConfig losing custom token IDs on save/load is unrelated to MXFP4 quantization falling back to bf16." + }, + { + "left": "issue:44568", + "right": "issue:45356", + "accept": false, + "reason": "Both are tokenizer regressions, but they concern different models and different mechanics (special-token insertion vs codec handling/warnings)." + }, + { + "left": "issue:43525", + "right": "issue:46291", + "accept": false, + "reason": "Missing pad_token_id on Llama4Config is unrelated to a kernels package version/revision requirement blocking transformers.pipelines import." + }, + { + "left": "issue:43404", + "right": "issue:45127", + "accept": false, + "reason": "Missing lm_head tying in Mistral3 is not the same as LoRA merge collapse with extended vocabulary." + }, + { + "left": "issue:44987", + "right": "issue:45538", + "accept": false, + "reason": "Model loading failure for physical-intelligence/fast is unrelated to CLIPTokenizer's model_max_length constant 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": 22, + "review_unit_index": 21, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:24643", + "issue:27957", + "issue:28282", + "issue:29127", + "issue:29159", + "issue:33357", + "issue:34567", + "issue:34689", + "issue:36331", + "issue:38175", + "issue:41720", + "issue:42175", + "issue:42915", + "issue:43257", + "issue:43317", + "issue:43425", + "issue:43452", + "issue:43540", + "issue:43577", + "issue:43582", + "issue:43697", + "issue:43742", + "issue:43749", + "issue:43756", + "issue:43824", + "issue:44062", + "issue:44291", + "issue:44479", + "issue:44488", + "issue:44560", + "issue:44568", + "issue:44821", + "issue:44849", + "issue:44898", + "issue:44987", + "issue:44991", + "issue:45137", + "issue:45357", + "issue:45381", + "issue:45405", + "issue:45488", + "issue:45538", + "issue:45701", + "issue:45907", + "issue:46291", + "issue:46512", + "issue:46693" + ], + "soft_pairs": [ + "issue:43697|issue:45357", + "issue:28282|issue:34689", + "issue:45137|issue:45405", + "issue:42915|issue:45538", + "issue:33357|issue:34567", + "issue:43425|issue:46291", + "issue:44479|issue:46693", + "issue:42175|issue:43824", + "issue:28282|issue:29127", + "issue:43742|issue:44291", + "issue:44062|issue:44991", + "issue:43452|issue:43582", + "issue:43577|issue:46512", + "issue:43317|issue:45538", + "issue:44849|issue:45701", + "issue:42915|issue:43749", + "issue:41720|issue:43540", + "issue:43540|issue:44560", + "issue:44568|issue:45488", + "issue:24643|issue:29159", + "issue:44987|issue:45907", + "issue:27957|issue:28282", + "issue:44479|issue:45381", + "issue:43540|issue:44479", + "issue:42175|issue:44488", + "issue:38175|issue:44898", + "issue:43257|issue:43756", + "issue:34567|issue:36331", + "issue:44291|issue:44821" + ], + "prepared_review_unit_hash": "77db16a41d8d4ecf55ef2c77165a2b0854e348120430f9701325be0f4dadc02f", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35770, + "estimated_input_tokens": 8943, + "estimated_eval_tokens": 18142 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly heterogeneous issues; only the Qwen video-input regression pair looks like the same underlying bug. The rest are distinct failures across unrelated subsystems.", + "confidence": 0.8, + "canonical_issue_reason": "issue:44479 is the broader, earlier regression report and best captures the underlying Qwen video-input bug covered by the accepted duplicate.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44479 is the strongest single issue to keep: it states the versioned regression, names multiple affected model families, and reads like the umbrella report for the duplicate pair.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43697", + "right": "issue:45357", + "accept": false, + "reason": "Different models and failures: RTDetrV2 output drift vs Qwen3.5 save_pretrained key mapping." + }, + { + "left": "issue:28282", + "right": "issue:34689", + "accept": false, + "reason": "Both are loading errors, but one is missing-PyTorch ImportError and the other is a Llama vision model load regression." + }, + { + "left": "issue:45137", + "right": "issue:45405", + "accept": false, + "reason": "DeepSpeed ZeRO3 deque error is unrelated to a PEFT version bump complaint." + }, + { + "left": "issue:42915", + "right": "issue:45538", + "accept": false, + "reason": "Qwen3Moe FP8 failure and CLIPTokenizer max_length are unrelated bugs." + }, + { + "left": "issue:33357", + "right": "issue:34567", + "accept": false, + "reason": "MacOS bus error in CLIP loading is unrelated to TrainerState token-count bookkeeping." + }, + { + "left": "issue:43425", + "right": "issue:46291", + "accept": false, + "reason": "Torch 2.10 incompatibility is not the same as kernels revision/version handling in pipelines import." + }, + { + "left": "issue:44479", + "right": "issue:46693", + "accept": false, + "reason": "Both mention Qwen/video or generation regressions, but one is a video-input correctness bug and the other is a long-context performance regression." + }, + { + "left": "issue:42175", + "right": "issue:43824", + "accept": false, + "reason": "Torch backend packaging issue is unrelated to a missing Qwen2_5_VL class export." + }, + { + "left": "issue:28282", + "right": "issue:29127", + "accept": false, + "reason": "Missing-PyTorch ImportError and LayoutLMv3 box-validation messaging are different problems." + }, + { + "left": "issue:43742", + "right": "issue:44291", + "accept": false, + "reason": "MobileLLM key loading and init_empty_weights unexpected argument are distinct load-path failures." + }, + { + "left": "issue:44062", + "right": "issue:44991", + "accept": false, + "reason": "Tokenizer load failure may be related, but the titles do not show the same concrete AddedToken duplication bug." + }, + { + "left": "issue:43452", + "right": "issue:43582", + "accept": false, + "reason": "gguf_file loading breakage and Apple Silicon caching_allocator_warmup TypeError are unrelated." + }, + { + "left": "issue:43577", + "right": "issue:46512", + "accept": false, + "reason": "BLIP2 dtype mismatch during loading is different from stale config.dtype after from_config(dtype=...)." + }, + { + "left": "issue:43317", + "right": "issue:45538", + "accept": false, + "reason": "Device-map/offload load failure and tokenizer max_length behavior are unrelated." + }, + { + "left": "issue:44849", + "right": "issue:45701", + "accept": false, + "reason": "Qwen3.5 hidden-states bug is unrelated to version-dependent tokenization changes." + }, + { + "left": "issue:42915", + "right": "issue:43749", + "accept": false, + "reason": "Qwen3Moe FP8 failure and FSDP CPU RAM efficient loading are different distributed-loading bugs." + }, + { + "left": "issue:41720", + "right": "issue:43540", + "accept": false, + "reason": "Qwen3 auto-device-map cuda assert and Qwen3OmniMoe video ValueError are different code paths." + }, + { + "left": "issue:43540", + "right": "issue:44560", + "accept": false, + "reason": "Both concern video inputs, but they target different Qwen model families and different failures." + }, + { + "left": "issue:44568", + "right": "issue:45488", + "accept": false, + "reason": "Tokenizer special-token regression and LlamaTokenizer pre-tokenizer override are separate tokenizer bugs." + }, + { + "left": "issue:24643", + "right": "issue:29159", + "accept": false, + "reason": "DeepSpeed training weight-dimension error is unrelated to slow/fast tokenizer inconsistency." + }, + { + "left": "issue:44987", + "right": "issue:45907", + "accept": false, + "reason": "Model-loading failure and list_repo_files revision forwarding are different issues." + }, + { + "left": "issue:27957", + "right": "issue:28282", + "accept": false, + "reason": "Flash Attention 2 XLMRoberta issue is unrelated to missing PyTorch ImportError." + }, + { + "left": "issue:44479", + "right": "issue:45381", + "accept": true, + "reason": "Both describe the same Qwen2.5-VL/Qwen video-input regression; the narrower vision_position_ids report fits the broader regression issue." + }, + { + "left": "issue:42175", + "right": "issue:44488", + "accept": false, + "reason": "Backend packaging with pip extras is unrelated to loading a specific community model." + }, + { + "left": "issue:38175", + "right": "issue:44898", + "accept": false, + "reason": "Siglip2 zero-probability output and Perceiver interpolation failure are different vision-model bugs." + }, + { + "left": "issue:43257", + "right": "issue:43756", + "accept": false, + "reason": "Qwen3 MOE weight conversion and Smollm3 RoPE layer count are unrelated model-specific issues." + }, + { + "left": "issue:34567", + "right": "issue:36331", + "accept": false, + "reason": "TrainerState token-count tracking and CustomTrainer compute_loss signature mismatch are different trainer API issues." + }, + { + "left": "issue:44291", + "right": "issue:44821", + "accept": false, + "reason": "init_empty_weights unexpected argument is unrelated to loading AutoImageProcessor from URL." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and conservative overall. The selected canonical issue (44479) is supported by the title as a broader Qwen video-input regression, and the one accepted soft pair (44479 vs 45381) is a plausible same-bug match based on the packet. The remaining rejects read cautiously and are not overstated." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 22, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:24643", + "issue:28282", + "issue:29127", + "issue:32090", + "issue:43404", + "issue:43452", + "issue:43526", + "issue:43531", + "issue:43540", + "issue:43577", + "issue:43697", + "issue:43716", + "issue:43720", + "issue:43742", + "issue:43756", + "issue:43906", + "issue:44062", + "issue:44488", + "issue:44568", + "issue:44589", + "issue:44625", + "issue:44737", + "issue:44792", + "issue:44821", + "issue:44877", + "issue:44898", + "issue:44913", + "issue:44964", + "issue:45042", + "issue:45072", + "issue:45081", + "issue:45216", + "issue:45325", + "issue:45357", + "issue:45405", + "issue:45479", + "issue:45588", + "issue:45600", + "issue:45823", + "issue:45920", + "issue:45987", + "issue:46121", + "issue:46399", + "issue:46463", + "issue:46468", + "issue:46491", + "issue:46693", + "issue:46701" + ], + "soft_pairs": [ + "issue:44792|issue:44964", + "issue:44062|issue:45081", + "issue:44737|issue:46463", + "issue:43756|issue:46399", + "issue:43716|issue:46701", + "issue:44568|issue:46491", + "issue:43756|issue:45325", + "issue:43742|issue:44589", + "issue:45042|issue:45081", + "issue:43577|issue:45072", + "issue:45823|issue:46463", + "issue:43906|issue:46468", + "issue:45405|issue:45600", + "issue:43531|issue:44821", + "issue:44737|issue:45588", + "issue:43404|issue:45987", + "issue:43906|issue:46693", + "issue:43452|issue:44488", + "issue:43526|issue:45479", + "issue:44625|issue:44877", + "issue:45357|issue:45920", + "issue:43540|issue:46693", + "issue:44913|issue:46121", + "issue:43720|issue:44898", + "issue:43697|issue:45216", + "issue:28282|issue:32090", + "issue:24643|issue:29127" + ], + "prepared_review_unit_hash": "641f90ac03fab3f4d0169e7b2215fcc82db2cc1342505f9e1c36d8aebda7b116", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 36222, + "estimated_input_tokens": 9056, + "estimated_eval_tokens": 18368 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Heterogeneous cluster: the soft pairs share only broad vocabulary (tokenizer/config/multimodal/dtype/loading) but not the same concrete bug or code path, so none should be merged.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue is safe here; the items are a grab-bag of unrelated bugs, so collapsing them would over-merge distinct problems.", + "canonical_pr_reason": null, + "best_issue_reason": "If one issue must be kept as the representative, issue:45920 is the most useful stand-alone anchor because it is a broad active regression report with inbound references; still, it is not a duplicate sink for the rest of the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44792", + "right": "issue:44964", + "accept": false, + "reason": "Both are multimodal-loading failures, but one is a janus test failure and the other is a Phi-4 model load error; different models and different failure paths." + }, + { + "left": "issue:44062", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer-related, but one is an AddedToken keyword collision and the other is a Mistral regex patch crash; not the same bug." + }, + { + "left": "issue:44737", + "right": "issue:46463", + "accept": false, + "reason": "XLNet CPU device placement versus T5 decoder-input validation; unrelated code paths and symptoms." + }, + { + "left": "issue:43756", + "right": "issue:46399", + "accept": false, + "reason": "Both mention rotary-related internals, but one is Smollm3 RoPE layer dropping and the other is a Qwen3.5 kernel registration error; distinct fixes." + }, + { + "left": "issue:43716", + "right": "issue:46701", + "accept": false, + "reason": "Both are multimodal, but one is a dtype mismatch between image preprocessor and model and the other is an image-text pipeline failure for DiffusionGemmaGenerationOutput." + }, + { + "left": "issue:44568", + "right": "issue:46491", + "accept": false, + "reason": "Tokenizer regressions, but one is BOS/EOS insertion and the other is CodeLlama leading-space round-trip behavior; different underlying bugs." + }, + { + "left": "issue:43756", + "right": "issue:45325", + "accept": false, + "reason": "Both involve RoPE/position handling, but one is missing layers in Smollm3 and the other is Qwen2.5-VL still-image temporal position scaling; not mergeable into one fix." + }, + { + "left": "issue:43742", + "right": "issue:44589", + "accept": false, + "reason": "Loading problems, but one is a key error for MobileLLM and the other is a missing Float8 storage type; different failure mechanisms." + }, + { + "left": "issue:45042", + "right": "issue:45081", + "accept": false, + "reason": "Image-processor torchvision dependency versus tokenizer regex patch crash; broad subsystem overlap only, no shared bug." + }, + { + "left": "issue:43577", + "right": "issue:45072", + "accept": false, + "reason": "Both are dtype mismatches, but they affect different model families and different forward paths; too broad to treat as the same bug." + }, + { + "left": "issue:45823", + "right": "issue:46463", + "accept": false, + "reason": "Gemma4 device-map forward mismatch and T5 decoder-input validation are unrelated issues." + }, + { + "left": "issue:43906", + "right": "issue:46468", + "accept": false, + "reason": "An isolated reproduction of one issue versus a broken official Nemotron example; no evidence of the same underlying bug." + }, + { + "left": "issue:45405", + "right": "issue:45600", + "accept": false, + "reason": "PEFT version pinning versus stale auto_mappings config references; one is dependency metadata, the other is config cleanup." + }, + { + "left": "issue:43531", + "right": "issue:44821", + "accept": false, + "reason": "Qwen3-MoE sliding-window behavior and AutoImageProcessor URL loading are unrelated." + }, + { + "left": "issue:44737", + "right": "issue:45588", + "accept": false, + "reason": "XLNet CPU arange bug versus flash_attention s_aux=None crash; different modules and triggers." + }, + { + "left": "issue:43404", + "right": "issue:45987", + "accept": false, + "reason": "Mistral3 lm_head tying and StaticCache.get_seq_length return type are unrelated implementation bugs." + }, + { + "left": "issue:43906", + "right": "issue:46693", + "accept": false, + "reason": "Reproduction of an old issue versus a FA2 long-context performance regression; no shared concrete code-path defect." + }, + { + "left": "issue:43452", + "right": "issue:44488", + "accept": false, + "reason": "Both are model-loading failures, but one is gguf_file handling and the other is a specific model not loading; too different to merge." + }, + { + "left": "issue:43526", + "right": "issue:45479", + "accept": false, + "reason": "Both mention labels, but one is an image processor returning the wrong array shape and the other is a sequence-classification zero-loss regression." + }, + { + "left": "issue:44625", + "right": "issue:44877", + "accept": false, + "reason": "Configuration-related, but one is num_labels propagation in Qwen3.5 and the other is strict config blocking granite_speech loading; not the same bug." + }, + { + "left": "issue:45357", + "right": "issue:45920", + "accept": false, + "reason": "One is save_pretrained key serialization for Qwen3.5, the other is wrong token IDs in AutoTokenizer across several models; different layers of the stack." + }, + { + "left": "issue:43540", + "right": "issue:46693", + "accept": false, + "reason": "Video-input processing crash versus FA2 long-context performance regression; unrelated symptoms and code paths." + }, + { + "left": "issue:44913", + "right": "issue:46121", + "accept": false, + "reason": "Both concern RoPE config handling, but one is silent reloading of rotary_pct and the other is a TypeError from JSON-loaded ignore_keys; not the same defect." + }, + { + "left": "issue:43720", + "right": "issue:44898", + "accept": false, + "reason": "Accelerate loading of packed BitNet weights versus Perceiver non-default-resolution interpolation failure; different models and mechanisms." + }, + { + "left": "issue:43697", + "right": "issue:45216", + "accept": false, + "reason": "RTDetrV2 output drift in v5 and Qwen3.5 checkpoint save regression are separate regression classes." + }, + { + "left": "issue:28282", + "right": "issue:32090", + "accept": false, + "reason": "Environment/import error for PyTorch versus Trainer broadcast handling of NoneType; unrelated." + }, + { + "left": "issue:24643", + "right": "issue:29127", + "accept": false, + "reason": "DeepSpeed weight-shape runtime error versus LayoutLMv3 boxes-validation messaging; different bugs and code paths." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 23, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:28282", + "issue:29159", + "issue:33357", + "issue:38175", + "issue:41762", + "issue:42898", + "issue:43295", + "issue:43452", + "issue:43525", + "issue:43531", + "issue:43577", + "issue:43643", + "issue:43646", + "issue:43653", + "issue:43688", + "issue:43697", + "issue:43742", + "issue:43749", + "issue:43756", + "issue:43819", + "issue:43994", + "issue:44361", + "issue:44442", + "issue:44464", + "issue:44466", + "issue:44493", + "issue:44568", + "issue:44743", + "issue:44811", + "issue:44843", + "issue:44898", + "issue:44936", + "issue:44991", + "issue:45005", + "issue:45200", + "issue:45356", + "issue:45372", + "issue:45375", + "issue:45542", + "issue:45704", + "issue:45823", + "issue:46050", + "issue:46170", + "issue:46178", + "issue:46402", + "issue:46463", + "issue:46491", + "issue:46701" + ], + "soft_pairs": [ + "issue:43643|issue:45375", + "issue:45200|issue:46463", + "issue:28282|issue:33357", + "issue:44466|issue:45005", + "issue:41762|issue:44936", + "issue:43688|issue:43749", + "issue:43452|issue:44843", + "issue:44361|issue:44442", + "issue:44493|issue:46050", + "issue:43525|issue:46701", + "issue:43819|issue:44811", + "issue:43653|issue:44361", + "issue:45372|issue:46170", + "issue:43994|issue:44898", + "issue:38175|issue:43295", + "issue:44936|issue:45823", + "issue:42898|issue:46491", + "issue:45542|issue:46178", + "issue:43295|issue:45356", + "issue:43531|issue:43742", + "issue:43653|issue:44568", + "issue:43452|issue:43577", + "issue:43577|issue:43697", + "issue:43646|issue:43749", + "issue:43756|issue:46402", + "issue:43688|issue:45704", + "issue:44843|issue:44991", + "issue:29159|issue:33357", + "issue:44464|issue:44743" + ], + "prepared_review_unit_hash": "85886fe8abea617f86fe74f8680524f2d02656ba2070b6828eff34dc44e43701", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36688, + "estimated_input_tokens": 9172, + "estimated_eval_tokens": 18600 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab bag of unrelated Transformers bug reports across tokenizers, model loading, generation, and backend-specific regressions. I would not merge any of the soft pairs as duplicates; they do not appear to share the same concrete code-path failure.", + "confidence": 0.95, + "canonical_issue_reason": "There is no strong single canonical duplicate here because the issues span different models, symptoms, and subsystems. If a representative anchor is needed, issue 44843 is the most concrete and broadly actionable loading-regression report in the set.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44843 is the best representative issue: it has a specific reproducible regression, a clear root area (_patch_mistral_regex / model_info), and a well-defined failure mode. The other issues are mostly model-specific or symptom-level reports that are not duplicate-equivalent.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43643", + "right": "issue:45375", + "accept": false, + "reason": "Both mention missing config fields, but they are different models and likely different code paths (trust_remote_code config loading vs @strict dropping a Qwen3 field)." + }, + { + "left": "issue:45200", + "right": "issue:46463", + "accept": false, + "reason": "Unrelated model-family issues: Gemma 4 mm_token_type_ids defaults vs T5 decoder-input validation/error messaging." + }, + { + "left": "issue:28282", + "right": "issue:33357", + "accept": false, + "reason": "Different failures entirely: missing PyTorch ImportError vs MacOS bus error loading a CLIP community model." + }, + { + "left": "issue:44466", + "right": "issue:45005", + "accept": false, + "reason": "Both involve tied weights, but one is a device-dependent serialization bug and the other is a broader translation-model issue; not clearly the same concrete defect." + }, + { + "left": "issue:41762", + "right": "issue:44936", + "accept": false, + "reason": "Different surfaces and code paths: ZeRO-3 loading IndexError vs trainer.evaluate() failing after training." + }, + { + "left": "issue:43688", + "right": "issue:43749", + "accept": false, + "reason": "Different subsystems: auxiliary-loss normalization in MoE models vs FSDP CPU RAM efficient loading." + }, + { + "left": "issue:43452", + "right": "issue:44843", + "accept": false, + "reason": "Both touch AutoTokenizer.from_pretrained, but one is about gguf_file handling and the other about an unconditional model_info() call breaking offline mode." + }, + { + "left": "issue:44361", + "right": "issue:44442", + "accept": false, + "reason": "Different tokenizer classes and failures; no evidence they share the same underlying bug." + }, + { + "left": "issue:44493", + "right": "issue:46050", + "accept": false, + "reason": "One is a position-id key regression, the other is quantization save/load corruption." + }, + { + "left": "issue:43525", + "right": "issue:46701", + "accept": false, + "reason": "AttributeError in Llama4Config pad_token_id vs a diffusion/image-text pipeline failure; unrelated." + }, + { + "left": "issue:43819", + "right": "issue:44811", + "accept": false, + "reason": "Different behaviors: DAC from_latents vs Whisper batch_decode skip_special_tokens handling." + }, + { + "left": "issue:43653", + "right": "issue:44361", + "accept": false, + "reason": "Tokenizer bugs in different models and different failure modes; not the same underlying issue." + }, + { + "left": "issue:45372", + "right": "issue:46170", + "accept": false, + "reason": "Different dependency problems: mistral_common ReasoningEffort import vs torch.distributed dependency on Windows/ROCm." + }, + { + "left": "issue:43994", + "right": "issue:44898", + "accept": false, + "reason": "Different vision-model failures with different symptoms and model families." + }, + { + "left": "issue:38175", + "right": "issue:43295", + "accept": false, + "reason": "Siglip2 zero-probabilities bug vs processor.tokenizer/image-passing regression; unrelated." + }, + { + "left": "issue:44936", + "right": "issue:45823", + "accept": false, + "reason": "Training/evaluation lifecycle failure vs Gemma4 device_map auto device mismatch." + }, + { + "left": "issue:42898", + "right": "issue:46491", + "accept": false, + "reason": "Both are tokenizer-behavior regressions, but one is a v5 global cleanup behavior change and the other is a CodeLlama-specific round-trip space loss." + }, + { + "left": "issue:45542", + "right": "issue:46178", + "accept": false, + "reason": "Both are backend availability issues, but for different backends and different error surfaces; not a duplicate bug." + }, + { + "left": "issue:43295", + "right": "issue:45356", + "accept": false, + "reason": "Different regressions affecting different tokenizer/processor paths and different model families." + }, + { + "left": "issue:43531", + "right": "issue:43742", + "accept": false, + "reason": "Different model-loading/generation issues with no shared concrete failure path." + }, + { + "left": "issue:43653", + "right": "issue:44568", + "accept": false, + "reason": "Different tokenizer bugs: special-token registration vs add_special_tokens not emitting BOS/EOS." + }, + { + "left": "issue:43452", + "right": "issue:43577", + "accept": false, + "reason": "Tokenizer/model loading with gguf_file vs BLIP2 dtype propagation; unrelated." + }, + { + "left": "issue:43577", + "right": "issue:43697", + "accept": false, + "reason": "Different problems: BLIP2 dtype staying float32 vs RTDetrV2 output differences." + }, + { + "left": "issue:43646", + "right": "issue:43749", + "accept": false, + "reason": "AutoConfig trust_remote_code field loss vs FSDP CPU RAM efficient loading; unrelated." + }, + { + "left": "issue:43756", + "right": "issue:46402", + "accept": false, + "reason": "Smollm3 RoPE layer issue vs Sam3Model text encoder weight loading; different model-specific bugs." + }, + { + "left": "issue:43688", + "right": "issue:45704", + "accept": false, + "reason": "Different issues in different models: MoE auxiliary-loss normalization vs T5 apex FusedRMSNorm memory leak." + }, + { + "left": "issue:44843", + "right": "issue:44991", + "accept": false, + "reason": "Both concern tokenizer loading, but one is an offline-mode regression from a specific regex patch and the other is a model-specific tokenizer load failure." + }, + { + "left": "issue:29159", + "right": "issue:33357", + "accept": false, + "reason": "Tokenizer inconsistency vs MacOS bus error; no shared code-path." + }, + { + "left": "issue:44464", + "right": "issue:44743", + "accept": false, + "reason": "Both are generation/caching consistency reports, but they describe different mechanisms and are not obviously mergeable into one fix." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 24, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:32090", + "issue:33666", + "issue:36010", + "issue:36331", + "issue:39692", + "issue:41720", + "issue:43262", + "issue:43278", + "issue:43295", + "issue:43493", + "issue:43504", + "issue:43550", + "issue:43577", + "issue:43606", + "issue:43643", + "issue:43653", + "issue:43688", + "issue:43720", + "issue:43742", + "issue:43782", + "issue:43825", + "issue:43927", + "issue:44361", + "issue:44464", + "issue:44466", + "issue:44479", + "issue:44484", + "issue:44514", + "issue:44554", + "issue:44561", + "issue:44610", + "issue:44625", + "issue:44704", + "issue:44734", + "issue:44821", + "issue:44964", + "issue:44977", + "issue:45290", + "issue:45405", + "issue:45440", + "issue:45491", + "issue:45701", + "issue:46002", + "issue:46050", + "issue:46209", + "issue:46365", + "issue:46612", + "issue:46693" + ], + "soft_pairs": [ + "issue:43653|issue:43927", + "issue:43688|issue:45701", + "issue:43688|issue:45440", + "issue:44484|issue:44704", + "issue:44479|issue:44625", + "issue:45491|issue:46002", + "issue:43550|issue:44361", + "issue:33666|issue:36010", + "issue:32090|issue:33666", + "issue:44734|issue:46612", + "issue:41720|issue:44464", + "issue:43742|issue:44821", + "issue:44466|issue:46050", + "issue:44464|issue:44977", + "issue:43504|issue:43550", + "issue:43504|issue:43720", + "issue:44977|issue:46693", + "issue:43643|issue:44561", + "issue:43504|issue:44610", + "issue:43577|issue:44554", + "issue:43720|issue:43927", + "issue:43782|issue:46693", + "issue:43606|issue:43927", + "issue:33666|issue:36331", + "issue:43278|issue:43688", + "issue:44514|issue:45290", + "issue:44554|issue:46365", + "issue:43262|issue:43295", + "issue:43720|issue:44361", + "issue:39692|issue:43493", + "issue:44964|issue:45405", + "issue:43493|issue:43825", + "issue:44977|issue:46209" + ], + "prepared_review_unit_hash": "ed5cc5b7e81635e6ce655abbbd0fdfef09b3f68e6dc3ce27907c226ce68880b9", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37716, + "estimated_input_tokens": 9429, + "estimated_eval_tokens": 19114 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a heterogeneous mix of unrelated Transformers bugs (trainer, tokenizers, multimodal processors, generation, loading, and model-specific regressions). The soft-similarity pairs mostly share broad keywords but not the same concrete failure mode or code path, so none look like true duplicates. No PRs are present.", + "confidence": 0.21, + "canonical_issue_reason": "Issue 43688 is the closest thing to a representative bug report here: it is a concrete model correctness regression with enough discussion to serve as a cluster anchor, even though the surrounding items are not true duplicates of it.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43688 is the best single issue to represent the set because it is a substantive, non-doc bug with a clear technical symptom; however, this is only a representative choice, not evidence of duplication.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43653", + "right": "issue:43927", + "accept": false, + "reason": "Both involve token/config handling, but one is a tokenizer special-token registration bug and the other is a config save/load field-loss bug; different root causes and code paths." + }, + { + "left": "issue:43688", + "right": "issue:45701", + "accept": false, + "reason": "Aux-loss normalization and tokenization-version changes are unrelated problems despite both being model behavior issues." + }, + { + "left": "issue:43688", + "right": "issue:45440", + "accept": false, + "reason": "Different models and different failures: auxiliary-loss normalization vs. DeepSeek implementation divergence." + }, + { + "left": "issue:44484", + "right": "issue:44704", + "accept": false, + "reason": "Save-pretrained shard size behavior is unrelated to AutoProcessor kwargs propagation." + }, + { + "left": "issue:44479", + "right": "issue:44625", + "accept": false, + "reason": "A multimodal video-input regression and a Qwen3.5 num_labels propagation bug are not the same defect." + }, + { + "left": "issue:45491", + "right": "issue:46002", + "accept": false, + "reason": "Both mention NaNs, but one is Gemma3 sliding-window attention on GPU and the other is classification-head init/checkout; different failure mechanisms." + }, + { + "left": "issue:43550", + "right": "issue:44361", + "accept": false, + "reason": "torch.compile/SDPA on Bamba is unrelated to an MLukeTokenizer AttributeError." + }, + { + "left": "issue:33666", + "right": "issue:36010", + "accept": false, + "reason": "Multi-GPU Qwen2-VL training and a GenerationMixin import error are unrelated." + }, + { + "left": "issue:32090", + "right": "issue:33666", + "accept": false, + "reason": "Both touch Trainer/multi-GPU contexts, but one is a NoneType broadcast error and the other is a Qwen2-VL training request; not the same bug." + }, + { + "left": "issue:44734", + "right": "issue:46612", + "accept": false, + "reason": "Both involve generation caches, but one is KV-cache continuation tensor indexing in serve and the other is beam-search cache reorder being skipped for specific model classes." + }, + { + "left": "issue:41720", + "right": "issue:44464", + "accept": false, + "reason": "Auto device-map cudaErrorAssert on Qwen3 is unrelated to compiled-forward chunked generation inconsistency." + }, + { + "left": "issue:43742", + "right": "issue:44821", + "accept": false, + "reason": "Model loading KeyError and AutoImageProcessor-from-URL failure are different loading paths and symptoms." + }, + { + "left": "issue:44466", + "right": "issue:46050", + "accept": false, + "reason": "Both concern save/load behavior, but one is tied-weight serialization and the other is quantized sinq round-tripping; different concrete bugs." + }, + { + "left": "issue:44464", + "right": "issue:44977", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency is not the same as a flash-attention-specific Qwen3.5 generation failure." + }, + { + "left": "issue:43504", + "right": "issue:43550", + "accept": false, + "reason": "A legacy-field model-load failure and a Bamba torch.compile/SDPA failure are unrelated." + }, + { + "left": "issue:43504", + "right": "issue:43720", + "accept": false, + "reason": "Both are loading-related, but one is a Beit legacy config field issue and the other is packed-weight unpacking during accelerate loading." + }, + { + "left": "issue:44977", + "right": "issue:46693", + "accept": false, + "reason": "One is a correctness bug in Qwen3.5 generation; the other is a performance regression with FA2 long-context generation." + }, + { + "left": "issue:43643", + "right": "issue:44561", + "accept": false, + "reason": "Both mention trust_remote_code, but one is missing fields in AutoConfig and the other is removal of is_torch_fx_available breaking remote-code models." + }, + { + "left": "issue:43504", + "right": "issue:44610", + "accept": false, + "reason": "Legacy config loading and processor/model input-size mismatch are unrelated." + }, + { + "left": "issue:43577", + "right": "issue:44554", + "accept": false, + "reason": "A dtype-loading bug in BLIP2 is unrelated to an MPS attention correctness issue." + }, + { + "left": "issue:43720", + "right": "issue:43927", + "accept": false, + "reason": "BitNet packed-weight loading and DiaConfig custom token ID persistence are different model/config issues." + }, + { + "left": "issue:43782", + "right": "issue:46693", + "accept": false, + "reason": "Qwen3VL weight_only loading error is unrelated to the FA2 long-context performance regression." + }, + { + "left": "issue:43606", + "right": "issue:43927", + "accept": false, + "reason": "CPU offload device mismatch and config custom-token-ID loss are unrelated." + }, + { + "left": "issue:33666", + "right": "issue:36331", + "accept": false, + "reason": "Qwen2-VL multi-GPU training and CustomTrainer.compute_loss signature mismatch are different Trainer issues." + }, + { + "left": "issue:43278", + "right": "issue:43688", + "accept": false, + "reason": "Embedding dtype drift across train/eval is not the same as auxiliary-loss normalization." + }, + { + "left": "issue:44514", + "right": "issue:45290", + "accept": false, + "reason": "Both are apply_chat_template crashes, but the triggers differ materially: batched padding=False vs. assistant messages with tool calls and no content." + }, + { + "left": "issue:44554", + "right": "issue:46365", + "accept": false, + "reason": "An MPS attention correctness issue is unrelated to an import conflict preventing Trainer from loading." + }, + { + "left": "issue:43262", + "right": "issue:43295", + "accept": false, + "reason": "Audio chat-template sampling-rate defaulting and processor.tokenizer/image-tokenization regression are different processor bugs." + }, + { + "left": "issue:44964", + "right": "issue:45405", + "accept": false, + "reason": "A Phi-4 multimodal load failure and a bumped MIN_PEFT_VERSION are unrelated." + }, + { + "left": "issue:43493", + "right": "issue:43825", + "accept": false, + "reason": "SigLIP2 implementation divergence and a pipeline error-message regression are unrelated." + }, + { + "left": "issue:39692", + "right": "issue:43493", + "accept": false, + "reason": "One is a documentation example bug; the other is a core implementation discrepancy for SigLIP2. Different artifacts and fixes." + }, + { + "left": "issue:44977", + "right": "issue:46209", + "accept": false, + "reason": "Qwen3.5 flash-attention generation failure and FP8 inference failure on B200 are different model/runtime problems." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 25, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:28218", + "issue:33357", + "issue:33666", + "issue:39692", + "issue:43257", + "issue:43278", + "issue:43295", + "issue:43299", + "issue:43334", + "issue:43452", + "issue:43493", + "issue:43504", + "issue:43525", + "issue:43572", + "issue:43575", + "issue:43606", + "issue:43646", + "issue:43653", + "issue:43701", + "issue:43824", + "issue:43844", + "issue:43927", + "issue:43994", + "issue:44112", + "issue:44361", + "issue:44451", + "issue:44466", + "issue:44488", + "issue:44493", + "issue:44568", + "issue:44589", + "issue:44625", + "issue:44792", + "issue:44843", + "issue:44849", + "issue:45357", + "issue:45405", + "issue:45520", + "issue:45584", + "issue:45759", + "issue:45812", + "issue:45823", + "issue:46050", + "issue:46192", + "issue:46209", + "issue:46285", + "issue:46489" + ], + "soft_pairs": [ + "issue:43452|issue:44451", + "issue:43525|issue:43572", + "issue:44625|issue:45584", + "issue:45823|issue:46050", + "issue:28218|issue:33357", + "issue:43994|issue:46192", + "issue:43278|issue:43844", + "issue:43493|issue:43844", + "issue:39692|issue:43295", + "issue:44849|issue:46209", + "issue:44568|issue:46489", + "issue:33357|issue:33666", + "issue:43646|issue:44466", + "issue:44488|issue:44843", + "issue:44568|issue:45812", + "issue:43278|issue:43701", + "issue:43927|issue:44361", + "issue:44849|issue:45357", + "issue:44792|issue:45405", + "issue:43575|issue:45405", + "issue:43257|issue:43299", + "issue:43504|issue:43606", + "issue:44589|issue:45520", + "issue:43606|issue:44112", + "issue:43824|issue:44589", + "issue:43334|issue:43572", + "issue:43334|issue:45759", + "issue:44493|issue:46285", + "issue:43504|issue:43653" + ], + "prepared_review_unit_hash": "90a8e403a386e74e9e2dfb50c6f95cd59c129c66d38c1d7b4ec868e8155a91db", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35919, + "estimated_input_tokens": 8980, + "estimated_eval_tokens": 18216 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose set of unrelated tokenizer, config, loading, model, and CI regressions. None of the soft edges look like true duplicates, so there is no single canonical issue here.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43452", + "right": "issue:44451", + "accept": false, + "reason": "Different models and failure modes: gguf/from_pretrained breakage versus a specific ScandiBERT load failure." + }, + { + "left": "issue:43525", + "right": "issue:43572", + "accept": false, + "reason": "Both mention missing token-id fields, but they affect different configs/models and appear to be separate regressions." + }, + { + "left": "issue:44625", + "right": "issue:45584", + "accept": false, + "reason": "Unrelated bugs: Qwen3.5 label propagation versus Whisper empty-transcription generation after special-token alignment." + }, + { + "left": "issue:45823", + "right": "issue:46050", + "accept": false, + "reason": "Different subsystems: device-map forward mismatch versus quantized model save/load corruption." + }, + { + "left": "issue:28218", + "right": "issue:33357", + "accept": false, + "reason": "Tokenizer spacing bug versus MacOS bus error on a CLIP model; no shared code-path." + }, + { + "left": "issue:43994", + "right": "issue:46192", + "accept": false, + "reason": "SigLIP2 output quality issue versus return_tensors being ignored when text_kwargs is passed." + }, + { + "left": "issue:43278", + "right": "issue:43844", + "accept": false, + "reason": "Both involve training behavior, but one is a dtype flip at eval and the other is a ZeRO-3 gradient anomaly." + }, + { + "left": "issue:43493", + "right": "issue:43844", + "accept": false, + "reason": "SigLIP2 implementation discrepancy is unrelated to the ZeRO-3 gradient increase bug." + }, + { + "left": "issue:39692", + "right": "issue:43295", + "accept": false, + "reason": "Doc/example errors are not the same bug as the processor.tokenizer regression in custom model code." + }, + { + "left": "issue:44849", + "right": "issue:46209", + "accept": false, + "reason": "Different domains: Qwen3.5 hidden-state behavior versus FP8 inference failure on B200 hardware." + }, + { + "left": "issue:44568", + "right": "issue:46489", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different tokenizers and different concrete behaviors." + }, + { + "left": "issue:33357", + "right": "issue:33666", + "accept": false, + "reason": "MacOS bus error on CLIP loading is unrelated to Qwen2-VL multi-GPU training." + }, + { + "left": "issue:43646", + "right": "issue:44466", + "accept": false, + "reason": "Custom model initialization breakage is not the same bug as device-dependent lm_head serialization." + }, + { + "left": "issue:44488", + "right": "issue:44843", + "accept": false, + "reason": "Loading a specific model family is different from an offline-mode failure caused by an unconditional hub call." + }, + { + "left": "issue:44568", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer-related, but the affected models and symptoms differ: missing BOS/EOS versus wrong token IDs for Granite." + }, + { + "left": "issue:43278", + "right": "issue:43701", + "accept": false, + "reason": "Embedding dtype drift in eval is unrelated to resume_from_checkpoint key mismatch." + }, + { + "left": "issue:43927", + "right": "issue:44361", + "accept": false, + "reason": "Different root causes: lost custom token IDs in DiaConfig versus an AttributeError in MLukeTokenizer task handling." + }, + { + "left": "issue:44849", + "right": "issue:45357", + "accept": false, + "reason": "Qwen3.5 hidden-state bug and incorrect visual-encoder keys in save_pretrained are separate regressions." + }, + { + "left": "issue:44792", + "right": "issue:45405", + "accept": false, + "reason": "A failing janus test is not the same issue as the PEFT minimum-version bump." + }, + { + "left": "issue:43575", + "right": "issue:45405", + "accept": false, + "reason": "TP OOM on a large Qwen2 model is unrelated to the PEFT dependency version change." + }, + { + "left": "issue:43257", + "right": "issue:43299", + "accept": false, + "reason": "Both involve Qwen MOE models, but one is weight conversion during accelerate/deepspeed load and the other is model loading breakage in dev." + }, + { + "left": "issue:43504", + "right": "issue:43606", + "accept": false, + "reason": "BEiT legacy-field loading and Bark CPU-offload device mismatch are unrelated bugs." + }, + { + "left": "issue:44589", + "right": "issue:45520", + "accept": false, + "reason": "Float8 storage lookup failure is a different issue from the flash_attn import KeyError on Python 3.13." + }, + { + "left": "issue:43606", + "right": "issue:44112", + "accept": false, + "reason": "A real CPU-offload device mismatch is not the same as a stale CI test in GraniteSpeech." + }, + { + "left": "issue:43824", + "right": "issue:44589", + "accept": false, + "reason": "Importing a missing Qwen2.5-VL class is unrelated to the Float8 storage-object TypeError." + }, + { + "left": "issue:43334", + "right": "issue:43572", + "accept": false, + "reason": "Similar missing token-id theme, but different models and different attribute names imply different bugs." + }, + { + "left": "issue:43334", + "right": "issue:45759", + "accept": false, + "reason": "Missing pad_token_id in Qwen3-VL is not the same as failing to unwrap text_config in composite Qwen configs." + }, + { + "left": "issue:44493", + "right": "issue:46285", + "accept": false, + "reason": "Unexpected position-id keys across many models is a different structural regression than CLIP text-model class inconsistency." + }, + { + "left": "issue:43504", + "right": "issue:43653", + "accept": false, + "reason": "BEiT legacy-field loading and BigBirdTokenizer special-token registration 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": 22, + "review_unit_index": 26, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:15354", + "issue:28218", + "issue:29159", + "issue:29942", + "issue:30333", + "issue:32090", + "issue:34567", + "issue:37428", + "issue:38175", + "issue:42490", + "issue:43065", + "issue:43295", + "issue:43450", + "issue:43493", + "issue:43540", + "issue:43572", + "issue:43696", + "issue:43701", + "issue:43844", + "issue:43873", + "issue:43937", + "issue:43957", + "issue:43994", + "issue:44291", + "issue:44423", + "issue:44464", + "issue:44493", + "issue:44554", + "issue:44734", + "issue:44964", + "issue:44977", + "issue:45072", + "issue:45084", + "issue:45198", + "issue:45200", + "issue:45229", + "issue:45406", + "issue:45446", + "issue:45520", + "issue:45584", + "issue:45588", + "issue:45759", + "issue:45823", + "issue:46050", + "issue:46121", + "issue:46369", + "issue:46424", + "issue:46701" + ], + "soft_pairs": [ + "issue:45200|issue:45229", + "issue:43572|issue:44291", + "issue:44464|issue:46424", + "issue:43937|issue:46701", + "issue:45229|issue:45823", + "issue:44464|issue:45584", + "issue:43844|issue:44964", + "issue:43873|issue:46050", + "issue:43994|issue:44493", + "issue:45588|issue:46050", + "issue:29942|issue:30333", + "issue:42490|issue:43696", + "issue:29159|issue:32090", + "issue:44977|issue:46701", + "issue:45084|issue:45520", + "issue:43065|issue:46369", + "issue:28218|issue:34567", + "issue:44423|issue:44734", + "issue:44554|issue:45446", + "issue:15354|issue:38175", + "issue:45406|issue:46121", + "issue:43493|issue:43701", + "issue:45072|issue:45198", + "issue:45588|issue:45759", + "issue:37428|issue:43957", + "issue:43540|issue:45200", + "issue:43295|issue:43450", + "issue:45200|issue:45823" + ], + "prepared_review_unit_hash": "0a496ec75b7255cd03892a2ea15c91a164d851f9a02612ebd06f6938c9a1fbab", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "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": "The cluster is heterogeneous; the soft pairs are mostly broad topical similarities rather than the same underlying bug. I would not merge any of the suggested edges, and there is no clear canonical issue/PR for the whole cluster.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45200", + "right": "issue:45229", + "accept": false, + "reason": "Gemma 4 token-type defaults vs a multi-GPU OOM; different failure modes and code paths." + }, + { + "left": "issue:43572", + "right": "issue:44291", + "accept": false, + "reason": "Both are loading-related regressions, but one is a missing config field and the other is an unexpected init argument." + }, + { + "left": "issue:44464", + "right": "issue:46424", + "accept": false, + "reason": "Both mention caching, but one is compiled-forward generation mismatch and the other is a cache-length config limitation." + }, + { + "left": "issue:43937", + "right": "issue:46701", + "accept": false, + "reason": "GenerationConfig validation for GLM-5 is unrelated to a diffusion/image-text pipeline failure." + }, + { + "left": "issue:45229", + "right": "issue:45823", + "accept": false, + "reason": "Gemma 4 OOM during inference is not the same as the device mismatch bug in forward." + }, + { + "left": "issue:44464", + "right": "issue:45584", + "accept": false, + "reason": "Compiled-forward chunked generation mismatch is unrelated to Whisper empty-transcription handling." + }, + { + "left": "issue:43844", + "right": "issue:44964", + "accept": false, + "reason": "ZeRO-3 gradient growth and Phi-4 multimodal loading are unrelated bugs." + }, + { + "left": "issue:43873", + "right": "issue:46050", + "accept": false, + "reason": "Quantization offloading behavior and sinq save/load breakage are different defects." + }, + { + "left": "issue:43994", + "right": "issue:44493", + "accept": false, + "reason": "SigLIP2 wrong outputs and unexpected position-id keys are not the same code-path issue." + }, + { + "left": "issue:45588", + "right": "issue:46050", + "accept": false, + "reason": "Flash-attention s_aux=None crash is unrelated to sinq quantization persistence." + }, + { + "left": "issue:29942", + "right": "issue:30333", + "accept": false, + "reason": "Failing attention tests and MLflow reporting on job failure are unrelated." + }, + { + "left": "issue:42490", + "right": "issue:43696", + "accept": false, + "reason": "Import warning on trailing slash vs CUDA out-of-memory; no shared bug." + }, + { + "left": "issue:29159", + "right": "issue:32090", + "accept": false, + "reason": "Tokenizer slow/fast inconsistency is unrelated to a NoneType broadcast error in Trainer." + }, + { + "left": "issue:44977", + "right": "issue:46701", + "accept": false, + "reason": "Qwen3.5 flash-attention generation and DiffusionGemma pipeline failure are different models and paths." + }, + { + "left": "issue:45084", + "right": "issue:45520", + "accept": false, + "reason": "Template-node compilation error and flash_attn import lookup error are distinct issues." + }, + { + "left": "issue:43065", + "right": "issue:46369", + "accept": false, + "reason": "Dummy Conv2d in Sam3PixelDecoder is not the same as memory overflow in sam3-litetext." + }, + { + "left": "issue:28218", + "right": "issue:34567", + "accept": false, + "reason": "Added-token spacing behavior is unrelated to TrainerState token-count tracking." + }, + { + "left": "issue:44423", + "right": "issue:44734", + "accept": false, + "reason": "Continuous-batching multimodal crash and KV-cache continuation indexing bug are different serve failures." + }, + { + "left": "issue:44554", + "right": "issue:45446", + "accept": false, + "reason": "MPS attention correctness and flex_attention version-check import are unrelated." + }, + { + "left": "issue:15354", + "right": "issue:38175", + "accept": false, + "reason": "TorchScript GeneratorExp export failure is unrelated to SigLIP2 zero-probability outputs." + }, + { + "left": "issue:45406", + "right": "issue:46121", + "accept": false, + "reason": "Gemma4Processor missing _tokenizer and rope-params list handling are different bugs." + }, + { + "left": "issue:43493", + "right": "issue:43701", + "accept": false, + "reason": "SigLIP2 HF-vs-JAX discrepancy and checkpoint resume key mismatch are unrelated." + }, + { + "left": "issue:45072", + "right": "issue:45198", + "accept": false, + "reason": "bfloat16 dtype mismatch in inference and Wav2Vec2 save/tokenization failure are separate problems." + }, + { + "left": "issue:45588", + "right": "issue:45759", + "accept": false, + "reason": "Flash-attention sink handling and Qwen multimodal config unwrapping are not the same underlying bug." + }, + { + "left": "issue:37428", + "right": "issue:43957", + "accept": false, + "reason": "Flash-support import error and meta-device model loading breakage are different loading regressions." + }, + { + "left": "issue:43540", + "right": "issue:45200", + "accept": false, + "reason": "Video input processing error in Qwen3OmniMoe is unrelated to Gemma 4 token-type defaults." + }, + { + "left": "issue:43295", + "right": "issue:43450", + "accept": false, + "reason": "Processor.tokenizer regression and batched video shape bug do not share the same code path." + }, + { + "left": "issue:45200", + "right": "issue:45823", + "accept": false, + "reason": "Token-type-id defaults and PLE device mismatch are separate Gemma 4 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": 22, + "review_unit_index": 27, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:28218", + "issue:35141", + "issue:37428", + "issue:38175", + "issue:39692", + "issue:42490", + "issue:42915", + "issue:43232", + "issue:43262", + "issue:43278", + "issue:43295", + "issue:43334", + "issue:43450", + "issue:43452", + "issue:43493", + "issue:43525", + "issue:43550", + "issue:43572", + "issue:43575", + "issue:43653", + "issue:43824", + "issue:43844", + "issue:43856", + "issue:43873", + "issue:43950", + "issue:43957", + "issue:43975", + "issue:44112", + "issue:44186", + "issue:44466", + "issue:44521", + "issue:44554", + "issue:44610", + "issue:44849", + "issue:44898", + "issue:44987", + "issue:45127", + "issue:45290", + "issue:45356", + "issue:45381", + "issue:45412", + "issue:45520", + "issue:45823", + "issue:46178", + "issue:46192", + "issue:46209", + "issue:46285", + "issue:46512" + ], + "soft_pairs": [ + "issue:44186|issue:46192", + "issue:45823|issue:46192", + "issue:43295|issue:43873", + "issue:43295|issue:44610", + "issue:43550|issue:43844", + "issue:45520|issue:46178", + "issue:43450|issue:43575", + "issue:43232|issue:44849", + "issue:43572|issue:44987", + "issue:44554|issue:45356", + "issue:43572|issue:43957", + "issue:44466|issue:45127", + "issue:37428|issue:43824", + "issue:42490|issue:43493", + "issue:39692|issue:46192", + "issue:44521|issue:45290", + "issue:45381|issue:46209", + "issue:43653|issue:44112", + "issue:38175|issue:43493", + "issue:43278|issue:43856", + "issue:28218|issue:35141", + "issue:43950|issue:46285", + "issue:42915|issue:43824", + "issue:43452|issue:46512", + "issue:43334|issue:43525", + "issue:43824|issue:43975", + "issue:44898|issue:45412", + "issue:43262|issue:43450" + ], + "prepared_review_unit_hash": "78001f71b6c7f784cb9e13ed4a7fd5b9df2504d48a0a4bbc0dd79652afacfb66", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "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 mostly heterogeneous; the soft pairs share broad subsystems at most, but none look like the same concrete bug/change. I would use issue 44521 as the representative issue because it is the most central/high-signal open bug in the closest recurring theme here (`apply_chat_template` / multimodal processor behavior).", + "confidence": 0.81, + "canonical_issue_reason": "Issue 44521 is the best representative because it has the strongest discussion/inbound-reference signal and describes a broad, user-facing multimodal `apply_chat_template` failure, which is the closest repeated theme in the cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44521 is the most suitable anchor for this cluster: it is active, broadly applicable, and more central than the other candidates, which are mostly narrow, unrelated regressions.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44186", + "right": "issue:46192", + "accept": false, + "reason": "Both are tokenizer-related, but one is a LayoutLMv2 batched NER/padding crash and the other is `return_tensors` being ignored when `text_kwargs` is passed; different code paths." + }, + { + "left": "issue:45823", + "right": "issue:46192", + "accept": false, + "reason": "Gemma4 device-map forward mismatch vs processor kwargs handling; unrelated failures." + }, + { + "left": "issue:43295", + "right": "issue:43873", + "accept": false, + "reason": "Processor/tokenizer API regression and quantization offloading are different subsystems and bugs." + }, + { + "left": "issue:43295", + "right": "issue:44610", + "accept": false, + "reason": "Both involve processors, but one is `processor.tokenizer`/image passthrough regression and the other is an OmDet-Turbo input-size mismatch." + }, + { + "left": "issue:43550", + "right": "issue:43844", + "accept": false, + "reason": "torch.compile+SDPA correctness bug vs DeepSpeed ZeRO-3 gradient growth; no shared code path." + }, + { + "left": "issue:45520", + "right": "issue:46178", + "accept": false, + "reason": "Both are import/backend errors, but for different symbols and different backend registry problems." + }, + { + "left": "issue:43450", + "right": "issue:43575", + "accept": false, + "reason": "Video processor batch-shape bug vs tensor-parallel OOM when loading a model; unrelated." + }, + { + "left": "issue:43232", + "right": "issue:44849", + "accept": false, + "reason": "Generation kwargs update after `sync_gpus` vs Qwen3.5 hidden-states bug; different behaviors." + }, + { + "left": "issue:43572", + "right": "issue:44987", + "accept": false, + "reason": "A missing `pad_token_idx` config issue and a model-loading failure are too different to treat as the same bug." + }, + { + "left": "issue:44554", + "right": "issue:45356", + "accept": false, + "reason": "MPS attention correctness issue vs tokenizer codec/regex regression; unrelated." + }, + { + "left": "issue:43572", + "right": "issue:43957", + "accept": false, + "reason": "StableLm pad-token config bug vs meta-device loading regression; different failures." + }, + { + "left": "issue:44466", + "right": "issue:45127", + "accept": false, + "reason": "Both touch tied weights, but one is device-dependent serialization and the other is LoRA merge collapse after vocab extension; not the same bug." + }, + { + "left": "issue:37428", + "right": "issue:43824", + "accept": false, + "reason": "Different missing-import symbols from different modules; not duplicates." + }, + { + "left": "issue:42490", + "right": "issue:43493", + "accept": false, + "reason": "Trailing-slash local-path warning vs SigLIP2 implementation discrepancy; unrelated." + }, + { + "left": "issue:39692", + "right": "issue:46192", + "accept": false, + "reason": "SigLIP2 doc-example errors vs processor `return_tensors` handling; different issues." + }, + { + "left": "issue:44521", + "right": "issue:45290", + "accept": false, + "reason": "Both mention `apply_chat_template`, but one is all-zero assistant masks for multimodal inputs and the other is a crash on tool-call assistant messages with no content." + }, + { + "left": "issue:45381", + "right": "issue:46209", + "accept": false, + "reason": "Qwen2.5-VL video position IDs vs FP8 inference failure on B200; unrelated." + }, + { + "left": "issue:43653", + "right": "issue:44112", + "accept": false, + "reason": "BigBirdTokenizer special-token bug vs a stale GraniteSpeech CI test; no shared underlying defect." + }, + { + "left": "issue:38175", + "right": "issue:43493", + "accept": false, + "reason": "Both are SigLIP2-related, but one reports zero probabilities and the other a HF/JAX implementation discrepancy; not enough to call them the same concrete bug." + }, + { + "left": "issue:43278", + "right": "issue:43856", + "accept": false, + "reason": "Embedding dtype regression vs Qwen3 MoE memory inefficiency; different code paths." + }, + { + "left": "issue:28218", + "right": "issue:35141", + "accept": false, + "reason": "Tokenizer extra-space bug vs embedding resize/post-init reinitialization bug; unrelated." + }, + { + "left": "issue:43950", + "right": "issue:46285", + "accept": false, + "reason": "Non-persistent buffer corruption during `from_pretrained` vs CLIP text model structure change; different regressions." + }, + { + "left": "issue:42915", + "right": "issue:43824", + "accept": false, + "reason": "Qwen3Moe FP8 config failure vs VL class import error; unrelated." + }, + { + "left": "issue:43452", + "right": "issue:46512", + "accept": false, + "reason": "`gguf_file` loading failure vs stale `config.dtype` after `from_config`; different loading semantics." + }, + { + "left": "issue:43334", + "right": "issue:43525", + "accept": false, + "reason": "Both are missing-`pad_token_id` AttributeErrors, but on different model families and loaders; not clearly the same concrete bug." + }, + { + "left": "issue:43824", + "right": "issue:43975", + "accept": false, + "reason": "ImportError for a VL class vs DeepSeek tokenizer detokenization regression; unrelated." + }, + { + "left": "issue:44898", + "right": "issue:45412", + "accept": false, + "reason": "Perceiver resolution failure vs RT-DETR memory leak after deletion; different bugs." + }, + { + "left": "issue:43262", + "right": "issue:43450", + "accept": false, + "reason": "Audio `apply_chat_template` sampling-rate default vs video processor batch-shape bug; different modalities and code paths." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 28, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:29942", + "issue:32090", + "issue:37428", + "issue:38175", + "issue:41720", + "issue:43278", + "issue:43335", + "issue:43388", + "issue:43425", + "issue:43504", + "issue:43526", + "issue:43550", + "issue:43606", + "issue:43611", + "issue:43653", + "issue:43819", + "issue:43825", + "issue:43827", + "issue:43901", + "issue:43931", + "issue:44112", + "issue:44206", + "issue:44442", + "issue:44448", + "issue:44479", + "issue:44561", + "issue:44610", + "issue:44743", + "issue:44792", + "issue:44855", + "issue:44857", + "issue:45083", + "issue:45198", + "issue:45200", + "issue:45325", + "issue:45412", + "issue:45520", + "issue:45563", + "issue:45584", + "issue:45593", + "issue:45759", + "issue:46170", + "issue:46192", + "issue:46209", + "issue:46234", + "issue:46285", + "issue:46512" + ], + "soft_pairs": [ + "issue:44442|issue:45198", + "issue:45200|issue:45325", + "issue:44792|issue:45563", + "issue:43819|issue:44112", + "issue:44479|issue:46192", + "issue:43388|issue:43526", + "issue:43827|issue:43901", + "issue:44561|issue:46234", + "issue:44610|issue:44857", + "issue:41720|issue:46209", + "issue:43335|issue:43550", + "issue:43819|issue:45083", + "issue:45759|issue:46285", + "issue:44857|issue:45593", + "issue:44206|issue:45584", + "issue:38175|issue:45412", + "issue:45520|issue:46170", + "issue:43606|issue:45083", + "issue:43335|issue:43653", + "issue:29942|issue:32090", + "issue:43550|issue:44112", + "issue:37428|issue:43425", + "issue:43611|issue:43931", + "issue:44448|issue:44743", + "issue:45759|issue:46512", + "issue:43335|issue:43504", + "issue:44855|issue:45083", + "issue:43278|issue:43825" + ], + "prepared_review_unit_hash": "8153038d670726ead7b2a6ff7aa4a491d5ccb18e2c8ec1b16275b94b252da3e7", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35795, + "estimated_input_tokens": 8949, + "estimated_eval_tokens": 18154 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These are mostly unrelated closed issues that only share broad Transformers subsystems or similar keywords; none of the soft edges look like the same underlying bug/change, so I would not merge any pair as duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44442", + "right": "issue:45198", + "accept": false, + "reason": "Different tokenizer failures in different models: AutoTokenizer loading vs Wav2Vec2 save_pretrained/tokenization." + }, + { + "left": "issue:45200", + "right": "issue:45325", + "accept": false, + "reason": "Different Qwen2.5-VL bugs: missing mm_token_type_ids default vs RoPE position scaling." + }, + { + "left": "issue:44792", + "right": "issue:45563", + "accept": false, + "reason": "One is a Janus image-generation test failure; the other is a stale generate() warning." + }, + { + "left": "issue:43819", + "right": "issue:44112", + "accept": false, + "reason": "Different failures: DAC latent/forward mismatch vs a GraniteSpeech CI device-override test." + }, + { + "left": "issue:44479", + "right": "issue:46192", + "accept": false, + "reason": "Different regressions: multimodal video input handling vs return_tensors being ignored with text_kwargs." + }, + { + "left": "issue:43388", + "right": "issue:43526", + "accept": false, + "reason": "Both involve labels, but one is a distributed metric gathering bug and the other is a Beit image processor label-reduction bug." + }, + { + "left": "issue:43827", + "right": "issue:43901", + "accept": false, + "reason": "Both are docs issues, but they concern different APIs and different v5 behavior changes." + }, + { + "left": "issue:44561", + "right": "issue:46234", + "accept": false, + "reason": "Different missing utilities: removal of is_torch_fx_available vs missing is_falcon_mamba_ssm_available." + }, + { + "left": "issue:44610", + "right": "issue:44857", + "accept": false, + "reason": "Different code paths: OmDet-Turbo processor size mismatch vs LwDetrImageLoss AMP/CUDA crash." + }, + { + "left": "issue:41720", + "right": "issue:46209", + "accept": false, + "reason": "Different hardware/runtime failures: Qwen3 auto device mapping cudaErrorAssert vs FP8 inference on B200." + }, + { + "left": "issue:43335", + "right": "issue:43550", + "accept": false, + "reason": "Different bugs in different models: SwitchTransformers sparse-layer creation vs Bamba torch.compile+SDPA failure." + }, + { + "left": "issue:43819", + "right": "issue:45083", + "accept": false, + "reason": "Different helper/model bugs; DAC latent mismatch is unrelated to qwen3_omni_moe feature-length computation." + }, + { + "left": "issue:45759", + "right": "issue:46285", + "accept": false, + "reason": "Different structural issues: Qwen multimodal config unwrapping vs CLIP text-model attribute inconsistency." + }, + { + "left": "issue:44857", + "right": "issue:45593", + "accept": false, + "reason": "Different training-time failures: AMP loss crash vs auxiliary-loss logic when denoising is off." + }, + { + "left": "issue:44206", + "right": "issue:45584", + "accept": false, + "reason": "Different behaviors: feature extractor unsupported arg vs Whisper empty-transcription generation after token alignment." + }, + { + "left": "issue:38175", + "right": "issue:45412", + "accept": false, + "reason": "Different model problems: unexpected zero probabilities vs RT-DETR memory not being released." + }, + { + "left": "issue:45520", + "right": "issue:46170", + "accept": false, + "reason": "Different platform/import issues: flash_attn lookup on Python 3.13 vs Windows ROCm torch.distributed dependency failure." + }, + { + "left": "issue:43606", + "right": "issue:45083", + "accept": false, + "reason": "Different bugs: CPU offload device mismatch vs qwen3_omni_moe output-length helper behavior." + }, + { + "left": "issue:43335", + "right": "issue:43653", + "accept": false, + "reason": "Different components and symptoms: SwitchTransformers config bug vs BigBirdTokenizer special-token registration/empty decode bug." + }, + { + "left": "issue:29942", + "right": "issue:32090", + "accept": false, + "reason": "One is Flash Attention 2 test failures; the other is a Trainer broadcast TypeError with NoneType." + }, + { + "left": "issue:43550", + "right": "issue:44112", + "accept": false, + "reason": "Different issues: Bamba torch.compile with SDPA vs a stale GraniteSpeech CI device override test." + }, + { + "left": "issue:37428", + "right": "issue:43425", + "accept": false, + "reason": "Different scope: missing flash_attention utility import vs general Torch 2.10 incompatibility." + }, + { + "left": "issue:43611", + "right": "issue:43931", + "accept": false, + "reason": "Different loading bugs: base_model_prefix compatibility vs Qwen3-VL weight-shape mismatch." + }, + { + "left": "issue:44448", + "right": "issue:44743", + "accept": false, + "reason": "Different regressions in different models: Pegasus output drift across v4/v5 vs Qwen recurrent-state reset with cache." + }, + { + "left": "issue:45759", + "right": "issue:46512", + "accept": false, + "reason": "Different config/dtype problems: multimodal text_config unwrapping vs stale config.dtype after from_config(dtype=...)." + }, + { + "left": "issue:43335", + "right": "issue:43504", + "accept": false, + "reason": "Different SwitchTransformers configuration vs Beit pretrained-model legacy-field loading bug." + }, + { + "left": "issue:44855", + "right": "issue:45083", + "accept": false, + "reason": "Different failures: Python 3.13 JIT parsing/IndentationError vs qwen3_omni_moe feature-length helper behavior." + }, + { + "left": "issue:43278", + "right": "issue:43825", + "accept": false, + "reason": "Different issues: train/eval embedding dtype mismatch vs pipeline error messaging about translation support." + } + ] + }, + "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": 22, + "review_unit_index": 29, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:26413", + "issue:28218", + "issue:29127", + "issue:31501", + "issue:32090", + "issue:33357", + "issue:34689", + "issue:35141", + "issue:37428", + "issue:39692", + "issue:42490", + "issue:43335", + "issue:43526", + "issue:43606", + "issue:43701", + "issue:43720", + "issue:43819", + "issue:43854", + "issue:44112", + "issue:44206", + "issue:44351", + "issue:44360", + "issue:44448", + "issue:44493", + "issue:44514", + "issue:44521", + "issue:44561", + "issue:44617", + "issue:44779", + "issue:44857", + "issue:45072", + "issue:45083", + "issue:45127", + "issue:45200", + "issue:45245", + "issue:45290", + "issue:45310", + "issue:45459", + "issue:45478", + "issue:45588", + "issue:45593", + "issue:45823", + "issue:45859", + "issue:46234", + "issue:46326", + "issue:46431", + "issue:46566", + "issue:46697" + ], + "soft_pairs": [ + "issue:44514|issue:44521", + "issue:43720|issue:44112", + "issue:29127|issue:32090", + "issue:43335|issue:45083", + "issue:43335|issue:44112", + "issue:43335|issue:45593", + "issue:45127|issue:45245", + "issue:46431|issue:46566", + "issue:37428|issue:44351", + "issue:45290|issue:46326", + "issue:45588|issue:45859", + "issue:45200|issue:45588", + "issue:34689|issue:35141", + "issue:45459|issue:46234", + "issue:42490|issue:43854", + "issue:39692|issue:43526", + "issue:26413|issue:31501", + "issue:43819|issue:44493", + "issue:45310|issue:45478", + "issue:43606|issue:44857", + "issue:44448|issue:44779", + "issue:44617|issue:46697", + "issue:39692|issue:44493", + "issue:28218|issue:32090", + "issue:44206|issue:45823", + "issue:44360|issue:44561", + "issue:44857|issue:45072", + "issue:43701|issue:45083", + "issue:32090|issue:33357" + ], + "prepared_review_unit_hash": "bc3736562e708c5342864b60bdc30718b66e4b08cf3c88a1b761a3537aefb261", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "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 mostly heterogeneous and does not form a coherent duplicate set. The only clear duplicate pair is the repeated Qwen3.5 Moe `from_pretrained` regression reports (45310/45478).", + "confidence": 0.95, + "canonical_issue_reason": "No single global canonical exists for the whole cluster; if anchoring to the only true duplicate pair, `issue:45310` is the best canonical issue because it is earlier and `issue:45478` is the same reported Qwen3.5 Moe `from_pretrained` error.", + "canonical_pr_reason": null, + "best_issue_reason": "`issue:45310` is the best representative issue because it cleanly matches `issue:45478` and has the earliest report for that exact bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44514", + "right": "issue:44521", + "accept": false, + "reason": "Both are `apply_chat_template` problems, but one is a batching crash and the other is zero assistant masks for multimodal inputs; different failure modes." + }, + { + "left": "issue:43720", + "right": "issue:44112", + "accept": false, + "reason": "Different models and different bugs: packed-weight loading in BitNet vs a stale GraniteSpeech CI test." + }, + { + "left": "issue:29127", + "right": "issue:32090", + "accept": false, + "reason": "Unrelated subsystems: LayoutLMv3 error messaging vs Trainer `_gpu_broadcast_one` NoneType handling." + }, + { + "left": "issue:43335", + "right": "issue:45083", + "accept": false, + "reason": "Both involve model internals, but one is SwitchTransformers sparse-layer config creation and the other is a qwen3_omni_moe feature-length helper issue." + }, + { + "left": "issue:43335", + "right": "issue:44112", + "accept": false, + "reason": "Different code paths and symptoms; no shared underlying defect." + }, + { + "left": "issue:43335", + "right": "issue:45593", + "accept": false, + "reason": "SwitchTransformers config bug is unrelated to D-FINE auxiliary-loss behavior." + }, + { + "left": "issue:45127", + "right": "issue:45245", + "accept": false, + "reason": "LoRA/embedding collapse and the 2^24 categories limit are unrelated." + }, + { + "left": "issue:46431", + "right": "issue:46566", + "accept": false, + "reason": "Chat-template absence is unrelated to DiffusionGemma weights staying on meta with `device_map=\"auto\"`." + }, + { + "left": "issue:37428", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they concern different missing symbols and likely different fixes." + }, + { + "left": "issue:45290", + "right": "issue:46326", + "accept": false, + "reason": "Both touch chat templates, but one is tool-call tokenization with empty assistant content and the other is `continue_final_message` on mistral_common tokenizers." + }, + { + "left": "issue:45588", + "right": "issue:45859", + "accept": false, + "reason": "Different failures: flash-attention `s_aux=None` crash vs CPU-offload incompatibility in Qwen rotary embedding." + }, + { + "left": "issue:45200", + "right": "issue:45588", + "accept": false, + "reason": "Gemma4 `mm_token_type_ids` defaulting and flash_attention `s_aux=None` are unrelated issues." + }, + { + "left": "issue:34689", + "right": "issue:35141", + "accept": false, + "reason": "Different model-loading regression vs post-init reinitialization of output embeddings." + }, + { + "left": "issue:45459", + "right": "issue:46234", + "accept": false, + "reason": "Tokenizer protobuf error masking and a missing import-utils availability helper are unrelated." + }, + { + "left": "issue:42490", + "right": "issue:43854", + "accept": false, + "reason": "A trailing-slash local path warning is not the same bug as GLM-4.7-Flash unit-test load failure." + }, + { + "left": "issue:39692", + "right": "issue:43526", + "accept": false, + "reason": "SigLIP2 docs/example errors and BeitImageProcessorFast `reduce_labels` are different issues." + }, + { + "left": "issue:26413", + "right": "issue:31501", + "accept": false, + "reason": "The resume-from-checkpoint iterator error and FineWeb training not starting are not the same concrete bug." + }, + { + "left": "issue:43819", + "right": "issue:44493", + "accept": false, + "reason": "DAC `from_latents`/STE mismatch is unrelated to unexpected position-id keys." + }, + { + "left": "issue:45310", + "right": "issue:45478", + "accept": true, + "reason": "Same exact bug report: Qwen3.5 Moe `from_pretrained` error on transformers>=5.4.0; `45478` is a later duplicate of `45310`." + }, + { + "left": "issue:43606", + "right": "issue:44857", + "accept": false, + "reason": "Both are runtime failures, but CPU offload device mismatch and float16 AMP crash in LwDetrImageLoss are different code paths." + }, + { + "left": "issue:44448", + "right": "issue:44779", + "accept": false, + "reason": "Both mention v5 regressions, but one is Pegasus output drift and the other is Deepseek tokenizer correctness; not the same underlying bug." + }, + { + "left": "issue:44617", + "right": "issue:46697", + "accept": false, + "reason": "CUDA OOM on Sam3Video is unrelated to Swin2SR over-padding." + }, + { + "left": "issue:39692", + "right": "issue:44493", + "accept": false, + "reason": "Doc-example issues and position-id key warnings are not the same problem." + }, + { + "left": "issue:28218", + "right": "issue:32090", + "accept": false, + "reason": "Tokenizer spacing after added tokens is unrelated to Trainer broadcast type errors." + }, + { + "left": "issue:44206", + "right": "issue:45823", + "accept": false, + "reason": "Feature-extractor argument regression and Gemma4 PLE device mismatch are different failures." + }, + { + "left": "issue:44360", + "right": "issue:44561", + "accept": false, + "reason": "DSA ReLU discussion and `is_torch_fx_available` removal breakage are unrelated." + }, + { + "left": "issue:44857", + "right": "issue:45072", + "accept": false, + "reason": "Different models and different dtype/device mismatch manifestations; not the same bug." + }, + { + "left": "issue:43701", + "right": "issue:45083", + "accept": false, + "reason": "Checkpoint key mismatch and qwen3_omni_moe output-length helper behavior are unrelated." + }, + { + "left": "issue:32090", + "right": "issue:33357", + "accept": false, + "reason": "Trainer GPU broadcast NoneType error is unrelated to the MacOS CLIP bus error." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: the only clear duplicate pair is issue:45310 / issue:45478, which have identical titles and the earlier report is correctly chosen as the canonical anchor. The soft-edge rejections are conservative and do not overstate equivalence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 30, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:10000", + "issue:15354", + "issue:25251", + "issue:27957", + "issue:29942", + "issue:32090", + "issue:33290", + "issue:34689", + "issue:35141", + "issue:37428", + "issue:38175", + "issue:40444", + "issue:41720", + "issue:42175", + "issue:42490", + "issue:43278", + "issue:43329", + "issue:43381", + "issue:43526", + "issue:43550", + "issue:43575", + "issue:43746", + "issue:43873", + "issue:43927", + "issue:43976", + "issue:44448", + "issue:44466", + "issue:44493", + "issue:44561", + "issue:44568", + "issue:44779", + "issue:44855", + "issue:44857", + "issue:44977", + "issue:45083", + "issue:45229", + "issue:45290", + "issue:45381", + "issue:45440", + "issue:45491", + "issue:45588", + "issue:45698", + "issue:45823", + "issue:45920", + "issue:46192", + "issue:46209", + "issue:46399", + "issue:46655" + ], + "soft_pairs": [ + "issue:42490|issue:43575", + "issue:38175|issue:44493", + "issue:43550|issue:45083", + "issue:29942|issue:35141", + "issue:45920|issue:46192", + "issue:27957|issue:34689", + "issue:40444|issue:43526", + "issue:43278|issue:43329", + "issue:44448|issue:44977", + "issue:45290|issue:45381", + "issue:44855|issue:46399", + "issue:44977|issue:45229", + "issue:45290|issue:46655", + "issue:43278|issue:43381", + "issue:10000|issue:15354", + "issue:43927|issue:45698", + "issue:44857|issue:45588", + "issue:43746|issue:46209", + "issue:44561|issue:44779", + "issue:37428|issue:42175", + "issue:32090|issue:33290", + "issue:43873|issue:45229", + "issue:25251|issue:32090", + "issue:43976|issue:45440", + "issue:44448|issue:44466", + "issue:42490|issue:43873", + "issue:44568|issue:45491", + "issue:41720|issue:45229", + "issue:45588|issue:45823", + "issue:43278|issue:43575" + ], + "prepared_review_unit_hash": "a33f8ca64dca7977dce823b11db2fe673f8ade2077a2a7151d767eea3347f408", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36906, + "estimated_input_tokens": 9227, + "estimated_eval_tokens": 18710 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose thematic similarity set, not a true duplicate group: it mixes unrelated bugs across tokenizers, flash-attention, multimodal processors, loading/saving, device mapping, and OOMs. None of the soft pairs look like the same underlying bug/change, so all are rejected.", + "confidence": 0.89, + "canonical_issue_reason": "issue:45698 is the broadest open load/save regression in the set and has the highest discussion/inbound-reference signal, so it is the best anchor for triage.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45920 is the most general concrete v5 regression report, affecting multiple model families, so it is the most actionable representative issue.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42490", + "right": "issue:43575", + "accept": false, + "reason": "One is a path-trailing-slash import warning; the other is TP-related OOM on a specific model." + }, + { + "left": "issue:38175", + "right": "issue:44493", + "accept": false, + "reason": "Different failures: zero probabilities in SigLIP2 vs unexpected position-id keys." + }, + { + "left": "issue:43550", + "right": "issue:45083", + "accept": false, + "reason": "Unrelated model/runtime bugs: torch.compile SDPA on Bamba vs feature-length helper logic in qwen3_omni_moe." + }, + { + "left": "issue:29942", + "right": "issue:35141", + "accept": false, + "reason": "Flash Attention 2 test failure is unrelated to embedding reinitialization during post-init." + }, + { + "left": "issue:45920", + "right": "issue:46192", + "accept": false, + "reason": "Both touch tokenizer APIs, but one is a wrong-token-ID regression and the other is return_tensors being ignored." + }, + { + "left": "issue:27957", + "right": "issue:34689", + "accept": false, + "reason": "Different subsystems and symptoms: Flash Attention 2 support vs a model-loading breakage." + }, + { + "left": "issue:40444", + "right": "issue:43526", + "accept": false, + "reason": "Different areas: multimodal finetuning with multiple images vs BeitImageProcessorFast label reduction." + }, + { + "left": "issue:43278", + "right": "issue:43329", + "accept": false, + "reason": "Embedding dtype drift is unrelated to undefined variables in a video token-count helper." + }, + { + "left": "issue:44448", + "right": "issue:44977", + "accept": false, + "reason": "Pegasus v4/v5 output mismatch is not the same as Qwen3.5 flash-attention generation failure." + }, + { + "left": "issue:45290", + "right": "issue:45381", + "accept": false, + "reason": "Chat-template crash with tool-call messages is different from qwen2.5-vl video position-id handling." + }, + { + "left": "issue:44855", + "right": "issue:46399", + "accept": false, + "reason": "Import parsing/torch.jit comment issue is unrelated to a missing decorator around apply_rotary_pos_emb registration." + }, + { + "left": "issue:44977", + "right": "issue:45229", + "accept": false, + "reason": "Flash-attention generation bug and multi-GPU CUDA OOM are different concrete problems." + }, + { + "left": "issue:45290", + "right": "issue:46655", + "accept": false, + "reason": "Assistant tool-call chat templating crash is unrelated to SmolVLM video padding math." + }, + { + "left": "issue:43278", + "right": "issue:43381", + "accept": false, + "reason": "Training/eval dtype mismatch is not the same bug as gradient checkpointing in eval mode." + }, + { + "left": "issue:10000", + "right": "issue:15354", + "accept": false, + "reason": "German DistilBertModel issue and GeneratorExp torch.jit.script/export issue are unrelated." + }, + { + "left": "issue:43927", + "right": "issue:45698", + "accept": false, + "reason": "Both involve save/load, but one loses custom token IDs while the other loads the wrong custom module." + }, + { + "left": "issue:44857", + "right": "issue:45588", + "accept": false, + "reason": "Float16 AMP crash in LwDetrImageLoss is unrelated to flash_attention.py handling s_aux=None." + }, + { + "left": "issue:43746", + "right": "issue:46209", + "accept": false, + "reason": "PEFT checkpoint loading and FP8 inference on B200 are different code paths and failure modes." + }, + { + "left": "issue:44561", + "right": "issue:44779", + "accept": false, + "reason": "Removal of is_torch_fx_available breaks remote-code models; DeepSeek tokenizer regression is a separate tokenizer bug." + }, + { + "left": "issue:37428", + "right": "issue:42175", + "accept": false, + "reason": "Missing flash_attention utility symbol and TensorFlow missing from a torch install are unrelated packaging/runtime issues." + }, + { + "left": "issue:32090", + "right": "issue:33290", + "accept": false, + "reason": "NoneType broadcast error in Trainer is unrelated to AdaFactor OOM in DeepSpeed." + }, + { + "left": "issue:43873", + "right": "issue:45229", + "accept": false, + "reason": "Quantization offloading misbehavior and Gemma4 multi-GPU OOM are different memory-management issues." + }, + { + "left": "issue:25251", + "right": "issue:32090", + "accept": false, + "reason": "Pipeline top_k nesting behavior is unrelated to the Trainer _gpu_broadcast_one NoneType error." + }, + { + "left": "issue:43976", + "right": "issue:45440", + "accept": false, + "reason": "Python-version compatibility breakage is unrelated to DeepSeekV3MoE divergence from remote implementation." + }, + { + "left": "issue:44448", + "right": "issue:44466", + "accept": false, + "reason": "Different concrete bugs: model output regression vs inconsistent lm_head.weight serialization." + }, + { + "left": "issue:42490", + "right": "issue:43873", + "accept": false, + "reason": "Path-trailing-slash import warning does not match quantization offloading behavior." + }, + { + "left": "issue:44568", + "right": "issue:45491", + "accept": false, + "reason": "Tokenizer add_special_tokens regression is unrelated to Gemma3 NaN embeddings from sliding-window attention." + }, + { + "left": "issue:41720", + "right": "issue:45229", + "accept": false, + "reason": "Qwen3 cudaErrorAssert with auto device mapping is not the same as Gemma4 OOM." + }, + { + "left": "issue:45588", + "right": "issue:45823", + "accept": false, + "reason": "Flash-attention sinkless-model AttributeError and PLE device mismatch are different forward-path bugs." + }, + { + "left": "issue:43278", + "right": "issue:43575", + "accept": false, + "reason": "Embedding dtype change in eval and TP-related OOM on Qwen2-57B 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": 22, + "review_unit_index": 31, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:27957", + "issue:29127", + "issue:29942", + "issue:30064", + "issue:33357", + "issue:34567", + "issue:35141", + "issue:39692", + "issue:42490", + "issue:42915", + "issue:42994", + "issue:43065", + "issue:43066", + "issue:43381", + "issue:43412", + "issue:43421", + "issue:43575", + "issue:43611", + "issue:43653", + "issue:43701", + "issue:43756", + "issue:43994", + "issue:44186", + "issue:44206", + "issue:44265", + "issue:44361", + "issue:44442", + "issue:44462", + "issue:44479", + "issue:44743", + "issue:44871", + "issue:44977", + "issue:45198", + "issue:45245", + "issue:45372", + "issue:45406", + "issue:45412", + "issue:45440", + "issue:45459", + "issue:45491", + "issue:45563", + "issue:45588", + "issue:45593", + "issue:45712", + "issue:45803", + "issue:46121", + "issue:46369", + "issue:46693" + ], + "soft_pairs": [ + "issue:45372|issue:45588", + "issue:33357|issue:39692", + "issue:45198|issue:45803", + "issue:44977|issue:45563", + "issue:29127|issue:34567", + "issue:44871|issue:44977", + "issue:43381|issue:43701", + "issue:43412|issue:45588", + "issue:43653|issue:44743", + "issue:44186|issue:45198", + "issue:43611|issue:46121", + "issue:44361|issue:45198", + "issue:45372|issue:45406", + "issue:42915|issue:42994", + "issue:45459|issue:45712", + "issue:29942|issue:30064", + "issue:44206|issue:44479", + "issue:43756|issue:45440", + "issue:43653|issue:44442", + "issue:43065|issue:45412", + "issue:43065|issue:43994", + "issue:45491|issue:46693", + "issue:45372|issue:45459", + "issue:45245|issue:45563", + "issue:29127|issue:33357", + "issue:44265|issue:45198", + "issue:43066|issue:44462", + "issue:27957|issue:29942", + "issue:45412|issue:46369", + "issue:43994|issue:45412", + "issue:45440|issue:45593", + "issue:43575|issue:43701", + "issue:43421|issue:45491", + "issue:29127|issue:35141", + "issue:42490|issue:43381" + ], + "prepared_review_unit_hash": "20049dd8cb09397b1b5c46622156aa0863f0b93a0b6cf78b54d47cf6afae977a", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 38215, + "estimated_input_tokens": 9554, + "estimated_eval_tokens": 19364 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No soft duplicates look safe to merge: the pairs mostly share broad themes (Flash Attention, tokenizers, Gemma/Qwen, memory, export) but describe different concrete failures and code paths.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45372", + "right": "issue:45588", + "accept": false, + "reason": "Both involve Gemma/processor loading, but one is an import error from mistral_common and the other is a flash_attention.py AttributeError on s_aux=None; different failure modes and paths." + }, + { + "left": "issue:33357", + "right": "issue:39692", + "accept": false, + "reason": "Both mention vision/CLIP/SigLIP2, but one is a MacOS bus error in model loading and the other is a documentation example with model/processor mismatch and quantization issues." + }, + { + "left": "issue:45198", + "right": "issue:45803", + "accept": false, + "reason": "One is save_pretrained/tokenization failure for a specific Wav2Vec2 model; the other is a broad Fuyu conversion bug caused by a bare except swallowing errors." + }, + { + "left": "issue:44977", + "right": "issue:45563", + "accept": false, + "reason": "Both touch generation/flash-attention, but one is a Qwen3.5 correctness regression and the other is a stale warning for paged generate(); not the same bug." + }, + { + "left": "issue:29127", + "right": "issue:34567", + "accept": false, + "reason": "Unrelated: LayoutLMv3 error-message clarity versus TrainerState token counter not updating." + }, + { + "left": "issue:44871", + "right": "issue:44977", + "accept": false, + "reason": "Gemma eos_token_id configuration mismatch is unrelated to Qwen3.5 flash-attention generation failure." + }, + { + "left": "issue:43381", + "right": "issue:43701", + "accept": false, + "reason": "Gradient checkpointing in eval mode and resume_from_checkpoint key mismatch are separate trainer/runtime issues." + }, + { + "left": "issue:43412", + "right": "issue:45588", + "accept": false, + "reason": "Gemma3n export failures versus a flash-attention AttributeError on sink-less models; different models and code paths." + }, + { + "left": "issue:43653", + "right": "issue:44743", + "accept": false, + "reason": "BigBirdTokenizer special-token registration/empty decode is unrelated to Qwen recurrent-state reset behavior." + }, + { + "left": "issue:44186", + "right": "issue:45198", + "accept": false, + "reason": "Different tokenizers and bugs: LayoutLMv2 NER/padding crash versus Wav2Vec2 save/tokenization failure." + }, + { + "left": "issue:43611", + "right": "issue:46121", + "accept": false, + "reason": "Base_model_prefix loading breakage is unrelated to rope validation TypeError on JSON-loaded lists." + }, + { + "left": "issue:44361", + "right": "issue:45198", + "accept": false, + "reason": "MLukeTokenizer task AttributeError and Wav2Vec2 save/tokenization failure are different tokenizer-specific issues." + }, + { + "left": "issue:45372", + "right": "issue:45406", + "accept": false, + "reason": "Gemma4 processor loading import failure and Gemma4Processor missing _tokenizer in serve are related only by model name, not the same defect." + }, + { + "left": "issue:42915", + "right": "issue:42994", + "accept": false, + "reason": "Qwen3Moe FP8 config failure and quantized model saving failure are distinct quantization problems." + }, + { + "left": "issue:45459", + "right": "issue:45712", + "accept": false, + "reason": "Tokenizer error masking when protobuf is absent is unrelated to leftover dummy classes in dummy_pt_objects.py." + }, + { + "left": "issue:29942", + "right": "issue:30064", + "accept": false, + "reason": "Flash Attention 2 test failures and void segmentation map processing are separate subsystems and bug reports." + }, + { + "left": "issue:44206", + "right": "issue:44479", + "accept": false, + "reason": "Feature extractor center-arg crash and video input regression affect different APIs and model families." + }, + { + "left": "issue:43756", + "right": "issue:45440", + "accept": false, + "reason": "Smollm3 RoPE layer selection bug and DeepseekV3MoE divergence from remote implementation are different model-specific logic issues." + }, + { + "left": "issue:43653", + "right": "issue:44442", + "accept": false, + "reason": "BigBird tokenizer decode bug is unrelated to AutoTokenizer loading FastSpeech2ConformerTokenizer." + }, + { + "left": "issue:43065", + "right": "issue:45412", + "accept": false, + "reason": "Dummy Sam3PixelDecoder Conv2d and RT-DETR memory leak are different problems." + }, + { + "left": "issue:43065", + "right": "issue:43994", + "accept": false, + "reason": "Dummy Sam3PixelDecoder Conv2d is unrelated to SigLIP2 nonsensical outputs with AutoModel/pipeline." + }, + { + "left": "issue:45491", + "right": "issue:46693", + "accept": false, + "reason": "Gemma3 NaN embeddings from mixed-length batching differ from a performance regression in long-context generation." + }, + { + "left": "issue:45372", + "right": "issue:45459", + "accept": false, + "reason": "Gemma4 processor import failure and tokenizer protobuf exception handling are unrelated error paths." + }, + { + "left": "issue:45245", + "right": "issue:45563", + "accept": false, + "reason": "Category-cardinality RuntimeError and paged-generate stale warning are not the same defect." + }, + { + "left": "issue:29127", + "right": "issue:33357", + "accept": false, + "reason": "LayoutLMv3 error-message clarity and MacOS bus error on a CLIP model are unrelated." + }, + { + "left": "issue:44265", + "right": "issue:45198", + "accept": false, + "reason": "torch.export failure with torch_compilable_check is unrelated to Wav2Vec2 save/tokenization issues." + }, + { + "left": "issue:43066", + "right": "issue:44462", + "accept": false, + "reason": "Wrong tokenizer decoder type in Transformers v5 and ignoring tokenizer.json are different tokenizer-loading regressions." + }, + { + "left": "issue:27957", + "right": "issue:29942", + "accept": false, + "reason": "XLMRoberta FA2 support and failing FA2 tests are too broad to be the same concrete bug." + }, + { + "left": "issue:45412", + "right": "issue:46369", + "accept": false, + "reason": "RT-DETR GC/memory leak and sam3-litetext memory overflow are distinct memory issues in different models." + }, + { + "left": "issue:43994", + "right": "issue:45412", + "accept": false, + "reason": "SigLIP2 nonsensical results and RT-DETR memory release failure are unrelated." + }, + { + "left": "issue:45440", + "right": "issue:45593", + "accept": false, + "reason": "DeepseekV3MoE divergence from remote implementation is unrelated to D-FINE auxiliary-loss behavior." + }, + { + "left": "issue:43575", + "right": "issue:43701", + "accept": false, + "reason": "Qwen2-57B-A14B-Instruct TP OOM and resume_from_checkpoint key mismatch are different bugs." + }, + { + "left": "issue:43421", + "right": "issue:45491", + "accept": false, + "reason": "Runtime special-token post-processor updates and Gemma3 NaN embeddings are unrelated tokenizer versus attention/batching issues." + }, + { + "left": "issue:29127", + "right": "issue:35141", + "accept": false, + "reason": "LayoutLMv3 error messaging and output embedding reinitialization after resizing are unrelated." + }, + { + "left": "issue:42490", + "right": "issue:43381", + "accept": false, + "reason": "Local-path trailing-slash import warning and gradient-checkpointing eval-mode failure are different loading/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": 22, + "review_unit_index": 32, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:27957", + "issue:29127", + "issue:33290", + "issue:34567", + "issue:34689", + "issue:35141", + "issue:42490", + "issue:42915", + "issue:43262", + "issue:43334", + "issue:43381", + "issue:43526", + "issue:43531", + "issue:43575", + "issue:43611", + "issue:43646", + "issue:43720", + "issue:43761", + "issue:43824", + "issue:43873", + "issue:43931", + "issue:43994", + "issue:44077", + "issue:44186", + "issue:44206", + "issue:44279", + "issue:44387", + "issue:44442", + "issue:44514", + "issue:44743", + "issue:44811", + "issue:44898", + "issue:44977", + "issue:45003", + "issue:45072", + "issue:45245", + "issue:45250", + "issue:45381", + "issue:45440", + "issue:45588", + "issue:45600", + "issue:45834", + "issue:45859", + "issue:46121", + "issue:46369", + "issue:46399", + "issue:46640", + "issue:46693" + ], + "soft_pairs": [ + "issue:43575|issue:44387", + "issue:43526|issue:45834", + "issue:43526|issue:46693", + "issue:44442|issue:45072", + "issue:42915|issue:43931", + "issue:43824|issue:44279", + "issue:43526|issue:44186", + "issue:43994|issue:46369", + "issue:46121|issue:46399", + "issue:45250|issue:46640", + "issue:27957|issue:29127", + "issue:27957|issue:33290", + "issue:43334|issue:43531", + "issue:43761|issue:45859", + "issue:33290|issue:34689", + "issue:33290|issue:35141", + "issue:44514|issue:45381", + "issue:45588|issue:45600", + "issue:33290|issue:34567", + "issue:43381|issue:43873", + "issue:44977|issue:45245", + "issue:43262|issue:44206", + "issue:43646|issue:44077", + "issue:44743|issue:44811", + "issue:44442|issue:44898", + "issue:45003|issue:45440", + "issue:34689|issue:43611", + "issue:42490|issue:44743", + "issue:43720|issue:44442" + ], + "prepared_review_unit_hash": "4679bf83cdfd05bd7e21bfdb2e24dbe873e9cb2f9503c4d7beca595694c745c3", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36188, + "estimated_input_tokens": 9047, + "estimated_eval_tokens": 18350 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: most items are separate model-specific bugs or regressions that only share broad themes like OOM, Flash Attention, or loading failures. None of the soft pairs looks like the same concrete bug/change, so they should stay separate.", + "confidence": 0.94, + "canonical_issue_reason": "No single canonical issue stands out; the items span unrelated subsystems and model families, and the apparent overlaps are too broad to treat as one duplicate cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "No global best issue for deduping: none of these reports is a clear superset or central tracker for the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43575", + "right": "issue:44387", + "accept": false, + "reason": "Both mention OOM, but one is a Qwen2-57B TP loading issue and the other is an int4 quantization regression in Transformers 5.x; different causes and code paths." + }, + { + "left": "issue:43526", + "right": "issue:45834", + "accept": false, + "reason": "Different models and failures: BEiT image processor label reduction vs Kosmos2.5 long OCR input memory overflow." + }, + { + "left": "issue:43526", + "right": "issue:46693", + "accept": false, + "reason": "Unrelated bugs: BEiT label handling vs a Flash Attention 2 long-context performance regression." + }, + { + "left": "issue:44442", + "right": "issue:45072", + "accept": false, + "reason": "Different components and symptoms: tokenizer loading failure for FastSpeech2ConformerTokenizer vs dtype mismatch in SwitchTransformers/TimmWrapperModel bfloat16 inference." + }, + { + "left": "issue:42915", + "right": "issue:43931", + "accept": false, + "reason": "Different model families and failure modes: Qwen3Moe FP8 config issue vs Qwen3-VL weight shape mismatch on load." + }, + { + "left": "issue:43824", + "right": "issue:44279", + "accept": false, + "reason": "Too vague on the right and no concrete shared bug; an import error for Qwen2.5-VL is not the same as a generic dependency issue." + }, + { + "left": "issue:43526", + "right": "issue:44186", + "accept": false, + "reason": "Different tokenizer/processor paths: BEiT Fast image processor label reduction vs LayoutLMv2 tokenizer crashes on NER padding/truncation." + }, + { + "left": "issue:43994", + "right": "issue:46369", + "accept": false, + "reason": "Different subsystems: SigLIP2 inference corruption vs Sam3-LiteText memory overflow." + }, + { + "left": "issue:46121", + "right": "issue:46399", + "accept": false, + "reason": "Distinct rope-related bugs: JSON-loaded validation list TypeError vs missing decorator for apply_rotary_pos_emb registration in Qwen3.5 kernels." + }, + { + "left": "issue:45250", + "right": "issue:46640", + "accept": false, + "reason": "Only a broad Flash Attention 2 overlap; one is a generic title, the other is a sink-less model crash in flash_attention.py." + }, + { + "left": "issue:27957", + "right": "issue:29127", + "accept": false, + "reason": "Completely different problems: Flash Attention 2 support for XLM-Roberta vs a LayoutLMv3 box/clarity error message." + }, + { + "left": "issue:27957", + "right": "issue:33290", + "accept": false, + "reason": "Different bug classes: model/attention support vs DeepSpeed Adafactor OOM." + }, + { + "left": "issue:43334", + "right": "issue:43531", + "accept": false, + "reason": "Both are Qwen-related, but one is a pad_token_id load failure and the other is a sliding_window issue in Qwen3-MoE." + }, + { + "left": "issue:43761", + "right": "issue:45859", + "accept": false, + "reason": "Different model areas and bugs: CLIPVisionModel hidden_states regression vs Qwen3.5 rotary embedding CPU offload incompatibility." + }, + { + "left": "issue:33290", + "right": "issue:34689", + "accept": false, + "reason": "Both are OOM/loading complaints, but one is optimizer memory in DeepSpeed and the other is a specific model-loading regression for Llama 3.2 Vision." + }, + { + "left": "issue:33290", + "right": "issue:35141", + "accept": false, + "reason": "Unrelated: Adafactor OOM in DeepSpeed vs embedding reinitialization after resizing token embeddings." + }, + { + "left": "issue:44514", + "right": "issue:45381", + "accept": false, + "reason": "Both involve Qwen2.5-VL, but one crashes in batched apply_chat_template with padding=False and the other has wrong vision_position_ids for video input." + }, + { + "left": "issue:45588", + "right": "issue:45600", + "accept": false, + "reason": "Different files and failures: flash_attention.py AttributeError on s_aux=None vs auto_mappings.py referencing removed Sam3LiteText configs." + }, + { + "left": "issue:33290", + "right": "issue:34567", + "accept": false, + "reason": "Different trainer/optimizer issues: DeepSpeed Adafactor OOM vs TrainerState.num_input_tokens_seen not updating." + }, + { + "left": "issue:43381", + "right": "issue:43873", + "accept": false, + "reason": "Different behavior: gradient checkpointing disallowed in eval mode vs offloading not working as expected with quantization." + }, + { + "left": "issue:44977", + "right": "issue:45245", + "accept": false, + "reason": "Unrelated failures: Qwen3.5 flash-attention generation bug vs category-count runtime error." + }, + { + "left": "issue:43262", + "right": "issue:44206", + "accept": false, + "reason": "Different audio preprocessing bugs: apply_chat_template sampling-rate default vs unsupported center argument in LasrFeatureExtractor." + }, + { + "left": "issue:43646", + "right": "issue:44077", + "accept": false, + "reason": "Related to initialization only at a broad level, but not the same concrete bug: one breaks custom model init in v5, the other is about patchtsmixer optional post_init handling." + }, + { + "left": "issue:44743", + "right": "issue:44811", + "accept": false, + "reason": "Different pathways and symptoms: Qwen3_5 recurrent state reset bug vs Whisper batch_decode ignoring skip_special_tokens." + }, + { + "left": "issue:44442", + "right": "issue:44898", + "accept": false, + "reason": "Different models and failures: FastSpeech2ConformerTokenizer loading vs Perceiver image classification failure at non-default resolution." + }, + { + "left": "issue:45003", + "right": "issue:45440", + "accept": false, + "reason": "Distinct issues: unsafe sys.modules access in modeling_utils vs divergence of native DeepseekV3MoE from remote implementation." + }, + { + "left": "issue:34689", + "right": "issue:43611", + "accept": false, + "reason": "Both are model-loading regressions, but they affect different releases and different root causes: Llama 3.2 Vision vs base_model_prefix handling in Transformers 5.0.0." + }, + { + "left": "issue:42490", + "right": "issue:44743", + "accept": false, + "reason": "No real overlap: trailing-slash local model import warning vs Qwen3_5 recurrent-state reset bug." + }, + { + "left": "issue:43720", + "right": "issue:44442", + "accept": false, + "reason": "Different domains: BitNet packed-weight loading in accelerate vs tokenizer loading failure for FastSpeech2ConformerTokenizer." + } + ] + }, + "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": 22, + "review_unit_index": 33, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:25251", + "issue:27957", + "issue:28282", + "issue:32090", + "issue:42175", + "issue:42617", + "issue:42907", + "issue:43232", + "issue:43316", + "issue:43388", + "issue:43493", + "issue:43525", + "issue:43575", + "issue:43643", + "issue:43701", + "issue:43723", + "issue:43742", + "issue:43749", + "issue:43761", + "issue:43827", + "issue:43873", + "issue:43931", + "issue:44062", + "issue:44077", + "issue:44079", + "issue:44188", + "issue:44190", + "issue:44442", + "issue:44464", + "issue:44479", + "issue:44496", + "issue:44855", + "issue:44871", + "issue:44913", + "issue:44987", + "issue:45137", + "issue:45216", + "issue:45325", + "issue:45479", + "issue:45698", + "issue:45759", + "issue:45803", + "issue:45907", + "issue:46178", + "issue:46234", + "issue:46421", + "issue:46506", + "issue:46697" + ], + "soft_pairs": [ + "issue:42175|issue:46234", + "issue:43723|issue:45479", + "issue:45325|issue:46697", + "issue:43316|issue:43525", + "issue:44077|issue:44479", + "issue:43493|issue:43827", + "issue:44062|issue:45907", + "issue:43761|issue:44079", + "issue:43873|issue:44871", + "issue:43761|issue:44188", + "issue:43232|issue:43388", + "issue:43749|issue:43931", + "issue:43931|issue:45216", + "issue:43316|issue:45759", + "issue:44855|issue:45137", + "issue:43643|issue:44913", + "issue:44442|issue:45803", + "issue:44987|issue:46178", + "issue:44190|issue:44496", + "issue:42907|issue:45698", + "issue:43701|issue:46506", + "issue:42617|issue:43575", + "issue:43575|issue:43742", + "issue:25251|issue:28282", + "issue:27957|issue:32090", + "issue:44464|issue:46421" + ], + "prepared_review_unit_hash": "24c549e40539e693cf552a7c262bcbbac5e79cf8d7315e59e9050d65b8007eaf", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 26, + "serialized_chars": 35650, + "estimated_input_tokens": 8913, + "estimated_eval_tokens": 18082 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a set of superficially similar but actually different bug reports around loading, configs, generation, and model-specific regressions. I don\u2019t see any soft edge that looks like the same underlying fix.", + "confidence": 0.23, + "canonical_issue_reason": "issue:45698 is the broadest and most actively discussed loading/saving report in the set, so it\u2019s the best placeholder canonical even though the cluster is heterogeneous.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45698 has the widest scope and strongest discussion signal among these issues, but it should be treated as a representative, not a true duplicate hub.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42175", + "right": "issue:46234", + "accept": false, + "reason": "One is a backend/installation packaging problem; the other is a missing import-utils availability helper. Same general area, different bug." + }, + { + "left": "issue:43723", + "right": "issue:45479", + "accept": false, + "reason": "Tokenizer loading in v5 and a sequence-classification loss bug are unrelated code paths." + }, + { + "left": "issue:45325", + "right": "issue:46697", + "accept": false, + "reason": "Qwen2.5-VL rope/temporal position indexing is not the same bug as Swin2SR image over-padding." + }, + { + "left": "issue:43316", + "right": "issue:43525", + "accept": false, + "reason": "Both mention config API inconsistencies, but they affect different models and different missing fields." + }, + { + "left": "issue:44077", + "right": "issue:44479", + "accept": false, + "reason": "PatchTSMixer post-init policy and Qwen video-input regression are unrelated model-specific issues." + }, + { + "left": "issue:43493", + "right": "issue:43827", + "accept": false, + "reason": "A model implementation discrepancy and docs still referencing pipeline() are different kinds of problems." + }, + { + "left": "issue:44062", + "right": "issue:45907", + "accept": false, + "reason": "Tokenizer AddedToken duplication and list_repo_files revision forwarding are unrelated API bugs." + }, + { + "left": "issue:43761", + "right": "issue:44079", + "accept": false, + "reason": "Hidden-states missing from a model forward pass is not the same as ModelOutput key assignment when a value was previously None." + }, + { + "left": "issue:43873", + "right": "issue:44871", + "accept": false, + "reason": "Quantization/offloading behavior and Gemma eos_token_id configuration are separate issues." + }, + { + "left": "issue:43761", + "right": "issue:44188", + "accept": false, + "reason": "CLIPVision hidden_states regression and torch.compile attention-kernel branching are different failures." + }, + { + "left": "issue:43232", + "right": "issue:43388", + "accept": false, + "reason": "Generation kwargs update after sync_gpus and metric gathering on last-batch labels are different training/inference bugs." + }, + { + "left": "issue:43749", + "right": "issue:43931", + "accept": false, + "reason": "FSDP CPU RAM-efficient loading breakage is not the same as a specific Qwen3-VL weight-shape mismatch." + }, + { + "left": "issue:43931", + "right": "issue:45216", + "accept": false, + "reason": "Model load shape mismatch and save_pretrained checkpoint corruption are different failure modes." + }, + { + "left": "issue:43316", + "right": "issue:45759", + "accept": false, + "reason": "Gemma3 text-config API mismatch and Qwen multimodal from_config text_config unwrapping are not the same underlying bug." + }, + { + "left": "issue:44855", + "right": "issue:45137", + "accept": false, + "reason": "Python 3.13 torch.jit parsing/indentation issue and DeepSpeed ZeRO3 deque underflow are unrelated." + }, + { + "left": "issue:43643", + "right": "issue:44913", + "accept": false, + "reason": "Missing fields from trust_remote_code configs and GPTNeoX rotary_pct not persisting on reload are different config-loading bugs." + }, + { + "left": "issue:44442", + "right": "issue:45803", + "accept": false, + "reason": "AutoTokenizer failing on one tokenizer class and a bare-except swallowing errors in Fuyu conversion are unrelated." + }, + { + "left": "issue:44987", + "right": "issue:46178", + "accept": false, + "reason": "Loading a specific model repo and missing BACKENDS_MAPPING support for tensorflow_text are different backend-resolution issues." + }, + { + "left": "issue:44190", + "right": "issue:44496", + "accept": false, + "reason": "Local dataset loading in an example script and an unrecognized model/config error are not the same problem." + }, + { + "left": "issue:42907", + "right": "issue:45698", + "accept": false, + "reason": "Saving dequantized models and loading the wrong custom module after save_pretrained both touch serialization, but they are different code paths and failures." + }, + { + "left": "issue:43701", + "right": "issue:46506", + "accept": false, + "reason": "Checkpoint resume key mismatch and CPU/CUDA divergence in PPDocLayoutV3 are unrelated." + }, + { + "left": "issue:42617", + "right": "issue:43575", + "accept": false, + "reason": "3d_parallel.py startup failure and TP-induced OOM when loading Qwen2-57B are different parallelism issues." + }, + { + "left": "issue:43575", + "right": "issue:43742", + "accept": false, + "reason": "Qwen2-57B tensor-parallel OOM and MobileLLM loading key error are unrelated loading failures." + }, + { + "left": "issue:25251", + "right": "issue:28282", + "accept": false, + "reason": "Pipeline top_k output-shape behavior and missing PyTorch installation import error are completely different." + }, + { + "left": "issue:27957", + "right": "issue:32090", + "accept": false, + "reason": "Flash Attention support for XLMRoberta and a Trainer broadcast NoneType error do not share the same code path." + }, + { + "left": "issue:44464", + "right": "issue:46421", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency and cache lazy initialization causing recompilation are related at a high level, but 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": 22, + "review_unit_index": 34, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:16998", + "issue:22355", + "issue:42175", + "issue:43065", + "issue:43295", + "issue:43316", + "issue:43352", + "issue:43388", + "issue:43404", + "issue:43441", + "issue:43452", + "issue:43540", + "issue:43550", + "issue:43572", + "issue:43606", + "issue:43646", + "issue:43650", + "issue:43688", + "issue:43701", + "issue:43716", + "issue:43761", + "issue:43825", + "issue:44077", + "issue:44162", + "issue:44479", + "issue:44496", + "issue:44514", + "issue:44534", + "issue:44610", + "issue:44617", + "issue:44792", + "issue:44871", + "issue:44913", + "issue:45127", + "issue:45200", + "issue:45584", + "issue:45632", + "issue:45803", + "issue:46050", + "issue:46170", + "issue:46178", + "issue:46202", + "issue:46234", + "issue:46291", + "issue:46365", + "issue:46369", + "issue:46424", + "issue:46640" + ], + "soft_pairs": [ + "issue:43550|issue:43701", + "issue:46234|issue:46291", + "issue:43606|issue:43701", + "issue:43295|issue:43825", + "issue:16998|issue:22355", + "issue:43295|issue:44871", + "issue:44871|issue:45584", + "issue:43316|issue:43716", + "issue:42175|issue:43316", + "issue:44077|issue:44534", + "issue:44871|issue:46050", + "issue:45803|issue:46365", + "issue:43646|issue:44913", + "issue:43452|issue:44496", + "issue:44162|issue:44792", + "issue:45803|issue:46170", + "issue:45632|issue:46424", + "issue:43404|issue:43441", + "issue:44610|issue:46369", + "issue:43650|issue:46202", + "issue:43761|issue:44479", + "issue:43065|issue:44617", + "issue:43540|issue:44514", + "issue:45127|issue:45200", + "issue:43388|issue:43688", + "issue:43352|issue:46640", + "issue:43572|issue:46178" + ], + "prepared_review_unit_hash": "0e65eb2586586f85c6b123b69d74c3aa078bf202ea6a4fc5f84f14c9469b693b", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35625, + "estimated_input_tokens": 8907, + "estimated_eval_tokens": 18070 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose mix of unrelated Transformers issues; none of the soft pairs look like the same underlying bug or change. Most similarities are only at the subsystem/model-family level, not the concrete failure mode.", + "confidence": 0.9, + "canonical_issue_reason": "Issue 43646 is the broadest, most representative regression in the set: a general Transformers 5.0.0 custom model initialization breakage.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43646 is the best standalone representative because it describes a concrete, high-impact v5 regression with broad user-facing scope.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43550", + "right": "issue:43701", + "accept": false, + "reason": "Different failures: torch.compile/SDPA runtime bug vs checkpoint key mismatch." + }, + { + "left": "issue:46234", + "right": "issue:46291", + "accept": false, + "reason": "Unrelated import problems: missing utility symbol vs pipelines import constraint from kernels." + }, + { + "left": "issue:43606", + "right": "issue:43701", + "accept": false, + "reason": "CPU offload device mismatch is unrelated to resume-from-checkpoint key mapping." + }, + { + "left": "issue:43295", + "right": "issue:43825", + "accept": false, + "reason": "Both are regressions, but one is a processor/tokenizer API break and the other is a pipeline error-message issue." + }, + { + "left": "issue:16998", + "right": "issue:22355", + "accept": false, + "reason": "Model max length question and missing transformers.onnx module are unrelated." + }, + { + "left": "issue:43295", + "right": "issue:44871", + "accept": false, + "reason": "Different regressions: processor/tokenizer handling vs Gemma eos token config inconsistency." + }, + { + "left": "issue:44871", + "right": "issue:45584", + "accept": false, + "reason": "Both involve generation/token settings, but the concrete bugs differ: eos_token_id mismatch vs empty-transcription failure after token alignment." + }, + { + "left": "issue:43316", + "right": "issue:43716", + "accept": false, + "reason": "Gemma3 config API discrepancy and Mistral-3 image dtype mismatch are different issues." + }, + { + "left": "issue:42175", + "right": "issue:43316", + "accept": false, + "reason": "Backend packaging/install issue is unrelated to Gemma3 config API shape." + }, + { + "left": "issue:44077", + "right": "issue:44534", + "accept": false, + "reason": "One is an invalid optional post_init allowance; the other is a non-persistent buffer initialization regression." + }, + { + "left": "issue:44871", + "right": "issue:46050", + "accept": false, + "reason": "Gemma eos_token_id config mismatch is unrelated to quantized save/load breakage." + }, + { + "left": "issue:45803", + "right": "issue:46365", + "accept": false, + "reason": "Bare except in tensor conversion is unrelated to Trainer import failure caused by datasets." + }, + { + "left": "issue:43646", + "right": "issue:44913", + "accept": false, + "reason": "Both relate to model/config loading, but one is a broad initialization regression and the other is a specific config field not persisting on reload." + }, + { + "left": "issue:43452", + "right": "issue:44496", + "accept": false, + "reason": "GGUF loading path breakage and unrecognized model config are different model-loading problems." + }, + { + "left": "issue:44162", + "right": "issue:44792", + "accept": false, + "reason": "ESM2 model breakage and a janus image-generation test failure are unrelated." + }, + { + "left": "issue:45803", + "right": "issue:46170", + "accept": false, + "reason": "Different concerns: exception handling in Fuyu vs Windows/ROCm import dependency issues." + }, + { + "left": "issue:45632", + "right": "issue:46424", + "accept": false, + "reason": "Cache path collision for trust_remote_code is unrelated to max_cache_length behavior." + }, + { + "left": "issue:43404", + "right": "issue:43441", + "accept": false, + "reason": "Mistral3 lm_head tying bug and Ministral-3 FlashAttention support issue affect different code paths." + }, + { + "left": "issue:44610", + "right": "issue:46369", + "accept": false, + "reason": "Processor output size mismatch and Sam3-litetext memory overflow are different problems." + }, + { + "left": "issue:43650", + "right": "issue:46202", + "accept": false, + "reason": "Placeholder issue text and Gemma pipeline roundtrip failure do not match." + }, + { + "left": "issue:43761", + "right": "issue:44479", + "accept": false, + "reason": "CLIPVision hidden_states regression and Qwen video-input regression are unrelated." + }, + { + "left": "issue:43065", + "right": "issue:44617", + "accept": false, + "reason": "Dummy Conv2d in Sam3PixelDecoder is not the same as Sam3Video OOM." + }, + { + "left": "issue:43540", + "right": "issue:44514", + "accept": false, + "reason": "Both are Qwen multimodal bugs, but they hit different APIs and failure modes: video input processing vs batched chat-template padding." + }, + { + "left": "issue:45127", + "right": "issue:45200", + "accept": false, + "reason": "LoRA merge collapse with tied embeddings is unrelated to Gemma 4 mm_token_type_ids defaults." + }, + { + "left": "issue:43388", + "right": "issue:43688", + "accept": false, + "reason": "Metric gathering label truncation and auxiliary loss normalization are different training bugs." + }, + { + "left": "issue:43352", + "right": "issue:46640", + "accept": false, + "reason": "Both mention Flash Attention, but 43352 is model-specific unsupported behavior while 46640 is too generic to be the same bug." + }, + { + "left": "issue:43572", + "right": "issue:46178", + "accept": false, + "reason": "StableLmConfig pad_token_idx regression and tensorflow_text backend mapping error 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": 22, + "review_unit_index": 35, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:25251", + "issue:27957", + "issue:35141", + "issue:41950", + "issue:42490", + "issue:43335", + "issue:43475", + "issue:43504", + "issue:43525", + "issue:43531", + "issue:43550", + "issue:43644", + "issue:43673", + "issue:43761", + "issue:43827", + "issue:43931", + "issue:43994", + "issue:44060", + "issue:44190", + "issue:44242", + "issue:44479", + "issue:44534", + "issue:44617", + "issue:44734", + "issue:44743", + "issue:44792", + "issue:44855", + "issue:44913", + "issue:44933", + "issue:44964", + "issue:44977", + "issue:45362", + "issue:45381", + "issue:45406", + "issue:45542", + "issue:45563", + "issue:45584", + "issue:45588", + "issue:45600", + "issue:45632", + "issue:45698", + "issue:46082", + "issue:46234", + "issue:46369", + "issue:46421", + "issue:46617", + "issue:46697" + ], + "soft_pairs": [ + "issue:45542|issue:46234", + "issue:43504|issue:45698", + "issue:43673|issue:44743", + "issue:44792|issue:45600", + "issue:44792|issue:44977", + "issue:41950|issue:43475", + "issue:43525|issue:43531", + "issue:44190|issue:44964", + "issue:44734|issue:45406", + "issue:44534|issue:44913", + "issue:46082|issue:46697", + "issue:35141|issue:42490", + "issue:44855|issue:45588", + "issue:43931|issue:44060", + "issue:43761|issue:44242", + "issue:45632|issue:46617", + "issue:43827|issue:43994", + "issue:44617|issue:45584", + "issue:43673|issue:46421", + "issue:45563|issue:45584", + "issue:45381|issue:46369", + "issue:43644|issue:44913", + "issue:43335|issue:44855", + "issue:44479|issue:44734", + "issue:44933|issue:45362", + "issue:25251|issue:27957", + "issue:43550|issue:44190" + ], + "prepared_review_unit_hash": "bc7697b5d53c2b89b03280207f66a2f535c78c351a1c9530b1b135ca5d58c3cb", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35209, + "estimated_input_tokens": 8803, + "estimated_eval_tokens": 17862 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: most soft pairs only share broad themes like loading, caching, generation, or pipeline behavior, but they describe different bugs and code paths. No PRs are present.", + "confidence": 0.95, + "canonical_issue_reason": "No single issue truly represents this set because the items are not one duplicate cluster; if forced, #45698 is the broadest and most active loading-related issue, but it does not semantically cover the rest.", + "canonical_pr_reason": null, + "best_issue_reason": "Not a strong global best for duplicate triage because the pairings are mostly unrelated; #45698 is only the most active generic loading bug, not a true cluster anchor.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45542", + "right": "issue:46234", + "accept": false, + "reason": "Different failures: TensorBoard/TF backend import path vs a missing import-utils availability helper." + }, + { + "left": "issue:43504", + "right": "issue:45698", + "accept": false, + "reason": "Both involve model loading, but one is a legacy BEiT preset field bug and the other is wrong custom-module resolution after save/load." + }, + { + "left": "issue:43673", + "right": "issue:44743", + "accept": false, + "reason": "Cache-related, but different bugs: missing GenerationMixin cache in chunked prefill vs recurrent states resetting with seq_len>1." + }, + { + "left": "issue:44792", + "right": "issue:45600", + "accept": false, + "reason": "Unrelated: a Janus image-generation test failure vs auto_mappings referencing removed Sam3LiteText configs." + }, + { + "left": "issue:44792", + "right": "issue:44977", + "accept": false, + "reason": "Different models and paths: Janus image generation test failure vs Qwen3.5 flash-attention generation bug." + }, + { + "left": "issue:41950", + "right": "issue:43475", + "accept": false, + "reason": "Pipeline processor-selection bug vs a missing output attribute in Sam3 vision encoder output." + }, + { + "left": "issue:43525", + "right": "issue:43531", + "accept": false, + "reason": "Different config issues: missing pad_token_id on Llama4Config vs Qwen3-MoE sliding_window behavior." + }, + { + "left": "issue:44190", + "right": "issue:44964", + "accept": false, + "reason": "Different loading contexts: local dataset handling in a training script vs loading a multimodal model with latest transformers." + }, + { + "left": "issue:44734", + "right": "issue:45406", + "accept": false, + "reason": "Both touch serve, but one is KV-cache continuation indexing and the other is Gemma4Processor missing _tokenizer." + }, + { + "left": "issue:44534", + "right": "issue:44913", + "accept": false, + "reason": "Distinct bugs: non-persistent buffers being junk-filled vs GPTNeoXConfig rotary_pct not persisting on reload." + }, + { + "left": "issue:46082", + "right": "issue:46697", + "accept": false, + "reason": "Unrelated model/config validation vs image-processor padding behavior." + }, + { + "left": "issue:35141", + "right": "issue:42490", + "accept": false, + "reason": "Different areas: output embedding reinitialization after resize vs local-path trailing-slash import warning." + }, + { + "left": "issue:44855", + "right": "issue:45588", + "accept": false, + "reason": "Python 3.13 import parsing failure vs a flash_attention crash on s_aux=None." + }, + { + "left": "issue:43931", + "right": "issue:44060", + "accept": false, + "reason": "Different Qwen issues: VL weight-shape mismatch vs a tied-weights warning involving dt_bias." + }, + { + "left": "issue:43761", + "right": "issue:44242", + "accept": false, + "reason": "Both are v5/model behavior bugs, but one is CLIP hidden_states output and the other is MoE load-balancing loss gating." + }, + { + "left": "issue:45632", + "right": "issue:46617", + "accept": false, + "reason": "Related cache/loading area, but not the same bug: cache-path collision for local leaf names vs FileNotFoundError when loading from cache." + }, + { + "left": "issue:43827", + "right": "issue:43994", + "accept": false, + "reason": "Docs still referencing pipeline() is not the same as siglip2 producing bad outputs with AutoModel/pipeline." + }, + { + "left": "issue:44617", + "right": "issue:45584", + "accept": false, + "reason": "Completely different symptoms: Sam3Video CUDA OOM vs Whisper empty-transcription generation failure." + }, + { + "left": "issue:43673", + "right": "issue:46421", + "accept": false, + "reason": "Both mention cache/chunked prefill, but one is a missing cache regression and the other is lazy cache init causing recompilation." + }, + { + "left": "issue:45563", + "right": "issue:45584", + "accept": false, + "reason": "Different generation bugs: stale warning for num_return_sequences vs Whisper empty-transcription handling." + }, + { + "left": "issue:45381", + "right": "issue:46369", + "accept": false, + "reason": "Different regressions: Qwen2.5-VL video position IDs vs Sam3-LiteText memory overflow." + }, + { + "left": "issue:43644", + "right": "issue:44913", + "accept": false, + "reason": "Different model-state issues: junk non-persistent buffers vs rotary_pct reverting after reload." + }, + { + "left": "issue:43335", + "right": "issue:44855", + "accept": false, + "reason": "Unrelated: SwitchTransformers sparse-layer creation bug vs a DebertaV2 Python 3.13 indentation/parsing error." + }, + { + "left": "issue:44479", + "right": "issue:44734", + "accept": false, + "reason": "Both are regressions, but one is video-input handling and the other is serve KV-cache continuation indexing." + }, + { + "left": "issue:44933", + "right": "issue:45362", + "accept": false, + "reason": "Different failures: nonexistent image_utils import vs Qwen3.5 chat generation crash." + }, + { + "left": "issue:25251", + "right": "issue:27957", + "accept": false, + "reason": "Unrelated: pipeline top_k returning a nested list vs XLMRoberta Flash Attention 2 support." + }, + { + "left": "issue:43550", + "right": "issue:44190", + "accept": false, + "reason": "Different code paths: Bamba SDPA/torch.compile failure vs a local dataset-loading script 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": 22, + "review_unit_index": 36, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:39692", + "issue:41720", + "issue:42175", + "issue:42796", + "issue:43065", + "issue:43295", + "issue:43335", + "issue:43526", + "issue:43540", + "issue:43854", + "issue:44077", + "issue:44112", + "issue:44190", + "issue:44246", + "issue:44279", + "issue:44462", + "issue:44464", + "issue:44521", + "issue:44610", + "issue:44617", + "issue:44734", + "issue:44737", + "issue:44779", + "issue:44877", + "issue:44898", + "issue:44938", + "issue:44964", + "issue:45003", + "issue:45127", + "issue:45245", + "issue:45276", + "issue:45357", + "issue:45381", + "issue:45542", + "issue:45563", + "issue:45584", + "issue:45600", + "issue:45632", + "issue:45803", + "issue:45907", + "issue:46082", + "issue:46121", + "issue:46234", + "issue:46369", + "issue:46489", + "issue:46655", + "issue:46697" + ], + "soft_pairs": [ + "issue:41720|issue:45245", + "issue:44077|issue:44877", + "issue:43065|issue:44610", + "issue:43335|issue:44190", + "issue:44464|issue:45563", + "issue:45381|issue:46655", + "issue:45127|issue:45357", + "issue:44617|issue:46369", + "issue:44779|issue:46489", + "issue:43526|issue:46697", + "issue:39692|issue:44190", + "issue:44112|issue:45600", + "issue:44737|issue:45803", + "issue:43295|issue:45276", + "issue:43540|issue:44734", + "issue:44521|issue:45381", + "issue:44938|issue:45542", + "issue:45276|issue:45584", + "issue:44462|issue:45907", + "issue:45632|issue:46234", + "issue:44898|issue:46697", + "issue:44246|issue:45003", + "issue:46082|issue:46121", + "issue:43854|issue:44964", + "issue:44964|issue:45600", + "issue:45245|issue:45584", + "issue:42796|issue:44521", + "issue:42175|issue:44279", + "issue:44112|issue:44964" + ], + "prepared_review_unit_hash": "d9a8730effbba1dc2688b4fd791c2ce8ab868d3b4d5107e3b718cfee1da7e574", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35806, + "estimated_input_tokens": 8952, + "estimated_eval_tokens": 18160 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose soft-similarity set of unrelated issues across different models, processors, configs, and tooling. I would not merge any of the soft edges as duplicates.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue stands out: the items span different subsystems and symptoms, so the cluster is too heterogeneous for duplicate consolidation.", + "canonical_pr_reason": null, + "best_issue_reason": "No best issue for duplicate triage; if forced to pick a representative, none of these is meaningfully representative of the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41720", + "right": "issue:45245", + "accept": false, + "reason": "Qwen3 auto device-mapping CUDA assert and a category-count limit error are unrelated failures." + }, + { + "left": "issue:44077", + "right": "issue:44877", + "accept": false, + "reason": "Different problems: optional post_init on PatchTSMixer vs strict granite_speech config loading." + }, + { + "left": "issue:43065", + "right": "issue:44610", + "accept": false, + "reason": "Sam3PixelDecoder dummy Conv2d and OmDet-Turbo processor/model size mismatch are separate bugs." + }, + { + "left": "issue:43335", + "right": "issue:44190", + "accept": false, + "reason": "SwitchTransformers sparse-layer construction and local dataset loading in a training script do not match." + }, + { + "left": "issue:44464", + "right": "issue:45563", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency and a stale paged-generate warning are different issues." + }, + { + "left": "issue:45381", + "right": "issue:46655", + "accept": false, + "reason": "Wrong vision_position_ids in Qwen2.5-VL video input vs SmolVLM video padding calculation." + }, + { + "left": "issue:45127", + "right": "issue:45357", + "accept": false, + "reason": "LoRA merge collapse with tied embeddings is unrelated to incorrect visual encoder keys on save_pretrained." + }, + { + "left": "issue:44617", + "right": "issue:46369", + "accept": false, + "reason": "Same family, but one is Sam3Video OOM and the other is Sam3-LiteText memory overflow; not the same bug." + }, + { + "left": "issue:44779", + "right": "issue:46489", + "accept": false, + "reason": "Both involve DeepSeek tokenizers, but they target different models/versions and different incorrect-output causes." + }, + { + "left": "issue:43526", + "right": "issue:46697", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and Swin2SR over-padding are unrelated image-processing bugs." + }, + { + "left": "issue:39692", + "right": "issue:44190", + "accept": false, + "reason": "SigLIP2 docs example errors and a local-dataset training script issue are unrelated." + }, + { + "left": "issue:44112", + "right": "issue:45600", + "accept": false, + "reason": "Stale GraniteSpeech device-override test vs auto_mappings referencing removed Sam3LiteText configs." + }, + { + "left": "issue:44737", + "right": "issue:45803", + "accept": false, + "reason": "XLNet device placement in torch.arange and a bare-except swallowing errors are different code paths." + }, + { + "left": "issue:43295", + "right": "issue:45276", + "accept": false, + "reason": "Processor.tokenizer regression and Gemma4 embedding-resize propagation are separate API bugs." + }, + { + "left": "issue:43540", + "right": "issue:44734", + "accept": false, + "reason": "Qwen3OmniMoe video input validation and serve KV-cache continuation indexing are unrelated crashes." + }, + { + "left": "issue:44521", + "right": "issue:45381", + "accept": false, + "reason": "Both are multimodal, but assistant-mask generation and video vision_position_ids are distinct functions/issues." + }, + { + "left": "issue:44938", + "right": "issue:45542", + "accept": false, + "reason": "Python 3.14 import failure and a missing TensorFlow backend when only tensorboard is installed are different packaging/runtime problems." + }, + { + "left": "issue:45276", + "right": "issue:45584", + "accept": false, + "reason": "Gemma4 resize_token_embeddings propagation and Whisper empty-transcription handling are unrelated." + }, + { + "left": "issue:44462", + "right": "issue:45907", + "accept": false, + "reason": "AutoTokenizer ignoring tokenizer.json and list_repo_files not forwarding revision are different repository-access bugs." + }, + { + "left": "issue:45632", + "right": "issue:46234", + "accept": false, + "reason": "Cache-path collision for trust_remote_code and a missing import_utils helper are unrelated." + }, + { + "left": "issue:44898", + "right": "issue:46697", + "accept": false, + "reason": "Perceiver interpolate_pos_encoding failure and Swin2SR over-padding are separate image-processor issues." + }, + { + "left": "issue:44246", + "right": "issue:45003", + "accept": false, + "reason": "Slow import timing and unsafe sys.modules access are not the same problem." + }, + { + "left": "issue:46082", + "right": "issue:46121", + "accept": false, + "reason": "Llama head_dim validation and JSON-loaded rope-key list handling are unrelated config parsing bugs." + }, + { + "left": "issue:43854", + "right": "issue:44964", + "accept": false, + "reason": "Both are model-loading reports, but they concern different models and likely different failure causes." + }, + { + "left": "issue:44964", + "right": "issue:45600", + "accept": false, + "reason": "Phi-4 multimodal load failure and Sam3LiteText auto-mapping CI break are unrelated." + }, + { + "left": "issue:45245", + "right": "issue:45584", + "accept": false, + "reason": "Category-count overflow and Whisper empty transcription have no shared code path." + }, + { + "left": "issue:42796", + "right": "issue:44521", + "accept": false, + "reason": "Tokenizer BPE-marker handling and multimodal assistant-mask generation are different subsystems." + }, + { + "left": "issue:42175", + "right": "issue:44279", + "accept": false, + "reason": "TensorFlow missing from pip extras and a generic dependency issue are too vague and not the same concrete bug." + }, + { + "left": "issue:44112", + "right": "issue:44964", + "accept": false, + "reason": "GraniteSpeech CI device override and Phi-4 model loading are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 37, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:27957", + "issue:28282", + "issue:32090", + "issue:35141", + "issue:36246", + "issue:38617", + "issue:41628", + "issue:41950", + "issue:42907", + "issue:43295", + "issue:43381", + "issue:43404", + "issue:43441", + "issue:43479", + "issue:43526", + "issue:43531", + "issue:43643", + "issue:43650", + "issue:43673", + "issue:43824", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44062", + "issue:44186", + "issue:44220", + "issue:44279", + "issue:44295", + "issue:44336", + "issue:44496", + "issue:44683", + "issue:44734", + "issue:44792", + "issue:44933", + "issue:44987", + "issue:45042", + "issue:45083", + "issue:45216", + "issue:45335", + "issue:45375", + "issue:45446", + "issue:45588", + "issue:45593", + "issue:45600", + "issue:45759", + "issue:46424", + "issue:46506" + ], + "soft_pairs": [ + "issue:43643|issue:44496", + "issue:27957|issue:35141", + "issue:43404|issue:45216", + "issue:44186|issue:44734", + "issue:36246|issue:41628", + "issue:43650|issue:44220", + "issue:36246|issue:38617", + "issue:44295|issue:46506", + "issue:32090|issue:35141", + "issue:43479|issue:45375", + "issue:41950|issue:44016", + "issue:28282|issue:35141", + "issue:43673|issue:46424", + "issue:43531|issue:45600", + "issue:44016|issue:44220", + "issue:45083|issue:46506", + "issue:43295|issue:45335", + "issue:43976|issue:44987", + "issue:44683|issue:45446", + "issue:43526|issue:43901", + "issue:44496|issue:45042", + "issue:42907|issue:43441", + "issue:43673|issue:44336", + "issue:43381|issue:45593", + "issue:43901|issue:44496", + "issue:43976|issue:44279", + "issue:43441|issue:45588", + "issue:44016|issue:44062", + "issue:44496|issue:45759", + "issue:44792|issue:46424", + "issue:43824|issue:44933" + ], + "prepared_review_unit_hash": "287b6e4c060a4d22c933096cccfbea291f581553da60b151489fa9f031af44ad", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36484, + "estimated_input_tokens": 9121, + "estimated_eval_tokens": 18498 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These are mostly unrelated Transformers issues that only overlap at a high level (imports, configs, cache, attention, or model-loading). None of the proposed soft pairs look like the same underlying bug or change, so I would not merge any of them.", + "confidence": 0.91, + "canonical_issue_reason": "issue:44496 is the broadest, most reusable representative of the cluster\u2019s config/load-path failures and has a clear, user-facing reproduction.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44496 is the clearest general loading/configuration failure with broad impact, making it the best anchor issue in this loose cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43643", + "right": "issue:44496", + "accept": false, + "reason": "Both are loading/configuration failures, but one is about missing fields with `trust_remote_code=True` and the other is an unrecognized model/config `model_type`; different bugs." + }, + { + "left": "issue:27957", + "right": "issue:35141", + "accept": false, + "reason": "FlashAttention/model-support behavior and embedding reinitialization after resize are unrelated code paths." + }, + { + "left": "issue:43404", + "right": "issue:45216", + "accept": false, + "reason": "Both touch weight tying, but they affect different model families and different failure modes." + }, + { + "left": "issue:44186", + "right": "issue:44734", + "accept": false, + "reason": "Tokenizer padding/NER crashes are unrelated to the serve KV-cache tensor-indexing bug." + }, + { + "left": "issue:36246", + "right": "issue:41628", + "accept": false, + "reason": "Different missing exports: `Qwen2_5_VLImageProcessor` versus `AutoImageProcessor`." + }, + { + "left": "issue:43650", + "right": "issue:44220", + "accept": false, + "reason": "`ADD THE DATA` is not the same bug as `_torch_extract_fbank_features()` misbehavior." + }, + { + "left": "issue:36246", + "right": "issue:38617", + "accept": false, + "reason": "Both are import errors, but they are for different symbols in different modules." + }, + { + "left": "issue:44295", + "right": "issue:46506", + "accept": false, + "reason": "A buffer read/access issue is not the same as CPU/CUDA output divergence." + }, + { + "left": "issue:32090", + "right": "issue:35141", + "accept": false, + "reason": "Distributed broadcast `NoneType` handling and embedding reinitialization are unrelated." + }, + { + "left": "issue:43479", + "right": "issue:45375", + "accept": false, + "reason": "Default config initialization with `None` is not the same as a field being dropped by `@strict`." + }, + { + "left": "issue:41950", + "right": "issue:44016", + "accept": false, + "reason": "Pipeline processor lookup is unrelated to a notebook syntax error." + }, + { + "left": "issue:28282", + "right": "issue:35141", + "accept": false, + "reason": "Missing PyTorch dependency detection and token-embedding resize behavior are different bugs." + }, + { + "left": "issue:43673", + "right": "issue:46424", + "accept": false, + "reason": "Both mention cache, but one is missing cache support in chunked prefill and the other is `max_cache_length` only applying to Quantized Cache." + }, + { + "left": "issue:43531", + "right": "issue:45600", + "accept": false, + "reason": "Sliding-window handling in Qwen3-MoE is unrelated to stale `auto_mappings.py` references." + }, + { + "left": "issue:44016", + "right": "issue:44220", + "accept": false, + "reason": "A notebook syntax error is not the same as a feature-extraction helper issue." + }, + { + "left": "issue:45083", + "right": "issue:46506", + "accept": false, + "reason": "Helper output-length semantics and CPU/CUDA result differences are not the same concrete defect." + }, + { + "left": "issue:43295", + "right": "issue:45335", + "accept": false, + "reason": "Processor/tokenizer regression and `decoder.embed_tokens` not updating after resize are different paths." + }, + { + "left": "issue:43976", + "right": "issue:44987", + "accept": false, + "reason": "Python version compatibility and a specific model-loading failure are unrelated." + }, + { + "left": "issue:44683", + "right": "issue:45446", + "accept": false, + "reason": "A compiled `flex_attention` failure on newer torch and an incorrect version guard for `AuxRequest` are different bugs." + }, + { + "left": "issue:43526", + "right": "issue:43901", + "accept": false, + "reason": "Image-label reduction behavior and documentation wording about `return_all_scores` do not describe the same issue." + }, + { + "left": "issue:44496", + "right": "issue:45042", + "accept": false, + "reason": "Generic model/config loading failure and an unnecessary torchvision dependency in PIL image processors are different bugs." + }, + { + "left": "issue:42907", + "right": "issue:43441", + "accept": false, + "reason": "Saving dequantized Ministrals/Devstrals and FlashAttention failures in Ministral-3 are related only at the model family level, not the same bug." + }, + { + "left": "issue:43673", + "right": "issue:44336", + "accept": false, + "reason": "Cache behavior and ANSI escape handling in loading_report are unrelated." + }, + { + "left": "issue:43381", + "right": "issue:45593", + "accept": false, + "reason": "Gradient checkpointing in eval mode and missing auxiliary losses for D-FINE are different behaviors." + }, + { + "left": "issue:43901", + "right": "issue:44496", + "accept": false, + "reason": "A docs/runtime mismatch about `return_all_scores` is not the same as a config/model-type import error." + }, + { + "left": "issue:43976", + "right": "issue:44279", + "accept": false, + "reason": "Python 3.9+ compatibility and a vague dependency issue are not the same underlying bug." + }, + { + "left": "issue:43441", + "right": "issue:45588", + "accept": false, + "reason": "Both involve flash attention, but one is a model-level RC failure and the other is a specific `s_aux=None` AttributeError; too different to merge." + }, + { + "left": "issue:44016", + "right": "issue:44062", + "accept": false, + "reason": "Notebook syntax error and `AddedToken(..., special=...)` argument collision are unrelated." + }, + { + "left": "issue:44496", + "right": "issue:45759", + "accept": false, + "reason": "Unrecognized `model_type` and failure to unwrap `text_config` in composite multimodal configs are different loading bugs." + }, + { + "left": "issue:44792", + "right": "issue:46424", + "accept": false, + "reason": "A janus image-generation test failure is not the same as cache-length handling for quantized cache." + }, + { + "left": "issue:43824", + "right": "issue:44933", + "accept": false, + "reason": "Missing export of `Qwen2_5_VLForConditionalGeneration` and non-existent import from `image_utils` are different import errors." + }, + { + "left": "issue:45042", + "right": "issue:46506", + "accept": false, + "reason": "Processor torchvision dependency regression and CPU/CUDA discrepancy 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": 22, + "review_unit_index": 38, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:39692", + "issue:41950", + "issue:42796", + "issue:43295", + "issue:43299", + "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:44521", + "issue:44683", + "issue:44779", + "issue:45020", + "issue:45081", + "issue:45200", + "issue:45216", + "issue:45276", + "issue:45278", + "issue:45335", + "issue:45357", + "issue:45405", + "issue:45542", + "issue:45584", + "issue:45593", + "issue:45812", + "issue:45823", + "issue:46082", + "issue:46123", + "issue:46209", + "issue:46369", + "issue:46399", + "issue:46512", + "issue:46515", + "issue:46693" + ], + "soft_pairs": [ + "issue:46123|issue:46369", + "issue:43299|issue:45216", + "issue:44521|issue:46693", + "issue:44279|issue:46515", + "issue:45335|issue:45584", + "issue:46082|issue:46399", + "issue:44279|issue:45405", + "issue:43577|issue:45216", + "issue:44075|issue:44368", + "issue:39692|issue:43901", + "issue:43295|issue:44779", + "issue:43976|issue:45405", + "issue:42796|issue:43650", + "issue:44351|issue:45278", + "issue:46082|issue:46512", + "issue:43299|issue:45357", + "issue:44779|issue:45812", + "issue:44242|issue:45200", + "issue:41950|issue:45020", + "issue:43299|issue:43854", + "issue:43479|issue:43525", + "issue:44466|issue:45276", + "issue:41950|issue:45823", + "issue:43650|issue:46209", + "issue:43653|issue:45593", + "issue:44683|issue:45542", + "issue:44462|issue:45081", + "issue:44016|issue:44291" + ], + "prepared_review_unit_hash": "ef27cdcc7f63d5ea04ece6c6d7c701a9561e33c8142bebe9e1b83d0aca99e5e2", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35631, + "estimated_input_tokens": 8908, + "estimated_eval_tokens": 18072 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: the items cover unrelated regressions in tokenizers, multimodal configs, pipelines, docs, runtime/version compatibility, and model-specific save/load paths. None of the soft pairs look like the same concrete bug or change, so I would not collapse any of them together.", + "confidence": 0.96, + "canonical_issue_reason": "No canonical issue: the cluster spans many unrelated reports, and the soft pairs are not true same-bug duplicates.", + "canonical_pr_reason": null, + "best_issue_reason": "No single issue stands out as the global representative; the reports are too diverse in model, subsystem, and failure mode.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46123", + "right": "issue:46369", + "accept": false, + "reason": "Different bugs: partial-batch result dropping in MaskGenerationPipeline vs memory overflow in sam3-litetext." + }, + { + "left": "issue:43299", + "right": "issue:45216", + "accept": false, + "reason": "Both are Qwen-related regressions, but one is model loading for Qwen3VL MoE and the other is save_pretrained checkpoint serialization for Qwen3.5." + }, + { + "left": "issue:44521", + "right": "issue:46693", + "accept": false, + "reason": "Unrelated: multimodal assistant_masks are all zero vs a long-context FA2 performance regression." + }, + { + "left": "issue:44279", + "right": "issue:46515", + "accept": false, + "reason": "Different compatibility problems: a transformers dependency issue vs torch 2.4.1 incompatibility." + }, + { + "left": "issue:45335", + "right": "issue:45584", + "accept": false, + "reason": "Different code paths: resize_token_embeddings not updating t5gemma embeddings vs Whisper failing on empty transcription after align_special_tokens." + }, + { + "left": "issue:46082", + "right": "issue:46399", + "accept": false, + "reason": "Separate issues: LlamaConfig head_dim validation vs Qwen3.5 use_kernels rotary registration crash." + }, + { + "left": "issue:44279", + "right": "issue:45405", + "accept": false, + "reason": "Dependency/version complaints, but not the same bug: generic transformers dependency issue vs an unsupported MIN_PEFT_VERSION bump." + }, + { + "left": "issue:43577", + "right": "issue:45216", + "accept": false, + "reason": "Different failures: BLIP2 dtype propagation bug vs Qwen3.5 save_pretrained regression." + }, + { + "left": "issue:44075", + "right": "issue:44368", + "accept": false, + "reason": "Unrelated: SGD optimizer args ignored vs a Qwen3.5 LoRA warning about tie_word_embeddings." + }, + { + "left": "issue:39692", + "right": "issue:43901", + "accept": false, + "reason": "Both are docs issues, but one is a SigLIP2 example error and the other is stale TextClassificationPipeline docs about return_all_scores." + }, + { + "left": "issue:43295", + "right": "issue:44779", + "accept": false, + "reason": "Different tokenizer/processor problems: processor.tokenizer regression vs Deepseek tokenizer output corruption." + }, + { + "left": "issue:43976", + "right": "issue:45405", + "accept": false, + "reason": "Different topics: Python 3.9+ support regression vs a PEFT minimum-version pin issue." + }, + { + "left": "issue:42796", + "right": "issue:43650", + "accept": false, + "reason": "No relation: tokenizer BPE marker handling vs an empty/nonspecific issue titled 'ADD THE DATA'." + }, + { + "left": "issue:44351", + "right": "issue:45278", + "accept": false, + "reason": "Import failures, but not the same concrete problem: missing HybridCache export vs broad import errors after upgrade." + }, + { + "left": "issue:46082", + "right": "issue:46512", + "accept": false, + "reason": "Different behavior: config validation for head_dim vs from_config leaving model.config.dtype stale." + }, + { + "left": "issue:43299", + "right": "issue:45357", + "accept": false, + "reason": "Separate Qwen regressions: loading Qwen3VL MoE vs saving incorrect visual encoder keys for Qwen3.5." + }, + { + "left": "issue:44779", + "right": "issue:45812", + "accept": false, + "reason": "Both tokenizer regressions, but for different model families and different symptoms: Deepseek tokenization vs Granite token IDs." + }, + { + "left": "issue:44242", + "right": "issue:45200", + "accept": false, + "reason": "Different architecture bugs: missing load-balancing loss vs Gemma 4 mm_token_type_ids defaulting." + }, + { + "left": "issue:41950", + "right": "issue:45020", + "accept": false, + "reason": "Unrelated: video-classification pipeline using image processors vs remote_code models breaking in recent versions." + }, + { + "left": "issue:43299", + "right": "issue:43854", + "accept": false, + "reason": "Different models and failure modes: Qwen3VL loading regression vs GLM-4.7-Flash failing in unit tests." + }, + { + "left": "issue:43479", + "right": "issue:43525", + "accept": false, + "reason": "Different config bugs: Phi4MultimodalConfig default handling vs Llama4Config missing pad_token_id." + }, + { + "left": "issue:44466", + "right": "issue:45276", + "accept": false, + "reason": "Different serialization vs embedding-resize issues: lm_head.weight tied-weight inconsistency vs Gemma4 resize_token_embeddings not propagating." + }, + { + "left": "issue:41950", + "right": "issue:45823", + "accept": false, + "reason": "Different subsystems: pipeline processor lookup vs Gemma4 PLE device mismatch during forward." + }, + { + "left": "issue:43650", + "right": "issue:46209", + "accept": false, + "reason": "No overlap: a vague 'ADD THE DATA' issue vs FP8 inference failure on B200." + }, + { + "left": "issue:43653", + "right": "issue:45593", + "accept": false, + "reason": "Different model bugs: BigBirdTokenizer special-token registration vs D-FINE auxiliary losses with denoising off." + }, + { + "left": "issue:44683", + "right": "issue:45542", + "accept": false, + "reason": "Unrelated runtime issues: compiled flex_attention on torch>=2.9 vs tensorboard-only install causing TF backend errors." + }, + { + "left": "issue:44462", + "right": "issue:45081", + "accept": false, + "reason": "Both involve tokenizers, but one is AutoTokenizer ignoring tokenizer.json and the other is Mistral regex patching crashing on backend_tokenizer access." + }, + { + "left": "issue:44016", + "right": "issue:44291", + "accept": false, + "reason": "Completely different: a notebook syntax error vs init_empty_weights passing an unexpected _is_hf_initialized argument." + } + ] + }, + "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": 22, + "review_unit_index": 39, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:33290", + "issue:41628", + "issue:42175", + "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:43994", + "issue:44016", + "issue:44162", + "issue:44188", + "issue:44246", + "issue:44261", + "issue:44263", + "issue:44297", + "issue:44336", + "issue:44361", + "issue:44373", + "issue:44589", + "issue:44623", + "issue:44671", + "issue:44683", + "issue:44749", + "issue:44779", + "issue:45081", + "issue:45103", + "issue:45362", + "issue:45372", + "issue:45397", + "issue:45542", + "issue:45593", + "issue:45600", + "issue:45656", + "issue:45712", + "issue:45800", + "issue:45823", + "issue:45834", + "issue:45907", + "issue:46291", + "issue:46399" + ], + "soft_pairs": [ + "issue:43441|issue:45834", + "issue:45103|issue:45372", + "issue:44361|issue:45593", + "issue:33290|issue:45656", + "issue:45081|issue:45907", + "issue:44261|issue:44263", + "issue:42994|issue:43317", + "issue:45600|issue:45712", + "issue:44623|issue:44749", + "issue:44623|issue:45397", + "issue:42175|issue:43976", + "issue:44671|issue:44779", + "issue:43901|issue:43994", + "issue:43441|issue:44683", + "issue:43673|issue:43761", + "issue:43723|issue:44162", + "issue:43901|issue:44246", + "issue:46291|issue:46399", + "issue:45542|issue:45800", + "issue:43525|issue:43881", + "issue:43824|issue:43976", + "issue:44016|issue:45823", + "issue:44188|issue:44336", + "issue:43525|issue:43643", + "issue:44336|issue:44373", + "issue:44589|issue:45362", + "issue:41628|issue:43976", + "issue:44297|issue:45397" + ], + "prepared_review_unit_hash": "9f41bc2fd015c20deb9f90ecf2d62050f8aedd9a574ef01b434849302d46bf8d", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 35990, + "estimated_input_tokens": 8998, + "estimated_eval_tokens": 18252 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly unrelated issue reports with a loose Transformers v5/regression theme; none of the soft pairs look like the same underlying bug or change. I\u2019d use the broad tokenizer-loading regression as the closest canonical issue.", + "confidence": 0.41, + "canonical_issue_reason": "issue:43723 is the broadest, most umbrella-like report here: a v5 tokenizer-loading failure via AutoTokenizer.from_pretrained that best represents the shared loading/regression theme.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43723 is the strongest standalone representative because it is generic, user-facing, and closest to the cluster\u2019s recurring v5 loading/import regressions.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43441", + "right": "issue:45834", + "accept": false, + "reason": "Different models and failure paths; not the same bug." + }, + { + "left": "issue:45103", + "right": "issue:45372", + "accept": false, + "reason": "Docstring helper crash vs missing import in Gemma processor loading." + }, + { + "left": "issue:44361", + "right": "issue:45593", + "accept": false, + "reason": "Tokenizer AttributeError vs missing auxiliary losses; unrelated." + }, + { + "left": "issue:33290", + "right": "issue:45656", + "accept": false, + "reason": "Both mention deepspeed, but optimizer OOM and double-step are distinct bugs." + }, + { + "left": "issue:45081", + "right": "issue:45907", + "accept": false, + "reason": "Tokenizer regex crash vs list_repo_files revision forwarding; unrelated." + }, + { + "left": "issue:44261", + "right": "issue:44263", + "accept": false, + "reason": "Different model internals and different symptoms." + }, + { + "left": "issue:42994", + "right": "issue:43317", + "accept": false, + "reason": "Saving quantized models vs loading dequantized models are different code paths." + }, + { + "left": "issue:45600", + "right": "issue:45712", + "accept": false, + "reason": "Different CI/repo hygiene problems, not one underlying defect." + }, + { + "left": "issue:44623", + "right": "issue:44749", + "accept": false, + "reason": "Processor saving files issue vs performance regression; unrelated." + }, + { + "left": "issue:44623", + "right": "issue:45397", + "accept": false, + "reason": "Processor save bug vs gemma-4 Zero3 load bug; different failures." + }, + { + "left": "issue:42175", + "right": "issue:43976", + "accept": false, + "reason": "Backend dependency packaging vs Python version incompatibility; unrelated." + }, + { + "left": "issue:44671", + "right": "issue:44779", + "accept": false, + "reason": "Different models and different incorrect outputs." + }, + { + "left": "issue:43901", + "right": "issue:43994", + "accept": false, + "reason": "Docs mismatch vs model/pipeline inference error; unrelated." + }, + { + "left": "issue:43441", + "right": "issue:44683", + "accept": false, + "reason": "Both attention-related, but different kernels and distinct failures." + }, + { + "left": "issue:43673", + "right": "issue:43761", + "accept": false, + "reason": "Cache regression vs CLIP hidden_states bug; different code paths." + }, + { + "left": "issue:43723", + "right": "issue:44162", + "accept": false, + "reason": "Tokenizer loading issue vs ESM2 model breakage; not the same bug." + }, + { + "left": "issue:43901", + "right": "issue:44246", + "accept": false, + "reason": "Documentation issue vs slow import issue; unrelated." + }, + { + "left": "issue:46291", + "right": "issue:46399", + "accept": false, + "reason": "Both kernels-related, but import requirement failure and missing decorator are separate issues." + }, + { + "left": "issue:45542", + "right": "issue:45800", + "accept": false, + "reason": "TensorBoard/TensorFlow backend packaging vs torch/transformers incompatibility; unrelated." + }, + { + "left": "issue:43525", + "right": "issue:43881", + "accept": false, + "reason": "Different model loading problems with different root causes." + }, + { + "left": "issue:43824", + "right": "issue:43976", + "accept": false, + "reason": "Missing export/import error vs Python version support issue; unrelated." + }, + { + "left": "issue:44016", + "right": "issue:45823", + "accept": false, + "reason": "Notebook syntax error vs device mismatch in Gemma4; unrelated." + }, + { + "left": "issue:44188", + "right": "issue:44336", + "accept": false, + "reason": "Attention divergence bug vs ANSI logging output; unrelated." + }, + { + "left": "issue:43525", + "right": "issue:43643", + "accept": false, + "reason": "Missing config field in one model vs remote-code field loss in another path." + }, + { + "left": "issue:44336", + "right": "issue:44373", + "accept": false, + "reason": "Logging ANSI codes vs docstring text; unrelated." + }, + { + "left": "issue:44589", + "right": "issue:45362", + "accept": false, + "reason": "Storage type import failure vs chat runtime crash; different bugs." + }, + { + "left": "issue:41628", + "right": "issue:43976", + "accept": false, + "reason": "Import error for AutoImageProcessor vs Python version incompatibility; unrelated." + }, + { + "left": "issue:44297", + "right": "issue:45397", + "accept": false, + "reason": "Tokenizer save metadata mismatch vs gemma-4 Zero3 loading failure; different issues." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 40, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:36246", + "issue:41950", + "issue:42491", + "issue:42757", + "issue:42886", + "issue:43257", + "issue:43299", + "issue:43475", + "issue:43526", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43673", + "issue:43784", + "issue:43901", + "issue:44220", + "issue:44242", + "issue:44246", + "issue:44295", + "issue:44336", + "issue:44351", + "issue:44479", + "issue:44485", + "issue:44496", + "issue:44561", + "issue:44734", + "issue:44779", + "issue:44857", + "issue:44871", + "issue:45003", + "issue:45103", + "issue:45137", + "issue:45292", + "issue:45362", + "issue:45397", + "issue:45542", + "issue:45561", + "issue:45563", + "issue:45593", + "issue:45859", + "issue:46132", + "issue:46369", + "issue:46399", + "issue:46424", + "issue:46506", + "issue:46614", + "issue:46655", + "issue:46665" + ], + "soft_pairs": [ + "issue:43784|issue:45542", + "issue:44246|issue:44496", + "issue:45561|issue:45563", + "issue:46614|issue:46665", + "issue:45103|issue:46399", + "issue:44242|issue:45593", + "issue:42491|issue:43299", + "issue:43526|issue:45003", + "issue:43531|issue:45362", + "issue:44295|issue:45397", + "issue:44336|issue:44485", + "issue:44246|issue:44871", + "issue:42886|issue:46132", + "issue:44336|issue:46424", + "issue:43550|issue:46506", + "issue:36246|issue:42757", + "issue:41950|issue:44220", + "issue:36246|issue:44351", + "issue:44779|issue:45292", + "issue:45859|issue:46399", + "issue:45003|issue:45561", + "issue:44734|issue:46424", + "issue:43901|issue:45003", + "issue:43540|issue:46655", + "issue:43257|issue:45137", + "issue:43673|issue:44561", + "issue:44479|issue:46655", + "issue:44496|issue:45003", + "issue:43475|issue:46369", + "issue:44242|issue:44857" + ], + "prepared_review_unit_hash": "45488d8feb6367647cba632443c8782e1bfb36c16561b9b4482634cf0184dd2d", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36589, + "estimated_input_tokens": 9148, + "estimated_eval_tokens": 18552 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: most items are distinct bug reports across unrelated subsystems, and none of the soft pairs look like the same underlying issue. The broadest video-regression report is the best representative, but not a duplicate match for the others.", + "confidence": 0.71, + "canonical_issue_reason": "issue:44479 is the broadest, most umbrella-like report in the set: it covers a v5.3.0 video-input regression across multiple Qwen VL/VL-MoE variants, making it the strongest canonical issue candidate among these unrelated reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44479 is the most general and centrally phrased issue here, with the widest model coverage and clearest regression framing; it best represents the only loosely related theme in the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43784", + "right": "issue:45542", + "accept": false, + "reason": "Different import failures in different packages; no shared code path or concrete bug." + }, + { + "left": "issue:44246", + "right": "issue:44496", + "accept": false, + "reason": "One is an import-performance complaint, the other is model/config loading; unrelated." + }, + { + "left": "issue:45561", + "right": "issue:45563", + "accept": false, + "reason": "Different bugs: xdist race on captured_info.txt vs a stale warning in paged generation." + }, + { + "left": "issue:46614", + "right": "issue:46665", + "accept": false, + "reason": "Different tokenizer/cache behavior versus from_pretrained behavior in Python classes; not the same defect." + }, + { + "left": "issue:45103", + "right": "issue:46399", + "accept": false, + "reason": "Unrelated areas: auto-docstring annotation handling vs Qwen3.5 kernel registration crash." + }, + { + "left": "issue:44242", + "right": "issue:45593", + "accept": false, + "reason": "Both mention missing auxiliary losses, but they affect different models and mechanisms; too broad to be the same bug." + }, + { + "left": "issue:42491", + "right": "issue:43299", + "accept": false, + "reason": "Both involve Qwen3 MoE loading/version compatibility, but the concrete failures and code paths differ." + }, + { + "left": "issue:43526", + "right": "issue:45003", + "accept": false, + "reason": "BeitImageProcessorFast label reduction vs sys.modules access in modeling_utils are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:45362", + "accept": false, + "reason": "Same model family only at a high level; sliding-window behavior and chat crash are different issues." + }, + { + "left": "issue:44295", + "right": "issue:45397", + "accept": false, + "reason": "Different subsystems and symptoms: position_ids buffer access vs Gemma-4 ZeRO-3 from_pretrained." + }, + { + "left": "issue:44336", + "right": "issue:44485", + "accept": false, + "reason": "Loading-report ANSI output is unrelated to GLM-5 RoPE implementation." + }, + { + "left": "issue:44246", + "right": "issue:44871", + "accept": false, + "reason": "Import slowness and Gemma eos_token_id mismatch are unrelated." + }, + { + "left": "issue:42886", + "right": "issue:46132", + "accept": false, + "reason": "Offline tokenizer cache loading vs AttentionInterface.register behavior change are unrelated." + }, + { + "left": "issue:44336", + "right": "issue:46424", + "accept": false, + "reason": "ANSI handling in loading_report has nothing to do with cache max length behavior." + }, + { + "left": "issue:43550", + "right": "issue:46506", + "accept": false, + "reason": "Different model/runtime failures: torch.compile+SDPA vs CPU/CUDA numerical divergence." + }, + { + "left": "issue:36246", + "right": "issue:42757", + "accept": false, + "reason": "Both are import errors, but for unrelated symbols in different libraries." + }, + { + "left": "issue:41950", + "right": "issue:44220", + "accept": false, + "reason": "Video-classification processor lookup and fbank feature extraction are different bugs." + }, + { + "left": "issue:36246", + "right": "issue:44351", + "accept": false, + "reason": "Different missing imports from transformers; no evidence of the same symbol or code path." + }, + { + "left": "issue:44779", + "right": "issue:45292", + "accept": false, + "reason": "Tokenizer regression and embedding-resize behavior are different defects." + }, + { + "left": "issue:45859", + "right": "issue:46399", + "accept": false, + "reason": "Both are Qwen3.5 rotary-related, but one is CPU offload incompatibility and the other is a kernel registration/decorator crash; not mergeable as one fix." + }, + { + "left": "issue:45003", + "right": "issue:45561", + "accept": false, + "reason": "Sys.modules access and a pytest-xdist file race are unrelated." + }, + { + "left": "issue:44734", + "right": "issue:46424", + "accept": false, + "reason": "KV-cache continuation indexing and cache max-length configuration are distinct cache bugs." + }, + { + "left": "issue:43901", + "right": "issue:45003", + "accept": false, + "reason": "Docs/behavior mismatch in TextClassificationPipeline is unrelated to sys.modules access." + }, + { + "left": "issue:43540", + "right": "issue:46655", + "accept": false, + "reason": "Both are video-input failures, but they affect different models and likely different processors; too broad to treat as one bug." + }, + { + "left": "issue:43257", + "right": "issue:45137", + "accept": false, + "reason": "DeepSpeed-related issues, but one is Qwen3 MoE weight conversion and the other is a ZeRO3 deque error." + }, + { + "left": "issue:43673", + "right": "issue:44561", + "accept": false, + "reason": "Generation cache/chunked_prefill issue and trust_remote_code FX helper removal are unrelated." + }, + { + "left": "issue:44479", + "right": "issue:46655", + "accept": false, + "reason": "Both are video regressions, but 44479 is a broad Qwen VL-family regression while 46655 is a SmolVLM padding bug; different concrete fixes." + }, + { + "left": "issue:43475", + "right": "issue:46369", + "accept": false, + "reason": "SAM3 attribute error and SAM3 memory overflow are different failure modes." + }, + { + "left": "issue:44496", + "right": "issue:45003", + "accept": false, + "reason": "Unrecognized model/config parsing and sys.modules access are unrelated." + }, + { + "left": "issue:44734", + "right": "issue:46424", + "accept": false, + "reason": "Cache continuation crash and quantized-cache max length handling are not the same bug." + }, + { + "left": "issue:44242", + "right": "issue:44857", + "accept": false, + "reason": "Load-balancing loss omission and AMP/CUDA crash in LwDetrImageLoss 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": 22, + "review_unit_index": 41, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:36246", + "issue:42886", + "issue:43065", + "issue:43262", + "issue:43352", + "issue:43408", + "issue:43475", + "issue:43540", + "issue:43596", + "issue:43653", + "issue:43931", + "issue:44038", + "issue:44060", + "issue:44077", + "issue:44162", + "issue:44246", + "issue:44295", + "issue:44368", + "issue:44466", + "issue:44479", + "issue:44561", + "issue:44749", + "issue:44877", + "issue:44977", + "issue:44987", + "issue:44995", + "issue:45137", + "issue:45200", + "issue:45229", + "issue:45250", + "issue:45292", + "issue:45335", + "issue:45375", + "issue:45397", + "issue:45412", + "issue:45459", + "issue:45563", + "issue:45584", + "issue:45588", + "issue:45600", + "issue:45656", + "issue:45701", + "issue:45759", + "issue:46123", + "issue:46170", + "issue:46399", + "issue:46506" + ], + "soft_pairs": [ + "issue:44162|issue:45200", + "issue:44295|issue:44749", + "issue:45375|issue:45759", + "issue:43262|issue:44246", + "issue:45292|issue:45584", + "issue:45292|issue:45701", + "issue:44561|issue:46399", + "issue:45137|issue:45600", + "issue:44877|issue:45375", + "issue:36246|issue:43596", + "issue:44038|issue:46506", + "issue:44466|issue:45335", + "issue:44987|issue:45459", + "issue:43352|issue:45250", + "issue:45412|issue:46123", + "issue:44479|issue:46170", + "issue:44060|issue:44077", + "issue:44995|issue:45200", + "issue:43653|issue:44162", + "issue:44368|issue:44977", + "issue:43475|issue:43540", + "issue:42886|issue:45229", + "issue:45397|issue:45656", + "issue:44877|issue:46170", + "issue:44995|issue:45588", + "issue:43065|issue:43408", + "issue:44368|issue:45563", + "issue:43931|issue:44368" + ], + "prepared_review_unit_hash": "c96112fd6aa3e4edf6e4dc35f3f350ee601513ddbe0b8de3e10e8f038e0d9211", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35602, + "estimated_input_tokens": 8901, + "estimated_eval_tokens": 18058 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are mostly distinct bugs across unrelated models/subsystems; none of the soft pairs look like the same underlying issue, so all soft edges are rejected. The Gemma4 multi-GPU CUDA OOM report is the strongest anchor because it has the highest discussion activity and clear user impact.", + "confidence": 0.9, + "canonical_issue_reason": "issue:45229 has the highest discussion activity and inbound reference count in the set, with a concrete, reproducible failure and broad user impact.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45229 is the best overall representative of the cluster because it is the most active, clearly described, and high-impact issue among otherwise unrelated reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44162", + "right": "issue:45200", + "accept": false, + "reason": "ESM2 breakage and Gemma4 mm_token_type_ids defaulting are different models and different failure modes." + }, + { + "left": "issue:44295", + "right": "issue:44749", + "accept": false, + "reason": "One is a buffer/position_ids error; the other is a tokenization slowdown regression after version upgrade." + }, + { + "left": "issue:45375", + "right": "issue:45759", + "accept": false, + "reason": "Missing Qwen3.5MoeVisionConfig field vs AutoModelForCausalLM text_config unwrapping are separate config-path bugs." + }, + { + "left": "issue:43262", + "right": "issue:44246", + "accept": false, + "reason": "Audio chat-template sampling-rate handling is unrelated to intermittent import latency." + }, + { + "left": "issue:45292", + "right": "issue:45584", + "accept": false, + "reason": "resize_token_embeddings mismatch is not the same as Whisper empty-transcription failure after align_special_tokens." + }, + { + "left": "issue:45292", + "right": "issue:45701", + "accept": false, + "reason": "Embedding resize behavior and tokenization changes are different bugs." + }, + { + "left": "issue:44561", + "right": "issue:46399", + "accept": false, + "reason": "Removed helper breaking trust_remote_code is unrelated to Qwen3.5 kernel registration/decorator failure." + }, + { + "left": "issue:45137", + "right": "issue:45600", + "accept": false, + "reason": "DeepSpeed ZeRO3 deque crash and stale auto_mappings config references do not share a concrete code path." + }, + { + "left": "issue:44877", + "right": "issue:45375", + "accept": false, + "reason": "Granite speech strict-config loading and Qwen3.5MoeVisionConfig field loss are distinct config issues." + }, + { + "left": "issue:36246", + "right": "issue:43596", + "accept": false, + "reason": "Missing Qwen2_5_VLImageProcessor import is unrelated to the ZeRO3/BertModel index error." + }, + { + "left": "issue:44038", + "right": "issue:46506", + "accept": false, + "reason": "Qwen3-VL-Moe loading issues and PPDocLayoutV3 CPU/CUDA divergence are different subsystems and symptoms." + }, + { + "left": "issue:44466", + "right": "issue:45335", + "accept": false, + "reason": "Tied-weight serialization inconsistency is not the same bug as resize_token_embeddings not updating decoder.embed_tokens." + }, + { + "left": "issue:44987", + "right": "issue:45459", + "accept": false, + "reason": "Model loading failure with transformers>=5.1.0 is separate from tokenizer error masking when protobuf is absent." + }, + { + "left": "issue:43352", + "right": "issue:45250", + "accept": false, + "reason": "A model-specific Flash Attention 2 unsupported error is not the same as a generic flash-attention topic." + }, + { + "left": "issue:45412", + "right": "issue:46123", + "accept": false, + "reason": "RT-DETR memory retention and MaskGenerationPipeline stale partial-batch handling are unrelated bugs." + }, + { + "left": "issue:44479", + "right": "issue:46170", + "accept": false, + "reason": "Video-input regression in Qwen models is unrelated to the Windows/ROCm torch.distributed import problem." + }, + { + "left": "issue:44060", + "right": "issue:44077", + "accept": false, + "reason": "Incorrect tied-weights warnings and patchtsmixer post_init policy are different issues." + }, + { + "left": "issue:44995", + "right": "issue:45200", + "accept": false, + "reason": "GlmMoeDsa stale cache on second forward pass is unrelated to Gemma4 mm_token_type_ids defaults." + }, + { + "left": "issue:43653", + "right": "issue:44162", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and ESM2 breakage are not the same underlying defect." + }, + { + "left": "issue:44368", + "right": "issue:44977", + "accept": false, + "reason": "A tied-embeddings warning during LoRA fine-tuning is not the same as a Flash Attention generation bug." + }, + { + "left": "issue:43475", + "right": "issue:43540", + "accept": false, + "reason": "SAM3 video encoder output attribute loss and Qwen3OmniMoe video processing errors are separate model paths." + }, + { + "left": "issue:42886", + "right": "issue:45229", + "accept": false, + "reason": "Offline tokenizer cache loading and Gemma4 multi-GPU CUDA OOM are unrelated." + }, + { + "left": "issue:45397", + "right": "issue:45656", + "accept": false, + "reason": "Gemma-4 ZeRO3 from_pretrained failure and duplicate optimizer.step calls are different DeepSpeed bugs." + }, + { + "left": "issue:44877", + "right": "issue:46170", + "accept": false, + "reason": "Granite_speech config strictness has nothing to do with the Windows/ROCm distributed dependency failure." + }, + { + "left": "issue:44995", + "right": "issue:45588", + "accept": false, + "reason": "Stale indexer cache on second forward and flash_attention AttributeError on s_aux=None are different failure modes." + }, + { + "left": "issue:43065", + "right": "issue:43408", + "accept": false, + "reason": "Dummy Conv2d in Sam3PixelDecoder and model-type mismatch warning in sam3_tracker are not the same bug." + }, + { + "left": "issue:44368", + "right": "issue:45563", + "accept": false, + "reason": "Tied-embedding warnings and paged generate stale num_return_sequences warning are unrelated." + }, + { + "left": "issue:43931", + "right": "issue:44368", + "accept": false, + "reason": "Qwen3-VL-30B weight-shape mismatch and tied-embeddings warnings are distinct 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": 22, + "review_unit_index": 42, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:28282", + "issue:30990", + "issue:41628", + "issue:42371", + "issue:42994", + "issue:43065", + "issue:43232", + "issue:43475", + "issue:43482", + "issue:43525", + "issue:43643", + "issue:43646", + "issue:43673", + "issue:43701", + "issue:43906", + "issue:44038", + "issue:44162", + "issue:44188", + "issue:44297", + "issue:44361", + "issue:44483", + "issue:44623", + "issue:44749", + "issue:44869", + "issue:44871", + "issue:44933", + "issue:44964", + "issue:45092", + "issue:45137", + "issue:45230", + "issue:45310", + "issue:45356", + "issue:45375", + "issue:45381", + "issue:45397", + "issue:45478", + "issue:45584", + "issue:45656", + "issue:45685", + "issue:45987", + "issue:46082", + "issue:46123", + "issue:46132", + "issue:46170", + "issue:46202", + "issue:46421", + "issue:46431" + ], + "soft_pairs": [ + "issue:45987|issue:46082", + "issue:43906|issue:46202", + "issue:44188|issue:46202", + "issue:44483|issue:46202", + "issue:45685|issue:46431", + "issue:44188|issue:46421", + "issue:45584|issue:46170", + "issue:43065|issue:46123", + "issue:44297|issue:44623", + "issue:44361|issue:44749", + "issue:44749|issue:45656", + "issue:43646|issue:44162", + "issue:43701|issue:44749", + "issue:44623|issue:45656", + "issue:44871|issue:46132", + "issue:43232|issue:43673", + "issue:42994|issue:43482", + "issue:44869|issue:45356", + "issue:43475|issue:43643", + "issue:42371|issue:45092", + "issue:45375|issue:45478", + "issue:28282|issue:30990", + "issue:41628|issue:44933", + "issue:45230|issue:45397", + "issue:44038|issue:44623", + "issue:45310|issue:45375", + "issue:44964|issue:45137", + "issue:43475|issue:45381", + "issue:42994|issue:43525" + ], + "prepared_review_unit_hash": "e483375eb980969ba647bd02f321661ac0d257500ad5660946e75bb0630f6c4d", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35554, + "estimated_input_tokens": 8889, + "estimated_eval_tokens": 18034 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly a bag of unrelated Transformers issues. Only one clear duplicate pair stands out: the two identical Qwen3.5-Moe from_pretrained regression reports.", + "confidence": 0.86, + "canonical_issue_reason": "issue:45310 is the earliest clear report of the exact Qwen3.5-Moe from_pretrained failure and has an exact duplicate in issue:45478, making it the best canonical issue anchor.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45310 is the strongest representative issue because it names the concrete regression precisely and matches an exact duplicate, while the rest of the cluster is largely unrelated.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45987", + "right": "issue:46082", + "accept": false, + "reason": "Different bugs: cache length return-type mismatch vs LlamaConfig head_dim validation." + }, + { + "left": "issue:43906", + "right": "issue:46202", + "accept": false, + "reason": "No shared code-path or symptom; one is a reproduction of an older issue, the other is a Gemma tokenizer roundtrip failure." + }, + { + "left": "issue:44188", + "right": "issue:46202", + "accept": false, + "reason": "Different failures: torch.compile attention-kernel divergence vs Gemma roundtrip on a special token." + }, + { + "left": "issue:44483", + "right": "issue:46202", + "accept": false, + "reason": "Chat completions API rejection is unrelated to Gemma tokenizer roundtrip behavior." + }, + { + "left": "issue:45685", + "right": "issue:46431", + "accept": false, + "reason": "Unrelated MPS histogram error vs missing chat template in a processor." + }, + { + "left": "issue:44188", + "right": "issue:46421", + "accept": false, + "reason": "Both mention compile/cache, but the concrete bugs differ: attention-kernel divergence vs spurious recompilation from lazy cache init." + }, + { + "left": "issue:45584", + "right": "issue:46170", + "accept": false, + "reason": "Different subsystems and failures: Whisper empty-transcription generation vs Windows ROCm import/torch.distributed dependency issues." + }, + { + "left": "issue:43065", + "right": "issue:46123", + "accept": false, + "reason": "SAM3 pixel-decoder dummy layer issue is unrelated to MaskGenerationPipeline dropping partial-batch results." + }, + { + "left": "issue:44297", + "right": "issue:44623", + "accept": false, + "reason": "Both involve save_pretrained, but one is tokenizer config metadata and the other is processor file emission; not the same bug." + }, + { + "left": "issue:44361", + "right": "issue:44749", + "accept": false, + "reason": "Tokenizer AttributeError on tasks is unrelated to the reported performance regression in data filtering." + }, + { + "left": "issue:44749", + "right": "issue:45656", + "accept": false, + "reason": "Performance slowdown during filtering is not the same as DeepSpeed causing optimizer.step to run twice." + }, + { + "left": "issue:43646", + "right": "issue:44162", + "accept": false, + "reason": "Custom config initialization/field loss is unrelated to ESM2 being broken." + }, + { + "left": "issue:43701", + "right": "issue:44749", + "accept": false, + "reason": "Checkpoint key mismatch is a different failure mode from the filtering slowdown report." + }, + { + "left": "issue:44623", + "right": "issue:45656", + "accept": false, + "reason": "Processor.save_pretrained missing files is unrelated to duplicate optimizer stepping under DeepSpeed." + }, + { + "left": "issue:44871", + "right": "issue:46132", + "accept": false, + "reason": "Gemma eos_token_id mismatch is unrelated to AttentionInterface.register changing registered-function behavior." + }, + { + "left": "issue:43232", + "right": "issue:43673", + "accept": false, + "reason": "Both are generation-related, but the concrete bugs differ: sync_gpus/update_model_kwargs vs chunked_prefill cache missing." + }, + { + "left": "issue:42994", + "right": "issue:43482", + "accept": false, + "reason": "Quantized model saving failure is unrelated to Qwen2.5-GGUF loading under transformers v5." + }, + { + "left": "issue:44869", + "right": "issue:45356", + "accept": false, + "reason": "Whisper timestamp decoding crash is unrelated to Kimi-K2.5 tokenizer codec regression." + }, + { + "left": "issue:43475", + "right": "issue:43643", + "accept": false, + "reason": "SAM3 vision encoder missing attribute is a different bug from trust_remote_code returning incomplete configs." + }, + { + "left": "issue:42371", + "right": "issue:45092", + "accept": false, + "reason": "TF32 API guidance and InternVL2 meta-init incompatibility are unrelated." + }, + { + "left": "issue:45375", + "right": "issue:45478", + "accept": false, + "reason": "Related model family, but one is a specific missing config field and the other is a broader from_pretrained failure; not enough to confirm same underlying bug." + }, + { + "left": "issue:28282", + "right": "issue:30990", + "accept": false, + "reason": "AutoModel missing PyTorch import error is unrelated to Sentence Transformers hanging on load." + }, + { + "left": "issue:41628", + "right": "issue:44933", + "accept": false, + "reason": "Importing AutoImageProcessor and a non-existent import from image_utils are different import-path issues." + }, + { + "left": "issue:45230", + "right": "issue:45397", + "accept": false, + "reason": "One item is too generic to establish identity; the Gemma-4 ZeRO3 from_pretrained bug is specific and distinct." + }, + { + "left": "issue:44038", + "right": "issue:44623", + "accept": false, + "reason": "Qwen3-VL-Moe loading problems are not the same as processor.save_pretrained missing files." + }, + { + "left": "issue:45310", + "right": "issue:45375", + "accept": false, + "reason": "Same model family, but the first is a generic loading error while the second is a specific config-field omission; not clearly the same bug." + }, + { + "left": "issue:44964", + "right": "issue:45137", + "accept": false, + "reason": "Phi-4 multimodal loading failure and DeepSpeed ZeRO3 deque pop error are unrelated." + }, + { + "left": "issue:43475", + "right": "issue:45381", + "accept": false, + "reason": "Different models and symptoms: SAM3 vision encoder attribute error vs Qwen2.5-VL video position-id mismatch." + }, + { + "left": "issue:42994", + "right": "issue:43525", + "accept": false, + "reason": "Quantized model save failure and Llama4Config pad_token_id access are unrelated config/model 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": 22, + "review_unit_index": 43, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:30333", + "issue:30990", + "issue:33290", + "issue:35141", + "issue:36010", + "issue:41628", + "issue:42175", + "issue:42673", + "issue:42831", + "issue:42898", + "issue:42994", + "issue:43232", + "issue:43278", + "issue:43377", + "issue:43504", + "issue:43704", + "issue:43723", + "issue:43784", + "issue:43824", + "issue:43873", + "issue:43976", + "issue:44038", + "issue:44188", + "issue:44222", + "issue:44242", + "issue:44351", + "issue:44496", + "issue:44568", + "issue:44623", + "issue:44792", + "issue:44933", + "issue:44945", + "issue:45081", + "issue:45103", + "issue:45137", + "issue:45310", + "issue:45406", + "issue:45561", + "issue:45600", + "issue:45712", + "issue:45753", + "issue:46077", + "issue:46123", + "issue:46202", + "issue:46326", + "issue:46399", + "issue:46693", + "issue:46701" + ], + "soft_pairs": [ + "issue:43723|issue:46326", + "issue:43784|issue:46399", + "issue:43232|issue:44188", + "issue:45406|issue:46202", + "issue:43873|issue:44945", + "issue:44945|issue:45753", + "issue:46202|issue:46701", + "issue:41628|issue:44351", + "issue:45081|issue:45103", + "issue:42673|issue:43704", + "issue:42994|issue:43824", + "issue:44038|issue:45310", + "issue:44222|issue:45310", + "issue:36010|issue:41628", + "issue:43278|issue:45137", + "issue:42831|issue:44222", + "issue:43232|issue:44792", + "issue:43504|issue:45561", + "issue:42175|issue:44933", + "issue:44496|issue:45561", + "issue:45561|issue:45600", + "issue:30333|issue:33290", + "issue:43976|issue:44933", + "issue:42175|issue:45712", + "issue:46077|issue:46123", + "issue:30990|issue:35141", + "issue:42898|issue:44568", + "issue:43377|issue:44242", + "issue:44623|issue:45310", + "issue:42673|issue:46693" + ], + "prepared_review_unit_hash": "182b2fe847c41573675544740f26dd3261710f4ce0b4d1fed199fd7dda4371e4", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36559, + "estimated_input_tokens": 9140, + "estimated_eval_tokens": 18536 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a set of unrelated bug reports that happen to share broad themes like import/load regressions and generation issues. Only one pair is a clear duplicate: the Qwen3ForCausalLM VRAM leak report.", + "confidence": 0.89, + "canonical_issue_reason": "Issue 42673 is the clearest duplicate anchor: it is the older, more detailed report of the same Qwen3ForCausalLM VRAM leak in multi-dataloader-thread usage, and it matches issue 43704 exactly by title and problem description.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 42673 is the best issue representative for the only accepted duplicate pair because it is the earlier and fuller report, with the strongest discussion context for the same concrete bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43723", + "right": "issue:46326", + "accept": false, + "reason": "Both are tokenizer-related, but one is a tokenizer loading/import issue and the other is a continue_final_message behavior bug for mistral_common tokenizers; different code paths." + }, + { + "left": "issue:43784", + "right": "issue:46399", + "accept": false, + "reason": "Unrelated failures: one is a sentence-transformers import NameError, the other is a Qwen3.5 kernel registration error." + }, + { + "left": "issue:43232", + "right": "issue:44188", + "accept": false, + "reason": "Different generation bugs: sync_gpus/_update_model_kwargs_for_generation versus attention-kernel divergence under torch.compile." + }, + { + "left": "issue:45406", + "right": "issue:46202", + "accept": false, + "reason": "Different subsystems and symptoms: Gemma4Processor serve crash versus Gemma pipeline roundtrip/tokenization issue." + }, + { + "left": "issue:43873", + "right": "issue:44945", + "accept": false, + "reason": "Offloading/quantization behavior is not the same as incorrect outputs under pipeline parallelism." + }, + { + "left": "issue:44945", + "right": "issue:45753", + "accept": false, + "reason": "Pipeline-parallel output corruption is distinct from a Qwen3_5 infinite-loop-on-image issue." + }, + { + "left": "issue:46202", + "right": "issue:46701", + "accept": false, + "reason": "Both involve Gemma, but one is a tokenization roundtrip bug and the other is an image-text-to-text pipeline failure; not the same defect." + }, + { + "left": "issue:41628", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they concern different symbols and likely different missing exports." + }, + { + "left": "issue:45081", + "right": "issue:45103", + "accept": false, + "reason": "Completely different bugs: Mistral tokenizer regex patching versus auto-docstring processing with future annotations." + }, + { + "left": "issue:42673", + "right": "issue:43704", + "accept": true, + "reason": "Exact duplicate: same Qwen3ForCausalLM VRAM leak when used across multiple dataloader threads, with matching title and failure mode." + }, + { + "left": "issue:42994", + "right": "issue:43824", + "accept": false, + "reason": "Quantized model saving failure and an import error for Qwen2_5_VLForConditionalGeneration are unrelated." + }, + { + "left": "issue:44038", + "right": "issue:45310", + "accept": false, + "reason": "Related model family, but different versions and different reported failures; not enough evidence they are the same concrete bug." + }, + { + "left": "issue:44222", + "right": "issue:45310", + "accept": false, + "reason": "FP8 save_pretrained failure is a save-path bug, while the Qwen3.5 Moe report is a from_pretrained/load-path error." + }, + { + "left": "issue:36010", + "right": "issue:41628", + "accept": false, + "reason": "Both are import failures, but they affect different APIs and likely different export points." + }, + { + "left": "issue:43278", + "right": "issue:45137", + "accept": false, + "reason": "Embedding dtype drift in evaluate is unrelated to the DeepSpeed ZeRO3 deque IndexError." + }, + { + "left": "issue:42831", + "right": "issue:44222", + "accept": false, + "reason": "One is an accuracy regression in FineGrainedFP8; the other is a model saving bug with FP8/MoE." + }, + { + "left": "issue:43232", + "right": "issue:44792", + "accept": false, + "reason": "Generation-kwargs synchronization bug and janus image generation test failure are different issues." + }, + { + "left": "issue:43504", + "right": "issue:45561", + "accept": false, + "reason": "Model-loading preset failure and a pytest-xdist race in testing utilities are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:44933", + "accept": false, + "reason": "Backend dependency packaging issue versus a missing import from image_utils; different problems." + }, + { + "left": "issue:44496", + "right": "issue:45561", + "accept": false, + "reason": "Model config loading failure and a test-utils file race are unrelated." + }, + { + "left": "issue:45561", + "right": "issue:45600", + "accept": false, + "reason": "One is an xdist race on captured_info.txt; the other is stale auto_mappings references breaking CI." + }, + { + "left": "issue:30333", + "right": "issue:33290", + "accept": false, + "reason": "Training-job MLFlow reporting failure and DeepSpeed/Adafactor OOM are different failure modes." + }, + { + "left": "issue:43976", + "right": "issue:44933", + "accept": false, + "reason": "Python version compatibility problem versus an import-from-image_utils issue; no common underlying bug." + }, + { + "left": "issue:42175", + "right": "issue:45712", + "accept": false, + "reason": "Different packaging/test concerns: missing TensorFlow backend in extra install versus leftover dummy classes in dummy_pt_objects." + }, + { + "left": "issue:46077", + "right": "issue:46123", + "accept": false, + "reason": "Different behavior bugs: bidirectional attention masking versus pipeline final-batch result dropping." + }, + { + "left": "issue:30990", + "right": "issue:35141", + "accept": false, + "reason": "Sentence-transformers loading hang is unrelated to embedding reinitialization after resize_token_embeddings." + }, + { + "left": "issue:42898", + "right": "issue:44568", + "accept": false, + "reason": "Both are tokenizer regressions in v5, but one is cleanup-space behavior and the other is BOS/EOS insertion; different code paths." + }, + { + "left": "issue:43377", + "right": "issue:44242", + "accept": false, + "reason": "MIMI padding-mask batching bug and missing load-balancing loss when router logits are disabled are unrelated." + }, + { + "left": "issue:44623", + "right": "issue:45310", + "accept": false, + "reason": "processor.save_pretrained missing files is a serialization bug, not the same as Qwen3.5 Moe from_pretrained failure." + }, + { + "left": "issue:42673", + "right": "issue:46693", + "accept": false, + "reason": "Both mention generation performance, but VRAM leakage in threads and FA2 long-context regression are different bugs." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded overall. The accepted duplicate pair is well supported by the packet: the titles match exactly, and 42673 is earlier and has more discussion/body content than 43704. The non-duplicate verdicts are conservative based on the provided titles." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 44, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:22355", + "issue:27957", + "issue:28282", + "issue:30333", + "issue:34689", + "issue:35141", + "issue:36331", + "issue:42222", + "issue:42796", + "issue:42831", + "issue:42898", + "issue:43116", + "issue:43295", + "issue:43479", + "issue:43644", + "issue:43723", + "issue:43784", + "issue:43824", + "issue:43874", + "issue:44016", + "issue:44060", + "issue:44075", + "issue:44079", + "issue:44190", + "issue:44222", + "issue:44295", + "issue:44297", + "issue:44448", + "issue:44568", + "issue:44623", + "issue:44734", + "issue:44779", + "issue:44869", + "issue:44933", + "issue:45092", + "issue:45103", + "issue:45125", + "issue:45200", + "issue:45245", + "issue:45356", + "issue:45405", + "issue:45656", + "issue:45712", + "issue:45812", + "issue:46132", + "issue:46231", + "issue:46665" + ], + "soft_pairs": [ + "issue:43479|issue:44869", + "issue:42222|issue:45200", + "issue:44016|issue:46231", + "issue:42898|issue:45812", + "issue:43723|issue:44568", + "issue:43874|issue:45092", + "issue:43479|issue:44079", + "issue:42898|issue:44779", + "issue:44734|issue:45103", + "issue:43784|issue:43824", + "issue:42898|issue:45356", + "issue:27957|issue:30333", + "issue:45405|issue:45712", + "issue:42898|issue:43295", + "issue:30333|issue:34689", + "issue:44060|issue:45125", + "issue:44295|issue:45656", + "issue:43116|issue:44190", + "issue:44933|issue:45712", + "issue:30333|issue:36331", + "issue:44222|issue:44297", + "issue:44075|issue:46231", + "issue:30333|issue:35141", + "issue:22355|issue:28282", + "issue:44297|issue:45656", + "issue:42831|issue:44623", + "issue:44448|issue:46132", + "issue:42796|issue:43723", + "issue:44222|issue:44623", + "issue:43479|issue:43644", + "issue:45245|issue:46132", + "issue:43479|issue:46665" + ], + "prepared_review_unit_hash": "a137d930d0e12b10aec45bf06331ad151d2ae5e8835cae547a6962ea58ad21e1", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 32, + "serialized_chars": 36886, + "estimated_input_tokens": 9222, + "estimated_eval_tokens": 18700 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: it mixes unrelated tokenizer, model-loading, training, config, docs, and utility bugs. None of the soft pairs looks like the same underlying issue, so no duplicate merge is justified.", + "confidence": 0.92, + "canonical_issue_reason": "No single canonical issue stands out because the items do not share one concrete bug or code path; they are mostly separate regressions that only share broad Transformers terminology.", + "canonical_pr_reason": "No PRs are present in this cluster.", + "best_issue_reason": "No global best issue: the set is too mixed to choose one representative fix target without collapsing distinct bugs into one bucket.", + "best_pr_reason": "No PRs are present in this cluster.", + "soft_edge_verdicts": [ + { + "left": "issue:43479", + "right": "issue:44869", + "accept": false, + "reason": "Different multimodal config init bug vs Whisper decode crash; unrelated code paths." + }, + { + "left": "issue:42222", + "right": "issue:45200", + "accept": false, + "reason": "VitPose model breakage vs Gemma 4 token-type defaults; different failures." + }, + { + "left": "issue:44016", + "right": "issue:46231", + "accept": false, + "reason": "Notebook syntax error vs docs parameter-name mismatch; not the same bug." + }, + { + "left": "issue:42898", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions, but different models and symptoms." + }, + { + "left": "issue:43723", + "right": "issue:44568", + "accept": false, + "reason": "Loading failure vs missing BOS/EOS insertion; distinct tokenizer issues." + }, + { + "left": "issue:43874", + "right": "issue:45092", + "accept": false, + "reason": "Missing image-processor method vs remote-code/meta-init incompatibility; unrelated." + }, + { + "left": "issue:43479", + "right": "issue:44079", + "accept": false, + "reason": "Config defaulting bug vs ModelOutput key assignment bug; different components." + }, + { + "left": "issue:42898", + "right": "issue:44779", + "accept": false, + "reason": "Tokenizer regression for different models; no shared concrete fix path." + }, + { + "left": "issue:44734", + "right": "issue:45103", + "accept": false, + "reason": "Serve/KV-cache indexing crash vs auto-docstring annotations crash; unrelated." + }, + { + "left": "issue:43784", + "right": "issue:43824", + "accept": false, + "reason": "Different import errors with different missing symbols/dependencies." + }, + { + "left": "issue:42898", + "right": "issue:45356", + "accept": false, + "reason": "Separate tokenizer regressions for different models/code paths." + }, + { + "left": "issue:27957", + "right": "issue:30333", + "accept": false, + "reason": "Flash Attention model support vs MLflow reporting on failed training; unrelated." + }, + { + "left": "issue:45405", + "right": "issue:45712", + "accept": false, + "reason": "Dependency-version bump vs leftover dummy classes/check-repo cleanup; different issues." + }, + { + "left": "issue:42898", + "right": "issue:43295", + "accept": false, + "reason": "Both involve v5 behavior changes, but one is token IDs and the other is processor/tokenizer API access." + }, + { + "left": "issue:30333", + "right": "issue:34689", + "accept": false, + "reason": "Training-job reporting bug vs model-loading regression; unrelated." + }, + { + "left": "issue:44060", + "right": "issue:45125", + "accept": false, + "reason": "Wrong tied-weights warning vs missing tensor-parallel plan; different Qwen3 bugs." + }, + { + "left": "issue:44295", + "right": "issue:45656", + "accept": false, + "reason": "Buffer read error vs double optimizer step with deepspeed; unrelated." + }, + { + "left": "issue:43116", + "right": "issue:44190", + "accept": false, + "reason": "Different example-script failures in different pipelines." + }, + { + "left": "issue:44933", + "right": "issue:45712", + "accept": false, + "reason": "Missing import from image_utils vs leftover dummy classes; not the same defect." + }, + { + "left": "issue:30333", + "right": "issue:36331", + "accept": false, + "reason": "MLflow tracking on failure vs CustomTrainer signature mismatch; distinct training bugs." + }, + { + "left": "issue:44222", + "right": "issue:44297", + "accept": false, + "reason": "FP8 MoE save_pretrained issue vs tokenizer metadata mismatch; different save paths." + }, + { + "left": "issue:44075", + "right": "issue:46231", + "accept": false, + "reason": "Optimizer argument usage bug vs documentation name mismatch; unrelated." + }, + { + "left": "issue:30333", + "right": "issue:35141", + "accept": false, + "reason": "Failed-job reporting vs embedding resize/post_init reinit bug; different code paths." + }, + { + "left": "issue:22355", + "right": "issue:28282", + "accept": false, + "reason": "Different missing-import/dependency errors; no shared root cause." + }, + { + "left": "issue:44297", + "right": "issue:45656", + "accept": false, + "reason": "Tokenizer save metadata bug vs optimizer double-step bug; unrelated." + }, + { + "left": "issue:42831", + "right": "issue:44623", + "accept": false, + "reason": "FP8 accuracy regression vs processor save_pretrained missing files; different issues." + }, + { + "left": "issue:44448", + "right": "issue:46132", + "accept": false, + "reason": "Model output difference vs attention-registration behavior change; unrelated." + }, + { + "left": "issue:42796", + "right": "issue:43723", + "accept": false, + "reason": "Tokenizer encoding bug vs tokenizer-loading bug; distinct symptoms and fixes." + }, + { + "left": "issue:44222", + "right": "issue:44623", + "accept": false, + "reason": "Model save bug vs processor save bug; separate artifacts and code paths." + }, + { + "left": "issue:43479", + "right": "issue:43644", + "accept": false, + "reason": "Config None-handling bug vs non-persistent-buffer initialization bug; unrelated." + }, + { + "left": "issue:45245", + "right": "issue:46132", + "accept": false, + "reason": "Category-limit runtime error vs attention-registration behavior change; unrelated." + }, + { + "left": "issue:43479", + "right": "issue:46665", + "accept": false, + "reason": "Specific multimodal config default bug vs generic from_pretrained-in-classes inconsistency; 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": 22, + "review_unit_index": 45, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:30990", + "issue:38175", + "issue:41628", + "issue:42757", + "issue:42831", + "issue:42907", + "issue:43122", + "issue:43232", + "issue:43475", + "issue:43644", + "issue:43723", + "issue:44190", + "issue:44206", + "issue:44222", + "issue:44261", + "issue:44263", + "issue:44279", + "issue:44295", + "issue:44297", + "issue:44360", + "issue:44373", + "issue:44484", + "issue:44485", + "issue:44623", + "issue:44734", + "issue:44829", + "issue:44871", + "issue:44977", + "issue:45071", + "issue:45092", + "issue:45103", + "issue:45310", + "issue:45357", + "issue:45381", + "issue:45834", + "issue:46002", + "issue:46050", + "issue:46082", + "issue:46132", + "issue:46209", + "issue:46305", + "issue:46328", + "issue:46421", + "issue:46424", + "issue:46431", + "issue:46655", + "issue:46693" + ], + "soft_pairs": [ + "issue:41628|issue:42757", + "issue:45071|issue:46082", + "issue:44297|issue:45310", + "issue:44373|issue:44485", + "issue:44263|issue:45092", + "issue:43723|issue:44977", + "issue:46132|issue:46693", + "issue:44222|issue:44484", + "issue:44206|issue:45103", + "issue:44190|issue:46082", + "issue:43232|issue:46424", + "issue:43122|issue:46209", + "issue:43232|issue:45834", + "issue:44734|issue:46421", + "issue:46209|issue:46431", + "issue:46305|issue:46655", + "issue:43475|issue:46305", + "issue:43723|issue:44871", + "issue:44263|issue:44360", + "issue:44295|issue:44623", + "issue:44261|issue:45092", + "issue:45381|issue:46305", + "issue:45092|issue:45357", + "issue:30990|issue:38175", + "issue:44263|issue:44373", + "issue:43644|issue:46305", + "issue:42907|issue:46050", + "issue:42831|issue:45310", + "issue:44829|issue:46002", + "issue:44279|issue:46328" + ], + "prepared_review_unit_hash": "e2bf1dfc3418d0dada1bfe11271efaaa3ae77ce1486132bf3676757717991fce", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 35828, + "estimated_input_tokens": 8957, + "estimated_eval_tokens": 18170 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Loose similarity cluster of mostly Transformers v5 regression reports across loading, config validation, generation, quantization, multimodal, and cache paths; no true duplicate hub emerges.", + "confidence": 0.93, + "canonical_issue_reason": "No single canonical duplicate: the items span unrelated bugs across tokenization, config validation, model saving/loading, generation, and performance.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45071 is the broadest version-regression report and the closest thing to a central infrastructure bug, but it still does not subsume the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41628", + "right": "issue:42757", + "accept": false, + "reason": "Different missing imports from different modules: AutoImageProcessor vs is_offline_mode." + }, + { + "left": "issue:45071", + "right": "issue:46082", + "accept": false, + "reason": "Related to config handling, but one is a type-checking regression and the other is a head_dim validation error on LlamaConfig." + }, + { + "left": "issue:44297", + "right": "issue:45310", + "accept": false, + "reason": "Tokenizer save metadata mismatch vs MoE from_pretrained failure; different code paths and symptoms." + }, + { + "left": "issue:44373", + "right": "issue:44485", + "accept": false, + "reason": "Docstring issue for position_ids versus a GLM-5 RoPE implementation discussion." + }, + { + "left": "issue:44263", + "right": "issue:45092", + "accept": false, + "reason": "torch.split return handling in a DSA indexer is unrelated to InternVL2 meta-initialization incompatibility." + }, + { + "left": "issue:43723", + "right": "issue:44977", + "accept": false, + "reason": "Tokenizer loading failure and flash-attention generation failure are distinct bugs." + }, + { + "left": "issue:46132", + "right": "issue:46693", + "accept": false, + "reason": "API behavior change in AttentionInterface.register versus a long-context FA2 performance regression." + }, + { + "left": "issue:44222", + "right": "issue:44484", + "accept": false, + "reason": "Saving a quantized model is not the same as asking why save_pretrained defaults to 50GB shard size." + }, + { + "left": "issue:44206", + "right": "issue:45103", + "accept": false, + "reason": "Feature extractor center-arg crash and auto_docstring future-annotations crash are unrelated." + }, + { + "left": "issue:44190", + "right": "issue:46082", + "accept": false, + "reason": "Local dataset loading in a training script is unrelated to LlamaConfig head_dim validation." + }, + { + "left": "issue:43232", + "right": "issue:46424", + "accept": false, + "reason": "Generation after sync_gpus and cache max_cache_length semantics are different cache-related issues." + }, + { + "left": "issue:43122", + "right": "issue:46209", + "accept": false, + "reason": "Tokenizer version-difference bug versus FP8 inference failure on B200; different subsystems." + }, + { + "left": "issue:43232", + "right": "issue:45834", + "accept": false, + "reason": "Generation kwargs update after sync_gpus does not match a Kosmos2.5 OCR index error." + }, + { + "left": "issue:44734", + "right": "issue:46421", + "accept": false, + "reason": "KV-cache continuation tensor indexing and cache lazy initialization recompilation are not the same defect." + }, + { + "left": "issue:46209", + "right": "issue:46431", + "accept": false, + "reason": "FP8 hardware inference failure is unrelated to missing chat-template support in a processor." + }, + { + "left": "issue:46305", + "right": "issue:46655", + "accept": false, + "reason": "Different sliding-window/padding bugs in different models and processors." + }, + { + "left": "issue:43475", + "right": "issue:46305", + "accept": false, + "reason": "Missing SAM3 output attribute is unrelated to a Moonshine streaming mask off-by-one." + }, + { + "left": "issue:43723", + "right": "issue:44871", + "accept": false, + "reason": "Both involve loading/config, but tokenizer import failure and eos_token_id mismatch are different bugs." + }, + { + "left": "issue:44263", + "right": "issue:44360", + "accept": false, + "reason": "Different DSA indexer bugs: torch.split outputs versus missing ReLU." + }, + { + "left": "issue:44295", + "right": "issue:44623", + "accept": false, + "reason": "Position_ids buffer read error and processor.save_pretrained missing files are separate save/load issues." + }, + { + "left": "issue:44261", + "right": "issue:45092", + "accept": false, + "reason": "MLA layernorm epsilon precision issue is unrelated to old InternVL2 meta-init incompatibility." + }, + { + "left": "issue:45381", + "right": "issue:46305", + "accept": false, + "reason": "Qwen2.5-VL vision_position_ids bug and Moonshine mask off-by-one are different model-specific logic errors." + }, + { + "left": "issue:45092", + "right": "issue:45357", + "accept": false, + "reason": "Old remote-code checkpoint meta-init incompatibility is not the same as incorrect visual encoder keys on save_pretrained." + }, + { + "left": "issue:30990", + "right": "issue:38175", + "accept": false, + "reason": "Sentence Transformers loading hang and SIGLIP2 zero probabilities are unrelated." + }, + { + "left": "issue:44263", + "right": "issue:44373", + "accept": false, + "reason": "Code bug in DSA indexer versus a wrong docstring for position_ids." + }, + { + "left": "issue:43644", + "right": "issue:46305", + "accept": false, + "reason": "Non-persistent buffer junk in v5.0 is unrelated to a streaming sliding-window mask error." + }, + { + "left": "issue:42907", + "right": "issue:46050", + "accept": false, + "reason": "Both are save/load quantization issues, but they concern different quantization methods and different failure modes." + }, + { + "left": "issue:42831", + "right": "issue:45310", + "accept": false, + "reason": "FineGrainedFP8 accuracy regression is not the same as a Qwen3.5 from_pretrained load error." + }, + { + "left": "issue:44829", + "right": "issue:46002", + "accept": false, + "reason": "Degenerate training with FA3 and NaNs at checkout/init are separate training stability bugs." + }, + { + "left": "issue:44279", + "right": "issue:46328", + "accept": false, + "reason": "A dependency issue is unrelated to a SequenceFeatureExtractor.pad performance problem." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 46, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:22355", + "issue:27957", + "issue:28218", + "issue:30990", + "issue:33357", + "issue:39401", + "issue:41628", + "issue:42371", + "issue:42757", + "issue:42796", + "issue:42831", + "issue:42886", + "issue:42907", + "issue:42994", + "issue:43066", + "issue:43122", + "issue:43232", + "issue:43408", + "issue:43576", + "issue:43646", + "issue:43756", + "issue:43784", + "issue:43824", + "issue:43937", + "issue:43976", + "issue:44038", + "issue:44060", + "issue:44117", + "issue:44188", + "issue:44261", + "issue:44263", + "issue:44295", + "issue:44297", + "issue:44373", + "issue:44483", + "issue:44488", + "issue:44568", + "issue:44704", + "issue:44945", + "issue:44964", + "issue:45278", + "issue:45406", + "issue:45593", + "issue:45600", + "issue:45753", + "issue:46002", + "issue:46369", + "issue:46402" + ], + "soft_pairs": [ + "issue:44483|issue:45406", + "issue:22355|issue:27957", + "issue:42831|issue:44038", + "issue:41628|issue:43824", + "issue:43408|issue:46402", + "issue:42831|issue:45753", + "issue:22355|issue:28218", + "issue:39401|issue:44568", + "issue:42757|issue:45278", + "issue:42886|issue:43066", + "issue:27957|issue:30990", + "issue:42886|issue:42994", + "issue:43232|issue:45593", + "issue:44188|issue:44373", + "issue:43408|issue:46369", + "issue:44261|issue:44373", + "issue:42796|issue:43122", + "issue:42371|issue:44263", + "issue:43646|issue:46002", + "issue:42907|issue:44945", + "issue:43756|issue:44060", + "issue:30990|issue:33357", + "issue:43576|issue:43784", + "issue:44295|issue:44297", + "issue:43976|issue:44945", + "issue:39401|issue:44488", + "issue:44117|issue:44704", + "issue:42907|issue:44964", + "issue:41628|issue:45278", + "issue:43784|issue:45600", + "issue:42907|issue:43937" + ], + "prepared_review_unit_hash": "cbfde06f5ebc747af8d82e2f41f1be5205b77b4518ac076440621981b97d7add", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36388, + "estimated_input_tokens": 9097, + "estimated_eval_tokens": 18450 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These pairs mostly overlap only at a broad theme level (tokenizers, loading, version regressions, or import errors) but they describe different concrete bugs/models/code paths, so I would not merge any of them as duplicates.", + "confidence": 0.56, + "canonical_issue_reason": "issue:45278 is the broadest and most central representative of the version-regression/import-failure subset in this set; it is the best umbrella issue to anchor related reports, even though it is not a safe duplicate match for the specific pairs here.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 has the clearest cluster-level relevance because it captures the broad upgrade-regression/import breakage pattern that several other issues loosely resemble.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44483", + "right": "issue:45406", + "accept": false, + "reason": "Different failures: REST chat-completions request rejection vs Gemma4Processor attribute error in transformers serve." + }, + { + "left": "issue:22355", + "right": "issue:27957", + "accept": false, + "reason": "Unrelated problems: missing transformers.onnx module vs XLMRoberta Flash Attention 2 support." + }, + { + "left": "issue:42831", + "right": "issue:44038", + "accept": false, + "reason": "Both are model-specific, but one is FineGrainedFP8 accuracy and the other is a Qwen3-VL-Moe bug; not the same code path." + }, + { + "left": "issue:41628", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but for different classes/models: AutoImageProcessor vs Qwen2.5-VL generation class." + }, + { + "left": "issue:43408", + "right": "issue:46402", + "accept": false, + "reason": "Different Sam3 symptoms: model-type warning vs missing text-encoder weights loading." + }, + { + "left": "issue:42831", + "right": "issue:45753", + "accept": false, + "reason": "Different model bugs: FP8 accuracy regression vs Qwen3_5 infinite loop on a specific image." + }, + { + "left": "issue:22355", + "right": "issue:28218", + "accept": false, + "reason": "No shared underlying bug: missing transformers.onnx module vs tokenizer adding extra spaces after added tokens." + }, + { + "left": "issue:39401", + "right": "issue:44568", + "accept": false, + "reason": "Both tokenizer-related, but one is wrong offset_mapping and the other is add_special_tokens not adding BOS/EOS." + }, + { + "left": "issue:42757", + "right": "issue:45278", + "accept": false, + "reason": "The first is a specific is_offline_mode import failure; the second is a broad set of import errors after upgrade." + }, + { + "left": "issue:42886", + "right": "issue:43066", + "accept": false, + "reason": "Offline cache loading failure and wrong tokenizer decoder type are different tokenizer issues." + }, + { + "left": "issue:27957", + "right": "issue:30990", + "accept": false, + "reason": "Flash Attention 2 support for XLMRoberta is unrelated to Sentence Transformers getting stuck loading." + }, + { + "left": "issue:42886", + "right": "issue:42994", + "accept": false, + "reason": "One is offline tokenizer cache loading; the other is quantized model saving failure." + }, + { + "left": "issue:43232", + "right": "issue:45593", + "accept": false, + "reason": "Different generation/training concerns: sync_gpus kwargs update vs auxiliary losses omitted when denoising is off." + }, + { + "left": "issue:44188", + "right": "issue:44373", + "accept": false, + "reason": "Runtime attention-kernel divergence is unrelated to a docstring issue for position_ids." + }, + { + "left": "issue:43408", + "right": "issue:46369", + "accept": false, + "reason": "Both mention Sam3, but one is a model-type warning and the other is memory overflow in sam3-litetext." + }, + { + "left": "issue:44261", + "right": "issue:44373", + "accept": false, + "reason": "Precision bug in MLA q_a_layernorm is unrelated to a wrong docstring for position_ids." + }, + { + "left": "issue:42796", + "right": "issue:43122", + "accept": false, + "reason": "Both concern tokenization, but one is BPE marker handling for a specific model and the other is cross-version tokenization drift." + }, + { + "left": "issue:42371", + "right": "issue:44263", + "accept": false, + "reason": "TF32 configuration guidance is unrelated to torch.split return values in GlmMoeDsaIndexer." + }, + { + "left": "issue:43646", + "right": "issue:46002", + "accept": false, + "reason": "Custom model initialization breakage and NaNs in classification heads are different failure modes." + }, + { + "left": "issue:42907", + "right": "issue:44945", + "accept": false, + "reason": "Quantized/dequantized model saving failure is unrelated to incorrect LLM output under pipeline parallelism." + }, + { + "left": "issue:43756", + "right": "issue:44060", + "accept": false, + "reason": "Smollm3 RoPE-layer mismatch and Qwen3-Next tied-weights warning are different model-architecture bugs." + }, + { + "left": "issue:30990", + "right": "issue:33357", + "accept": false, + "reason": "Sentence-Transformers load hang and MacOS bus error with a CLIP model are not the same issue." + }, + { + "left": "issue:43576", + "right": "issue:43784", + "accept": false, + "reason": "Broken transformers env command and a NameError importing sentence-transformers are unrelated." + }, + { + "left": "issue:44295", + "right": "issue:44297", + "accept": false, + "reason": "Reading registered position_ids as a buffer and tokenizer_class mismatch on save_pretrained are different bugs." + }, + { + "left": "issue:43976", + "right": "issue:44945", + "accept": false, + "reason": "Python version support regression is unrelated to incorrect LLM output from pipeline parallelism." + }, + { + "left": "issue:39401", + "right": "issue:44488", + "accept": false, + "reason": "Wrong offset_mapping in Qwen3 is unrelated to loading cjvt/sleng-bert." + }, + { + "left": "issue:44117", + "right": "issue:44704", + "accept": false, + "reason": "A None handling bug in TOKENIZER_MAPPING_NAMES is different from AutoProcessor not forwarding kwargs to cached_file." + }, + { + "left": "issue:42907", + "right": "issue:44964", + "accept": false, + "reason": "Saving dequantized Ministrals/Devstrals and loading Phi-4 multimodal instruct are different model-loading/saving problems." + }, + { + "left": "issue:41628", + "right": "issue:45278", + "accept": false, + "reason": "Both are import-related, but this is a specific AutoImageProcessor import error versus a broad many-import-errors regression." + }, + { + "left": "issue:43784", + "right": "issue:45600", + "accept": false, + "reason": "Importing sentence-transformers fails with NameError, while the Sam3LiteText issue is broken auto_mappings references." + }, + { + "left": "issue:42907", + "right": "issue:43937", + "accept": false, + "reason": "Quantized model saving failure is unrelated to GLM-5 invalid GenerationConfig." + } + ] + }, + "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": 22, + "review_unit_index": 47, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:22355", + "issue:29127", + "issue:29159", + "issue:36296", + "issue:38617", + "issue:39401", + "issue:42175", + "issue:42673", + "issue:42757", + "issue:42898", + "issue:42907", + "issue:43066", + "issue:43116", + "issue:43122", + "issue:43278", + "issue:43408", + "issue:43452", + "issue:43475", + "issue:43493", + "issue:43611", + "issue:43824", + "issue:43937", + "issue:44038", + "issue:44188", + "issue:44222", + "issue:44295", + "issue:44297", + "issue:44351", + "issue:44360", + "issue:44485", + "issue:44521", + "issue:44625", + "issue:44704", + "issue:44779", + "issue:44829", + "issue:44938", + "issue:44945", + "issue:45072", + "issue:45200", + "issue:45216", + "issue:45278", + "issue:45290", + "issue:45478", + "issue:45753", + "issue:46002", + "issue:46326", + "issue:46328", + "issue:46614" + ], + "soft_pairs": [ + "issue:44360|issue:44485", + "issue:44704|issue:45290", + "issue:44625|issue:44829", + "issue:39401|issue:44521", + "issue:43122|issue:46326", + "issue:43122|issue:44521", + "issue:43408|issue:43475", + "issue:44038|issue:44297", + "issue:43116|issue:45200", + "issue:39401|issue:46614", + "issue:42907|issue:43452", + "issue:44038|issue:44295", + "issue:44038|issue:44222", + "issue:22355|issue:29159", + "issue:45216|issue:45478", + "issue:43824|issue:44351", + "issue:43122|issue:44779", + "issue:36296|issue:43116", + "issue:43278|issue:44945", + "issue:42757|issue:44351", + "issue:42175|issue:44938", + "issue:22355|issue:29127", + "issue:43493|issue:45753", + "issue:44188|issue:46328", + "issue:43937|issue:44485", + "issue:42898|issue:46614", + "issue:38617|issue:45278", + "issue:43066|issue:44779", + "issue:42673|issue:45072", + "issue:43611|issue:46002" + ], + "prepared_review_unit_hash": "910995c7ac3af1b97a9b88e77c87a3334e262c92f08449f7f9093b1503513bda", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36360, + "estimated_input_tokens": 9090, + "estimated_eval_tokens": 18436 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose set of Transformers issues with a few tokenizer/import-regression themes, but the soft pairs are still too heterogeneous to treat as duplicates. I would not merge any of the proposed soft edges.", + "confidence": 0.34, + "canonical_issue_reason": "If a representative issue is needed, #43122 is the broadest tokenizer-regression report (v4.57.3 to v5.0) and best matches the general theme of the cluster, but it is not a true duplicate anchor for most other items.", + "canonical_pr_reason": null, + "best_issue_reason": "#43122 is the most umbrella-like issue in the set: it describes a general tokenizer behavior regression across versions, which is more central than the narrower model-specific reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44360", + "right": "issue:44485", + "accept": false, + "reason": "Different bugs: a DSA indexer discussion versus GLM-5 RoPE implementation; same broad area, not the same defect." + }, + { + "left": "issue:44704", + "right": "issue:45290", + "accept": false, + "reason": "Different code paths: cached_file kwargs forwarding versus chat template crashes with tool-call assistant messages." + }, + { + "left": "issue:44625", + "right": "issue:44829", + "accept": false, + "reason": "Unrelated failures: num_labels propagation in Qwen3.5 config versus degenerate training with flash_attention_3." + }, + { + "left": "issue:39401", + "right": "issue:44521", + "accept": false, + "reason": "Both involve tokenization-related behavior, but one is wrong offset mappings and the other is multimodal assistant masks; not the same bug." + }, + { + "left": "issue:43122", + "right": "issue:46326", + "accept": false, + "reason": "Both are tokenizer regressions, but one is general version-to-version tokenization drift and the other is continue_final_message handling for mistral_common tokenizers." + }, + { + "left": "issue:43122", + "right": "issue:44521", + "accept": false, + "reason": "Different features and symptoms: general tokenization changes versus multimodal chat template masks." + }, + { + "left": "issue:43408", + "right": "issue:43475", + "accept": false, + "reason": "Same SAM 3 family, but one is a model-type warning and the other is a missing attribute on vision encoder output." + }, + { + "left": "issue:44038", + "right": "issue:44297", + "accept": false, + "reason": "Different issues: Qwen3-VL-Moe loading bug versus tokenizer save_pretrained class-name mismatch." + }, + { + "left": "issue:43116", + "right": "issue:45200", + "accept": false, + "reason": "One is a multi-label classification example script issue; the other is Gemma 4 multimodal token type IDs defaulting to zeros." + }, + { + "left": "issue:39401", + "right": "issue:46614", + "accept": false, + "reason": "Both tokenizer-adjacent, but offset_mapping correctness and stale phonemizer language backend are distinct defects." + }, + { + "left": "issue:42907", + "right": "issue:43452", + "accept": false, + "reason": "Different save/load failures: dequantized model saving versus gguf_file breaking from_pretrained resolution." + }, + { + "left": "issue:44038", + "right": "issue:44295", + "accept": false, + "reason": "Different regressions: Qwen3-VL-Moe loading versus position_ids buffer access." + }, + { + "left": "issue:44038", + "right": "issue:44222", + "accept": false, + "reason": "Both mention FP8/MoE contexts, but the bugs are different: loading behavior versus save_pretrained on moe." + }, + { + "left": "issue:22355", + "right": "issue:29159", + "accept": false, + "reason": "An import error for transformers.onnx is unrelated to slow/fast tokenizer inconsistency." + }, + { + "left": "issue:45216", + "right": "issue:45478", + "accept": false, + "reason": "Related to Qwen3.5/Qwen3.5-Moe version regressions, but one is save_pretrained checkpoint correctness and the other is from_pretrained loading error." + }, + { + "left": "issue:43824", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they concern different missing symbols from different modules." + }, + { + "left": "issue:43122", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer regressions, but one is a broad version drift report and the other is a specific Deepseek tokenizer failure; not enough evidence they are the same bug." + }, + { + "left": "issue:36296", + "right": "issue:43116", + "accept": false, + "reason": "Tensor parallel training bug and example-script multi-label classification failure are unrelated." + }, + { + "left": "issue:43278", + "right": "issue:44945", + "accept": false, + "reason": "Different distributed/evaluation behaviors: dtype change in evaluate versus incorrect output under pipeline parallelism." + }, + { + "left": "issue:42757", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they are for different missing imports and different dependency surfaces." + }, + { + "left": "issue:42175", + "right": "issue:44938", + "accept": false, + "reason": "Backend packaging/install issue versus Python 3.14 load failure; not the same underlying problem." + }, + { + "left": "issue:22355", + "right": "issue:29127", + "accept": false, + "reason": "No meaningful overlap: ONNX import failure versus LayoutLMv3 error-message clarity." + }, + { + "left": "issue:43493", + "right": "issue:45753", + "accept": false, + "reason": "Both involve Qwen3.5, but one is implementation discrepancy versus infinite-loop behavior on a specific image." + }, + { + "left": "issue:44188", + "right": "issue:46328", + "accept": false, + "reason": "Attention-kernel divergence under torch.compile is unrelated to SequenceFeatureExtractor.pad performance." + }, + { + "left": "issue:43937", + "right": "issue:44485", + "accept": false, + "reason": "GenerationConfig validation for GLM-5 is unrelated to GLM-5 RoPE implementation discussion." + }, + { + "left": "issue:42898", + "right": "issue:46614", + "accept": false, + "reason": "v5 tokenization-space cleanup behavior and phonemizer language state are different mechanisms." + }, + { + "left": "issue:38617", + "right": "issue:45278", + "accept": false, + "reason": "Different import breakages: layer_type_validation missing versus many import errors after upgrading to 5.5.0." + }, + { + "left": "issue:43066", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer regressions, but one is decoder-type mismatch in v5 and the other is Deepseek tokenization output changes; not the same concrete bug." + }, + { + "left": "issue:42673", + "right": "issue:45072", + "accept": false, + "reason": "VRAM leak in Qwen3ForCausalLM multi-threaded use is unrelated to dtype mismatches in CI inference." + }, + { + "left": "issue:43611", + "right": "issue:46002", + "accept": false, + "reason": "base_model_prefix loading regression and NaNs in classification heads are distinct 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": 22, + "review_unit_index": 48, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:28348", + "issue:33453", + "issue:39401", + "issue:42222", + "issue:42371", + "issue:42673", + "issue:42796", + "issue:43064", + "issue:43116", + "issue:43441", + "issue:43493", + "issue:43673", + "issue:43825", + "issue:43827", + "issue:43874", + "issue:43937", + "issue:43976", + "issue:43992", + "issue:44038", + "issue:44075", + "issue:44117", + "issue:44291", + "issue:44355", + "issue:44380", + "issue:44466", + "issue:44625", + "issue:44704", + "issue:44779", + "issue:44936", + "issue:44995", + "issue:45071", + "issue:45083", + "issue:45356", + "issue:45357", + "issue:45405", + "issue:45446", + "issue:45478", + "issue:45563", + "issue:45588", + "issue:46002", + "issue:46077", + "issue:46231", + "issue:46328", + "issue:46614", + "issue:46640", + "issue:46655", + "issue:46697" + ], + "soft_pairs": [ + "issue:44355|issue:46231", + "issue:43116|issue:43493", + "issue:42371|issue:43874", + "issue:43976|issue:45446", + "issue:44038|issue:45478", + "issue:44625|issue:46002", + "issue:45357|issue:45478", + "issue:45083|issue:45446", + "issue:42222|issue:43441", + "issue:44075|issue:44936", + "issue:43116|issue:43825", + "issue:43992|issue:44704", + "issue:28348|issue:46640", + "issue:44995|issue:46697", + "issue:43064|issue:43673", + "issue:42673|issue:42796", + "issue:43827|issue:45446", + "issue:33453|issue:39401", + "issue:43937|issue:45563", + "issue:42673|issue:46655", + "issue:45356|issue:46614", + "issue:44380|issue:46077", + "issue:43673|issue:46328", + "issue:44117|issue:45588", + "issue:45071|issue:46614", + "issue:44291|issue:45071", + "issue:45405|issue:45446", + "issue:39401|issue:44779", + "issue:44466|issue:46002" + ], + "prepared_review_unit_hash": "d7c976b2609a53d24401d94a2ba3551fbf5f02201aafb649f26d50feb8874463", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35688, + "estimated_input_tokens": 8922, + "estimated_eval_tokens": 18100 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab-bag of unrelated transformer bugs and regressions. None of the soft pairs look like the same underlying issue strongly enough to merge.", + "confidence": 0.24, + "canonical_issue_reason": "No single issue cleanly canonicalizes this cluster; the items span unrelated subsystems (tokenizers, attention, training, vision, version checks) with no clear duplicate nucleus.", + "canonical_pr_reason": null, + "best_issue_reason": "If one issue must stand in for the cluster, 33453 is the broadest and most general tokenizer-regression report, but it is still not a true duplicate of the other items.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44355", + "right": "issue:46231", + "accept": false, + "reason": "Unrelated problems: compiled Python file handling vs set_training() docs parameter mismatch." + }, + { + "left": "issue:43116", + "right": "issue:43493", + "accept": false, + "reason": "Different bugs: classification example returns empty results vs SigLIP2 implementation discrepancy." + }, + { + "left": "issue:42371", + "right": "issue:43874", + "accept": false, + "reason": "Different subsystems and failures: TF32 API settings vs missing image-patch method in GLM46V fast processor." + }, + { + "left": "issue:43976", + "right": "issue:45446", + "accept": false, + "reason": "Both involve version checks, but for different imports/components and different failure modes." + }, + { + "left": "issue:44038", + "right": "issue:45478", + "accept": false, + "reason": "Related model family, but not the same concrete bug; Qwen3-VL-Moe vs Qwen3.5 Moe from_pretrained errors are not clearly the same code path." + }, + { + "left": "issue:44625", + "right": "issue:46002", + "accept": false, + "reason": "Different behaviors: num_labels propagation vs NaNs in classification heads." + }, + { + "left": "issue:45357", + "right": "issue:45478", + "accept": false, + "reason": "Different regression reports for different Qwen3.5 issues; not enough evidence of one shared bug." + }, + { + "left": "issue:45083", + "right": "issue:45446", + "accept": false, + "reason": "Different functionality: qwen3_omni_moe feature-extract length helper vs flex_attention PyTorch version gating." + }, + { + "left": "issue:42222", + "right": "issue:43441", + "accept": false, + "reason": "Both touch Ministral/vitpose model support, but the failures are unrelated model-specific breakages." + }, + { + "left": "issue:44075", + "right": "issue:44936", + "accept": false, + "reason": "Different training/optimizer bug vs trainer.evaluate() post-train failure." + }, + { + "left": "issue:43116", + "right": "issue:43825", + "accept": false, + "reason": "Both mention pipeline/example behavior, but the reported issues are different (classification output vs translation-task error messaging)." + }, + { + "left": "issue:43992", + "right": "issue:44704", + "accept": false, + "reason": "Different from_pretrained/caching issues affecting different APIs and code paths." + }, + { + "left": "issue:28348", + "right": "issue:46640", + "accept": false, + "reason": "One is a request to add FA2 support for GPT2LMHeadModel; the other is a generic flash-attention failure with no clear shared bug." + }, + { + "left": "issue:44995", + "right": "issue:46697", + "accept": false, + "reason": "Unrelated model bugs: stale indexer cache on second forward pass vs image over-padding." + }, + { + "left": "issue:43064", + "right": "issue:43673", + "accept": false, + "reason": "Different training-state optimizer issue vs GenerationMixin cache missing during chunked prefill." + }, + { + "left": "issue:42673", + "right": "issue:42796", + "accept": false, + "reason": "Different problems: VRAM leak in Qwen3ForCausalLM threads vs tokenizer BPE marker handling." + }, + { + "left": "issue:43827", + "right": "issue:45446", + "accept": false, + "reason": "Documentation/pipeline deprecation issue vs PyTorch version check bug; not the same artifact." + }, + { + "left": "issue:33453", + "right": "issue:39401", + "accept": false, + "reason": "Both are tokenizer-related, but one is a loading regression and the other is wrong offset_mapping on Qwen3." + }, + { + "left": "issue:43937", + "right": "issue:45563", + "accept": false, + "reason": "Different generation issues: invalid GenerationConfig vs stale warning for num_return_sequences." + }, + { + "left": "issue:42673", + "right": "issue:46655", + "accept": false, + "reason": "Different model/component: VRAM leak in Qwen3ForCausalLM vs video padding bug in SmolVLM." + }, + { + "left": "issue:45356", + "right": "issue:46614", + "accept": false, + "reason": "Different tokenization/runtime issues: Kimi-K2.5 codec regression vs stale phonemizer backend state." + }, + { + "left": "issue:44380", + "right": "issue:46077", + "accept": false, + "reason": "Both are attention-related, but one is GPT2 scaling under SDPA/FA and the other is Gemma4 bidirectional mask construction." + }, + { + "left": "issue:43673", + "right": "issue:46328", + "accept": false, + "reason": "Different areas: generation cache behavior vs feature extractor padding performance." + }, + { + "left": "issue:44117", + "right": "issue:45588", + "accept": false, + "reason": "Tokenizer mapping None bug vs flash_attention.py s_aux=None AttributeError; unrelated code paths." + }, + { + "left": "issue:45071", + "right": "issue:46614", + "accept": false, + "reason": "Different issues: PretrainedConfig type checking vs tokenizer phonemizer backend state." + }, + { + "left": "issue:44291", + "right": "issue:45071", + "accept": false, + "reason": "Different regressions: init_empty_weights argument handling vs PretrainedConfig type checking." + }, + { + "left": "issue:45405", + "right": "issue:45446", + "accept": false, + "reason": "Different versioning problems in different dependencies; not the same bug." + }, + { + "left": "issue:39401", + "right": "issue:44779", + "accept": false, + "reason": "Both tokenizer regressions, but one is offset_mapping and the other is general incorrect results for Deepseek; not clearly the same defect." + }, + { + "left": "issue:44466", + "right": "issue:46002", + "accept": false, + "reason": "Inconsistent lm_head serialization vs NaNs in classification heads are separate problems." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 49, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:38617", + "issue:39401", + "issue:41628", + "issue:42222", + "issue:42371", + "issue:42617", + "issue:43066", + "issue:43097", + "issue:43122", + "issue:43408", + "issue:43450", + "issue:43673", + "issue:43742", + "issue:43874", + "issue:43937", + "issue:43986", + "issue:44075", + "issue:44117", + "issue:44220", + "issue:44261", + "issue:44263", + "issue:44410", + "issue:44484", + "issue:44485", + "issue:44493", + "issue:44496", + "issue:44829", + "issue:44877", + "issue:44933", + "issue:44977", + "issue:45071", + "issue:45081", + "issue:45216", + "issue:45278", + "issue:45310", + "issue:45341", + "issue:45362", + "issue:45478", + "issue:45542", + "issue:45561", + "issue:45656", + "issue:46178", + "issue:46209", + "issue:46231", + "issue:46298", + "issue:46328", + "issue:46512", + "issue:46617" + ], + "soft_pairs": [ + "issue:44410|issue:45310", + "issue:43937|issue:44977", + "issue:43450|issue:43986", + "issue:45341|issue:45561", + "issue:45341|issue:46328", + "issue:45071|issue:46512", + "issue:44829|issue:45216", + "issue:44117|issue:45081", + "issue:42617|issue:45341", + "issue:38617|issue:41628", + "issue:42371|issue:44261", + "issue:43122|issue:46298", + "issue:42222|issue:46178", + "issue:45362|issue:46328", + "issue:43408|issue:43742", + "issue:44877|issue:46231", + "issue:42371|issue:46617", + "issue:44484|issue:45310", + "issue:39401|issue:46209", + "issue:43937|issue:44496", + "issue:44933|issue:45278", + "issue:44410|issue:45478", + "issue:44075|issue:45656", + "issue:38617|issue:44220", + "issue:44263|issue:44485", + "issue:43066|issue:43122", + "issue:42222|issue:45542", + "issue:43874|issue:44261", + "issue:43673|issue:45341", + "issue:43097|issue:44493" + ], + "prepared_review_unit_hash": "c68a91efbdb52eb9c0f8cfbfa1c7f9c61705096dc740a3839c51ead6aedddf1a", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36277, + "estimated_input_tokens": 9070, + "estimated_eval_tokens": 18396 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab bag of unrelated Transformers regressions and support issues, with a common surface theme of version-5 breakages, tokenizer/config/import problems, and a few model-specific bugs. I do not see a true duplicate core spanning the set.", + "confidence": 0.44, + "canonical_issue_reason": "issue:45278 is the broadest umbrella report here, covering multiple import failures after upgrading to 5.5.0, so it is the best representative of the cluster\u2019s shared upgrade/regression theme.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 best represents the cluster because it is the most general and highest-level regression report; however, the rest of the items are too heterogeneous to treat as one duplicate family.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44410", + "right": "issue:45310", + "accept": false, + "reason": "Both are Qwen-family failures, but one is missing layer projections and the other is a from_pretrained error for Qwen3.5 MoE; different concrete bugs." + }, + { + "left": "issue:43937", + "right": "issue:44977", + "accept": false, + "reason": "GLM generation-config validation and Qwen3.5 flash-attention generation issues are unrelated code paths." + }, + { + "left": "issue:43450", + "right": "issue:43986", + "accept": false, + "reason": "One is a video-processor shape bug for batched input; the other is a dependency-related AutoProcessor crash without torchvision." + }, + { + "left": "issue:45341", + "right": "issue:45561", + "accept": false, + "reason": "Both mention testing utils, but one is a generic bug report and the other is a pytest-xdist race on captured_info.txt; not the same bug." + }, + { + "left": "issue:45341", + "right": "issue:46328", + "accept": false, + "reason": "testing_utils versus SequenceFeatureExtractor.pad performance; unrelated issues." + }, + { + "left": "issue:45071", + "right": "issue:46512", + "accept": false, + "reason": "Both concern config/dtype behavior, but one is type-checking breakage and the other is stale model.config.dtype after from_config; different fixes." + }, + { + "left": "issue:44829", + "right": "issue:45216", + "accept": false, + "reason": "Flash-attention training degeneration and Qwen3.5 save_pretrained regression are distinct problems." + }, + { + "left": "issue:44117", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer mapping None-handling and Mistral regex patch backend_tokenizer crash are separate tokenizer-loading defects." + }, + { + "left": "issue:42617", + "right": "issue:45341", + "accept": false, + "reason": "3d_parallel.py execution failure is unrelated to a testing_utils bug." + }, + { + "left": "issue:38617", + "right": "issue:41628", + "accept": false, + "reason": "Two different import errors for different symbols; no shared underlying bug shown." + }, + { + "left": "issue:42371", + "right": "issue:44261", + "accept": false, + "reason": "TF32 settings guidance and MLA rms_norm_eps precision issue do not overlap in code path." + }, + { + "left": "issue:43122", + "right": "issue:46298", + "accept": false, + "reason": "Tokenizer behavior regression and a noisy logging message are unrelated." + }, + { + "left": "issue:42222", + "right": "issue:46178", + "accept": false, + "reason": "VitPose model breakage and missing tensorflow_text backend mapping are different issues." + }, + { + "left": "issue:45362", + "right": "issue:46328", + "accept": false, + "reason": "Chat crash for Qwen3.5-35B and a pad() performance issue are unrelated." + }, + { + "left": "issue:43408", + "right": "issue:43742", + "accept": false, + "reason": "Model-type warning for sam3_* and MobileLLM key error are not the same defect." + }, + { + "left": "issue:44877", + "right": "issue:46231", + "accept": false, + "reason": "Strict config loading for granite_speech and a docs parameter-name mismatch are unrelated." + }, + { + "left": "issue:42371", + "right": "issue:46617", + "accept": false, + "reason": "TF32 backend settings and trust_remote_code cache FileNotFoundError are unrelated." + }, + { + "left": "issue:44484", + "right": "issue:45310", + "accept": false, + "reason": "save_pretrained shard-size question and Qwen3.5 MoE from_pretrained failure are unrelated." + }, + { + "left": "issue:39401", + "right": "issue:46209", + "accept": false, + "reason": "Tokenizer offset mapping and FP8 inference on B200 are completely different bugs." + }, + { + "left": "issue:43937", + "right": "issue:44496", + "accept": false, + "reason": "Invalid generation config for GLM-5 and unrecognized model config for Olmo-Hybrid are separate issues." + }, + { + "left": "issue:44933", + "right": "issue:45278", + "accept": false, + "reason": "Both are import-related, but one is a specific missing image_utils import while the other is a broad upgrade regression; not the same concrete failure." + }, + { + "left": "issue:44410", + "right": "issue:45478", + "accept": false, + "reason": "Different Qwen model families and different failure symptoms." + }, + { + "left": "issue:44075", + "right": "issue:45656", + "accept": false, + "reason": "SGD args ignored and optimizer.step called twice with deepspeed are different optimizer bugs." + }, + { + "left": "issue:38617", + "right": "issue:44220", + "accept": false, + "reason": "ImportError for configuration_utils and _torch_extract_fbank_features are unrelated." + }, + { + "left": "issue:44263", + "right": "issue:44485", + "accept": false, + "reason": "GLM tensor split issue and GLM-5 RoPE implementation issue are separate model-specific bugs." + }, + { + "left": "issue:43066", + "right": "issue:43122", + "accept": false, + "reason": "Both involve tokenizer changes in v5, but one is a decoder-type mismatch and the other is changed tokenization output; not the same bug." + }, + { + "left": "issue:42222", + "right": "issue:45542", + "accept": false, + "reason": "VitPose breakage and TensorFlow backend detection are unrelated." + }, + { + "left": "issue:43874", + "right": "issue:44261", + "accept": false, + "reason": "GLM46V image-processor attribute error and MLA rms_norm_eps precision issue do not share a code path." + }, + { + "left": "issue:43673", + "right": "issue:45341", + "accept": false, + "reason": "Chunked prefill cache issue and testing_utils bug are unrelated." + }, + { + "left": "issue:43097", + "right": "issue:44493", + "accept": false, + "reason": "Both mention config/key changes, but one is removal of a config option and the other is unexpected position-id keys; not the same underlying bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 50, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:27957", + "issue:38617", + "issue:41628", + "issue:42175", + "issue:42222", + "issue:42757", + "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:44117", + "issue:44261", + "issue:44263", + "issue:44315", + "issue:44360", + "issue:44410", + "issue:44483", + "issue:44484", + "issue:44485", + "issue:44829", + "issue:44933", + "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:46123", + "issue:46172", + "issue:46231", + "issue:46369", + "issue:46655", + "issue:46664" + ], + "soft_pairs": [ + "issue:42175|issue:42222", + "issue:41628|issue:45071", + "issue:45071|issue:45542", + "issue:44263|issue:46172", + "issue:43931|issue:44410", + "issue:44360|issue:46172", + "issue:43012|issue:43577", + "issue:44483|issue:45464", + "issue:44484|issue:45478", + "issue:27957|issue:43352", + "issue:43441|issue:44829", + "issue:43931|issue:44038", + "issue:43097|issue:43452", + "issue:45003|issue:45341", + "issue:43986|issue:45341", + "issue:43874|issue:44263", + "issue:44933|issue:46231", + "issue:45753|issue:46369", + "issue:44261|issue:44485", + "issue:44117|issue:44987", + "issue:45440|issue:46172", + "issue:43673|issue:46123", + "issue:43531|issue:45341", + "issue:43097|issue:45588", + "issue:43064|issue:44315", + "issue:43986|issue:45372", + "issue:38617|issue:42757", + "issue:46655|issue:46664", + "issue:44410|issue:45375" + ], + "prepared_review_unit_hash": "11fc709e71e003df22f92aebcb0a41e8908f3998fc962653dc12c69dd29ea260", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36193, + "estimated_input_tokens": 9049, + "estimated_eval_tokens": 18354 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous: the soft pairs are broad topical matches (same model family, same subsystem, or same class of error) rather than the same concrete bug/change. I don\u2019t see any pair that should be merged as true duplicates.", + "confidence": 0.82, + "canonical_issue_reason": "No clear canonical issue: the items span unrelated import regressions, model-specific loading bugs, training/runtime issues, and documentation problems. The soft links are too broad to indicate one shared underlying defect.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45372 is the strongest standalone issue in the set: it has a clear regression, concrete import error, and active discussion, making it the most useful representative issue even though it is not a duplicate of the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42175", + "right": "issue:42222", + "accept": false, + "reason": "Both are model/backend complaints, but one is a TensorFlow packaging/backend issue and the other is a ViTPose model breakage; different code paths and fixes." + }, + { + "left": "issue:41628", + "right": "issue:45071", + "accept": false, + "reason": "An AutoImageProcessor import failure and a PreTrainedConfig type-checking regression are unrelated bugs." + }, + { + "left": "issue:45071", + "right": "issue:45542", + "accept": false, + "reason": "These both involve API/backend validation, but one is a config typing regression and the other is a TensorBoard/TensorFlow backend install problem." + }, + { + "left": "issue:44263", + "right": "issue:46172", + "accept": false, + "reason": "Different model internals and failure modes: torch.split indexing vs a dtype/transpose bug in deepseek-v4." + }, + { + "left": "issue:43931", + "right": "issue:44410", + "accept": false, + "reason": "Both mention Qwen models, but one is a weight-shape load error and the other is missing projections in layer 0; not the same defect." + }, + { + "left": "issue:44360", + "right": "issue:46172", + "accept": false, + "reason": "A missing ReLU in a DSA indexer is unrelated to a tensor dtype/transpose issue." + }, + { + "left": "issue:43012", + "right": "issue:43577", + "accept": false, + "reason": "Both are dtype-related, but one is a compilation warning and the other is incorrect dtype propagation during Blip2 loading." + }, + { + "left": "issue:44483", + "right": "issue:45464", + "accept": false, + "reason": "Both touch chat/completions, but one is a general endpoint request rejection and the other is a Qwen3.5 streaming inference failure." + }, + { + "left": "issue:44484", + "right": "issue:45478", + "accept": false, + "reason": "A save_pretrained shard-size question is unrelated to a Qwen3.5 MoE from_pretrained error." + }, + { + "left": "issue:27957", + "right": "issue:43352", + "accept": false, + "reason": "Both are Flash Attention 2 support issues, but they concern different models and likely different implementation gaps." + }, + { + "left": "issue:43441", + "right": "issue:44829", + "accept": false, + "reason": "Both mention flash attention, but one is a model load failure in v5 RC and the other is degenerate training with FA3; different problems." + }, + { + "left": "issue:43931", + "right": "issue:44038", + "accept": false, + "reason": "Same broad Qwen3-VL/MoE area, but one is a concrete weight-shape mismatch and the other is a general version/model bug report." + }, + { + "left": "issue:43097", + "right": "issue:43452", + "accept": false, + "reason": "A removed config method and a gguf_file tokenizer/model loading break are different API regressions." + }, + { + "left": "issue:45003", + "right": "issue:45341", + "accept": false, + "reason": "Unsafe sys.modules access in modeling_utils is unrelated to a small testing_utils bug." + }, + { + "left": "issue:43986", + "right": "issue:45341", + "accept": false, + "reason": "A video-model AutoProcessor crash without torchvision is unrelated to a testing utility bug." + }, + { + "left": "issue:43874", + "right": "issue:44263", + "accept": false, + "reason": "Different model families and failures: missing image-patch counting vs torch.split return-value handling." + }, + { + "left": "issue:44933", + "right": "issue:46231", + "accept": false, + "reason": "A missing import from image_utils and a documentation parameter mismatch are not the same bug." + }, + { + "left": "issue:45753", + "right": "issue:46369", + "accept": false, + "reason": "Infinite-loop behavior on a specific image is unrelated to a memory overflow in another model." + }, + { + "left": "issue:44261", + "right": "issue:44485", + "accept": false, + "reason": "A missing rms_norm_eps in MLA q_a_layernorm is different from a GLM-5 RoPE implementation discussion." + }, + { + "left": "issue:44117", + "right": "issue:44987", + "accept": false, + "reason": "A tokenizer-mapping None issue and a model-specific loading failure overlap only at a high level; no shared concrete code path is evident." + }, + { + "left": "issue:45440", + "right": "issue:46172", + "accept": false, + "reason": "Both are DeepSeek-related, but one is divergence from the remote implementation and the other is a tensor dtype/transpose bug." + }, + { + "left": "issue:43673", + "right": "issue:46123", + "accept": false, + "reason": "A generation cache regression during chunked prefill is unrelated to MaskGenerationPipeline's final partial-batch flag handling." + }, + { + "left": "issue:43531", + "right": "issue:45341", + "accept": false, + "reason": "A Qwen3-MoE sliding_window bug is unrelated to a testing utility issue." + }, + { + "left": "issue:43097", + "right": "issue:45588", + "accept": false, + "reason": "A removed embedding-tie method and a flash_attention AttributeError on sink-less models are different code paths." + }, + { + "left": "issue:43064", + "right": "issue:44315", + "accept": false, + "reason": "FSDP2/PEFT optimizer-state corruption and Liger Kernel application during model_init are unrelated training bugs." + }, + { + "left": "issue:43986", + "right": "issue:45372", + "accept": false, + "reason": "A torchvision-dependent video processor crash and a mistral_common import error affecting Gemma 4 processor loading are separate dependency issues." + }, + { + "left": "issue:38617", + "right": "issue:42757", + "accept": false, + "reason": "Both are ImportErrors from dependency churn, but they involve different missing symbols in different packages." + }, + { + "left": "issue:46655", + "right": "issue:46664", + "accept": false, + "reason": "Both are SmolVLM video-processor issues, but one is padding calculation and the other is the meaning of num_frames; related area, not the same bug." + }, + { + "left": "issue:44410", + "right": "issue:45375", + "accept": false, + "reason": "Both are Qwen3.5/MoE vision-related, but one is missing attention projections and the other is a missing config field silently dropped by strict validation." + } + ] + }, + "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": 22, + "review_unit_index": 51, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:36683", + "issue:38617", + "issue:42890", + "issue:42907", + "issue:43012", + "issue:43066", + "issue:43097", + "issue:43208", + "issue:43352", + "issue:43502", + "issue:43550", + "issue:43582", + "issue:43825", + "issue:43873", + "issue:43906", + "issue:43976", + "issue:43986", + "issue:43992", + "issue:44016", + "issue:44038", + "issue:44062", + "issue:44077", + "issue:44117", + "issue:44162", + "issue:44222", + "issue:44351", + "issue:44368", + "issue:44380", + "issue:44410", + "issue:44484", + "issue:44655", + "issue:44704", + "issue:44936", + "issue:44938", + "issue:45103", + "issue:45478", + "issue:45479", + "issue:45584", + "issue:45712", + "issue:45753", + "issue:46050", + "issue:46132", + "issue:46172", + "issue:46178", + "issue:46298", + "issue:46365", + "issue:46421", + "issue:46701" + ], + "soft_pairs": [ + "issue:43097|issue:46050", + "issue:44038|issue:44410", + "issue:43986|issue:44484", + "issue:42890|issue:46298", + "issue:38617|issue:44351", + "issue:44062|issue:44117", + "issue:44117|issue:46701", + "issue:43097|issue:43873", + "issue:44936|issue:46365", + "issue:43502|issue:43992", + "issue:44704|issue:45103", + "issue:44380|issue:46421", + "issue:44938|issue:46178", + "issue:43986|issue:44117", + "issue:43986|issue:44016", + "issue:43208|issue:43550", + "issue:42907|issue:44655", + "issue:43825|issue:44655", + "issue:42907|issue:45753", + "issue:43012|issue:43550", + "issue:45712|issue:46298", + "issue:43012|issue:44368", + "issue:43066|issue:43906", + "issue:43992|issue:44484", + "issue:43352|issue:44380", + "issue:43352|issue:43976", + "issue:43097|issue:46298", + "issue:44162|issue:46172", + "issue:45584|issue:46365", + "issue:44380|issue:46132", + "issue:44077|issue:46365", + "issue:44222|issue:45478", + "issue:44380|issue:45479", + "issue:36683|issue:43582" + ], + "prepared_review_unit_hash": "3b1d945546bac57eadab231f378b8a55d78f8643d5f02c19cc398371adaf91d0", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37631, + "estimated_input_tokens": 9408, + "estimated_eval_tokens": 19072 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are mostly unrelated standalone Transformers issues spanning different models, loaders, serialization paths, and runtime backends. I found no soft-edge pair that clearly represents the same underlying bug/change, so no pair is accepted.", + "confidence": 0.31, + "canonical_issue_reason": "issue:43873 has the strongest general signal in this set (high discussion and inbound references) and is a broad, active bug report in a core feature area, making it the best representative placeholder canonical issue.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43873 is the most prominent issue here by activity and scope; among otherwise unrelated reports, it is the most suitable single issue to anchor the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43097", + "right": "issue:46050", + "accept": false, + "reason": "Both involve v5/serialization-adjacent behavior, but one is an API removal warning and the other is quantized save/load breakage; different failure modes and code paths." + }, + { + "left": "issue:44038", + "right": "issue:44410", + "accept": false, + "reason": "Both mention Qwen/VL-family models, but one is a transformers 5.0 compatibility bug and the other is missing projections in qwen3next; not the same defect." + }, + { + "left": "issue:43986", + "right": "issue:44484", + "accept": false, + "reason": "One is a loading crash without torchvision, the other is a save_pretrained shard-size question; unrelated problems." + }, + { + "left": "issue:42890", + "right": "issue:46298", + "accept": false, + "reason": "A flaky test due to missing set_seed is unrelated to a noisy logging message." + }, + { + "left": "issue:38617", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they concern different missing symbols from different migration paths; not one shared bug." + }, + { + "left": "issue:44062", + "right": "issue:44117", + "accept": false, + "reason": "Different tokenizer bugs: AddedToken argument handling versus TOKENIZER_MAPPING_NAMES returning None." + }, + { + "left": "issue:44117", + "right": "issue:46701", + "accept": false, + "reason": "Tokenizer mapping lookup failure is unrelated to a DiffusionGemma image-text pipeline failure." + }, + { + "left": "issue:43097", + "right": "issue:43873", + "accept": false, + "reason": "Tie-embeddings removal and offloading-with-quantization are separate regressions with different root causes." + }, + { + "left": "issue:44936", + "right": "issue:46365", + "accept": false, + "reason": "trainer.evaluate() after train() is unrelated to import-time Trainer breakage caused by datasets." + }, + { + "left": "issue:43502", + "right": "issue:43992", + "accept": false, + "reason": "One is incorrect network access under local_files_only, the other is a missing UMT5Encoder weight during loading." + }, + { + "left": "issue:44704", + "right": "issue:45103", + "accept": false, + "reason": "AutoProcessor kwargs forwarding and auto_docstring future-annotations crash are different subsystems." + }, + { + "left": "issue:44380", + "right": "issue:46421", + "accept": false, + "reason": "GPT2 attention scaling with SDPA/FlashAttention and cache lazy-init recompilation are separate backend issues." + }, + { + "left": "issue:44938", + "right": "issue:46178", + "accept": false, + "reason": "Python 3.14 import failure and tensorflow_text backend validation are unrelated compatibility errors." + }, + { + "left": "issue:43986", + "right": "issue:44117", + "accept": false, + "reason": "Video-model loading without torchvision does not match TOKENIZER_MAPPING_NAMES returning None." + }, + { + "left": "issue:43986", + "right": "issue:44016", + "accept": false, + "reason": "A missing torchvision dependency crash is unrelated to a notebook syntax error." + }, + { + "left": "issue:43208", + "right": "issue:43550", + "accept": false, + "reason": "xLSTM training bugs and Bamba torch.compile/SDPA failure are different model/backend problems." + }, + { + "left": "issue:42907", + "right": "issue:44655", + "accept": false, + "reason": "Dequantized model saving failure is not the same as pipeline save_pretrained limitations." + }, + { + "left": "issue:43825", + "right": "issue:44655", + "accept": false, + "reason": "A v5 pipeline error message issue is unrelated to saving Pipeline objects." + }, + { + "left": "issue:42907", + "right": "issue:45753", + "accept": false, + "reason": "Model dequantized-save failure and Qwen3_5 infinite loop on a specific image are different bugs." + }, + { + "left": "issue:43012", + "right": "issue:43550", + "accept": false, + "reason": "bf16 compile warning and Bamba SDPA compile failure are not the same concrete defect." + }, + { + "left": "issue:45712", + "right": "issue:46298", + "accept": false, + "reason": "Leftover dummy classes leaking into dir(transformers) is unrelated to a noisy logging message." + }, + { + "left": "issue:43012", + "right": "issue:44368", + "accept": false, + "reason": "PyTorch precision warning and a tie_word_embeddings warning are distinct warning sources." + }, + { + "left": "issue:43066", + "right": "issue:43906", + "accept": false, + "reason": "Tokenizer decoder type mismatch and a reproduction of issue 38071 do not appear to be the same bug." + }, + { + "left": "issue:43992", + "right": "issue:44484", + "accept": false, + "reason": "Missing embedding weights on load and a max_shard_size question are unrelated." + }, + { + "left": "issue:43352", + "right": "issue:44380", + "accept": false, + "reason": "Nemotron FA2 unsupported and GPT2 attention scaling under SDPA/FlashAttention are different backend/model issues." + }, + { + "left": "issue:43352", + "right": "issue:43976", + "accept": false, + "reason": "Unsupported Flash Attention 2 for Nemotron is unrelated to the Python version compatibility report." + }, + { + "left": "issue:43097", + "right": "issue:46298", + "accept": false, + "reason": "A removed config method and a noisy log message are not the same underlying issue." + }, + { + "left": "issue:44162", + "right": "issue:46172", + "accept": false, + "reason": "ESM2 breakage in workflows and a DeepSeek-V4 tensor transpose bug are unrelated model-specific failures." + }, + { + "left": "issue:45584", + "right": "issue:46365", + "accept": false, + "reason": "Whisper empty-transcription generation failure is unrelated to importing Trainer." + }, + { + "left": "issue:44380", + "right": "issue:46132", + "accept": false, + "reason": "GPT2 backend-specific attention scaling and AttentionInterface.register behavior change are different code paths." + }, + { + "left": "issue:44077", + "right": "issue:46365", + "accept": false, + "reason": "patchtsmixer post_init validation and Trainer import failure are unrelated." + }, + { + "left": "issue:44222", + "right": "issue:45478", + "accept": false, + "reason": "FP8 save_pretrained for MoE and Qwen3.5 MoE from_pretrained error are both MoE-related but not the same concrete bug or mergeable fix." + }, + { + "left": "issue:44380", + "right": "issue:45479", + "accept": false, + "reason": "Attention scaling ignored in specific backends is unrelated to single-label-classification zero-loss degeneration." + }, + { + "left": "issue:36683", + "right": "issue:43582", + "accept": false, + "reason": "Gemma3Config missing vocab_size and Apple Silicon caching_allocator_warmup TypeError 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": 22, + "review_unit_index": 52, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:36010", + "issue:36683", + "issue:38617", + "issue:42175", + "issue:42490", + "issue:42757", + "issue:42898", + "issue:42994", + "issue:43012", + "issue:43064", + "issue:43066", + "issue:43208", + "issue:43299", + "issue:43352", + "issue:43475", + "issue:43502", + "issue:43525", + "issue:43526", + "issue:43611", + "issue:43618", + "issue:43756", + "issue:43761", + "issue:43824", + "issue:43867", + "issue:43901", + "issue:43937", + "issue:43976", + "issue:44038", + "issue:44077", + "issue:44230", + "issue:44380", + "issue:44483", + "issue:44485", + "issue:44514", + "issue:44625", + "issue:44655", + "issue:44683", + "issue:44938", + "issue:45216", + "issue:45276", + "issue:45290", + "issue:45600", + "issue:45698", + "issue:46077", + "issue:46082", + "issue:46172", + "issue:46298", + "issue:46365" + ], + "soft_pairs": [ + "issue:42175|issue:46365", + "issue:43064|issue:44380", + "issue:42490|issue:43012", + "issue:42757|issue:43824", + "issue:43012|issue:44683", + "issue:43208|issue:43526", + "issue:45600|issue:46298", + "issue:44483|issue:44514", + "issue:43618|issue:44380", + "issue:45698|issue:46172", + "issue:44038|issue:44230", + "issue:43618|issue:43761", + "issue:38617|issue:43502", + "issue:43756|issue:44485", + "issue:43299|issue:43867", + "issue:42994|issue:44655", + "issue:36683|issue:43475", + "issue:44625|issue:46365", + "issue:45276|issue:46077", + "issue:42898|issue:43066", + "issue:36010|issue:42757", + "issue:44483|issue:45290", + "issue:43352|issue:43937", + "issue:43611|issue:44655", + "issue:43901|issue:44655", + "issue:36683|issue:44077", + "issue:43976|issue:44938", + "issue:43867|issue:45216", + "issue:44625|issue:46077", + "issue:46077|issue:46082", + "issue:36683|issue:43525", + "issue:43618|issue:45216", + "issue:36683|issue:44625" + ], + "prepared_review_unit_hash": "04efc25ef308e3e23522f86126aa95033fe71510d5d27e368cce2209ae6c1ec6", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37204, + "estimated_input_tokens": 9301, + "estimated_eval_tokens": 18858 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a set of superficial transformer-regression lookalikes (import errors, save/load bugs, config mismatches, chat-template issues, backend-specific model bugs). I don\u2019t see any soft pair that is clearly the same underlying bug/change, so all candidate merges should be rejected. If one issue must act as a representative, the save/load regression in 45698 is the broadest fit, but it is still not a true duplicate umbrella.", + "confidence": 0.89, + "canonical_issue_reason": "45698 is the closest broad representative because it centers on save_pretrained/from_pretrained serialization behavior, which is the only theme that recurs across multiple items; however, the cluster is too heterogeneous for it to be a true canonical duplicate target.", + "canonical_pr_reason": null, + "best_issue_reason": "45698 is the best single issue to represent the cluster because it is active, broad, and aligned with the most repeated save/load-related theme in the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42175", + "right": "issue:46365", + "accept": false, + "reason": "Both mention import/training availability, but one is backend packaging and the other is a datasets-installed Trainer import conflict; unrelated root causes." + }, + { + "left": "issue:43064", + "right": "issue:44380", + "accept": false, + "reason": "Different code paths: distributed optimizer-state corruption vs attention-scaling being ignored under SDPA/FlashAttention." + }, + { + "left": "issue:42490", + "right": "issue:43012", + "accept": false, + "reason": "One is a local-path import warning, the other is a PyTorch precision warning while compiling bfloat16 models." + }, + { + "left": "issue:42757", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but for different symbols and different dependency surfaces (huggingface_hub vs transformers model export)." + }, + { + "left": "issue:43012", + "right": "issue:44683", + "accept": false, + "reason": "Both involve PyTorch/compile behavior, but one is a warning and the other is a hard failure on newer torch; not the same bug." + }, + { + "left": "issue:43208", + "right": "issue:43526", + "accept": false, + "reason": "xLSTM training bugs and BeitImageProcessorFast label reduction are unrelated model/data-processing issues." + }, + { + "left": "issue:45600", + "right": "issue:46298", + "accept": false, + "reason": "auto_mappings CI breakage from removed config references is not the same as a noisy logging message." + }, + { + "left": "issue:44483", + "right": "issue:44514", + "accept": false, + "reason": "Both touch chat/completions, but one is an API acceptance problem and the other is a processor crash on batched tool-call inputs." + }, + { + "left": "issue:43618", + "right": "issue:44380", + "accept": false, + "reason": "CLIP output-attention regression and GPT2 attention scaling under backend kernels are different model behaviors." + }, + { + "left": "issue:45698", + "right": "issue:46172", + "accept": false, + "reason": "Save/load custom-module resolution is unrelated to the deepseek-v4 transpose/dtype bug." + }, + { + "left": "issue:44038", + "right": "issue:44230", + "accept": false, + "reason": "Both concern Qwen3-VL/MoE, but one is a loading bug and the other is fp8 inference support; not the same fix." + }, + { + "left": "issue:43618", + "right": "issue:43761", + "accept": false, + "reason": "Both are CLIP regressions, but one loses attentions and the other loses hidden_states; they are separate outputs and fixes." + }, + { + "left": "issue:38617", + "right": "issue:43502", + "accept": false, + "reason": "Importing layer_type_validation and ignoring local_files_only are unrelated dependency vs network-policy bugs." + }, + { + "left": "issue:43756", + "right": "issue:44485", + "accept": false, + "reason": "Smollm3 RoPE-layer counting and GLM-5 RoPE implementation are different model-specific architecture questions." + }, + { + "left": "issue:43299", + "right": "issue:43867", + "accept": false, + "reason": "Qwen3VL MoE loading regression and sorted-state_dict load failure are distinct loading problems." + }, + { + "left": "issue:42994", + "right": "issue:44655", + "accept": false, + "reason": "Quantized-model save failure and inability to save Pipeline objects both involve saving, but the affected objects and failure modes differ." + }, + { + "left": "issue:36683", + "right": "issue:43475", + "accept": false, + "reason": "Gemma3Config missing vocab_size and SAM3 vision output missing fpn_position_embeddings are separate missing-attribute bugs in different models." + }, + { + "left": "issue:44625", + "right": "issue:46365", + "accept": false, + "reason": "Qwen3.5 num_labels propagation and Trainer import conflicts with datasets are unrelated." + }, + { + "left": "issue:45276", + "right": "issue:46077", + "accept": false, + "reason": "Both are Gemma4-related, but one is embedding resize propagation and the other is attention-mask construction; separate code paths." + }, + { + "left": "issue:42898", + "right": "issue:43066", + "accept": false, + "reason": "Tokenizer cleanup behavior in v5 and wrong tokenizer decoder type are different tokenizer regressions." + }, + { + "left": "issue:36010", + "right": "issue:42757", + "accept": false, + "reason": "Different missing imports from different modules and different dependency updates." + }, + { + "left": "issue:44483", + "right": "issue:45290", + "accept": false, + "reason": "Both involve chat-template / chat-completions, but one is request validation and the other is tokenization crash on assistant tool-call messages." + }, + { + "left": "issue:43352", + "right": "issue:43937", + "accept": false, + "reason": "Flash Attention 2 support for NemotronH and invalid GenerationConfig for GLM-5 are unrelated model-specific issues." + }, + { + "left": "issue:43611", + "right": "issue:44655", + "accept": false, + "reason": "Model loading with base_model_prefix and saving Pipelines are different serialization behaviors." + }, + { + "left": "issue:43901", + "right": "issue:44655", + "accept": false, + "reason": "Docs mismatch for return_all_scores and Pipeline save_pretrained failure are not the same issue." + }, + { + "left": "issue:36683", + "right": "issue:44077", + "accept": false, + "reason": "Missing Gemma config attribute and patchtsmixer post_init validity are unrelated model config constraints." + }, + { + "left": "issue:43976", + "right": "issue:44938", + "accept": false, + "reason": "Python version support breakage in transformers 5.1.0 and Python 3.14 load failures are different compatibility issues." + }, + { + "left": "issue:43867", + "right": "issue:45216", + "accept": false, + "reason": "Sorted state_dict loading error and incorrect Qwen3.5 saved checkpoint are both save/load-related, but the concrete failure modes do not match." + }, + { + "left": "issue:44625", + "right": "issue:46077", + "accept": false, + "reason": "Different Qwen3.5/Gemma4 model families and different bugs." + }, + { + "left": "issue:46077", + "right": "issue:46082", + "accept": false, + "reason": "Gemma4 attention-mask behavior and LlamaConfig head_dim validation are unrelated config/model-shape issues." + }, + { + "left": "issue:36683", + "right": "issue:43525", + "accept": false, + "reason": "Two AttributeErrors, but for different config fields in different models." + }, + { + "left": "issue:43618", + "right": "issue:45216", + "accept": false, + "reason": "CLIP attention output regression and Qwen3.5 checkpoint corruption are separate bugs." + }, + { + "left": "issue:36683", + "right": "issue:44625", + "accept": false, + "reason": "Gemma3Config missing vocab_size and Qwen3.5 num_labels propagation are different config-synchronization 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": 22, + "review_unit_index": 53, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:28282", + "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:43874", + "issue:43976", + "issue:43992", + "issue:44062", + "issue:44075", + "issue:44077", + "issue:44112", + "issue:44222", + "issue:44246", + "issue:44355", + "issue:44360", + "issue:44492", + "issue:44556", + "issue:44734", + "issue:44855", + "issue:44857", + "issue:44938", + "issue:44964", + "issue:45071", + "issue:45125", + "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" + ], + "soft_pairs": [ + "issue:43992|issue:45310", + "issue:43064|issue:44492", + "issue:44075|issue:46365", + "issue:43010|issue:45278", + "issue:44355|issue:44855", + "issue:44938|issue:45071", + "issue:43867|issue:45375", + "issue:43867|issue:46202", + "issue:36683|issue:43064", + "issue:42890|issue:45561", + "issue:41669|issue:44246", + "issue:43010|issue:45850", + "issue:36296|issue:43575", + "issue:43867|issue:45357", + "issue:43618|issue:44077", + "issue:43010|issue:43976", + "issue:43992|issue:45478", + "issue:28282|issue:33453", + "issue:43618|issue:46285", + "issue:44360|issue:45440", + "issue:43618|issue:44857", + "issue:42890|issue:44964", + "issue:42890|issue:44112", + "issue:42898|issue:46664", + "issue:42890|issue:45600", + "issue:43502|issue:44062", + "issue:43874|issue:46617", + "issue:43723|issue:44556", + "issue:44222|issue:45125", + "issue:44734|issue:46617" + ], + "prepared_review_unit_hash": "ff9ba39342a7407dc307aa18a48639a5d48533bd9d82c58be1df391966d23ec7", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36255, + "estimated_input_tokens": 9064, + "estimated_eval_tokens": 18384 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a broad cluster of mostly unrelated v5-era import/load/regression reports. The soft pairs are mostly only loosely similar on surface wording, and none meet a strict duplicate threshold.", + "confidence": 0.57, + "canonical_issue_reason": "Issue 45278 is the broadest representative symptom in the set: a general breakage report after upgrading from 4.57.0 to 5.5.0, matching the cluster\u2019s overall import/load regression theme better than the narrower model-specific bugs.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45278 is the best single anchor for the cluster because it captures the widest user-visible regression among the listed issues and is most likely to subsume nearby upgrade/import failures.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43992", + "right": "issue:45310", + "accept": false, + "reason": "Both are loading errors, but they concern different models and different failure modes (missing embed_tokens.weight vs Qwen3.5 Moe from_pretrained error)." + }, + { + "left": "issue:43064", + "right": "issue:44492", + "accept": false, + "reason": "Optimizer-state corruption in training is unrelated to a cache-strategy typo." + }, + { + "left": "issue:44075", + "right": "issue:46365", + "accept": false, + "reason": "One is about ignored SGD args; the other is an import dependency interaction with datasets." + }, + { + "left": "issue:43010", + "right": "issue:45278", + "accept": false, + "reason": "Cache/update no_grad behavior is a concrete code-path issue; the other is a broad import regression report, not the same bug." + }, + { + "left": "issue:44355", + "right": "issue:44855", + "accept": false, + "reason": "Generic compiled-file import failures are different from the specific DebertaV2 JIT parsing/comment issue." + }, + { + "left": "issue:44938", + "right": "issue:45071", + "accept": false, + "reason": "Python 3.14 load failure and PretrainedConfig type-checking regression are different problems." + }, + { + "left": "issue:43867", + "right": "issue:45375", + "accept": false, + "reason": "A state_dict loading-order problem is not the same as a missing vision-config field being dropped by strict validation." + }, + { + "left": "issue:43867", + "right": "issue:46202", + "accept": false, + "reason": "Model load failure and Gemma token roundtrip failure are unrelated." + }, + { + "left": "issue:36683", + "right": "issue:43064", + "accept": false, + "reason": "Gemma3Config missing vocab_size is unrelated to the FSDP2/PEFT optimizer-state bug." + }, + { + "left": "issue:42890", + "right": "issue:45561", + "accept": false, + "reason": "Both are test failures, but one is seed-related flakiness and the other is an xdist file race." + }, + { + "left": "issue:41669", + "right": "issue:44246", + "accept": false, + "reason": "Both mention import slowness, but one is a specific import-star cleanup proposal and the other is an intermittent import slowdown report without the same concrete cause." + }, + { + "left": "issue:43010", + "right": "issue:45850", + "accept": false, + "reason": "A cache no_grad fix and a broad 'too many models break' complaint are not the same underlying change or bug." + }, + { + "left": "issue:36296", + "right": "issue:43575", + "accept": false, + "reason": "Tensor-parallel training bug and tensor-parallel OOM on model load are distinct symptoms with different likely causes." + }, + { + "left": "issue:43867", + "right": "issue:45357", + "accept": false, + "reason": "Loading error from sorted state_dict keys is not the same as saving incorrect visual encoder keys." + }, + { + "left": "issue:43618", + "right": "issue:44077", + "accept": false, + "reason": "CLIPOutput attentions assignment and patchtsmixer post_init policy are unrelated." + }, + { + "left": "issue:43010", + "right": "issue:43976", + "accept": false, + "reason": "Cache no_grad behavior is unrelated to a Python-version compatibility failure." + }, + { + "left": "issue:43992", + "right": "issue:45478", + "accept": false, + "reason": "Different models and different load errors; not the same duplicate." + }, + { + "left": "issue:28282", + "right": "issue:33453", + "accept": false, + "reason": "Missing PyTorch causes an ImportError, while tokenizer loading regression is a separate issue." + }, + { + "left": "issue:43618", + "right": "issue:46285", + "accept": false, + "reason": "Both are CLIP-related, but one is about attentions fields and the other is an architectural inconsistency in text_model presence." + }, + { + "left": "issue:44360", + "right": "issue:45440", + "accept": false, + "reason": "A DSA/ReLU discussion is unrelated to DeepseekV3MoE divergence from remote implementation." + }, + { + "left": "issue:43618", + "right": "issue:44857", + "accept": false, + "reason": "CLIP attentions handling and a float16 AMP crash in LwDetrImageLoss are unrelated." + }, + { + "left": "issue:42890", + "right": "issue:44964", + "accept": false, + "reason": "Seed-related test flakiness is not the same as a Phi-4 multimodal loading failure." + }, + { + "left": "issue:42890", + "right": "issue:44112", + "accept": false, + "reason": "Both are CI/test issues, but they have different roots and different failing components." + }, + { + "left": "issue:42898", + "right": "issue:46664", + "accept": false, + "reason": "Tokenization cleanup behavior and video num_frames semantics are different API changes." + }, + { + "left": "issue:43502", + "right": "issue:44062", + "accept": false, + "reason": "Offline-loading API requests leaking is unrelated to AddedToken receiving duplicate special kwargs." + }, + { + "left": "issue:43874", + "right": "issue:46617", + "accept": false, + "reason": "A missing image-patch-count method and a cache FileNotFoundError are different load failures." + }, + { + "left": "issue:43723", + "right": "issue:44556", + "accept": false, + "reason": "Both are v5 loading regressions, but one is tokenizer loading and the other is checkpoint reload compatibility; too broad to merge." + }, + { + "left": "issue:44222", + "right": "issue:45125", + "accept": false, + "reason": "FP8 save_pretrained behavior and missing tensor-parallel plan are unrelated." + }, + { + "left": "issue:44734", + "right": "issue:46617", + "accept": false, + "reason": "KV-cache continuation tensor indexing and trust_remote_code cache file lookup 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": 22, + "review_unit_index": 54, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:27957", + "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:43475", + "issue:43502", + "issue:43526", + "issue:43576", + "issue:43824", + "issue:43827", + "issue:43901", + "issue:44222", + "issue:44230", + "issue:44315", + "issue:44351", + "issue:44355", + "issue:44360", + "issue:44387", + "issue:44485", + "issue:44556", + "issue:44655", + "issue:44749", + "issue:44908", + "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" + ], + "soft_pairs": [ + "issue:44387|issue:44556", + "issue:43010|issue:43065", + "issue:43824|issue:46617", + "issue:43502|issue:44485", + "issue:43576|issue:46132", + "issue:45125|issue:45478", + "issue:30333|issue:33453", + "issue:44355|issue:44749", + "issue:45125|issue:45375", + "issue:42548|issue:44351", + "issue:43010|issue:43232", + "issue:30990|issue:31795", + "issue:43502|issue:44222", + "issue:43502|issue:44961", + "issue:43827|issue:44961", + "issue:30333|issue:36296", + "issue:43299|issue:44556", + "issue:44355|issue:45071", + "issue:44355|issue:44655", + "issue:33453|issue:34689", + "issue:27957|issue:33453", + "issue:42757|issue:43576", + "issue:43576|issue:45071", + "issue:44315|issue:44360", + "issue:31795|issue:38617", + "issue:43576|issue:43824", + "issue:44961|issue:45468", + "issue:44485|issue:44961", + "issue:44351|issue:46617", + "issue:44908|issue:44961", + "issue:43576|issue:44938", + "issue:36296|issue:43901", + "issue:46143|issue:46144", + "issue:44222|issue:44230", + "issue:36296|issue:43526", + "issue:43475|issue:46617", + "issue:44315|issue:46209", + "issue:41628|issue:42548", + "issue:44485|issue:46664" + ], + "prepared_review_unit_hash": "9aaba30b02073f4b5300869b96fc9c49eb8793237766bae54a988205265d4910", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 39, + "serialized_chars": 38108, + "estimated_input_tokens": 9527, + "estimated_eval_tokens": 19310 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No convincing duplicate cluster emerges here. The titles overlap mostly at a generic level (v5, loading, import errors, docs), but the underlying bugs and code paths are different, so I would not merge any soft pair.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue is justified: this set spans unrelated regressions across imports, model loading, docs, training, tensor parallelism, FP8, and vision/audio components.", + "canonical_pr_reason": null, + "best_issue_reason": "If forced to pick a representative issue, 43576 is the broadest v5 regression report with the clearest cross-cutting impact, but it still is not a duplicate hub for the rest.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44387", + "right": "issue:44556", + "accept": false, + "reason": "Both mention v5 regressions, but one is int4 OOM/memory growth and the other is checkpoint reload failure; different failure modes and code paths." + }, + { + "left": "issue:43010", + "right": "issue:43065", + "accept": false, + "reason": "Cache no_grad decoration and Sam3PixelDecoder dummy Conv2d are unrelated component bugs." + }, + { + "left": "issue:43824", + "right": "issue:46617", + "accept": false, + "reason": "ImportError for a specific Qwen2.5-VL class is not the same as a FileNotFoundError when loading a trust_remote_code model from cache." + }, + { + "left": "issue:43502", + "right": "issue:44485", + "accept": false, + "reason": "local_files_only network leakage and GLM-5 RoPE implementation are different issues." + }, + { + "left": "issue:43576", + "right": "issue:46132", + "accept": false, + "reason": "Broken env command and AttentionInterface.register behavior are separate CLI vs API problems." + }, + { + "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 from_pretrained loading regression." + }, + { + "left": "issue:30333", + "right": "issue:33453", + "accept": false, + "reason": "MLflow reporting on failed training and tokenizer loading regression are unrelated." + }, + { + "left": "issue:44355", + "right": "issue:44749", + "accept": false, + "reason": "Compiled Python file errors and the Chinese report about slower filtering after upgrade are not the same bug." + }, + { + "left": "issue:45125", + "right": "issue:45375", + "accept": false, + "reason": "Tensor-parallel plan missing and strict-dropping of a Qwen3.5MoeVisionConfig field are different model/config issues." + }, + { + "left": "issue:42548", + "right": "issue:44351", + "accept": false, + "reason": "Missing PreTrainedModel export and missing HybridCache export are distinct import errors." + }, + { + "left": "issue:43010", + "right": "issue:43232", + "accept": false, + "reason": "Both touch generation internals, but cache update no_grad and post-sync_gpus kwargs updating are different code paths." + }, + { + "left": "issue:30990", + "right": "issue:31795", + "accept": false, + "reason": "Sentence Transformers loading hang and documentation confusion about forward arguments are unrelated." + }, + { + "left": "issue:43502", + "right": "issue:44222", + "accept": false, + "reason": "Unexpected API requests with local_files_only and FP8 save_pretrained behavior are different concerns." + }, + { + "left": "issue:43502", + "right": "issue:44961", + "accept": false, + "reason": "A network-usage bug and an empty-title issue are not related." + }, + { + "left": "issue:43827", + "right": "issue:44961", + "accept": false, + "reason": "Docs still referencing pipeline() and the unrelated 'racoon' issue do not match." + }, + { + "left": "issue:30333", + "right": "issue:36296", + "accept": false, + "reason": "MLflow reporting and tensor-parallel training are separate problems." + }, + { + "left": "issue:43299", + "right": "issue:44556", + "accept": false, + "reason": "Both are upgrade/loading regressions, but they affect different models and symptoms, so they are not the same concrete bug." + }, + { + "left": "issue:44355", + "right": "issue:45071", + "accept": false, + "reason": "Compiled-file load errors and PretrainedConfig type-checking breakage are unrelated." + }, + { + "left": "issue:44355", + "right": "issue:44655", + "accept": false, + "reason": "Compiled Python file failures and inability to save Pipeline objects are different issues." + }, + { + "left": "issue:33453", + "right": "issue:34689", + "accept": false, + "reason": "Tokenizer-loading regression and Llama 3.2 Vision model loading breakage are separate load bugs." + }, + { + "left": "issue:27957", + "right": "issue:33453", + "accept": false, + "reason": "Flash Attention 2 support for XLMRoberta is unrelated to tokenizer loading regression." + }, + { + "left": "issue:42757", + "right": "issue:43576", + "accept": false, + "reason": "Missing huggingface_hub export and broken transformers env command are different API/CLI regressions." + }, + { + "left": "issue:43576", + "right": "issue:45071", + "accept": false, + "reason": "Broken env command and PretrainedConfig type checking are unrelated v5 issues." + }, + { + "left": "issue:44315", + "right": "issue:44360", + "accept": false, + "reason": "Liger Kernel application in model_init and DSA indexer ReLU discussion are unrelated." + }, + { + "left": "issue:31795", + "right": "issue:38617", + "accept": false, + "reason": "Documentation confusion and an import error for layer_type_validation are not the same bug." + }, + { + "left": "issue:43576", + "right": "issue:43824", + "accept": false, + "reason": "CLI env regression and Qwen2.5-VL import error are different failures." + }, + { + "left": "issue:44961", + "right": "issue:45468", + "accept": false, + "reason": "The 'racoon' issue contains no evidence of sharing the Gemma-4 audio positional encoding bug." + }, + { + "left": "issue:44485", + "right": "issue:44961", + "accept": false, + "reason": "GLM-5 RoPE implementation and the unrelated 'racoon' issue are not duplicates." + }, + { + "left": "issue:44351", + "right": "issue:46617", + "accept": false, + "reason": "Importing HybridCache from transformers and loading a trust_remote_code model from cache are different error classes and code paths." + }, + { + "left": "issue:44908", + "right": "issue:44961", + "accept": false, + "reason": "Scheduler kwarg handling and the unrelated 'racoon' issue do not match." + }, + { + "left": "issue:43576", + "right": "issue:44938", + "accept": false, + "reason": "Broken env command in v5 and Python 3.14 load failure are distinct compatibility issues." + }, + { + "left": "issue:36296", + "right": "issue:43901", + "accept": false, + "reason": "Tensor parallel training bug and docs mentioning return_all_scores are unrelated." + }, + { + "left": "issue:46143", + "right": "issue:46144", + "accept": false, + "reason": "Both are RoFormer-related, but one is about kwargs propagation and the other about attention-interface usage; not the same concrete bug." + }, + { + "left": "issue:44222", + "right": "issue:44230", + "accept": false, + "reason": "FP8 save_pretrained/moe and FP8 inference support for Qwen3-VL/Qwen3.5 MoE are related theme-wise but not the same code-path problem." + }, + { + "left": "issue:36296", + "right": "issue:43526", + "accept": false, + "reason": "Tensor parallel training and BeitImageProcessorFast label reduction are unrelated." + }, + { + "left": "issue:43475", + "right": "issue:46617", + "accept": false, + "reason": "Missing Sam3VisionEncoderOutput attribute and cache FileNotFoundError are different bugs." + }, + { + "left": "issue:44315", + "right": "issue:46209", + "accept": false, + "reason": "Liger Kernel application and FP8 inference on B200 are unrelated issues." + }, + { + "left": "issue:41628", + "right": "issue:42548", + "accept": false, + "reason": "AutoImageProcessor import failure and PreTrainedModel import failure are different missing exports." + }, + { + "left": "issue:44485", + "right": "issue:46664", + "accept": false, + "reason": "GLM-5 RoPE implementation and SmolVLM num_frames semantics 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": 22, + "review_unit_index": 55, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:31795", + "issue:33453", + "issue:41669", + "issue:42371", + "issue:42890", + "issue:43352", + "issue:43408", + "issue:43525", + "issue:43572", + "issue:43576", + "issue:43673", + "issue:43827", + "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: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:45381", + "issue:45478", + "issue:45542", + "issue:45753", + "issue:45850", + "issue:46095", + "issue:46218", + "issue:46328", + "issue:46664" + ], + "soft_pairs": [ + "issue:45362|issue:45850", + "issue:43673|issue:44492", + "issue:44273|issue:45542", + "issue:44360|issue:46664", + "issue:44623|issue:45230", + "issue:44038|issue:45125", + "issue:44273|issue:45003", + "issue:44393|issue:45753", + "issue:44393|issue:45071", + "issue:45381|issue:46664", + "issue:43576|issue:44556", + "issue:41669|issue:43673", + "issue:43525|issue:45850", + "issue:44556|issue:45092", + "issue:43572|issue:45850", + "issue:45278|issue:45850", + "issue:44230|issue:46095", + "issue:44492|issue:45850", + "issue:42371|issue:44492", + "issue:41669|issue:46218", + "issue:42890|issue:44908", + "issue:43408|issue:44315", + "issue:44246|issue:44273", + "issue:43352|issue:44393", + "issue:43576|issue:44908", + "issue:44704|issue:44908", + "issue:43976|issue:44393", + "issue:44908|issue:45071", + "issue:44295|issue:45230", + "issue:44297|issue:45230", + "issue:41669|issue:44351", + "issue:44492|issue:46328", + "issue:41669|issue:45341", + "issue:43976|issue:45850", + "issue:44393|issue:45478", + "issue:43976|issue:46218", + "issue:44077|issue:44861", + "issue:43827|issue:44908", + "issue:31795|issue:33453", + "issue:44315|issue:44829", + "issue:43576|issue:44273" + ], + "prepared_review_unit_hash": "a49ae9501034ade8cae99cc43ca1f6a338b074cf3548ac6354887173573d5cbd", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 41, + "serialized_chars": 39443, + "estimated_input_tokens": 9861, + "estimated_eval_tokens": 19978 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a mix of unrelated Transformers v5 regressions, model-specific failures, and import/load issues. I do not see any soft pair that is clearly the same underlying bug/change, so I would not merge any of the candidate edges.", + "confidence": 0.62, + "canonical_issue_reason": "issue:45850 is the broadest umbrella report ('Since 5.0 version it breaks too many models') and best reflects the shared v5-regression theme across the cluster, even though it is not a strict duplicate of the others.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45850 is the best cluster representative because it is the most general summary of the widespread v5 breakage reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45362", + "right": "issue:45850", + "accept": false, + "reason": "Generic v5 breakage is too broad; this is a specific Qwen3.5 chat crash, not the same bug." + }, + { + "left": "issue:43673", + "right": "issue:44492", + "accept": false, + "reason": "Chunked-prefill cache failure and a cache-strategy typo are different issues." + }, + { + "left": "issue:44273", + "right": "issue:45542", + "accept": false, + "reason": "Lazy loading regressions and tensorboard-without-TF backend errors are unrelated." + }, + { + "left": "issue:44360", + "right": "issue:46664", + "accept": false, + "reason": "DSA indexer/ReLU discussion is unrelated to SmolVLM video frame semantics." + }, + { + "left": "issue:44623", + "right": "issue:45230", + "accept": false, + "reason": "Missing processor save files is not the same as a vague bug report." + }, + { + "left": "issue:44038", + "right": "issue:45125", + "accept": false, + "reason": "Both involve Qwen3/MoE, but one is a model bug and the other is tensor-parallel config support." + }, + { + "left": "issue:44273", + "right": "issue:45003", + "accept": false, + "reason": "Lazy loading and unsafe sys.modules access are different failure modes." + }, + { + "left": "issue:44393", + "right": "issue:45753", + "accept": false, + "reason": "Both are Qwen3-VL related, but one is 2D box output hallucination and the other is an infinite loop on a specific image." + }, + { + "left": "issue:44393", + "right": "issue:45071", + "accept": false, + "reason": "A vision-model output bug is not the same as a PretrainedConfig type-checking regression." + }, + { + "left": "issue:45381", + "right": "issue:46664", + "accept": false, + "reason": "Video position-id corruption and num_frames semantics are distinct processor bugs." + }, + { + "left": "issue:43576", + "right": "issue:44556", + "accept": false, + "reason": "Broken env command and checkpoint reload incompatibility are different v5 regressions." + }, + { + "left": "issue:41669", + "right": "issue:43673", + "accept": false, + "reason": "Import-star/lazy-loading performance issues are not the same as cache-missing during chunked prefill." + }, + { + "left": "issue:43525", + "right": "issue:45850", + "accept": false, + "reason": "Missing pad_token_id on Llama4Config is a specific config bug, not the broad v5 breakage report." + }, + { + "left": "issue:44556", + "right": "issue:45092", + "accept": false, + "reason": "Both mention v5 compatibility, but one is generic checkpoint reload failure and the other is remote-code/meta-init incompatibility." + }, + { + "left": "issue:43572", + "right": "issue:45850", + "accept": false, + "reason": "StableLmConfig pad_token_idx missing is a model-specific config regression, not the umbrella v5 report." + }, + { + "left": "issue:45278", + "right": "issue:45850", + "accept": false, + "reason": "General import errors after upgrade are broader, but still not the same concrete issue as the umbrella complaint." + }, + { + "left": "issue:44230", + "right": "issue:46095", + "accept": false, + "reason": "FP8 inference support for Qwen models is unrelated to a DeepSeek weight-conversion script request." + }, + { + "left": "issue:44492", + "right": "issue:45850", + "accept": false, + "reason": "A cache-strategy typo is not the same as broad v5 breakage." + }, + { + "left": "issue:42371", + "right": "issue:44492", + "accept": false, + "reason": "TF32 API guidance and a cache-strategy typo have no direct overlap." + }, + { + "left": "issue:41669", + "right": "issue:46218", + "accept": false, + "reason": "Import-star removal/import slowdown is not the same as a specific Qwen3_5ForCausalLM import failure." + }, + { + "left": "issue:42890", + "right": "issue:44908", + "accept": false, + "reason": "Missing set_seed in tests and inverse_sqrt scheduler kwargs are unrelated." + }, + { + "left": "issue:43408", + "right": "issue:44315", + "accept": false, + "reason": "A model-type warning and Liger Kernel application in model_init are different problems." + }, + { + "left": "issue:44246", + "right": "issue:44273", + "accept": false, + "reason": "Import performance and lazy-loading correctness are related only loosely, not the same bug." + }, + { + "left": "issue:43352", + "right": "issue:44393", + "accept": false, + "reason": "Flash Attention 2 support for Nemotron is unrelated to Qwen3-VL 2D box output errors." + }, + { + "left": "issue:43576", + "right": "issue:44908", + "accept": false, + "reason": "Environment CLI breakage and scheduler kwargs handling are different code paths." + }, + { + "left": "issue:44704", + "right": "issue:44908", + "accept": false, + "reason": "AutoProcessor cached_file kwargs and scheduler kwargs are unrelated." + }, + { + "left": "issue:43976", + "right": "issue:44393", + "accept": false, + "reason": "Python version support regression is unrelated to a vision-model output bug." + }, + { + "left": "issue:44908", + "right": "issue:45071", + "accept": false, + "reason": "Scheduler kwargs and config type checking do not share the same underlying defect." + }, + { + "left": "issue:44295", + "right": "issue:45230", + "accept": false, + "reason": "Position_ids buffer handling is not the same as an unspecified bug report." + }, + { + "left": "issue:44297", + "right": "issue:45230", + "accept": false, + "reason": "Tokenizer save metadata mismatch is a specific bug, not a generic report." + }, + { + "left": "issue:41669", + "right": "issue:44351", + "accept": false, + "reason": "Import-star/lazy-loading performance problems are not the same as missing HybridCache exports." + }, + { + "left": "issue:44492", + "right": "issue:46328", + "accept": false, + "reason": "Cache-strategy typo and SequenceFeatureExtractor.pad inefficiency are unrelated." + }, + { + "left": "issue:41669", + "right": "issue:45341", + "accept": false, + "reason": "Import-time slowdown is unrelated to a testing_utils bug." + }, + { + "left": "issue:43976", + "right": "issue:45850", + "accept": false, + "reason": "Python compatibility regression is not the same as generic v5 model breakage." + }, + { + "left": "issue:44393", + "right": "issue:45478", + "accept": false, + "reason": "Qwen3-VL bounding-box hallucination is not the same as Qwen3.5 MoE from_pretrained failure." + }, + { + "left": "issue:43976", + "right": "issue:46218", + "accept": false, + "reason": "Python support issues are unrelated to a missing Qwen3_5ForCausalLM import." + }, + { + "left": "issue:44077", + "right": "issue:44861", + "accept": false, + "reason": "Optional post_init policy and tied-weight-key attribute errors are different model-internals issues." + }, + { + "left": "issue:43827", + "right": "issue:44908", + "accept": false, + "reason": "Docs referencing pipeline() and scheduler kwargs handling are unrelated." + }, + { + "left": "issue:31795", + "right": "issue:33453", + "accept": false, + "reason": "Documentation confusion and tokenizer loading regression are different issues." + }, + { + "left": "issue:44315", + "right": "issue:44829", + "accept": false, + "reason": "Liger Kernel not applied in model_init and flash_attention_3 training degeneration are unrelated." + }, + { + "left": "issue:43576", + "right": "issue:44273", + "accept": false, + "reason": "Broken env command and lazy loading correctness are different v5-related regressions, not one bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 22, + "review_unit_index": 56, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:28348", + "issue:30333", + "issue:31515", + "issue:31795", + "issue:33453", + "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:44230", + "issue:44261", + "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:46172", + "issue:46218", + "issue:46655" + ], + "soft_pairs": [ + "issue:43519|issue:46655", + "issue:44360|issue:45468", + "issue:44655|issue:44861", + "issue:44273|issue:44908", + "issue:45230|issue:45310", + "issue:43576|issue:46218", + "issue:43519|issue:44485", + "issue:43519|issue:44908", + "issue:41669|issue:43704", + "issue:46018|issue:46172", + "issue:31795|issue:34689", + "issue:31515|issue:31795", + "issue:43704|issue:44485", + "issue:44485|issue:45468", + "issue:43502|issue:43519", + "issue:42371|issue:43704", + "issue:45850|issue:46218", + "issue:43519|issue:44623", + "issue:43606|issue:45468", + "issue:44263|issue:45468", + "issue:44261|issue:45468", + "issue:43298|issue:46095", + "issue:43519|issue:43704", + "issue:44351|issue:46218", + "issue:43352|issue:46218", + "issue:42913|issue:43066", + "issue:28348|issue:45250", + "issue:44908|issue:44998", + "issue:44492|issue:44998", + "issue:43618|issue:44998", + "issue:44998|issue:45850", + "issue:44938|issue:44998", + "issue:44861|issue:45290", + "issue:31515|issue:38617", + "issue:43525|issue:44861", + "issue:43572|issue:44861", + "issue:28348|issue:43352", + "issue:44230|issue:44623", + "issue:30333|issue:31515", + "issue:31515|issue:36296", + "issue:44230|issue:45310", + "issue:31515|issue:33453", + "issue:43298|issue:44230" + ], + "prepared_review_unit_hash": "612e541b6a4ad3789237a84ef682e150478c3b9c0c9f17d05296cf3f70e2a5f5", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 43, + "serialized_chars": 39011, + "estimated_input_tokens": 9753, + "estimated_eval_tokens": 19762 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No soft edge looks like a true duplicate: the set is a heterogeneous mix of model-specific bugs, loading regressions, docs issues, feature requests, and unrelated failures. I would not merge any of the suggested pairs.", + "confidence": 0.95, + "canonical_issue_reason": "issue:33453 is the clearest concrete regression among the loading-related subset, but the overall cluster is too heterogeneous for a strong canonical beyond that.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:33453 is the best available anchor because it is a concrete, self-contained loading regression and more reusable than the model-specific or unrelated reports; however, it is still only a loose fit for the full cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43519", + "right": "issue:46655", + "accept": false, + "reason": "Different processor bugs in different models: timestamp calculation vs padding calculation." + }, + { + "left": "issue:44360", + "right": "issue:45468", + "accept": false, + "reason": "Unrelated subsystems and failures: DSA indexer ReLU vs Gemma-4 audio positional encoding." + }, + { + "left": "issue:44655", + "right": "issue:44861", + "accept": false, + "reason": "Different code paths: saving Pipeline objects vs _get_tied_weight_keys AttributeError." + }, + { + "left": "issue:44273", + "right": "issue:44908", + "accept": false, + "reason": "Lazy loading behavior is unrelated to inverse_sqrt scheduler kwargs handling." + }, + { + "left": "issue:45230", + "right": "issue:45310", + "accept": false, + "reason": "Generic bug report vs a specific Qwen3.5 MoE from_pretrained error; no shared concrete bug." + }, + { + "left": "issue:43576", + "right": "issue:46218", + "accept": false, + "reason": "CLI env command breakage is not the same as a model import failure." + }, + { + "left": "issue:43519", + "right": "issue:44485", + "accept": false, + "reason": "Different models and bugs: Qwen3VL timestamp logic vs GLM-5 RoPE implementation." + }, + { + "left": "issue:43519", + "right": "issue:44908", + "accept": false, + "reason": "Processor timestamp math is unrelated to scheduler parameter plumbing." + }, + { + "left": "issue:41669", + "right": "issue:43704", + "accept": false, + "reason": "Import-* cleanup/perf issue is unrelated to the VRAM leak in dataloader threads." + }, + { + "left": "issue:46018", + "right": "issue:46172", + "accept": false, + "reason": "DeepSeek-V4 shared-expert gating and a tensor transpose bug are different code-path problems." + }, + { + "left": "issue:31795", + "right": "issue:34689", + "accept": false, + "reason": "Documentation confusion is not the same as a model-loading regression." + }, + { + "left": "issue:31515", + "right": "issue:31795", + "accept": false, + "reason": "Loading slowdown and documentation ambiguity are different issues." + }, + { + "left": "issue:43704", + "right": "issue:44485", + "accept": false, + "reason": "VRAM leak in multi-threaded inference does not match a RoPE implementation discussion." + }, + { + "left": "issue:44485", + "right": "issue:45468", + "accept": false, + "reason": "Different models and different math paths; not the same bug." + }, + { + "left": "issue:43502", + "right": "issue:43519", + "accept": false, + "reason": "local_files_only network leakage is unrelated to timestamp calculation." + }, + { + "left": "issue:42371", + "right": "issue:43704", + "accept": false, + "reason": "TF32 API guidance and a VRAM leak are unrelated." + }, + { + "left": "issue:45850", + "right": "issue:46218", + "accept": false, + "reason": "Too broad/vague v5 breakage is not the same concrete bug as a Qwen import error." + }, + { + "left": "issue:43519", + "right": "issue:44623", + "accept": false, + "reason": "Timestamp calculation and missing save_pretrained files are different failures." + }, + { + "left": "issue:43606", + "right": "issue:45468", + "accept": false, + "reason": "CPU offload device mismatch is unrelated to Gemma-4 audio positional encoding." + }, + { + "left": "issue:44263", + "right": "issue:45468", + "accept": false, + "reason": "torch.split return values bug does not match Gemma-4 audio encoding." + }, + { + "left": "issue:44261", + "right": "issue:45468", + "accept": false, + "reason": "Precision error in MLA layernorm is unrelated to Gemma-4 audio positional encoding." + }, + { + "left": "issue:43298", + "right": "issue:46095", + "accept": false, + "reason": "Continuous batching audio support and a DeepSeek weight conversion question are different feature areas." + }, + { + "left": "issue:43519", + "right": "issue:43704", + "accept": false, + "reason": "Timestamp calculation bug and VRAM leak are unrelated." + }, + { + "left": "issue:44351", + "right": "issue:46218", + "accept": false, + "reason": "Both are import errors, but for different symbols/modules and likely different fixes." + }, + { + "left": "issue:43352", + "right": "issue:46218", + "accept": false, + "reason": "Flash Attention 2.0 support for Nemotron is unrelated to a Qwen import failure." + }, + { + "left": "issue:42913", + "right": "issue:43066", + "accept": false, + "reason": "Both concern tokenizer behavior, but one is a behavior difference and the other a wrong decoder type; not the same bug." + }, + { + "left": "issue:28348", + "right": "issue:45250", + "accept": false, + "reason": "Specific GPT2 Flash Attention 2.0 support request is not the same as a generic Flash Attention 2.0 issue." + }, + { + "left": "issue:44908", + "right": "issue:44998", + "accept": false, + "reason": "Scheduler kwargs bug is unrelated to the spam/invalid report." + }, + { + "left": "issue:44492", + "right": "issue:44998", + "accept": false, + "reason": "Typo in cache strategies is unrelated to the invalid title." + }, + { + "left": "issue:43618", + "right": "issue:44998", + "accept": false, + "reason": "CLIPOutput attentions regression is unrelated to the invalid report." + }, + { + "left": "issue:44998", + "right": "issue:45850", + "accept": false, + "reason": "Invalid/spam report and a broad v5 breakage complaint are not duplicates." + }, + { + "left": "issue:44938", + "right": "issue:44998", + "accept": false, + "reason": "Python 3.14 load failure is unrelated to the invalid report." + }, + { + "left": "issue:44861", + "right": "issue:45290", + "accept": false, + "reason": "Tied-weight key crash and apply_chat_template tool-call crash are different code paths." + }, + { + "left": "issue:31515", + "right": "issue:38617", + "accept": false, + "reason": "Slow from_pretrained loading and a missing import name are different problems." + }, + { + "left": "issue:43525", + "right": "issue:44861", + "accept": false, + "reason": "Missing pad_token_id in Llama4Config is unrelated to tied-weight key handling." + }, + { + "left": "issue:28348", + "right": "issue:43352", + "accept": false, + "reason": "Both mention Flash Attention 2.0, but they are different models and distinct support gaps." + }, + { + "left": "issue:44230", + "right": "issue:44623", + "accept": false, + "reason": "FP8 support request is unrelated to missing processor.save_pretrained files." + }, + { + "left": "issue:30333", + "right": "issue:31515", + "accept": false, + "reason": "MLFlow reporting on failed jobs is unrelated to checkpoint loading performance." + }, + { + "left": "issue:31515", + "right": "issue:36296", + "accept": false, + "reason": "Loading slowness and tensor-parallel training bugs are not the same issue." + }, + { + "left": "issue:44230", + "right": "issue:45310", + "accept": false, + "reason": "FP8 support request and Qwen3.5 MoE from_pretrained failure are different problems." + }, + { + "left": "issue:31515", + "right": "issue:33453", + "accept": false, + "reason": "Both are loading-related, but one is slow checkpoint loading and the other is a tokenizer-loading regression; not the same bug." + }, + { + "left": "issue:43298", + "right": "issue:44230", + "accept": false, + "reason": "Continuous batching for audio models and FP8 support 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": 22, + "review_unit_index": 57, + "review_unit_count": 57, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-476", + "nodes": [ + "issue:37934", + "issue:42936", + "issue:43240", + "issue:43668", + "issue:43717", + "issue:43813", + "issue:44031", + "issue:46018", + "issue:46144" + ], + "soft_pairs": [ + "issue:43717|issue:46144", + "issue:42936|issue:46018", + "issue:43668|issue:44031", + "issue:42936|issue:43240", + "issue:43240|issue:46144", + "issue:42936|issue:44031", + "issue:42936|issue:46144", + "issue:42936|issue:43813", + "issue:43240|issue:43668", + "issue:43668|issue:43813", + "issue:37934|issue:43668", + "issue:42936|issue:43668" + ], + "prepared_review_unit_hash": "1c55e55ec313330ef91410557d0892bb6d6ca92a284e6a9892e5a8bfc53ed1c2", + "component_budget": { + "node_count": 476, + "item_count": 476, + "soft_pair_count": 1755, + "serialized_chars": 909280, + "estimated_input_tokens": 227320, + "estimated_eval_tokens": 454896 + }, + "budget": { + "node_count": 9, + "item_count": 9, + "soft_pair_count": 12, + "serialized_chars": 8610, + "estimated_input_tokens": 2153, + "estimated_eval_tokens": 4562 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is not a true duplicate cluster: the issues span unrelated models, configs, tokenizer warnings, and small correctness/typing fixes. None of the soft pairs describe the same underlying bug or change.", + "confidence": 0.96, + "canonical_issue_reason": "Issue #42936 is the strongest standalone bug report in the set: it describes a concrete implementation omission with an active discussion, while the other items are unrelated to it.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #42936 is the best anchor for the cluster because it is the most concrete and actionable report; the rest are mostly unrelated or narrower cleanup items.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43717", + "right": "issue:46144", + "accept": false, + "reason": "Mamba/Mamba-2 init_weights behavior and RoFormer attention interface are different code paths and bugs." + }, + { + "left": "issue:42936", + "right": "issue:46018", + "accept": false, + "reason": "Mask2Former ignore_value handling and DeepSeek-V4 shared-expert gating are unrelated model issues." + }, + { + "left": "issue:43668", + "right": "issue:44031", + "accept": false, + "reason": "A type-hint mistake in ModernBERTConfig is not the same as tokenizer regex warning behavior." + }, + { + "left": "issue:42936", + "right": "issue:43240", + "accept": false, + "reason": "One is about Mask2Former ignore_value usage; the other is about kwargs not reaching a loss function." + }, + { + "left": "issue:43240", + "right": "issue:46144", + "accept": false, + "reason": "Loss-kwargs propagation and RoFormer attention interface are different bugs in different subsystems." + }, + { + "left": "issue:42936", + "right": "issue:44031", + "accept": false, + "reason": "Mask2Former loss masking and tokenizer regex warnings are unrelated failure modes." + }, + { + "left": "issue:42936", + "right": "issue:46144", + "accept": false, + "reason": "Mask2Former ignore_value and RoFormer attention interface do not describe the same underlying problem." + }, + { + "left": "issue:42936", + "right": "issue:43813", + "accept": false, + "reason": "A Mask2Former missing-use bug is unrelated to a quantization typo in another area." + }, + { + "left": "issue:43240", + "right": "issue:43668", + "accept": false, + "reason": "Passing kwargs to loss code and fixing an incorrect type hint are not the same change." + }, + { + "left": "issue:43668", + "right": "issue:43813", + "accept": false, + "reason": "ModernBERTConfig typing and a quantization-operation typo are distinct cleanup issues." + }, + { + "left": "issue:37934", + "right": "issue:43668", + "accept": false, + "reason": "Llama4TextL2Norm RMS-norm behavior and ModernBERTConfig norm_eps typing concern different components and symptoms." + }, + { + "left": "issue:42936", + "right": "issue:43668", + "accept": false, + "reason": "Mask2Former ignore_value usage and ModernBERTConfig norm_eps typing are unrelated issues." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 22, + "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": "455ecc031685a3d30704ea0885304de515496f1de3785c20ab7fce581cf7338a", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 2877, + "estimated_input_tokens": 720, + "estimated_eval_tokens": 1696 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 2877, + "estimated_input_tokens": 720, + "estimated_eval_tokens": 1696 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The three issues are about unrelated feature requests: tokenizer truncation behavior, ReduceLROnPlateau-like scheduling, and Wav2Vec2 word time offsets. There is no duplicate cluster here.", + "confidence": 0.98, + "canonical_issue_reason": "No canonical issue: the items describe different products/features and do not share the same underlying bug or change.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 10519 is the most concrete and self-contained request, but it is not a duplicate of the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:10519", + "right": "issue:11307", + "accept": false, + "reason": "Tokenizer truncation direction and Wav2Vec2 word timestamp extraction are unrelated features in different code paths." + }, + { + "left": "issue:10519", + "right": "issue:11005", + "accept": false, + "reason": "Tokenizer truncation behavior and learning-rate scheduler functionality are completely different requests." + }, + { + "left": "issue:11005", + "right": "issue:11307", + "accept": false, + "reason": "ReduceLROnPlateau-like training control and speech word boundary offsets do not describe the same problem." + } + ] + }, + "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": 22, + "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": "d4e0e7b79e724df758e2fa16f9f2872655bbb1b9d12b599fc03601d3bbd46598", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 2478, + "estimated_input_tokens": 620, + "estimated_eval_tokens": 1496 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 2478, + "estimated_input_tokens": 620, + "estimated_eval_tokens": 1496 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster does not look like a true duplicate set: one issue is about Tapas tokenization, one about private-model auth/download errors, and one about loading bert-base-uncased. The only partial overlap is between the two model-loading issues, but their titles point to different failure modes.", + "confidence": 0.89, + "canonical_issue_reason": "16618 is the broadest and most central model-loading report, but it still appears distinct from the private-download/auth issue and the Tapas tokenization bug.", + "canonical_pr_reason": null, + "best_issue_reason": "16618 is the best anchor for the cluster because it is the most general and most discussed issue, even though it is not a true duplicate of the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:13244", + "right": "issue:16618", + "accept": false, + "reason": "Tapas tokenization mismatches are unrelated to a failure to load bert-base-uncased; different subsystem and different bug." + }, + { + "left": "issue:15980", + "right": "issue:16618", + "accept": false, + "reason": "Both involve model loading/downloads, but one is specifically private-model access without login while the other is a generic bert-base-uncased load failure; 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": 4, + "component_count": 22, + "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": "db645671ddd09d1f8db6428429ce66a4b8c6a9b50f16a2fb4c2a7e1e89dcc76f", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1590, + "estimated_input_tokens": 398, + "estimated_eval_tokens": 1052 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1590, + "estimated_input_tokens": 398, + "estimated_eval_tokens": 1052 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are two unrelated feature-request issues: one for Flash Attention 2 support on more architectures, the other for GGUF support on more architectures. They share only a broad \u201ccommunity contribution\u201d framing, not the same underlying bug or change.", + "confidence": 0.98, + "canonical_issue_reason": "Issue 26350 is the clearer canonical issue because it was created earlier and has substantially more discussion, indicating it is the more established thread for its own feature request.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 26350 is the best issue representative for the cluster only in the sense that it is the older, more active thread; however, the cluster is not truly duplicate-worthy since the requests are different.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:26350", + "right": "issue:33260", + "accept": false, + "reason": "Reject: these request different capabilities for different architectures (Flash Attention 2 vs GGUF). They are not the same underlying bug/change and should not be merged." + } + ] + }, + "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": 22, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-27088-8", + "nodes": [ + "issue:27088", + "issue:28936", + "issue:38435", + "issue:43834", + "issue:44365", + "issue:44370", + "issue:46222", + "issue:46226" + ], + "soft_pairs": [ + "issue:27088|issue:44370", + "issue:27088|issue:44365", + "issue:27088|issue:43834", + "issue:27088|issue:38435", + "issue:27088|issue:28936", + "issue:28936|issue:43834", + "issue:28936|issue:44365", + "issue:28936|issue:38435", + "issue:28936|issue:44370", + "issue:38435|issue:44370", + "issue:38435|issue:44365", + "issue:38435|issue:43834", + "issue:38435|issue:46222", + "issue:38435|issue:46226", + "issue:43834|issue:44370", + "issue:43834|issue:44365", + "issue:43834|issue:46222", + "issue:43834|issue:46226", + "issue:44365|issue:44370", + "issue:44365|issue:46222", + "issue:44365|issue:46226", + "issue:44370|issue:46222", + "issue:44370|issue:46226", + "issue:46222|issue:46226" + ], + "prepared_review_unit_hash": "4cd1acf5ac6a5bfdca6518cd29c9bcf6818f4eabc01886a36779f33b5baea9bf", + "component_budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 24, + "serialized_chars": 13527, + "estimated_input_tokens": 3382, + "estimated_eval_tokens": 7020 + }, + "budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 24, + "serialized_chars": 13527, + "estimated_input_tokens": 3382, + "estimated_eval_tokens": 7020 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster splits into two real duplicate pockets: the English placeholder-title issues and the identical Thai-title pair. The language-specific translation requests remain separate, and cross-pocket similarity is not enough to merge them into one duplicate set.", + "confidence": 0.9, + "canonical_issue_reason": "No single canonical issue for the whole cluster: the evidence supports two separate duplicate pockets rather than one unified set. If a representative is needed, it should be chosen within the English placeholder pocket, not across pockets.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43834 is the best global representative because it is the earliest item in the strongest English placeholder subgroup and matches the other placeholder issues closely; the Thai pair should be handled as a separate duplicate pocket.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:27088", + "right": "issue:44370", + "accept": false, + "reason": "Turkish translation request vs generic placeholder template; similar theme, but not the same issue." + }, + { + "left": "issue:27088", + "right": "issue:44365", + "accept": false, + "reason": "Different concrete language request versus placeholder title; no evidence of a true duplicate." + }, + { + "left": "issue:27088", + "right": "issue:43834", + "accept": false, + "reason": "Turkish request is distinct from the placeholder subgroup and should not be merged with it." + }, + { + "left": "issue:27088", + "right": "issue:38435", + "accept": false, + "reason": "Different target languages (Turkish vs Romanian), so these are separate requests." + }, + { + "left": "issue:27088", + "right": "issue:28936", + "accept": false, + "reason": "Different target languages (Turkish vs Spanish), so these are separate requests." + }, + { + "left": "issue:28936", + "right": "issue:43834", + "accept": false, + "reason": "Spanish translation request versus placeholder template; related area, but not the same request." + }, + { + "left": "issue:28936", + "right": "issue:44365", + "accept": false, + "reason": "Spanish translation request versus placeholder template; not enough to call them duplicates." + }, + { + "left": "issue:28936", + "right": "issue:38435", + "accept": false, + "reason": "Different target languages (Spanish vs Romanian), so these are separate issues." + }, + { + "left": "issue:28936", + "right": "issue:44370", + "accept": false, + "reason": "Spanish translation request versus placeholder template; similar wording alone is insufficient." + }, + { + "left": "issue:38435", + "right": "issue:44370", + "accept": false, + "reason": "Romanian request versus placeholder issue; same broad theme, but not the same concrete task." + }, + { + "left": "issue:38435", + "right": "issue:44365", + "accept": false, + "reason": "Romanian request versus placeholder issue; no direct evidence they are duplicates." + }, + { + "left": "issue:38435", + "right": "issue:43834", + "accept": false, + "reason": "Romanian-specific request is distinct from the generic placeholder subgroup." + }, + { + "left": "issue:38435", + "right": "issue:46222", + "accept": false, + "reason": "Romanian request versus Thai-language issue; soft similarity is not enough to merge." + }, + { + "left": "issue:38435", + "right": "issue:46226", + "accept": false, + "reason": "Romanian request versus Thai-language issue; same broad topic, different concrete issue." + }, + { + "left": "issue:43834", + "right": "issue:44370", + "accept": true, + "reason": "Near-identical placeholder translation issues with the same template and highly similar bodies." + }, + { + "left": "issue:43834", + "right": "issue:44365", + "accept": true, + "reason": "Same placeholder-title template and almost identical body content; same underlying issue." + }, + { + "left": "issue:43834", + "right": "issue:46222", + "accept": false, + "reason": "Different duplicate pocket: English placeholder template versus Thai title; similarity alone is not enough." + }, + { + "left": "issue:43834", + "right": "issue:46226", + "accept": false, + "reason": "Different duplicate pocket: English placeholder template versus Thai title; treat as separate cluster." + }, + { + "left": "issue:44365", + "right": "issue:44370", + "accept": true, + "reason": "Same placeholder subgroup with nearly identical content; clearly the same underlying request." + }, + { + "left": "issue:44365", + "right": "issue:46222", + "accept": false, + "reason": "Cross-pocket match between English placeholder and Thai issue; too weak to merge conservatively." + }, + { + "left": "issue:44365", + "right": "issue:46226", + "accept": false, + "reason": "Cross-pocket match between English placeholder and Thai issue; not enough evidence for a duplicate." + }, + { + "left": "issue:44370", + "right": "issue:46222", + "accept": false, + "reason": "Cross-pocket match between English placeholder and Thai issue; identical body length is not sufficient." + }, + { + "left": "issue:44370", + "right": "issue:46226", + "accept": false, + "reason": "Cross-pocket match between English placeholder and Thai issue; same theme, different duplicate pocket." + }, + { + "left": "issue:46222", + "right": "issue:46226", + "accept": true, + "reason": "Exact same Thai title and identical body length; strongest duplicate match in the cluster." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The cluster is only well-supported as two separate near-duplicate pockets: the English placeholder-title issues (43834/44365/44370) and the identical Thai-title pair (46222/46226). The accepted cross-pairs between those two pockets overstate the evidence; the packet only shows soft similarity and identical body lengths, not enough to conservatively treat them as the same duplicate set. The language-specific issues (27088/28936/38435) should remain separate as you noted, but the cross-subgroup duplicate verdicts should be rejected." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 6, + "component_count": 22, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-36817-18", + "nodes": [ + "issue:36817", + "issue:36879", + "issue:37015", + "issue:41115", + "issue:41211", + "issue:42503", + "issue:42738", + "issue:43671", + "issue:43837", + "issue:43909", + "issue:44121", + "issue:44205", + "issue:45027", + "issue:45177", + "issue:45322", + "issue:46133", + "issue:46645", + "issue:46653" + ], + "soft_pairs": [ + "issue:36817|issue:41211", + "issue:36817|issue:43671", + "issue:36817|issue:36879", + "issue:36879|issue:41211", + "issue:36879|issue:43909", + "issue:36879|issue:44121", + "issue:36879|issue:43837", + "issue:36879|issue:43671", + "issue:37015|issue:43837", + "issue:37015|issue:44121", + "issue:37015|issue:43909", + "issue:36879|issue:37015", + "issue:37015|issue:42503", + "issue:41115|issue:43837", + "issue:41115|issue:44121", + "issue:41115|issue:42503", + "issue:41115|issue:45027", + "issue:41211|issue:43671", + "issue:42503|issue:43837", + "issue:42503|issue:43671", + "issue:42503|issue:42738", + "issue:42738|issue:45027", + "issue:42738|issue:46653", + "issue:43671|issue:43837", + "issue:43671|issue:45027", + "issue:43671|issue:45177", + "issue:43671|issue:44121", + "issue:43671|issue:43909", + "issue:43837|issue:43909", + "issue:43837|issue:45322", + "issue:43837|issue:44121", + "issue:43909|issue:44205", + "issue:43909|issue:46645", + "issue:44121|issue:45322", + "issue:43837|issue:45027", + "issue:37015|issue:45027", + "issue:43837|issue:45177", + "issue:45322|issue:46133", + "issue:42503|issue:46653", + "issue:45027|issue:46653" + ], + "prepared_review_unit_hash": "9e81a3001fc65823eac29f12ef2e6f1563885c092f93eb53f9c71bb533a69e80", + "component_budget": { + "node_count": 18, + "item_count": 18, + "soft_pair_count": 40, + "serialized_chars": 24142, + "estimated_input_tokens": 6036, + "estimated_eval_tokens": 12328 + }, + "budget": { + "node_count": 18, + "item_count": 18, + "soft_pair_count": 40, + "serialized_chars": 24142, + "estimated_input_tokens": 6036, + "estimated_eval_tokens": 12328 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This is a loose similarity cluster of many different model-add/support requests. They share broad wording, but they target different concrete models or features, so there are no true duplicates here.", + "confidence": 0.79, + "canonical_issue_reason": "issue:43671 is the most central and representative concrete request in this noisy similarity cluster, so it is the least-bad anchor even though the items are not duplicates.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43671 is the best standalone representative of the cluster because it is a clear model-support request and has the strongest connectivity, but the cluster is too broad for true deduping.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:36817", + "right": "issue:41211", + "accept": false, + "reason": "Different models: EuroBert vs DEIMv2." + }, + { + "left": "issue:36817", + "right": "issue:43671", + "accept": false, + "reason": "Different model requests: EuroBert config support vs Qwen3-TTS." + }, + { + "left": "issue:36817", + "right": "issue:36879", + "accept": false, + "reason": "Different model families: EuroBert vs RF-DETR." + }, + { + "left": "issue:36879", + "right": "issue:41211", + "accept": false, + "reason": "RF-DETR and DEIMv2 are unrelated model additions." + }, + { + "left": "issue:36879", + "right": "issue:43909", + "accept": false, + "reason": "RF-DETR is not the same as LFM2.5 Audio 1.5B." + }, + { + "left": "issue:36879", + "right": "issue:44121", + "accept": false, + "reason": "Object detection model vs weight-sparse transformer request." + }, + { + "left": "issue:36879", + "right": "issue:43837", + "accept": false, + "reason": "RF-DETR is unrelated to Qwen3-ASR support." + }, + { + "left": "issue:36879", + "right": "issue:43671", + "accept": false, + "reason": "RF-DETR and Qwen3-TTS are different concrete additions." + }, + { + "left": "issue:37015", + "right": "issue:43837", + "accept": false, + "reason": "NeoBERT and Qwen3-ASR are different model requests." + }, + { + "left": "issue:37015", + "right": "issue:44121", + "accept": false, + "reason": "NeoBERT vs OpenAI weight-sparse transformer are distinct changes." + }, + { + "left": "issue:37015", + "right": "issue:43909", + "accept": false, + "reason": "NeoBERT is unrelated to LFM2.5 Audio 1.5B." + }, + { + "left": "issue:36879", + "right": "issue:37015", + "accept": false, + "reason": "RF-DETR and NeoBERT are different models and code paths." + }, + { + "left": "issue:37015", + "right": "issue:42503", + "accept": false, + "reason": "NeoBERT and ModernVBERT are separate model additions." + }, + { + "left": "issue:41115", + "right": "issue:43837", + "accept": false, + "reason": "MiniCPM3 architecture request vs Qwen3-ASR support." + }, + { + "left": "issue:41115", + "right": "issue:44121", + "accept": false, + "reason": "MiniCPM3 and the weight-sparse transformer request are not the same change." + }, + { + "left": "issue:41115", + "right": "issue:42503", + "accept": false, + "reason": "MiniCPM3 architecture and ModernVBERT models are distinct additions." + }, + { + "left": "issue:41115", + "right": "issue:45027", + "accept": false, + "reason": "MiniCPM3 and Voxtral TTS are unrelated model support requests." + }, + { + "left": "issue:41211", + "right": "issue:43671", + "accept": false, + "reason": "DEIMv2 and Qwen3-TTS are different model targets." + }, + { + "left": "issue:42503", + "right": "issue:43837", + "accept": false, + "reason": "ModernVBERT models and Qwen3-ASR support are unrelated." + }, + { + "left": "issue:42503", + "right": "issue:43671", + "accept": false, + "reason": "ModernVBERT and Qwen3-TTS are different concrete requests." + }, + { + "left": "issue:42503", + "right": "issue:42738", + "accept": false, + "reason": "Specific ModernVBERT support vs broad BERT-like RoPE request." + }, + { + "left": "issue:42738", + "right": "issue:45027", + "accept": false, + "reason": "BERT-like RoPE models and Voxtral TTS are not the same underlying change." + }, + { + "left": "issue:42738", + "right": "issue:46653", + "accept": false, + "reason": "Model architecture request vs an unrelated personal-name issue." + }, + { + "left": "issue:43671", + "right": "issue:43837", + "accept": false, + "reason": "Related family, but TTS and ASR are different concrete support changes." + }, + { + "left": "issue:43671", + "right": "issue:45027", + "accept": false, + "reason": "Different models: Qwen3-TTS vs Voxtral-4B-TTS-2603." + }, + { + "left": "issue:43671", + "right": "issue:45177", + "accept": false, + "reason": "Qwen3-TTS support and DeepSeek-OCR2 are unrelated model requests." + }, + { + "left": "issue:43671", + "right": "issue:44121", + "accept": false, + "reason": "Qwen3-TTS support is not the same as adding the weight-sparse transformer." + }, + { + "left": "issue:43671", + "right": "issue:43909", + "accept": false, + "reason": "Qwen3-TTS and LFM2.5 Audio are different audio models." + }, + { + "left": "issue:43837", + "right": "issue:43909", + "accept": false, + "reason": "Qwen3-ASR and LFM2.5 Audio 1.5B are different speech models." + }, + { + "left": "issue:43837", + "right": "issue:45322", + "accept": false, + "reason": "ASR support and EUPE are unrelated model additions." + }, + { + "left": "issue:43837", + "right": "issue:44121", + "accept": false, + "reason": "Qwen3-ASR and the sparse transformer request are different code paths." + }, + { + "left": "issue:43909", + "right": "issue:44205", + "accept": false, + "reason": "Audio model request vs SAM3-LiteText are unrelated." + }, + { + "left": "issue:43909", + "right": "issue:46645", + "accept": false, + "reason": "LFM2.5 Audio support and Granite SWA+sink option are different changes." + }, + { + "left": "issue:44121", + "right": "issue:45322", + "accept": false, + "reason": "Weight-sparse transformer request vs EUPE model request." + }, + { + "left": "issue:43837", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-ASR and Voxtral TTS are different speech tasks/models." + }, + { + "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 different model/architecture requests." + }, + { + "left": "issue:42503", + "right": "issue:46653", + "accept": false, + "reason": "ModernVBERT models and the name-only issue are unrelated." + }, + { + "left": "issue:45027", + "right": "issue:46653", + "accept": false, + "reason": "Voxtral TTS support and the name-only issue are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 7, + "component_count": 22, + "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": "9c8fffc9c79ddd21269b01aa91967d7e553ba3f6fa507dd5d81b6a22fd1ec0a5", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1506, + "estimated_input_tokens": 377, + "estimated_eval_tokens": 1010 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1506, + "estimated_input_tokens": 377, + "estimated_eval_tokens": 1010 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are not duplicates: one is a closed docs request about adding Pruna as an optimization framework, and the other is an open technical issue about a Vision RoPE refactor. They concern entirely different areas and change types.", + "confidence": 0.97, + "canonical_issue_reason": "Issue #46443 is the better canonical issue for this cluster because it is the substantive open technical problem; #38740 is a closed documentation/proposal request and not the same topic.", + "canonical_pr_reason": null, + "best_issue_reason": "#46443 is the strongest issue artifact here because it describes a concrete code-path/refactor area, whereas #38740 is only a docs addition request.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:38740", + "right": "issue:46443", + "accept": false, + "reason": "Different subjects and scopes: docs proposal for Pruna vs Vision RoPE refactor. No shared code path, filenames, or explicit linkage indicating 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": 22, + "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": "4c3e365f69b4ec0737373e39d782cc4bd99c0bf9b34c823a605b25697501c7ce", + "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": "The two issues are unrelated: one concerns block decoding, while the other requests DeepSeek v3.2 support. They should not be clustered as duplicates.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41084", + "right": "issue:43037", + "accept": false, + "reason": "Different topics with no shared code-path or explicit linkage: \"Set Block Decoding\" vs \"DeepSeek v3.2 support\"." + } + ] + }, + "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": 22, + "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": "fcb5bd8c129d8a56bbda2046be157b1ca5f102812e9fad984006c319dbc163c6", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1584, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1584, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are related only at a very high level (Trainer customization), but they describe different problems: one is about `remove_unused_column` behavior, the other about FLOPS estimation for custom trainers. They should not be clustered as duplicates.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 43598 is the more concrete, narrowly scoped request: it names a specific Trainer behavior (`remove_unused_column`) and asks for better customizability.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43598 is the best representative of the cluster because it is the clearest actionable code-path change, whereas 43600 is a separate metric/estimation concern.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43598", + "right": "issue:43600", + "accept": false, + "reason": "They share only the broad Trainer/customization theme. One concerns input column pruning, the other FLOPS estimation for custom trainers; these are different underlying bugs and would not be fixed by one change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 10, + "component_count": 22, + "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": "9e869e2e04e7ee202cf17780abcf8e3184f23555916f4f823c946bd45ecada1e", + "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 two issues are only superficially similar: both ask for new Qwen model support, but one is about AutoModel/AutoModelForConditionalGeneration registration for Qwen3-Omni, while the other is a separate AutoModelForSequenceClassification feature request for Qwen3.5. They should not be treated as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": "43684 is the broader, more general registration request and is the better cluster anchor if one must be chosen; 44405 targets a different model family and a different task head.", + "canonical_pr_reason": null, + "best_issue_reason": "43684 is the most representative of the only common theme here (model registration/support), but it does not subsume 44405's distinct sequence-classification request.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43684", + "right": "issue:44405", + "accept": false, + "reason": "Different model variants and different capabilities: Qwen3-Omni registration for AutoModel/AutoModelForConditionalGeneration vs Qwen3.5 AutoModelForSequenceClassification support. Shared wording is not enough to indicate the same underlying change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 11, + "component_count": 22, + "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": "c4c23fde8091e907088ff85b527fa386f7149d7f6516ceaff981461b65715fb9", + "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": "43845/43846/43859 look like near-identical vague issues, but there isn\u2019t enough concrete detail to safely collapse them; 43835 appears unrelated. No PRs are present.", + "confidence": 0.42, + "canonical_issue_reason": "issue:43845 is the earliest of the three 'huggingface' issues and the best representative if one must be chosen, but the evidence is too thin to treat the others as confirmed duplicates.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43845 is the strongest cluster representative because it is one of the repeated 'huggingface' reports and has the earliest timestamp among the closest matches.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43835", + "right": "issue:43845", + "accept": false, + "reason": "Different titles ('Fett' vs 'huggingface'); no substantive evidence of the same underlying issue." + }, + { + "left": "issue:43835", + "right": "issue:43846", + "accept": false, + "reason": "Different titles ('Fett' vs 'huggingface'); no shared concrete details to justify a duplicate merge." + }, + { + "left": "issue:43835", + "right": "issue:43859", + "accept": false, + "reason": "Different titles ('Fett' vs 'huggingface'); too little information to link them as the same issue." + }, + { + "left": "issue:43845", + "right": "issue:43846", + "accept": false, + "reason": "The titles match, but they are too generic and there is no concrete description or target to confirm the same bug." + }, + { + "left": "issue:43845", + "right": "issue:43859", + "accept": false, + "reason": "Near-identical vague titles are not enough on their own to prove the same underlying issue." + }, + { + "left": "issue:43846", + "right": "issue:43859", + "accept": false, + "reason": "Same generic 'huggingface' title, but no substantive evidence that these are the same report." + } + ] + }, + "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": 22, + "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": "83f066781ba4ad7db8e4d71290d81bc7cac7f7aaa99c67be40453caae4cdadb1", + "component_budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 9, + "serialized_chars": 5791, + "estimated_input_tokens": 1448, + "estimated_eval_tokens": 3152 + }, + "budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 9, + "serialized_chars": 5791, + "estimated_input_tokens": 1448, + "estimated_eval_tokens": 3152 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is dominated by low-information/spam-like issues with no actionable bug or change description. The two non-identical titles also look unrelated, so there is no safe duplicate consolidation here.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "44371 is the earliest of the identical '' issues and is the least arbitrary representative, but none of these issues describe a real underlying bug/change well enough to serve as a meaningful canonical target.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44371", + "right": "issue:44572", + "accept": false, + "reason": "Same spam-like placeholder title, but no substantive bug/change is described, so this is not a safe duplicate match." + }, + { + "left": "issue:44371", + "right": "issue:44573", + "accept": false, + "reason": "Same spam-like placeholder title, but the content is too thin to confirm a shared underlying issue." + }, + { + "left": "issue:44371", + "right": "issue:44418", + "accept": false, + "reason": "One is a spam placeholder and the other is a documentation enhancement request; they do not look like the same issue." + }, + { + "left": "issue:44371", + "right": "issue:44937", + "accept": false, + "reason": "Spam placeholder versus a different spammy-sounding title; no concrete shared bug or change is visible." + }, + { + "left": "issue:44418", + "right": "issue:44572", + "accept": false, + "reason": "Documentation enhancement and spam placeholder are not the same underlying issue." + }, + { + "left": "issue:44418", + "right": "issue:44573", + "accept": false, + "reason": "Documentation enhancement and spam placeholder are not the same underlying issue." + }, + { + "left": "issue:44572", + "right": "issue:44573", + "accept": false, + "reason": "Both are identical spam placeholders, but there is no real issue description to justify treating them as the same bug/change." + }, + { + "left": "issue:44572", + "right": "issue:44937", + "accept": false, + "reason": "Spam placeholder versus a distinct titled issue; no evidence of one concrete shared problem." + }, + { + "left": "issue:44573", + "right": "issue:44937", + "accept": false, + "reason": "Spam placeholder versus a distinct titled issue; no evidence of one concrete shared problem." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 13, + "component_count": 22, + "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": "baed8f40dca467624ca5e2ce5bc2bbf2fb5009c8af83fae05b901a949c2cae9e", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1644, + "estimated_input_tokens": 411, + "estimated_eval_tokens": 1078 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1644, + "estimated_input_tokens": 411, + "estimated_eval_tokens": 1078 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two closed issues look very similar and may be near-duplicates, but the evidence is only soft similarity (same title, close body length, Jaccard overlap) rather than direct content confirmation. #45145 is the better representative because it is earlier and already has inbound references.", + "confidence": 0.79, + "canonical_issue_reason": "Issue #45145 is the better canonical candidate because it was created first and already has inbound references, while #45151 appears to be a later near-duplicate report.", + "canonical_pr_reason": null, + "best_issue_reason": "#45145 is the best cluster representative because it is the earliest report and has the most established reference activity.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45145", + "right": "issue:45151", + "accept": false, + "reason": "Same title and similar body metrics suggest a likely duplicate, but the evidence provided is only soft_similarity, so I can't confirm they are the same underlying issue with enough confidence." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The packet only supports a soft similarity match, not a definitive duplicate. Same title and similar body lengths/Jaccard justify a likely-duplicate assessment, but the analyst states the issues \"are duplicates\" and that the body content is \"highly similar\" without direct body evidence. The canonical choice of the earlier issue is fine, but the soft-edge verdict and summary are a bit too strong for the available evidence." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 14, + "component_count": 22, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45295-2", + "nodes": [ + "issue:45295", + "issue:45373" + ], + "soft_pairs": [], + "prepared_review_unit_hash": "f5fd95fada2f530ab0999e343c17c16266cde0c7182bb22509972ca44827cd49", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 0, + "serialized_chars": 1286, + "estimated_input_tokens": 322, + "estimated_eval_tokens": 900 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 0, + "serialized_chars": 1286, + "estimated_input_tokens": 322, + "estimated_eval_tokens": 900 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues describe the same missing Gemma 4 sequence-classification support; the later, more specific open issue is the better cluster representative.", + "confidence": 0.94, + "canonical_issue_reason": "issue #45373 is the clearest canonical tracker: it is open, explicitly names the missing class `Gemma4ForSequenceClassification`, and states the module-level gap relative to Gemma 2/3.", + "canonical_pr_reason": null, + "best_issue_reason": "issue #45373 best captures the concrete request and likely remaining work, while #45295 is a shorter earlier duplicate with less detail and is already closed.", + "best_pr_reason": null, + "soft_edge_verdicts": [] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 15, + "component_count": 22, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45672-2", + "nodes": [ + "issue:45672", + "issue:45674" + ], + "soft_pairs": [ + "issue:45672|issue:45674" + ], + "prepared_review_unit_hash": "0c7f5dfa0831aa9f63e91589c8f1b44c3f2b0054472445c04077e0af40d1a155", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1673, + "estimated_input_tokens": 419, + "estimated_eval_tokens": 1094 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1673, + "estimated_input_tokens": 419, + "estimated_eval_tokens": 1094 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are two different 4-bit inference bugs: one is a Gemma4/audio-module uint8 torch.finfo TypeError, the other is a BitsAndBytesConfig lm_head exclusion regression caused by llm_int8_skip_modules. They share the same general quantized-inference context but not the same underlying failure path.", + "confidence": 0.91, + "canonical_issue_reason": "No single canonical issue: the titles and described failures point to distinct root causes in different code paths, so they should not be deduplicated.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45674 is the better representative issue because it describes a broader, library-level regression with a clearer config-driven failure mode; issue:45672 is a narrower model-specific symptom.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45672", + "right": "issue:45674", + "accept": false, + "reason": "Both involve 4-bit inference, but the failures are different: Gemma4/audio-module uint8 torch.finfo TypeError versus BitsAndBytesConfig lm_head exclusion being cleared by llm_int8_skip_modules. Different bug, not a duplicate." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 16, + "component_count": 22, + "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": "3daf4129d24e51b0865c6246b2b9fe3b3f23bc99b8001942c5b67afbaac149f6", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1476, + "estimated_input_tokens": 369, + "estimated_eval_tokens": 994 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1476, + "estimated_input_tokens": 369, + "estimated_eval_tokens": 994 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two closed issues are unrelated personal-name/date entries and do not appear to describe the same underlying bug or request. Do not deduplicate or cluster them together.", + "confidence": 0.98, + "canonical_issue_reason": "No canonical issue: the titles reference different people and dates, with no body content, links, or other evidence of the same problem.", + "canonical_pr_reason": null, + "best_issue_reason": "Neither issue is a good cluster representative because the items appear unrelated rather than duplicate reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45762", + "right": "issue:45763", + "accept": false, + "reason": "Different names and dates; no shared description, links, or evidence of the same underlying issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 17, + "component_count": 22, + "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": "499e3e37b286d7d1aeaa2c55149b7d3ca6b08189a5530afbc4e0bfbdac6704f5", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1480, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1480, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two closed issues are not a duplicate pair; they have different person-name titles and only weak surface similarity (shared surname/date formatting). No code, targets, or discussion evidence suggests a common underlying bug or change.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "issue:45764 is the slightly better representative only because it was created first; there is no substantive evidence that either issue should be canonical for the other.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45764", + "right": "issue:45765", + "accept": false, + "reason": "Reject: the titles refer to different individuals and the only overlap is surname/date formatting, which is not enough to indicate the same underlying issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 18, + "component_count": 22, + "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": "ae8e21904cd3f9c7dc67750c5660bd6aaf9c4bc58ac234346d6f6bd4b8160cf7", + "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": "These two issues are not duplicates: one requests a new Qwen3_5 token-classification/value-model class, the other requests FP8 training support. They concern different features and different code paths.", + "confidence": 0.97, + "canonical_issue_reason": "No canonical issue: issue 45810 is a model-class feature request, while issue 46295 is a training/runtime precision request; they are unrelated.", + "canonical_pr_reason": null, + "best_issue_reason": "No single best issue for deduplication here because the cluster does not represent one underlying bug or change.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45810", + "right": "issue:46295", + "accept": false, + "reason": "Reject: the requests are about different capabilities (new token classification model class vs FP8 training support) and do not describe the same underlying problem 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": 19, + "component_count": 22, + "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": "772e727a8a37ed70dd1876aad0f4ee720d2ccada5f00305f14dde55afe6557ae", + "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 likely duplicate pairs stand out: 45870/45871 and 45872/45873. 45869 is related by topic but appears to be a different request, so it should stay separate.", + "confidence": 0.93, + "canonical_issue_reason": "45872 is a reasonable canonical issue for the 45872/45873 pair because the titles match exactly and the body lengths match, so it can stand in for that same apparent request.", + "canonical_pr_reason": null, + "best_issue_reason": "45872 is the best overall issue candidate in this cluster because it anchors an exact-duplicate pair (45873) and has a specific title that distinguishes it from the more generic example issue.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45869", + "right": "issue:45870", + "accept": false, + "reason": "Related financial/DataForge topic, but the titles point to different requests: dataset/sample offer versus an example prediction issue." + }, + { + "left": "issue:45869", + "right": "issue:45871", + "accept": false, + "reason": "Same as 45869 vs 45870: similar theme, but not the same apparent request." + }, + { + "left": "issue:45870", + "right": "issue:45871", + "accept": true, + "reason": "Exact same title and matching body length; they look like the same issue reported twice." + }, + { + "left": "issue:45870", + "right": "issue:45872", + "accept": false, + "reason": "Different apparent requests: an example for financial time series prediction versus a financial time series + scraping service request." + }, + { + "left": "issue:45870", + "right": "issue:45873", + "accept": false, + "reason": "Different apparent requests; only the broad DataForge/financial theme overlaps." + }, + { + "left": "issue:45871", + "right": "issue:45872", + "accept": false, + "reason": "Different titles and different apparent request types, so not a likely duplicate pair." + }, + { + "left": "issue:45871", + "right": "issue:45873", + "accept": false, + "reason": "Different apparent requests; shared topic is too broad to treat as a duplicate." + }, + { + "left": "issue:45872", + "right": "issue:45873", + "accept": true, + "reason": "Exact same title and matching body length; they appear to be duplicate copies of the same issue." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The duplicate-pair calls are mostly well-grounded: 45870/45871 and 45872/45873 have identical titles and body lengths, so treating them as likely duplicates is conservative. However, the canonical/best-issue reasoning overstates the evidence: the packet does not contain bodies, so phrases like \"duplicated verbatim\" and \"more substantive\" are not supported, and \"strongest global issue candidate\" is not evidenced beyond title/body-length similarity. Please keep the language to \"likely duplicate\" / \"same apparent request\" and avoid claiming content-level distinctions that aren't in the packet." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 20, + "component_count": 22, + "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": "543df39bec430c1d13a1ac7c4d276c5cc715cc9564d07efa4662236cccfa665c", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1637, + "estimated_input_tokens": 410, + "estimated_eval_tokens": 1076 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1637, + "estimated_input_tokens": 410, + "estimated_eval_tokens": 1076 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are plausibly related because both point to the static-cache path and `linear_attention`, but the evidence here is only title-level similarity. That supports a tentative duplicate/cluster link, not a strong claim that they share the exact same root cause.", + "confidence": 0.58, + "canonical_issue_reason": "Issue 46439 is the better canonical issue because it describes the broader failure mode in the static-cache/`early_initialization` path affecting linear-attention layers, whereas 46441 reads as a more model-specific crash symptom.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 46439 is the strongest standalone report since it names the mechanism and the general corruption/runtime-error behavior, making it the best representative for this cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46439", + "right": "issue:46441", + "accept": true, + "reason": "Tentatively accept: both titles involve static cache and `linear_attention`, so they are a plausible same-bug cluster candidate. However, the packet only provides soft similarity and titles, so this should be treated as provisional rather than a proven shared root cause." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The linkage is plausible, but the analyst overstates the evidence available in the packet. We only have titles plus soft similarity, which support a conservative note that both issues involve static-cache and linear_attention, not a confident claim that they share the same root cause or that Qwen3.5/early_initialization applies to both. The soft-edge acceptance should be more tentative." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 21, + "component_count": 22, + "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": "dac62016dca354642595b0144a96235e40e9f4d012b0fbb95079319fb05155b3", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1707, + "estimated_input_tokens": 427, + "estimated_eval_tokens": 1110 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1707, + "estimated_input_tokens": 427, + "estimated_eval_tokens": 1110 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues are the same report with identical title/body about Wav2Vec2CTCTokenizer v5 mishandling the word delimiter as a special token. They should be clustered together, with the open issue as the better canonical record.", + "confidence": 0.99, + "canonical_issue_reason": "Issue 46552 is open and already has inbound references, so it is the better live tracker for the duplicated bug report.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 46552 is the strongest issue representative because it is the active duplicate target and carries the same exact report text as 46551.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46551", + "right": "issue:46552", + "accept": true, + "reason": "Same title, same body, same underlying tokenizer bug report; this is a direct duplicate." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The clustering is well grounded in the packet: both issues have identical titles, identical body lengths, and the pair has Jaccard 1.0 with only soft_similarity evidence. Choosing the open issue (46552) as the canonical tracker is also conservative and supported by its inbound references. No overstatement beyond the available evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 22, + "component_count": 22, + "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": "0050c78ae0f5b864f7eb05c083f52ffdc1b4dfe7bee60f2cb689ad7a086e2883", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1566, + "estimated_input_tokens": 392, + "estimated_eval_tokens": 1040 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1566, + "estimated_input_tokens": 392, + "estimated_eval_tokens": 1040 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues do not look like the same bug: one reports a tensor shape mismatch during object-detection training, while the other reports GPU memory exhaustion after several epochs in RT-DETR v2 training.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "If a single issue must represent the cluster, 46657 is the better anchor because it names a concrete, diagnosable failure mode; 46692 is a separate OOM problem with different remediation.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46657", + "right": "issue:46692", + "accept": false, + "reason": "Different underlying failures: tensor size mismatch vs out-of-memory. They share a training context, but not the same 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": 1, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-39895-14", + "nodes": [ + "pull_request:39895", + "pull_request:41855", + "pull_request:42668", + "pull_request:43251", + "pull_request:43254", + "pull_request:43424", + "pull_request:43823", + "pull_request:43995", + "pull_request:44827", + "pull_request:45170", + "pull_request:45213", + "pull_request:45655", + "pull_request:46048", + "pull_request:46267" + ], + "soft_pairs": [ + "pull_request:39895|pull_request:43424", + "pull_request:41855|pull_request:43823", + "pull_request:41855|pull_request:43424", + "pull_request:42668|pull_request:43823", + "pull_request:43251|pull_request:43823", + "pull_request:43251|pull_request:44827", + "pull_request:43424|pull_request:44827", + "pull_request:43424|pull_request:43823", + "pull_request:43823|pull_request:45655", + "pull_request:43823|pull_request:43995", + "pull_request:43995|pull_request:45213", + "pull_request:43995|pull_request:44827", + "pull_request:44827|pull_request:45170", + "pull_request:44827|pull_request: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": "e2f1e58824852d3668674cbc54b28681493307842d3a338cde4eb1706727f361", + "component_budget": { + "node_count": 14, + "item_count": 14, + "soft_pair_count": 19, + "serialized_chars": 38155, + "estimated_input_tokens": 9539, + "estimated_eval_tokens": 19334 + }, + "budget": { + "node_count": 14, + "item_count": 14, + "soft_pair_count": 19, + "serialized_chars": 38155, + "estimated_input_tokens": 9539, + "estimated_eval_tokens": 19334 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This set is mostly unrelated PRs from different models/features. The only true duplicate-like pair is the two `fixed_cross_entropy` patches for issue #43240; the rest are only loosely related by subsystem or generic \u201csoft similarity\u201d and should stay separate.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 43251 is the best canonical PR for the actual duplicate pair: it explicitly targets issue #43240 and directly fixes `fixed_cross_entropy` by forwarding supported `cross_entropy` kwargs. PR 43254 is a near-identical alternate patch to the same code path.", + "best_issue_reason": null, + "best_pr_reason": "PR 43251 is the strongest representative of the only real duplicate cluster because it has the explicit issue link, touches the same file, and implements the same concrete fix in the cleanest form.", + "soft_edge_verdicts": [ + { + "left": "pull_request:39895", + "right": "pull_request:43424", + "accept": false, + "reason": "Unrelated model-addition PR vs. an executorch test change; no shared bug or mergeable code-path overlap." + }, + { + "left": "pull_request:41855", + "right": "pull_request:43823", + "accept": false, + "reason": "Tokenizer utility patch vs. adding a new model; same broad area of Transformers, but not the same underlying change." + }, + { + "left": "pull_request:41855", + "right": "pull_request:43424", + "accept": false, + "reason": "Mistral tokenizer methods and executorch export tests are different concerns." + }, + { + "left": "pull_request:42668", + "right": "pull_request:43823", + "accept": false, + "reason": "Processor auto-registration / robustness changes are unrelated to adding MobileLLM." + }, + { + "left": "pull_request:43251", + "right": "pull_request:43823", + "accept": false, + "reason": "Cross-entropy kwargs fix is unrelated to introducing MobileLLM." + }, + { + "left": "pull_request:43251", + "right": "pull_request:44827", + "accept": false, + "reason": "Different code paths: loss utility kwargs handling vs. Mistral4 test/model changes." + }, + { + "left": "pull_request:43424", + "right": "pull_request:44827", + "accept": false, + "reason": "Executorch dynamic-shape tests and Mistral4 fixes do not describe the same bug." + }, + { + "left": "pull_request:43424", + "right": "pull_request:43823", + "accept": false, + "reason": "Executorch export tests are not the same change as MobileLLM model addition." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45655", + "accept": false, + "reason": "Model addition vs. class-resolution fix; no common underlying defect." + }, + { + "left": "pull_request:43823", + "right": "pull_request:43995", + "accept": false, + "reason": "MobileLLM model scaffolding and Falcon output-interface refactoring are distinct changes." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45213", + "accept": false, + "reason": "Model addition and a model-creation skill draft are not the same artifact or fix." + }, + { + "left": "pull_request:43995", + "right": "pull_request:44827", + "accept": false, + "reason": "Falcon output collection refactor and Mistral4 test fixes target different model internals." + }, + { + "left": "pull_request:43995", + "right": "pull_request:45213", + "accept": false, + "reason": "Falcon model refactor is unrelated to the model-creation skill draft." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45170", + "accept": false, + "reason": "Mistral4 test/model changes vs. `layrnorm` renaming across CLIP-like models; not the same bug." + }, + { + "left": "pull_request:44827", + "right": "pull_request:46267", + "accept": false, + "reason": "Mistral4 test fixes and chat parsing updates are unrelated." + }, + { + "left": "pull_request:44827", + "right": "pull_request:46048", + "accept": false, + "reason": "Mistral4 fixes and TDT loss kernel integration address different subsystems and failure modes." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45655", + "accept": false, + "reason": "Mistral4 test fixes are not the same issue as config-resolution ordering in `modeling_utils`." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45213", + "accept": false, + "reason": "Mistral4 PR and model-creation skill draft are unrelated." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45213", + "accept": false, + "reason": "MobileLLM model addition and the model-creation skill draft are not the same underlying change." + }, + { + "left": "pull_request:43251", + "right": "pull_request:43254", + "accept": true, + "reason": "Both patches fix the same concrete `fixed_cross_entropy` bug in `src/transformers/loss/loss_utils.py` by forwarding supported kwargs to `nn.functional.cross_entropy`, and one can plausibly subsume the other." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary stays grounded in the packet and is appropriately conservative. The only accepted pair, PR 43251/43254, is a plausible single-bugfix duplicate because both change `src/transformers/loss/loss_utils.py` for the same issue target (#43240) and implement the same `fixed_cross_entropy` kwargs forwarding. The remaining pairs are correctly left as unrelated soft similarities without overstating evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 2, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-40546-2", + "nodes": [ + "pull_request:40546", + "pull_request:42781" + ], + "soft_pairs": [ + "pull_request:40546|pull_request:42781" + ], + "prepared_review_unit_hash": "9f6aca24fcf6c672600960b36b1abceaedba3adbab3987124966948a93974620", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 7225, + "estimated_input_tokens": 1807, + "estimated_eval_tokens": 3870 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 7225, + "estimated_input_tokens": 1807, + "estimated_eval_tokens": 3870 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs are related to VibeVoice, but they are not duplicates: #40546 is the core VibeVoice implementation, while #42781 adds a separate realtime/acoustic-tokenizer extension and related generation changes.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "#40546 is the more central artifact: it introduces the base VibeVoice model, docs, auto-class wiring, and core conversion/modeling pieces. #42781 builds on the same model family but targets a different realtime variant with additional tokenizer/feature-extraction work.", + "best_issue_reason": null, + "best_pr_reason": "#40546 is the best representative for the cluster because it is the primary end-to-end VibeVoice implementation. #42781 is a distinct follow-on change for realtime support, not a duplicate of the base model PR.", + "soft_edge_verdicts": [ + { + "left": "pull_request:40546", + "right": "pull_request:42781", + "accept": false, + "reason": "Shared VibeVoice filenames are not enough to call these duplicates. #40546 implements the base model, while #42781 adds realtime support plus acoustic tokenizer and generation updates; they address different concrete changes and would not plausibly be merged as one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 3, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-40755-2", + "nodes": [ + "pull_request:40755", + "pull_request:41224" + ], + "soft_pairs": [ + "pull_request:40755|pull_request:41224" + ], + "prepared_review_unit_hash": "f9e3ccdffbfd980ef252b29e8ba5cbbeb7e4e5f1c6787b7b10088350f2cb90e4", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4676, + "estimated_input_tokens": 1169, + "estimated_eval_tokens": 2594 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4676, + "estimated_input_tokens": 1169, + "estimated_eval_tokens": 2594 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two PRs are unrelated feature additions in different model areas: TimesFM covariate forecasting vs. DINOv3 image classification support. The similarity is only superficial (model/docs/test support), so they should not be clustered as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "No single best PR for duplicate triage: the PRs target different models and different concrete capabilities, so neither is a canonical representative of the other.", + "soft_edge_verdicts": [ + { + "left": "pull_request:40755", + "right": "pull_request:41224", + "accept": false, + "reason": "Different underlying changes and code paths: TimesFM forecasting with covariates vs DINOv3 image classification support. Too broad/superficial to treat as the same PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 4, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-41356-2", + "nodes": [ + "pull_request:41356", + "pull_request:44339" + ], + "soft_pairs": [ + "pull_request:41356|pull_request:44339" + ], + "prepared_review_unit_hash": "aead1308c167b0bae6ff11bf9358340baa5e76f816618383418b505a10a9cd01", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6050, + "estimated_input_tokens": 1513, + "estimated_eval_tokens": 3282 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6050, + "estimated_input_tokens": 1513, + "estimated_eval_tokens": 3282 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two DEIMv2 pull requests target the same model family, but they are not the same change: PR 41356 is an early/basic model + image processor add, while PR 44339 is a much broader follow-up that adds loss code, auto-mappings, conversion tooling, modular files, and more integration work.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44339 is the more complete and representative DEIMv2 implementation; PR 41356 looks like an earlier partial version of the same model bring-up.", + "best_issue_reason": null, + "best_pr_reason": "PR 44339 is the strongest single artifact for the cluster because it subsumes the basic model work and includes the broader integration needed for a full Transformers addition.", + "soft_edge_verdicts": [ + { + "left": "pull_request:41356", + "right": "pull_request:44339", + "accept": false, + "reason": "Same model area, but not the same concrete change: 44339 expands the implementation substantially and adds new code paths beyond the basic PR 41356. They look like related iterations, not mergeable duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 5, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-42975-2", + "nodes": [ + "pull_request:42975", + "pull_request:42976" + ], + "soft_pairs": [ + "pull_request:42975|pull_request:42976" + ], + "prepared_review_unit_hash": "f59cdad395855e3446ca2a8e1cf08467b45872d6ac0935efe1cc11a05659f922", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5935, + "estimated_input_tokens": 1484, + "estimated_eval_tokens": 3224 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5935, + "estimated_input_tokens": 1484, + "estimated_eval_tokens": 3224 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are overlapping workflow-maintenance PRs, but they do not fix the same underlying problem. One is a Node 24 compatibility sweep for core GitHub Actions, while the other is a broader latest-version bump centered on docker/build-push-action and related workflow actions.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "No single PR is a clear canonical representative because the upgrades target different action families and different maintenance goals.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42975", + "right": "pull_request:42976", + "accept": false, + "reason": "Different upgrade tracks: 42975 is specifically for Node 24 compatibility of actions like checkout/cache/upload-artifact, while 42976 updates docker/build-push-action and other actions. They touch similar workflows but do not appear to be 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": 6, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-42978-7", + "nodes": [ + "pull_request:42978", + "pull_request:43448", + "pull_request:43451", + "pull_request:45490", + "pull_request:46562", + "pull_request:46658", + "pull_request:46705" + ], + "soft_pairs": [ + "pull_request:42978|pull_request:43448", + "pull_request:42978|pull_request:43451", + "pull_request:42978|pull_request:45490", + "pull_request:42978|pull_request:46705", + "pull_request:43448|pull_request:43451", + "pull_request:43448|pull_request:45490", + "pull_request:43448|pull_request:46705", + "pull_request:43451|pull_request:45490", + "pull_request:43451|pull_request:46705", + "pull_request:45490|pull_request:46562", + "pull_request:45490|pull_request:46705", + "pull_request:46562|pull_request:46658" + ], + "prepared_review_unit_hash": "87524226e9a0ba46d9cf2283eba982a9c62905bf51b01eff62b1b96bcf835409", + "component_budget": { + "node_count": 7, + "item_count": 7, + "soft_pair_count": 12, + "serialized_chars": 24366, + "estimated_input_tokens": 6092, + "estimated_eval_tokens": 12440 + }, + "budget": { + "node_count": 7, + "item_count": 7, + "soft_pair_count": 12, + "serialized_chars": 24366, + "estimated_input_tokens": 6092, + "estimated_eval_tokens": 12440 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous model-addition and infrastructure PRs. The only clear duplicate pair is the Step 3.7 Flash work: #46562 and #46658 target the same model/codepath, while the other PRs are separate model introductions or an unrelated distributed-config change.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "#46562 is the best canonical PR for the only true duplicate pair: it is the more complete, model-facing Step 3.7 Flash implementation, while #46658 is a WIP/mini-model companion rather than a separate change.", + "best_issue_reason": null, + "best_pr_reason": "#46562 is the strongest representative PR in this cluster because it corresponds to the actual Step 3.7 Flash model addition and is the cleanest merge target among the duplicate pair.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42978", + "right": "pull_request:43448", + "accept": false, + "reason": "Both add different vision-language models (ViT NEPA vs Molmo); shared docs/auto-registration files are incidental, not the same change." + }, + { + "left": "pull_request:42978", + "right": "pull_request:43451", + "accept": false, + "reason": "Different model additions with distinct architectures and files; overlap in auto docs is too broad to indicate duplication." + }, + { + "left": "pull_request:42978", + "right": "pull_request:45490", + "accept": false, + "reason": "ViT NEPA and CTSM are unrelated model additions; same repo plumbing does not make them the same underlying bug or feature." + }, + { + "left": "pull_request:42978", + "right": "pull_request:46705", + "accept": false, + "reason": "Model addition PR versus distributed-config infrastructure PR; no shared concrete code-path problem." + }, + { + "left": "pull_request:43448", + "right": "pull_request:43451", + "accept": false, + "reason": "Molmo and Molmo2 are separate model families/releases with different files and behavior; not mergeable as one PR." + }, + { + "left": "pull_request:43448", + "right": "pull_request:45490", + "accept": false, + "reason": "Molmo and CTSM are distinct model additions; shared framework registration is not sufficient for duplicate triage." + }, + { + "left": "pull_request:43448", + "right": "pull_request:46705", + "accept": false, + "reason": "A model-addition PR and a distributed configuration PR are unrelated changes." + }, + { + "left": "pull_request:43451", + "right": "pull_request:45490", + "accept": false, + "reason": "Molmo2 and CTSM are unrelated model additions; overlap is only in generic auto/model registry files." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46705", + "accept": false, + "reason": "Molmo2 model work is unrelated to distributed config support." + }, + { + "left": "pull_request:45490", + "right": "pull_request:46562", + "accept": false, + "reason": "CTSM and Step 3.7 Flash are different models; similar auto-registration edits do not imply the same bug or feature." + }, + { + "left": "pull_request:45490", + "right": "pull_request:46705", + "accept": false, + "reason": "CTSM model addition versus distributed config infrastructure; no duplicate relationship." + }, + { + "left": "pull_request:46562", + "right": "pull_request:46658", + "accept": true, + "reason": "Both PRs implement Step 3.7 Flash, share the same model registration path, and appear to be two versions of the same underlying feature/change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 7, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43291-8", + "nodes": [ + "pull_request:43291", + "pull_request:43498", + "pull_request:43665", + "pull_request:44815", + "pull_request:45114", + "pull_request:46380", + "pull_request:46539", + "pull_request:46579" + ], + "soft_pairs": [ + "pull_request:43291|pull_request:43665", + "pull_request:43291|pull_request:44815", + "pull_request:43291|pull_request:46380", + "pull_request:43498|pull_request:46579", + "pull_request:43665|pull_request:46539", + "pull_request:43665|pull_request:44815", + "pull_request:43665|pull_request:46380", + "pull_request:43665|pull_request:46579", + "pull_request:44815|pull_request:46539", + "pull_request:44815|pull_request:46380", + "pull_request:44815|pull_request:46579", + "pull_request:45114|pull_request:46579", + "pull_request:46380|pull_request:46539", + "pull_request:46380|pull_request:46579" + ], + "prepared_review_unit_hash": "2c57b19c10c70688314220c5fb5ab2940eb3aa54ed57980b6b2a676d765eb72d", + "component_budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 14, + "serialized_chars": 23165, + "estimated_input_tokens": 5792, + "estimated_eval_tokens": 11840 + }, + "budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 14, + "serialized_chars": 23165, + "estimated_input_tokens": 5792, + "estimated_eval_tokens": 11840 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is heterogeneous: it mixes unrelated test fixes, a Whisper tokenizer change, a deprecated API alias, a generation-validation fix, CI infrastructure changes, and a typo cleanup. None of the soft edges look like the same underlying bug or change, so no merges should be made.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46579 is the broadest, most representative code-cleanup PR in the set, but the cluster is not actually a duplicate group; the other PRs are unrelated fixes in different subsystems.", + "best_issue_reason": null, + "best_pr_reason": "PR 46579 is the best single representative only in a nominal sense because it spans many code files and is broader than the others, but it should not be treated as a duplicate target for the rest of the cluster.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43291", + "right": "pull_request:43665", + "accept": false, + "reason": "Both touch tests, but they target different models and different problems; not the same underlying bug." + }, + { + "left": "pull_request:43291", + "right": "pull_request:44815", + "accept": false, + "reason": "Whisper tokenizer/test fix is unrelated to the FP8 dequantization bug fix." + }, + { + "left": "pull_request:43291", + "right": "pull_request:46380", + "accept": false, + "reason": "Whisper test changes and generation watermarking validation are unrelated code paths." + }, + { + "left": "pull_request:43498", + "right": "pull_request:46579", + "accept": false, + "reason": "A backward-compatibility alias for tie_weights is unrelated to a broad typo-cleanup PR." + }, + { + "left": "pull_request:43665", + "right": "pull_request:44815", + "accept": false, + "reason": "Test assertion changes for CLIP/ViT do not match the FP8 dequantization fix." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46380", + "accept": false, + "reason": "Different subsystems and different failure modes; only superficial test/bug-fix similarity." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46539", + "accept": false, + "reason": "Model test edits are unrelated to the Phimoe CI integration fix." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46579", + "accept": false, + "reason": "Model test tweaks are unrelated to code typo cleanup across many files." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46380", + "accept": false, + "reason": "FP8 quantization/loading changes and generation config validation fixes are distinct bugs." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46539", + "accept": false, + "reason": "Different problems: dequantization logic versus Phimoe integration CI behavior." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46579", + "accept": false, + "reason": "FP8 kernel/loading fixes are unrelated to typo-only cleanup." + }, + { + "left": "pull_request:45114", + "right": "pull_request:46579", + "accept": false, + "reason": "Doctest formatting cleanup is not the same underlying change as code typo fixes." + }, + { + "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:46579", + "accept": false, + "reason": "A generation config bug fix is unrelated to broad typo cleanup." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 8, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43488-2", + "nodes": [ + "pull_request:43488", + "pull_request:43532" + ], + "soft_pairs": [ + "pull_request:43488|pull_request:43532" + ], + "prepared_review_unit_hash": "e3099d4305c8fd845343735220549c70f26434e55f80ce349d0247e434c116a0", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4563, + "estimated_input_tokens": 1141, + "estimated_eval_tokens": 2538 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4563, + "estimated_input_tokens": 1141, + "estimated_eval_tokens": 2538 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both items are unrelated PRs that only overlap on a shared Vit test filename; they do not look like the same underlying change or bug fix.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "43532 is the more representative PR in this cluster: it makes the broader CI/workflow and test-harness updates, while 43488 is a small repo-bot/check-style change with a version/dependency tweak.", + "best_issue_reason": null, + "best_pr_reason": "43532 is the best overall PR candidate because it targets the main CI/test execution path and is more substantive than 43488, which looks like a throwaway validation PR.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43488", + "right": "pull_request:43532", + "accept": false, + "reason": "Reject: the PRs have different purposes and code paths. 43488 is a small setup/version/test tweak, while 43532 changes workflow automation and testing utilities; the shared Vit test file is not enough to indicate 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": "pull_request", + "component_index": 9, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43612-2", + "nodes": [ + "pull_request:43612", + "pull_request:43613" + ], + "soft_pairs": [ + "pull_request:43612|pull_request:43613" + ], + "prepared_review_unit_hash": "9e434d983d855ba8a2a67d8eff95524cc9b8164ca9304b918d27b4a60b82ef69", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6081, + "estimated_input_tokens": 1521, + "estimated_eval_tokens": 3298 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6081, + "estimated_input_tokens": 1521, + "estimated_eval_tokens": 3298 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two distinct PRs adding separate segmentation pipelines: one for promptable concept segmentation and one for promptable visual segmentation. They share documentation and pipeline scaffolding, but the underlying task/model support differs enough that they should not be clustered as duplicates.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #43613 is the broader and more representative pipeline addition in this cluster: it adds promptable visual segmentation with more model/test coverage and a fuller integration surface.", + "best_issue_reason": null, + "best_pr_reason": "PR #43613 is the strongest standalone candidate because it appears to implement the more complete, broader pipeline feature and has the richer set of supporting changes.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43612", + "right": "pull_request:43613", + "accept": false, + "reason": "Reject: these are different features targeting different tasks/code paths (concept segmentation vs visual segmentation). Shared docs and pipeline plumbing are not enough to treat them as the same change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 10, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43651-2", + "nodes": [ + "pull_request:43651", + "pull_request:43663" + ], + "soft_pairs": [ + "pull_request:43651|pull_request:43663" + ], + "prepared_review_unit_hash": "6d2af6a1f09caec203bdd5806e103583150e314a05e3d728f9a97c953c97dd9e", + "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": "Both pull requests modify `src/transformers/trainer.py`, but they introduce different trainer extensibility hooks in different code paths: one changes gradient-accumulation loss scaling, the other extracts signature-column filtering into an overridable helper. They are related to Trainer customization, but not the same underlying change.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #43663 is the better representative of the cluster because it refactors a core Trainer helper into an overridable method and is the broader API-style change.", + "best_issue_reason": null, + "best_pr_reason": "PR #43663 is the stronger standalone candidate: it cleanly exposes a reusable hook for custom trainers and is more generally applicable than the loss-scaling tweak.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43651", + "right": "pull_request:43663", + "accept": false, + "reason": "Different trainer hooks and different behaviors: gradient-accumulation loss scaling vs. signature-column filtering. Same file/subsystem, but not the same bug or mergeable change set." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 11, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43656-2", + "nodes": [ + "pull_request:43656", + "pull_request:43842" + ], + "soft_pairs": [ + "pull_request:43656|pull_request:43842" + ], + "prepared_review_unit_hash": "beda9940a0dad8634dced4c5402b2ed57e6ae56ddcb122871f6db6bb6eba3a75", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 3664, + "estimated_input_tokens": 916, + "estimated_eval_tokens": 2088 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 3664, + "estimated_input_tokens": 916, + "estimated_eval_tokens": 2088 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two PRs are related because they both touch `src/transformers/cli/serve.py` and address annotation-related `NameError`s, but they appear to fix different runtime paths: one broadly enables postponed annotation evaluation and updates several signatures, while the other specifically guards `TypeAdapter` behind `TYPE_CHECKING` for the pydantic-optional case. Related, yes; same fix cluster, not clearly.", + "confidence": 0.58, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #43656 is the broader serve-CLI annotation fix and is the better representative of the shared file-level problem; #43842 is a narrower follow-up for a specific `TypeAdapter` import/evaluation case.", + "best_issue_reason": null, + "best_pr_reason": "PR #43656 is the best single representative because it addresses the more general annotation-evaluation issue in `serve.py`, while #43842 is a targeted variant tied to `TypeAdapter` and optional pydantic handling.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43656", + "right": "pull_request:43842", + "accept": false, + "reason": "Both PRs are annotation-related and in the same file, but they solve different concrete issues: one uses `from __future__ import annotations` and signature rewrites, the other adds a `TYPE_CHECKING` import for `TypeAdapter`. That is related, but not enough to say they are the same underlying bugfix or merge into one PR." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The relationship is plausible, but the evidence only clearly supports that both PRs touch `src/transformers/cli/serve.py` and both mitigate annotation-related NameErrors. That is weaker than saying they target the same underlying code path or the same bugfix. One PR adds `from __future__ import annotations` and adjusts several model/processor annotations; the other specifically handles `TypeAdapter` under `TYPE_CHECKING` for the pydantic-optional case. I\u2019d keep these as related, but not assert they are the same fix cluster with this level of confidence." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 12, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43779-2", + "nodes": [ + "pull_request:43779", + "pull_request:43816" + ], + "soft_pairs": [ + "pull_request:43779|pull_request:43816" + ], + "prepared_review_unit_hash": "e7c674adcc031aa8883109a5dfc678b4407fb20b10764302092e384d031ba83a", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4362, + "estimated_input_tokens": 1091, + "estimated_eval_tokens": 2438 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4362, + "estimated_input_tokens": 1091, + "estimated_eval_tokens": 2438 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both pull requests appear to implement the same SwanLab integration change: adding run id/resume support in `integration_utils.py` for issue #43698, with one variant also expanding the docs. They are strong duplicate PR candidates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #43816 is the cleaner representative: it carries the same SwanLab id/resume support plus the accompanying documentation updates, making it the more complete version of the shared change.", + "best_issue_reason": null, + "best_pr_reason": "PR #43816 is the best global representative because it appears to include the full feature change and docs for SwanLab run id/resume support, while #43779 is the narrower implementation-focused variant.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43779", + "right": "pull_request:43816", + "accept": true, + "reason": "Same underlying SwanLab callback enhancement for id/resume support in the same file and same issue target; the doc differences do not change the core code-path change, so these are duplicate PRs." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: both PRs target issue #43698, touch the same file, and the previews indicate the same SwanLab id/resume support with #43816 adding docs. The duplicate/mergeable verdict is conservative enough given the shared target and overlapping code change." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 13, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43996-28", + "nodes": [ + "pull_request:43996", + "pull_request:43997", + "pull_request:44007", + "pull_request:44010", + "pull_request:44013", + "pull_request:44017", + "pull_request:44018", + "pull_request:44019", + "pull_request: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:44116", + "pull_request:44068|pull_request:44722", + "pull_request:44018|pull_request:44116", + "pull_request:44071|pull_request:44722", + "pull_request:44018|pull_request:44085", + "pull_request:44018|pull_request:44086", + "pull_request:44068|pull_request:44086", + "pull_request:44071|pull_request:44085", + "pull_request:44059|pull_request:44086", + "pull_request:44085|pull_request:44086", + "pull_request:44098|pull_request:44101", + "pull_request:44059|pull_request:44722", + "pull_request:44018|pull_request:44071", + "pull_request:44017|pull_request:44161", + "pull_request:44017|pull_request:44722", + "pull_request:44017|pull_request:44073", + "pull_request:44068|pull_request:44071", + "pull_request:43997|pull_request:44073", + "pull_request:44071|pull_request:44086", + "pull_request:44017|pull_request:44071", + "pull_request:44073|pull_request:44101", + "pull_request:44073|pull_request:44161", + "pull_request:44059|pull_request:44071", + "pull_request:44056|pull_request:44101", + "pull_request:44018|pull_request:44059", + "pull_request:44059|pull_request:44068", + "pull_request:44101|pull_request:44161", + "pull_request:43997|pull_request:44017", + "pull_request:43997|pull_request:44056", + "pull_request:44071|pull_request:44076", + "pull_request:44074|pull_request:44101", + "pull_request:44056|pull_request:44161", + "pull_request:43997|pull_request:44018", + "pull_request:43997|pull_request:44161", + "pull_request:44074|pull_request:44161", + "pull_request:44018|pull_request:44161", + "pull_request:43997|pull_request:44071", + "pull_request:44010|pull_request:44071", + "pull_request:44010|pull_request:44073", + "pull_request:44018|pull_request:44129", + "pull_request:43997|pull_request:44074", + "pull_request:44010|pull_request:44018", + "pull_request:43997|pull_request:44129", + "pull_request:44071|pull_request:44129", + "pull_request:43996|pull_request:44161", + "pull_request:43996|pull_request:44017", + "pull_request:44010|pull_request:44044", + "pull_request:44044|pull_request:44074", + "pull_request:44066|pull_request:44722", + "pull_request:44085|pull_request:44722", + "pull_request:44085|pull_request:44129", + "pull_request:43996|pull_request:44007", + "pull_request:43996|pull_request:44129", + "pull_request:44044|pull_request:44129", + "pull_request:44044|pull_request:44154" + ], + "prepared_review_unit_hash": "adf3ac6614a6cd9cffe15d9d9ef338d1a23e3ed997df208aae45b29032a331df", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 78, + "serialized_chars": 84904, + "estimated_input_tokens": 21226, + "estimated_eval_tokens": 42708 + }, + "budget": { + "node_count": 27, + "item_count": 27, + "soft_pair_count": 72, + "serialized_chars": 75099, + "estimated_input_tokens": 18775, + "estimated_eval_tokens": 37806 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a collection of separate model-specific output-tracing refactors, not one broad duplicate set. The only clear duplicate-like subcluster is the GPT-J/CodeGen thread, plus a smaller GPT-Neo pair.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": "44722 is the best representative of the GPT-J/CodeGen subcluster: it directly targets #43979, touches the same output-tracing code path as the closest neighbors, and is the broadest/latest variant in that thread.", + "best_issue_reason": null, + "best_pr_reason": "44722 is the strongest single PR anchor here because it is central to the only real overlap cluster and has the clearest direct relationship to the shared output-tracing refactor work.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43996", + "right": "pull_request:44085", + "accept": false, + "reason": "Different model families and no shared concrete code path; both are only loosely about output tracing." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44044", + "accept": false, + "reason": "Different models and different refactors; shared issue target is not enough." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44085", + "accept": true, + "reason": "Both are GPT-J output-tracing refactors in the same code path; 44066 just bundles an extra CodeGen change." + }, + { + "left": "pull_request:44007", + "right": "pull_request:44072", + "accept": false, + "reason": "Different model implementations; same theme, but not the same bug/change." + }, + { + "left": "pull_request:44072", + "right": "pull_request:44722", + "accept": false, + "reason": "Different models; output-tracing refactors are separate per architecture." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44072", + "accept": false, + "reason": "CodeGen/GPT-J work versus EfficientNet; unrelated concrete code paths." + }, + { + "left": "pull_request:44013", + "right": "pull_request:44044", + "accept": false, + "reason": "MobileNetV2 versus DeBERTa-v2; only a broad standardization theme is shared." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-J/CodeGen versus MGP-STR; different model code paths." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-J/CodeGen versus MPT; same infrastructure theme, different change." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44068", + "accept": true, + "reason": "Same GPT-Neo files and same output-tracing refactor; these look like the same underlying change." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT-J/CodeGen versus GPT-Neo; not the same model or code path." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44066", + "accept": false, + "reason": "GPT-Neo versus GPT-J/CodeGen; only the output-tracing pattern matches." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44722", + "accept": false, + "reason": "ResNet versus GPT-J/CodeGen; unrelated model refactors." + }, + { + "left": "pull_request:44086", + "right": "pull_request:44722", + "accept": false, + "reason": "MGP-STR versus GPT-J/CodeGen; different concrete code paths." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44085", + "accept": false, + "reason": "ResNet versus GPT-J; different model families." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44019", + "accept": false, + "reason": "GPT-Neo and ResNet are unrelated implementations." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44071", + "accept": false, + "reason": "ResNet versus MPT; same broad refactor style, different model code." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo versus Flaubert/XLM; not the same underlying change." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT-Neo versus GPT-J/CodeGen; distinct model-specific refactors." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo versus Flaubert/XLM; no shared code path." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44722", + "accept": false, + "reason": "MPT versus GPT-J/CodeGen; different architectures." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44085", + "accept": false, + "reason": "GPT-Neo versus GPT-J; not the same model or change." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo versus MGP-STR; unrelated code paths." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo versus MGP-STR; only the refactor style is similar." + }, + { + "left": "pull_request:44098", + "right": "pull_request:44101", + "accept": false, + "reason": "ViLT versus Flaubert/XLM; different model-specific changes." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-2 versus MPT; not the same underlying bug/change." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44086", + "accept": false, + "reason": "MPT versus MGP-STR; different code paths." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44161", + "accept": false, + "reason": "SegFormer versus LongT5; different refactors in different models." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44017", + "accept": false, + "reason": "RegNet versus SegFormer; same output-tracing theme, different model code." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44056", + "accept": false, + "reason": "RegNet versus MPNet; unrelated concrete changes." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44101", + "accept": false, + "reason": "TextNet versus Flaubert/XLM; not the same change." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44161", + "accept": false, + "reason": "MPNet versus LongT5; different model implementations." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44018", + "accept": false, + "reason": "RegNet versus GPT-Neo; different code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44161", + "accept": false, + "reason": "RegNet versus LongT5; unrelated model refactors." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44161", + "accept": false, + "reason": "TextNet versus LongT5; different concrete changes." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-Neo versus MPT; different model code." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44071", + "accept": false, + "reason": "SqueezeBERT versus MPT; same refactor pattern, different implementation." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44073", + "accept": false, + "reason": "SqueezeBERT versus VisualBERT; unrelated models." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-Neo versus SpeechT5; not the same code path." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44101", + "accept": false, + "reason": "VisualBERT versus Flaubert/XLM; different model changes." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT-2 versus GPT-Neo; only the output-tracing theme overlaps." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44101", + "accept": false, + "reason": "MPNet versus Flaubert/XLM; different implementations." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44059", + "accept": false, + "reason": "GPT-Neo versus GPT-2; same broad subsystem, but not the same change." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-2 versus MGP-STR; unrelated models." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44076", + "accept": false, + "reason": "MPT versus ImageGPT; separate refactors." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44061", + "accept": false, + "reason": "No matching concrete duplicate change is evident." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44044", + "accept": false, + "reason": "SqueezeBERT versus DeBERTa-v2; different model-specific work." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44074", + "accept": false, + "reason": "DeBERTa-v2 versus TextNet; unrelated changes." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44722", + "accept": true, + "reason": "Same GPT-J/CodeGen output-tracing refactor thread, with nearly identical code path and issue target." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44722", + "accept": true, + "reason": "Both are GPT-J output-tracing refactors in the same file family; 44722 is the broader version." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-J versus SpeechT5; different models and code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44007", + "accept": false, + "reason": "CVT/FNet versus RegNet/ResNet/RT-DETR-ResNet; different model families." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44129", + "accept": false, + "reason": "CVT/FNet versus SpeechT5; unrelated changes." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44129", + "accept": false, + "reason": "DeBERTa-v2 versus SpeechT5; different concrete implementations." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44154", + "accept": false, + "reason": "DeBERTa-v2 versus ViTS; only the output-tracing pattern matches, not the change." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is well grounded in the packet: it correctly identifies the GPT-J/CodeGen thread as the only strong duplicate-like cluster, the GPT-Neo pair as a smaller subcluster, and treats the rest as separate model-specific refactors. The accepted soft-edge verdicts are conservative and look mergeable as same-bugfix PRs within the same code path." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 13, + "component_count": 36, + "review_unit_index": 2, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43996-28", + "nodes": [ + "pull_request:43997", + "pull_request:44010", + "pull_request: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": "0d925e397e58ecbb477ba55dfa52b11beb35cc38545565ce26370a216ae3bede", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 78, + "serialized_chars": 84904, + "estimated_input_tokens": 21226, + "estimated_eval_tokens": 42708 + }, + "budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 6, + "serialized_chars": 17126, + "estimated_input_tokens": 4282, + "estimated_eval_tokens": 8820 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These PRs are all model-specific refactors to the new standardized output-capture/tracing machinery, but they touch different model implementations and do not look mergeable as duplicates of the same concrete fix.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44073 is the cleanest representative of the cluster: it clearly introduces the standardized output-capture pattern in a single model, with explicit _can_record_outputs wiring and the new decorator imports.", + "best_issue_reason": null, + "best_pr_reason": "PR 44073 is the best overall representative because its title and diff most directly express the shared refactor theme without being unusually narrow or noisy.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44024", + "right": "pull_request:44076", + "accept": false, + "reason": "Both are similar output-tracing refactors, but they apply to different models (FocalNet vs ImageGPT) and distinct code paths, so they are not the same underlying change." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44074", + "accept": false, + "reason": "VisualBert and TextNet are separate model implementations; these are parallel refactors, not one shared bugfix or one PR that could plausibly subsume the other." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44073", + "accept": false, + "reason": "MPNet and VisualBert each add output-capture plumbing in different model files; same broad subsystem, but not the same concrete change." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44074", + "accept": false, + "reason": "MPNet and TextNet are unrelated model-specific refactors, even though both migrate to the same decorator-based output collection style." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44010", + "accept": false, + "reason": "RegNet and SqueezeBert changes are separate model migrations to the new output tracing API, not a duplicate fix on the same code path." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44129", + "accept": false, + "reason": "SqueezeBert and SpeechT5 affect different model stacks and different output-tracing implementations; they should stay separate." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 14, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43998-13", + "nodes": [ + "pull_request:43998", + "pull_request:43999", + "pull_request:44000", + "pull_request:44001", + "pull_request:44002", + "pull_request:44003", + "pull_request:44004", + "pull_request:44025", + "pull_request:44026", + "pull_request:44027", + "pull_request:44028", + "pull_request:44029", + "pull_request:44030" + ], + "soft_pairs": [ + "pull_request:43998|pull_request:44002", + "pull_request:43998|pull_request:44000", + "pull_request:43998|pull_request:44001", + "pull_request:43998|pull_request:44003", + "pull_request:43998|pull_request:44029", + "pull_request:43999|pull_request:44001", + "pull_request:43999|pull_request:44002", + "pull_request:43999|pull_request:44025", + "pull_request:43999|pull_request:44026", + "pull_request:43999|pull_request:44027", + "pull_request:44000|pull_request:44001", + "pull_request:44000|pull_request:44002", + "pull_request:44000|pull_request:44003", + "pull_request:44000|pull_request:44025", + "pull_request:44001|pull_request:44002", + "pull_request:44001|pull_request:44003", + "pull_request:44001|pull_request:44025", + "pull_request:44002|pull_request:44029", + "pull_request:44002|pull_request:44003", + "pull_request:44003|pull_request:44028", + "pull_request:44003|pull_request:44004", + "pull_request:43998|pull_request:44004", + "pull_request:44001|pull_request:44004", + "pull_request:44002|pull_request:44004", + "pull_request:44000|pull_request:44004", + "pull_request:44025|pull_request:44026", + "pull_request:44025|pull_request:44027", + "pull_request:44025|pull_request:44028", + "pull_request:44025|pull_request:44029", + "pull_request:44025|pull_request:44030", + "pull_request:44026|pull_request:44029", + "pull_request:44026|pull_request:44027", + "pull_request:44026|pull_request:44028", + "pull_request:44026|pull_request:44030", + "pull_request:44027|pull_request:44028", + "pull_request:44027|pull_request:44029", + "pull_request:44027|pull_request:44030", + "pull_request:44028|pull_request:44029", + "pull_request:44028|pull_request:44030", + "pull_request:44029|pull_request:44030" + ], + "prepared_review_unit_hash": "9c14ec8944788469923035968933b472224a9ec24a55c8a09d8b9be3835908cb", + "component_budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 40, + "serialized_chars": 40181, + "estimated_input_tokens": 10046, + "estimated_eval_tokens": 20348 + }, + "budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 40, + "serialized_chars": 40181, + "estimated_input_tokens": 10046, + "estimated_eval_tokens": 20348 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is a set of separate per-model output-tracing refactors, not duplicate fixes. The similar titles and diff patterns come from a shared refactor template, but each PR targets a different architecture and code path.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "No true canonical duplicate exists here; the PRs are model-specific refactors rather than the same change. If one representative is needed, PR 44003 is the broadest example because it spans both mamba and falcon_mamba.", + "best_issue_reason": null, + "best_pr_reason": "PR 44003 is the best cluster representative because it has the broadest scope and the most substantial output-tracing rewrite among the candidates, but it is still not a duplicate of the others.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43998", + "right": "pull_request:44002", + "accept": false, + "reason": "Different model families and forward paths; same refactor motif only." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44000", + "accept": false, + "reason": "Timm backbone vs vision-text dual encoder; unrelated code paths." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44001", + "accept": false, + "reason": "Timm backbone vs UnivNet; shared output-tracing theme is not enough." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44003", + "accept": false, + "reason": "Different architectures (timm_backbone vs mamba/falcon_mamba); not the same bug/change." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44029", + "accept": false, + "reason": "Timm backbone and RWKV are distinct models with separate implementation paths." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44001", + "accept": false, + "reason": "MobileNetV1 and UnivNet are unrelated; only the refactor pattern matches." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44002", + "accept": false, + "reason": "MobileNetV1 vs UperNet; different model code paths." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44025", + "accept": false, + "reason": "MobileNetV1 and Depth Anything are separate architectures; not mergeable as one PR." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44026", + "accept": false, + "reason": "MobileNetV1 vs VisionEncoderDecoder; same template, different change." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44027", + "accept": false, + "reason": "MobileNetV1 and SpeechEncoderDecoder are different code paths." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44001", + "accept": false, + "reason": "Vision-text dual encoder and UnivNet are unrelated models." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44002", + "accept": false, + "reason": "Different architectures and call stacks; no shared underlying defect." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44003", + "accept": false, + "reason": "Vision-text dual encoder vs mamba/falcon_mamba; not the same concrete change." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44025", + "accept": false, + "reason": "Different model families; only broadly similar output-handling refactors." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44002", + "accept": false, + "reason": "UnivNet and UperNet are separate code paths with different outputs." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44003", + "accept": false, + "reason": "UnivNet and mamba/falcon_mamba do not share the same bug or fix." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44025", + "accept": false, + "reason": "Different architectures; same refactor wording is not enough for duplicate status." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44029", + "accept": false, + "reason": "UperNet and RWKV are distinct model implementations." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44003", + "accept": false, + "reason": "UperNet vs mamba/falcon_mamba; different concrete changes." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44028", + "accept": false, + "reason": "Mamba/FalconMamba and SuperPoint are unrelated models." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44004", + "accept": false, + "reason": "Different subsystems (mamba vs codegen) and different forward semantics." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44004", + "accept": false, + "reason": "Timm backbone and CodeGen are unrelated code paths." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44004", + "accept": false, + "reason": "UnivNet and CodeGen are different model families." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44004", + "accept": false, + "reason": "UperNet and CodeGen do not share a single underlying bug/change." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44004", + "accept": false, + "reason": "Vision-text dual encoder vs CodeGen; same refactor theme only." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44026", + "accept": false, + "reason": "Depth Anything and VisionEncoderDecoder are separate architectures." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44027", + "accept": false, + "reason": "Depth Anything vs SpeechEncoderDecoder; not the same code-path problem." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44028", + "accept": false, + "reason": "Depth Anything and SuperPoint are unrelated implementations." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44029", + "accept": false, + "reason": "Depth Anything and RWKV are distinct models with different refactors." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44030", + "accept": false, + "reason": "Depth Anything vs DPR; different code paths and outputs." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44029", + "accept": false, + "reason": "VisionEncoderDecoder and RWKV are unrelated models." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44027", + "accept": false, + "reason": "VisionEncoderDecoder and SpeechEncoderDecoder are different architectures." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44028", + "accept": false, + "reason": "VisionEncoderDecoder vs SuperPoint; same refactor motif only." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44030", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44028", + "accept": false, + "reason": "SpeechEncoderDecoder and SuperPoint are unrelated implementations." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44029", + "accept": false, + "reason": "SpeechEncoderDecoder and RWKV do not share a concrete fix." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44030", + "accept": false, + "reason": "SpeechEncoderDecoder vs DPR; different underlying changes." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44029", + "accept": false, + "reason": "SuperPoint and RWKV are different model code paths." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44030", + "accept": false, + "reason": "SuperPoint vs DPR; only the output-tracing pattern overlaps." + }, + { + "left": "pull_request:44029", + "right": "pull_request:44030", + "accept": false, + "reason": "RWKV and DPR are separate architectures with different fixes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 15, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-44662-2", + "nodes": [ + "pull_request:44662", + "pull_request:45401" + ], + "soft_pairs": [ + "pull_request:44662|pull_request:45401" + ], + "prepared_review_unit_hash": "471ad25dd9aaa744d9cc6a244e8bfb453617b7bfb512c0295465ddc4fb64b990", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6551, + "estimated_input_tokens": 1638, + "estimated_eval_tokens": 3532 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6551, + "estimated_input_tokens": 1638, + "estimated_eval_tokens": 3532 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are two separate model-implementation PRs: one adds PenguinVL, the other adds Voxtral TTS. They share standard transformers scaffolding, but not the same underlying bug or change.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "If a representative PR is needed, 45401 is the cleaner one to anchor on because it has an explicit issue target and a narrower, self-contained TTS feature scope.", + "best_issue_reason": null, + "best_pr_reason": "45401 is the better global representative among the two because it is explicitly tied to an issue and is more narrowly scoped; 44662 is a larger standalone model addition with no issue linkage.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44662", + "right": "pull_request:45401", + "accept": false, + "reason": "Different model additions (PenguinVL vs Voxtral TTS) with different code paths and goals; shared registration/docs boilerplate is not enough to treat them as the same change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 16, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-44979-2", + "nodes": [ + "pull_request:44979", + "pull_request:45363" + ], + "soft_pairs": [ + "pull_request:44979|pull_request:45363" + ], + "prepared_review_unit_hash": "4f813547662fe17b524bd147e5b827ffffa8e3377f87073f58e15b4bb03b4057", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4407, + "estimated_input_tokens": 1102, + "estimated_eval_tokens": 2460 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4407, + "estimated_input_tokens": 1102, + "estimated_eval_tokens": 2460 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are related but distinct pull requests: one introduces a generic Module Fusion API, while the other wires up n-to-1 kernel fusion through KernelConfig and adds an example. They do not look like the same mergeable change.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #44979 is the more canonical representative because it introduces the foundational Module Fusion API; PR #45363 is an application/integration on top of that API rather than the same change.", + "best_issue_reason": null, + "best_pr_reason": "PR #44979 is the best cluster representative since it defines the core abstraction and new module implementation, whereas PR #45363 mainly consumes that work to enable a specific kernel-fusion path.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44979", + "right": "pull_request:45363", + "accept": false, + "reason": "Different underlying changes: #44979 adds the generic fusion API/module, while #45363 integrates n-to-1 kernel fusion via KernelConfig and example code. They are related but not the same concrete fix/change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 17, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45086-2", + "nodes": [ + "pull_request:45086", + "pull_request:45317" + ], + "soft_pairs": [ + "pull_request:45086|pull_request:45317" + ], + "prepared_review_unit_hash": "8114a26b465bc41c06f67a8b949f8a8c89fb41322276beb6af5243829c38e5f4", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4414, + "estimated_input_tokens": 1104, + "estimated_eval_tokens": 2464 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4414, + "estimated_input_tokens": 1104, + "estimated_eval_tokens": 2464 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both pull requests address the same Mistral tokenizer bug in `_patch_mistral_regex`: using `tokenizer.backend_tokenizer.pre_tokenizer` causes an `AttributeError`, and both switch to `tokenizer.pre_tokenizer`. PR 45317 is the more complete fix because it includes test coverage.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45317 is the better canonical PR because it implements the same fix as 45086 and adds tests, making it the more complete and reviewable change.", + "best_issue_reason": null, + "best_pr_reason": "PR 45317 is the strongest representative of the cluster: same concrete bug fix, same file, same tracked issue, plus tests that validate the regression.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45086", + "right": "pull_request:45317", + "accept": true, + "reason": "Same underlying AttributeError fix in `_patch_mistral_regex`; the second PR is essentially a superset with tests and the same code-path change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 18, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45115-4", + "nodes": [ + "pull_request:45115", + "pull_request:45569", + "pull_request:46332", + "pull_request:46565" + ], + "soft_pairs": [ + "pull_request:45115|pull_request:45569", + "pull_request:45569|pull_request:46565", + "pull_request:46332|pull_request:46565" + ], + "prepared_review_unit_hash": "632cc621e59a96cb262169c87d55237dabc795c8d8b0c07b3a55bb6ceebd9fd3", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 3, + "serialized_chars": 13757, + "estimated_input_tokens": 3440, + "estimated_eval_tokens": 7136 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 3, + "serialized_chars": 13757, + "estimated_input_tokens": 3440, + "estimated_eval_tokens": 7136 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is heterogeneous: two NemotronH refactor/architecture PRs and two unrelated Nemotron ASR streaming PRs. None of the soft pairs look like the same underlying bug or change, so there is no true duplicate/canonical merge candidate across the cluster.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "46565 is the strongest standalone PR in the cluster because it is the broadest/latest ASR-streaming feature change, but it is still not a duplicate of the other PRs and only weakly represents the cluster.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45115", + "right": "pull_request:45569", + "accept": false, + "reason": "Both are NemotronH-related, but one is a GranitemoeHybrid inheritance/refactor and the other adds a dense/sparse split plus docs/auto wiring. Similar subsystem, different change." + }, + { + "left": "pull_request:45569", + "right": "pull_request:46565", + "accept": false, + "reason": "These touch overlapping auto/docs files, but they target different models and problems: NemotronH refactor vs Nemotron ASR streaming. Not the same code-path fix." + }, + { + "left": "pull_request:46332", + "right": "pull_request:46565", + "accept": false, + "reason": "Both are ASR streaming PRs, but they cover different model families and different implementation scopes. They are related features, not the same underlying change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 19, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45332-2", + "nodes": [ + "pull_request:45332", + "pull_request:45333" + ], + "soft_pairs": [ + "pull_request:45332|pull_request:45333" + ], + "prepared_review_unit_hash": "d3a7fbe432f48fcd8ee23438fe62d74a0f3d6926c48daf640f1f1d25561aaf9b", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5465, + "estimated_input_tokens": 1367, + "estimated_eval_tokens": 2990 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5465, + "estimated_input_tokens": 1367, + "estimated_eval_tokens": 2990 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are related feature PRs for heterogeneous model support, but they split the work into a config-layer foundation and a broader modeling/integration change. They share files and scope, yet they do not look like the same concrete change set or a single fix that should be deduplicated.", + "confidence": 0.79, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45332 is the better representative because it is the broader end-to-end feature PR: it includes the per-layer config integration plus cache/modeling changes needed to actually use heterogeneous configs across the stack.", + "best_issue_reason": null, + "best_pr_reason": "PR #45332 is the strongest single PR for the cluster because it covers the full heterogeneous model support path, while #45333 is the narrower config-only foundation.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45332", + "right": "pull_request:45333", + "accept": false, + "reason": "Related but not duplicates: #45333 adds per-layer configuration infrastructure, while #45332 builds on that with cache/modeling support and other integrations. They touch overlapping files, but they do not appear to fix the same concrete code-path problem and would not naturally collapse into one PR without losing separation of concerns." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 20, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45415-2", + "nodes": [ + "pull_request:45415", + "pull_request:45425" + ], + "soft_pairs": [ + "pull_request:45415|pull_request:45425" + ], + "prepared_review_unit_hash": "003cf4a47dc8f8c69f173d6a66f6fb55c3ed2ccca7eb20e10e87fdeb9eaf8e42", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5514, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5514, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are both typing-related PRs in transformers, but they are not the same change: one is a broad type-checking sweep across many files, while the other is a focused typing-support patch for modeling_utils and related integrations.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45415 is the broader umbrella typing pass across many files, while PR 45425 is a narrower follow-up touching a subset of shared typing helpers.", + "best_issue_reason": null, + "best_pr_reason": "PR 45415 is the better cluster representative because it is the more general, cross-cutting typing cleanup, but it still does not match 45425 as a duplicate.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45415", + "right": "pull_request:45425", + "accept": false, + "reason": "Reject: the overlap is only in general typing infrastructure. 45415 is a broad repo-wide type-checking sweep; 45425 adds specific protocols/aliases for modeling_utils and PEFT-related code, so they do not fix the same concrete change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 21, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45513-2", + "nodes": [ + "pull_request:45513", + "pull_request:46084" + ], + "soft_pairs": [ + "pull_request:45513|pull_request:46084" + ], + "prepared_review_unit_hash": "9786ea98f9d976b14f711abcd84ac77996fbeab8c7668d46cb316f9bb68348fd", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4949, + "estimated_input_tokens": 1238, + "estimated_eval_tokens": 2732 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4949, + "estimated_input_tokens": 1238, + "estimated_eval_tokens": 2732 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are two unrelated PRs that both adjust cached forward behavior for sequence-length > 1, but in different model families and code paths. They should not be deduplicated.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #46084 is the better canonical PR for this cluster because it is narrowly scoped, has an explicit issue target, and directly fixes a concrete use_cache bug for seq_len > 1. PR #45513 is broader and spans multiple Qwen/Olmo model files.", + "best_issue_reason": null, + "best_pr_reason": "#46084 is the best representative PR: it has the clearest bug statement, an explicit linked issue, and a focused patch. #45513 is more sprawling and appears to cover a different model-specific caching defect.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45513", + "right": "pull_request:46084", + "accept": false, + "reason": "Different underlying bugs in different model implementations: Qwen3.5/Olmo hybrid linear attention cached forward vs Mamba2 use_cache with seq_len > 1. Similar theme, but not the same fixable code-path." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 22, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45548-2", + "nodes": [ + "pull_request:45548", + "pull_request:45662" + ], + "soft_pairs": [ + "pull_request:45548|pull_request:45662" + ], + "prepared_review_unit_hash": "ce2384b1d5570fac25ad2d811949d841a3b4fa0c67e5d6c31617ca2343434cb0", + "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 parallelism, but they fix different bugs on different loading paths: one targets DeepSpeed ZeRO-3/accelerate launch config handling, the other fixes expert parameters being overwritten during FSDP2/distributed loading. They should not be merged as duplicates.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45662 is the better canonical PR because it fixes a concrete expert-parameter overwrite bug in the distributed loading path and has the broader, more direct code-path impact.", + "best_issue_reason": null, + "best_pr_reason": "PR 45662 is the strongest representative of the cluster: it addresses a specific, user-visible EP failure mode with a focused fix across loading and tensor-parallel integration.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45548", + "right": "pull_request:45662", + "accept": false, + "reason": "They share the broad EP theme, but 45548 is about DeepSpeed ZeRO-3 loading/config serialization via accelerate launch, while 45662 fixes experts being silently overwritten by rank-0 broadcast during FSDP2/distributed loading. Different concrete bugs and code paths." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 23, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45639-2", + "nodes": [ + "pull_request:45639", + "pull_request:45645" + ], + "soft_pairs": [ + "pull_request:45639|pull_request:45645" + ], + "prepared_review_unit_hash": "9330c25b42182b406fdfb91d6274229093a873a82a1f78f85cefc2edbc65e0b8", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4589, + "estimated_input_tokens": 1148, + "estimated_eval_tokens": 2552 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4589, + "estimated_input_tokens": 1148, + "estimated_eval_tokens": 2552 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs address the same xdist-related captured_info logging collision in testing_utils; PR 45645 is the broader, more complete fix that also updates CI to read multiple log files.", + "confidence": 0.84, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45645 is the better canonical PR because it contains the core xdist-safe logging fix plus the CI/workflow adjustments needed to consume the new per-worker artifacts.", + "best_issue_reason": null, + "best_pr_reason": "PR 45645 is the strongest representative of the change set: same underlying bug, broader coverage, and the supporting workflow updates make it the more complete fix.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45639", + "right": "pull_request:45645", + "accept": true, + "reason": "Both PRs fix the same concrete problem: xdist collisions around captured_info output in testing_utils. 45645 appears to extend 45639 with the same core fix plus CI handling for multiple output files, so they are plausibly mergeable into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 24, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45670-2", + "nodes": [ + "pull_request:45670", + "pull_request:46320" + ], + "soft_pairs": [ + "pull_request:45670|pull_request:46320" + ], + "prepared_review_unit_hash": "da4ca3f9439594ae1425fd31d61b760794e161976890f6ba44b4b533c6866dd8", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4375, + "estimated_input_tokens": 1094, + "estimated_eval_tokens": 2444 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4375, + "estimated_input_tokens": 1094, + "estimated_eval_tokens": 2444 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two PRs are unrelated: one updates AutoModel registration for glmasr/dia, while the other adds RecurrentGemma attention-dispatch support plus docs/tests. They do not appear to be the same underlying change or a single mergeable fix.", + "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:45670", + "right": "pull_request:46320", + "accept": false, + "reason": "Different features and code paths: PR 45670 only changes auto-model class mapping, while PR 46320 modifies RecurrentGemma attention implementation dispatch and related docs/tests. They are not the same bug/change and would not be merged into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 25, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45720-2", + "nodes": [ + "pull_request:45720", + "pull_request:46259" + ], + "soft_pairs": [ + "pull_request:45720|pull_request:46259" + ], + "prepared_review_unit_hash": "73224546847048c491f4fc84f40209fdd3c8effd059623808a285d6ce8ab5fd0", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 7248, + "estimated_input_tokens": 1812, + "estimated_eval_tokens": 3880 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 7248, + "estimated_input_tokens": 1812, + "estimated_eval_tokens": 3880 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These PRs are only superficially similar: both are linter-driven, but one adds tuple-return handling for model forwards (TRF003) while the other expands weight-initialization modeling rules (TRF018). They address different code paths and should not be merged as duplicates.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: 45720 and 46259 implement different lint rules and fix different behaviors, so they are separate changes.", + "best_issue_reason": null, + "best_pr_reason": "Neither PR is a good global best for the cluster because they solve unrelated problems; if forced to pick a representative, 46259 is broader, but the cluster should remain split.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45720", + "right": "pull_request:46259", + "accept": false, + "reason": "Different rules and fixes: 45720 adds can_return_tuple/return_dict handling across model forwards, while 46259 refactors _init_weights behavior for many modules. Shared files are incidental, not the same underlying change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 26, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45737-2", + "nodes": [ + "pull_request:45737", + "pull_request:45738" + ], + "soft_pairs": [ + "pull_request:45737|pull_request:45738" + ], + "prepared_review_unit_hash": "cc2ade73a49ec6a8883a7a24c433988645588bcd049e8aad2bce56b1b24daa1c", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4131, + "estimated_input_tokens": 1033, + "estimated_eval_tokens": 2322 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4131, + "estimated_input_tokens": 1033, + "estimated_eval_tokens": 2322 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs address the same MusicgenMelody cache initialization bug in `forward()`: replacing `EncoderDecoderCache(DynamicCache, DynamicCache)` with a single `DynamicCache`. PR #45737 is the more complete duplicate because it bundles the fix with a regression test, while #45738 is the fix-only variant.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45737 is the best canonical PR because it contains the same code fix as #45738 plus a regression test, making it the most complete representation of the bug fix.", + "best_issue_reason": null, + "best_pr_reason": "PR #45737 is the strongest overall candidate: it fixes the concrete cache bug and adds a targeted regression test, whereas #45738 only changes the implementation.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45737", + "right": "pull_request:45738", + "accept": true, + "reason": "Same underlying code-path bug in `musicgen_melody` cache setup; the diffs match on the core fix and the test-only addition in #45737 could plausibly live in the same PR as #45738." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is grounded in the packet: both PRs make the same `musicgen_melody` cache initialization change, and #45737 only adds a regression test on top of that fix. The soft-edge accept is conservative enough because the diffs are mergeable into one PR for the same bugfix." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 27, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45932-2", + "nodes": [ + "pull_request:45932", + "pull_request:46100" + ], + "soft_pairs": [ + "pull_request:45932|pull_request:46100" + ], + "prepared_review_unit_hash": "57ceb34b13313f7cd456e992b4bd2930d7132d5e070f4d3441193e98aded017f", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5160, + "estimated_input_tokens": 1290, + "estimated_eval_tokens": 2836 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5160, + "estimated_input_tokens": 1290, + "estimated_eval_tokens": 2836 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two PRs are not duplicates: one is a model/workflow change with telemetry and architecture updates, while the other is a test-migration PR adding DDP/kernel test infrastructure and model tests. They touch different code paths and should not be clustered together as the same change.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: PR #45932 and PR #46100 have different goals, touched files, and code paths, so they are not the same underlying change.", + "best_issue_reason": null, + "best_pr_reason": "PR #45932 is the better representative PR only in the sense that it is a more substantive code-change PR; PR #46100 is primarily test migration/infrastructure.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45932", + "right": "pull_request:46100", + "accept": false, + "reason": "Reject: these are unrelated \"do not merge\" PRs. #45932 changes model code and CI secrets/workflow; #46100 adds distributed test config and test migrations. They do not fix the same concrete bug or implement 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": 28, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46126-2", + "nodes": [ + "pull_request:46126", + "pull_request:46269" + ], + "soft_pairs": [ + "pull_request:46126|pull_request:46269" + ], + "prepared_review_unit_hash": "93b79d783fd5a92cce4a5c58b817695381ebceeaf7b83f34cb0af2445ffcfcd9", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5524, + "estimated_input_tokens": 1381, + "estimated_eval_tokens": 3018 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5524, + "estimated_input_tokens": 1381, + "estimated_eval_tokens": 3018 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These PRs overlap in distributed/EP-related files, but they do not look like the same change: PR 46126 is a focused EP + Trainer integration feature, while PR 46269 is a broad distributed branch base touching many unrelated areas. The shared filenames are not enough to treat them as duplicates.", + "confidence": 0.92, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46126 is the more focused, concrete change: it targets EP + Trainer integration and has a narrow set of distributed-model updates. PR 46269 is a broad branch-base PR with many unrelated edits, so it is not a good canonical representative for this cluster.", + "best_issue_reason": null, + "best_pr_reason": "PR 46126 is the best representative because it describes a specific code-path change and has a coherent scope. PR 46269 is too broad and mixed to serve as the cluster\u2019s main PR.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46126", + "right": "pull_request:46269", + "accept": false, + "reason": "Shared distributed filenames do not indicate the same underlying bug or change. 46126 is a targeted EP/Trainer integration PR, while 46269 is a large base branch with many additional distributed and unrelated modifications; 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": 29, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46138-9", + "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" + ], + "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:46504", + "pull_request:46138|pull_request:46432", + "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:46558|pull_request:46563", + "pull_request:46504|pull_request:46563", + "pull_request:46432|pull_request:46563" + ], + "prepared_review_unit_hash": "c739833276fa42339bfc7099f48d8947cffdea11b4227a3205d04dc0733edb8c", + "component_budget": { + "node_count": 9, + "item_count": 9, + "soft_pair_count": 26, + "serialized_chars": 23358, + "estimated_input_tokens": 5840, + "estimated_eval_tokens": 11936 + }, + "budget": { + "node_count": 9, + "item_count": 9, + "soft_pair_count": 26, + "serialized_chars": 23358, + "estimated_input_tokens": 5840, + "estimated_eval_tokens": 11936 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Mostly unrelated workflow maintenance PRs. The only tight textual cluster is three changes to `pr-ci-security-gate.yml`, but they alter different behaviors and should not be merged as duplicates.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "No true duplicate PR canonical emerges: the set spans several distinct workflow edits, and the `pr-ci-security-gate.yml` PRs are separate behavior changes rather than the same fix. If a representative is needed, `#46554` is the most substantive security-gate change, but it is not a duplicate of the others.", + "best_issue_reason": null, + "best_pr_reason": "`#46554` is the strongest representative of the cluster because it makes the most concrete security-gate/code-path change, but it still stands alone and should not absorb the other PRs as duplicates.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46138", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflow files and different fixes: stdin handling for self-comment CI vs new security-gate logic." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46558", + "accept": false, + "reason": "Unrelated changes in different workflows; one updates self-comment CI, the other tightens PR CI security-gate permissions/approval flow." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46563", + "accept": false, + "reason": "Different workflow and behavior: self-comment CI input handling vs switching the security gate to a reusable workflow." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46504", + "accept": false, + "reason": "Same broad area of CI workflows, but different scripts and fixes; not the same underlying bug or change." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46432", + "accept": false, + "reason": "Different workflow files and purposes: comment parsing in one PR, SSH runner hardening and env cleanup in the other." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46503", + "accept": false, + "reason": "Distinct workflow changes: SSH runner hardening vs allowing benchmark_v2 to run on workflow_dispatch." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46501", + "accept": false, + "reason": "Different workflow files and unrelated updates; no shared concrete code-path problem." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46502", + "accept": false, + "reason": "SSH runner security/hardening changes are unrelated to assign-reviewers checkout behavior." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflows and different fixes; one hardens SSH runner download/env handling, the other changes PR CI security gating logic." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46558", + "accept": false, + "reason": "Not the same change: SSH runner workflow hardening vs PR CI gate permission/approval tweaks." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46563", + "accept": false, + "reason": "Separate workflow concerns; no common concrete bug to merge as one PR." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46503", + "accept": false, + "reason": "Both are workflow chores, but they change different files and different behaviors." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46502", + "accept": false, + "reason": "Unrelated workflow edits: anti-slop permissions vs reviewer assignment checkout ref." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46504", + "accept": false, + "reason": "Different workflow files and separate fixes; not duplicate change requests." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46554", + "accept": false, + "reason": "No shared underlying bug: anti-slop permission cleanup vs security-gate file scanning and checkout changes." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46503", + "accept": false, + "reason": "Different workflows and distinct behavior changes." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46504", + "accept": false, + "reason": "Assign-reviewers checkout ref fix is unrelated to build-ci-docker-images shell input handling." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflow code paths; not the same underlying issue." + }, + { + "left": "pull_request:46503", + "right": "pull_request:46504", + "accept": false, + "reason": "Benchmark dispatch support and build-ci docker trigger handling are separate changes." + }, + { + "left": "pull_request:46503", + "right": "pull_request:46554", + "accept": false, + "reason": "Unrelated workflows and fixes; no plausible single PR merge target." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46554", + "accept": false, + "reason": "Both touch CI workflow scripts, but they address different jobs and different problems." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46558", + "accept": false, + "reason": "Different workflow files and unrelated logic changes." + }, + { + "left": "pull_request:46554", + "right": "pull_request:46558", + "accept": false, + "reason": "Same workflow file, but different behavior changes: 46554 adds file-diff security checks, while 46558 adjusts permissions and approval gating." + }, + { + "left": "pull_request:46554", + "right": "pull_request:46563", + "accept": false, + "reason": "Same workflow file, but one adds local security checks and the other replaces the job with a reusable workflow; different change sets." + }, + { + "left": "pull_request:46558", + "right": "pull_request:46563", + "accept": false, + "reason": "Both edit `pr-ci-security-gate.yml`, but they implement different mechanisms and are not the same fix." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46563", + "accept": false, + "reason": "Different workflow files and different maintenance changes." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46563", + "accept": false, + "reason": "SSH runner hardening and PR CI gate refactor are unrelated." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflows and distinct code paths; not duplicate." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 30, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46152-2", + "nodes": [ + "pull_request:46152", + "pull_request:46207" + ], + "soft_pairs": [ + "pull_request:46152|pull_request:46207" + ], + "prepared_review_unit_hash": "32fc0c5e924c7544bf607006f0a20fb0a4c290a5986a7b0998429d2efad86bcf", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4276, + "estimated_input_tokens": 1069, + "estimated_eval_tokens": 2394 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4276, + "estimated_input_tokens": 1069, + "estimated_eval_tokens": 2394 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both items are PRs about replacing raw file-open source reads with a safer source-reading mechanism, but they target different code paths and different bugs, so they should not be deduplicated.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "PR #46207 is the stronger standalone change: it targets a specific helper in modeling_utils, includes tests, and addresses a concrete source-reading path rather than a broader import-utils refactor.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46152", + "right": "pull_request:46207", + "accept": false, + "reason": "They are related only at a high level (avoiding open() for source reading). One changes import_utils module path resolution; the other changes _can_set_attn_implementation in modeling_utils. Different code paths, different failure modes, not one mergeable 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": 31, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46154-2", + "nodes": [ + "pull_request:46154", + "pull_request:46196" + ], + "soft_pairs": [ + "pull_request:46154|pull_request:46196" + ], + "prepared_review_unit_hash": "29bb69f5e8fda1d7d42d40980e6c8e13e4d5471b6b453c1d5394a3df7dc7a3d5", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5698, + "estimated_input_tokens": 1425, + "estimated_eval_tokens": 3106 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5698, + "estimated_input_tokens": 1425, + "estimated_eval_tokens": 3106 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs add SDPA dispatch support, but they target different model families and separate code paths (T5 vs wav2vec2-family speech models). They are related in theme, not duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #46196 is the better representative because it covers a broader, more central SDPA dispatch update across multiple speech-model implementations and has the larger change set/activity.", + "best_issue_reason": null, + "best_pr_reason": "PR #46196 is the strongest standalone PR in this cluster due to wider scope and clearer cross-model impact; PR #46154 is a narrower T5-specific addition.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46154", + "right": "pull_request:46196", + "accept": false, + "reason": "Both concern SDPA dispatch, but they fix different model stacks and likely different attention implementations; they do not look like the same concrete bug/change and would not plausibly be merged into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 32, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46173-2", + "nodes": [ + "pull_request:46173", + "pull_request:46474" + ], + "soft_pairs": [ + "pull_request:46173|pull_request:46474" + ], + "prepared_review_unit_hash": "8ac88de891156efc041ae6eac2ff2b6f957f37c7ee42f44803b224f91f9b44ec", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 3631, + "estimated_input_tokens": 908, + "estimated_eval_tokens": 2072 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 3631, + "estimated_input_tokens": 908, + "estimated_eval_tokens": 2072 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both pull requests make the same code change in `src/transformers/cache_utils.py` to widen `StaticLayer.get_seq_length` from `int` to `int | torch.Tensor`; they are near-duplicate implementations of the same fix for issue #45987.", + "confidence": 0.99, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #46173 is the better canonical representative because it is the earlier, more explicitly titled version of the same fix and already points to the tracking issue #45987.", + "best_issue_reason": null, + "best_pr_reason": "PR #46173 is the strongest standalone representative of the change: it contains the same minimal patch, the same target issue, and a clearer description of the intended fix.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46173", + "right": "pull_request:46474", + "accept": true, + "reason": "Same underlying bug fix, same file, same diff hunk, and same explicit issue target (#45987); these are functionally duplicate PRs." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: both PRs touch the same file, make the same type-annotation change for `StaticLayer.get_seq_length`, and both explicitly target issue #45987. The canonical-pr rationale is also conservative and supported by the titles/metadata. No overstatement beyond the visible diff evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 33, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46310-2", + "nodes": [ + "pull_request:46310", + "pull_request:46546" + ], + "soft_pairs": [ + "pull_request:46310|pull_request:46546" + ], + "prepared_review_unit_hash": "04fad04f796d7c11018c653a5b6eb2fb038bc8882eb45143664ca7a43a528af9", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4851, + "estimated_input_tokens": 1213, + "estimated_eval_tokens": 2682 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4851, + "estimated_input_tokens": 1213, + "estimated_eval_tokens": 2682 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs fix the same Moonshine streaming sliding-window right-boundary off-by-one bug (shared issue target 46305, identical moonshine_streaming mask change). PR 46546 is broader because it also carries the Gemma4 counterpart and tests.", + "confidence": 0.92, + "canonical_issue_reason": "No issue artifact is present in the cluster; the shared target 46305 appears to be the underlying bug report for the right-window boundary off-by-one.", + "canonical_pr_reason": "PR 46546 is the better canonical PR because it includes the Moonshine fix plus the corresponding Gemma4 fix and test coverage, making it the more complete change set.", + "best_issue_reason": null, + "best_pr_reason": "PR 46546 is the strongest representative of the fix: it contains the same Moonshine boundary correction as 46310 and adds the related Gemma4 patch and tests.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46310", + "right": "pull_request:46546", + "accept": true, + "reason": "Same underlying Moonshine right-window boundary bug; identical fix in the shared Moonshine files and same explicit issue target 46305, with 46546 adding a related companion fix/test coverage." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 34, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46376-2", + "nodes": [ + "pull_request:46376", + "pull_request:46417" + ], + "soft_pairs": [ + "pull_request:46376|pull_request:46417" + ], + "prepared_review_unit_hash": "6a21810767d7019b34ed13cfd87d3e564896e7f1e692760de4a5ff3c4c12d3f5", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5937, + "estimated_input_tokens": 1485, + "estimated_eval_tokens": 3226 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5937, + "estimated_input_tokens": 1485, + "estimated_eval_tokens": 3226 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The cluster contains two unrelated pull requests: one adds multimodal support to continuous batching, and the other adds the HunYuan VL model and its docs/tests. They do not appear to fix the same code-path problem.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "pull_request:46417 is the stronger standalone PR: it is a self-contained model addition with docs, code, and tests, whereas pull_request:46376 is a draft infrastructure change and not a duplicate of the other PR.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46376", + "right": "pull_request:46417", + "accept": false, + "reason": "Different goals and code paths: continuous batching/multimodal cache work versus adding a new HunYuan VL model stack. They are not mergeable as one PR and do not represent 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": "pull_request", + "component_index": 35, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46574-2", + "nodes": [ + "pull_request:46574", + "pull_request:46580" + ], + "soft_pairs": [ + "pull_request:46574|pull_request:46580" + ], + "prepared_review_unit_hash": "573e85a73aa5aea08f6fd6e66d29cd314131cb266ac77a55dca53423f11c33a8", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4410, + "estimated_input_tokens": 1103, + "estimated_eval_tokens": 2462 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4410, + "estimated_input_tokens": 1103, + "estimated_eval_tokens": 2462 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two PRs are unrelated tokenizer fixes: one changes CodeLlama decode behavior to preserve leading spaces, while the other adjusts AutoTokenizer routing for specific DeepSeek checkpoints. They should not be clustered as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "pull_request:46574", + "right": "pull_request:46580", + "accept": false, + "reason": "Different models and different code paths: CodeLlama decoder stripping vs DeepSeek checkpoint-to-tokenizer-class selection. They do not appear mergeable into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 36, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46603-2", + "nodes": [ + "pull_request:46603", + "pull_request:46604" + ], + "soft_pairs": [ + "pull_request:46603|pull_request:46604" + ], + "prepared_review_unit_hash": "1b822f7769f8a5da1dec56c6db8cba99639bf0091398d137539f617a6dfffb35", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5495, + "estimated_input_tokens": 1374, + "estimated_eval_tokens": 3004 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5495, + "estimated_input_tokens": 1374, + "estimated_eval_tokens": 3004 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs touch the Mistral tokenizer area and even share several files, but they are different changes: one moves `MistralConverter` into a dedicated `integrations/mistral` package, while the other adds tekken tokenizer detection, conversion, and save utilities. They are related follow-up/refactor work, not the same underlying bug or feature change.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #46604 is the more substantive feature change, adding tekken tokenizer detection/conversion/save support; #46603 is a narrower packaging refactor that extracts `MistralConverter`.", + "best_issue_reason": null, + "best_pr_reason": "PR #46604 best represents the cluster because it contains the concrete new tokenizer functionality, whereas #46603 is primarily an internal move of existing Mistral conversion code.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46603", + "right": "pull_request:46604", + "accept": false, + "reason": "Shared Mistral tokenizer files do not make these the same change: #46603 relocates `MistralConverter`, while #46604 implements tekken tokenizer support and utilities. They would not plausibly merge into one single PR as duplicate code-path fixes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + } + ] +} diff --git a/snapshots/20260617T060052Z/analysis-runs/hybrid-model-20260617t060052z/manifest.json b/snapshots/20260617T060052Z/analysis-runs/hybrid-model-20260617t060052z/manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..aec5bbe8b6f267d0a100e8c2723ccbf61b32335f --- /dev/null +++ b/snapshots/20260617T060052Z/analysis-runs/hybrid-model-20260617t060052z/manifest.json @@ -0,0 +1,14 @@ +{ + "analysis_id": "hybrid-model-20260617t060052z", + "artifacts": { + "hybrid": "snapshots/20260617T060052Z/analysis-runs/hybrid-model-20260617t060052z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260617T060052Z/analysis-runs/hybrid-model-20260617t060052z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "model": null, + "published_at": "2026-06-17T06:10:53Z", + "repo": "huggingface/transformers", + "schema_version": 1, + "snapshot_id": "20260617T060052Z", + "variant": "hybrid" +} diff --git a/snapshots/20260617T060052Z/manifest.json b/snapshots/20260617T060052Z/manifest.json index 869ca1616e433c01c6d7afefd5ca76ed337e86b7..35f8e7f770f52e2746e9acd92b26ce04fad691f2 100644 --- a/snapshots/20260617T060052Z/manifest.json +++ b/snapshots/20260617T060052Z/manifest.json @@ -32,6 +32,24 @@ "timeline_events": 314 }, "extracted_at": "2026-06-17T06:00:52Z", + "published_analysis": { + "canonical_analysis_id": "hybrid-model-20260617t060052z", + "runs": { + "hybrid-model-20260617t060052z": { + "analysis_id": "hybrid-model-20260617t060052z", + "artifacts": { + "hybrid": "snapshots/20260617T060052Z/analysis-runs/hybrid-model-20260617t060052z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260617T060052Z/analysis-runs/hybrid-model-20260617t060052z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "manifest_path": "snapshots/20260617T060052Z/analysis-runs/hybrid-model-20260617t060052z/manifest.json", + "model": null, + "published_at": "2026-06-17T06:10:53Z", + "variant": "hybrid" + } + }, + "schema_version": 1 + }, "repo": "huggingface/transformers", "snapshot_id": "20260617T060052Z", "watermark": {