diff --git a/analysis/current/analysis-report-hybrid.json b/analysis/current/analysis-report-hybrid.json index c836de90cb8340495bee4138bbc91cb566f46153..215cd30a6313d65840600937b65e6da9a8ac47b3 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": "20260508T180023Z", - "generated_at": "2026-05-08T18:13:49Z", + "snapshot_id": "20260509T000027Z", + "generated_at": "2026-05-09T00:11:01Z", "evidence_quality": "full", "llm_enrichment": true, "meta_bugs": [ diff --git a/analysis/current/analysis-report-hybrid.llm-reviews.json b/analysis/current/analysis-report-hybrid.llm-reviews.json index 8f4510708ee0f91b6c78912e10c6f471bd5c403c..2c72b69356db580c3cd6279a4a610405378a3c64 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": "20260508T180023Z", - "generated_at": "2026-05-08T18:13:49Z", + "snapshot_id": "20260509T000027Z", + "generated_at": "2026-05-09T00:11:01Z", "model": "gpt-5.4-mini?service_tier=flex", "reviews": [ { @@ -26,46 +26,46 @@ "issue:13244|issue:16618", "issue:15980|issue:16618" ], - "prepared_review_unit_hash": "ac2e1a526e420f5a304e9a048ce322a5cc4b5c1f40526934f32332253e0f1a76", + "prepared_review_unit_hash": "4929c4a3069ad42a997018a65be24c829a02d2274479dc5fb99cda6877b28426", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 2476, - "estimated_input_tokens": 619, - "estimated_eval_tokens": 1494 + "serialized_chars": 2477, + "estimated_input_tokens": 620, + "estimated_eval_tokens": 1496 }, "budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 2476, - "estimated_input_tokens": 619, - "estimated_eval_tokens": 1494 + "serialized_chars": 2477, + "estimated_input_tokens": 620, + "estimated_eval_tokens": 1496 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These issues are only loosely related by Hugging Face model/tokenizer loading, but they describe different problems: tokenization mismatch, private-model auth/error messaging, and a model load failure for bert-base-uncased. No pair looks like the same underlying bug.", - "confidence": 0.91, - "canonical_issue_reason": null, + "summary": "The cluster is not a true duplicate set. Issue 16618 is the broad model-loading failure, while 15980 is specifically about a bad auth error for private model downloads and 13244 is about Tapas tokenization differences.", + "confidence": 0.9, + "canonical_issue_reason": "#16618 is the broadest and most central user-facing model-load failure in the set; the others are about different root causes and code paths.", "canonical_pr_reason": null, - "best_issue_reason": "issue:16618 is the broadest and most central of the three, focusing on a concrete model-loading failure; the other two are about tokenizer behavior and private-model login/error messaging.", + "best_issue_reason": "#16618 best represents the cluster\u2019s shared model-loading theme, even though the other issues are not duplicates of it.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:13244", "right": "issue:16618", "accept": false, - "reason": "Tokenization differences vs model loading failure are different code paths and symptoms." + "reason": "Tapas tokenization behavior is unrelated to a generic BERT model load failure; different subsystem and bug." }, { "left": "issue:15980", "right": "issue:16618", "accept": false, - "reason": "Private-model login/error-message handling is distinct from failing to load bert-base-uncased." + "reason": "Private-model download auth/error-message handling is a different failure mode from 'can't load bert-base-uncased'; not the same underlying bug." } ] }, @@ -188,8 +188,8 @@ "issue:30277|issue:36331", "issue:34928|issue:36010", "issue:43039|issue:44805", - "issue:41762|issue:43596", "issue:43901|issue:44512", + "issue:41762|issue:43596", "issue:39290|issue:41720", "issue:44841|issue:45823", "issue:43039|issue:44560", @@ -199,8 +199,8 @@ "issue:41093|issue:43638", "issue:43425|issue:45070", "issue:43596|issue:45800", - "issue:45362|issue:45464", "issue:43792|issue:45070", + "issue:45362|issue:45464", "issue:43404|issue:43866", "issue:39290|issue:45823", "issue:45464|issue:45685", @@ -211,358 +211,418 @@ "issue:45800|issue:45823", "issue:41093|issue:41762" ], - "prepared_review_unit_hash": "35361b1ace66a000ae7500f9a1172674cf95f90777542d62d6ebc2a76b7e1b6e", + "prepared_review_unit_hash": "83671578cf33ce3c9529ee4bef7423820903edbe840e6cb4e4d37193ba8486f1", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 71, - "serialized_chars": 48868, - "estimated_input_tokens": 12217, - "estimated_eval_tokens": 24690 + "serialized_chars": 48869, + "estimated_input_tokens": 12218, + "estimated_eval_tokens": 24692 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly false-positive similarity across unrelated model/runtime/docs issues. Only one duplicate family is clear: the DeepSpeed ZeRO-3 BertModel index-out-of-bounds reports (issue:43596 and issue:43638).", - "confidence": 0.86, - "canonical_issue_reason": "issue:43596 is the cleanest representative of the only clear duplicate family: the same ZeRO-3/BertModel index-out-of-bounds load failure as issue:43638.", + "summary": "This cluster is mostly a set of unrelated issue reports with similar-sounding error strings. The only clear duplicate is the DeepSpeed ZeRO-3/Bert index-out-of-bounds pair; the rest differ by model, backend, or feature area.", + "confidence": 0.83, + "canonical_issue_reason": "issue:43596 is the best anchor for the one clear duplicate set: it names the ZeRO-3 init index-out-of-bounds failure concisely and covers the shared BertModel code path.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43596 is the best single issue to keep because it states the core ZeRO-3/BertModel failure clearly and is the earliest of the near-identical reports.", + "best_issue_reason": "issue:43596 is the strongest representative issue in this set because it is the clearest, most general description of the shared ZeRO-3/Bert failure.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44322", "right": "issue:45464", "accept": false, - "reason": "Different Qwen3.5 failure modes: config attribute error vs streaming chat/completions error." + "reason": "Different Qwen3.5 failures: one is a config attribute error, the other is a streaming chat/completions runtime issue." }, { "left": "issue:45237", "right": "issue:45685", "accept": false, - "reason": "Unrelated device/runtime issues: AMD GPU failure vs MPS histogram op not implemented." + "reason": "Different models and backends: AMD GPU failure for GPT-OSS vs an MPS histogram op error for moe." }, { "left": "issue:43716", "right": "issue:45237", "accept": false, - "reason": "Different model-specific bugs: image preprocessor dtype mismatch vs AMD GPU runtime failure." + "reason": "Different model families and symptoms; this is a dtype mismatch in Mistral-3 image preprocessing, not the GPT-OSS AMD GPU crash." }, { "left": "issue:43828", "right": "issue:45237", "accept": false, - "reason": "Autocast dtype mismatch vs AMD GPU device-map failure; not the same bug." + "reason": "Both are runtime/backend errors, but the concrete failures and affected models are unrelated." }, { "left": "issue:45237", "right": "issue:45507", "accept": false, - "reason": "Different problems: device support/runtime failure vs invalid method call in GraniteMoEHybrid." + "reason": "Different code paths: GPT-OSS AMD GPU failure vs GraniteMoEHybrid calling an invalid method." }, { "left": "issue:44292", "right": "issue:45237", "accept": false, - "reason": "Different model/runtime contexts: NVFP4 execution error vs AMD GPU issue." + "reason": "Different model/quantization issue; NVFP4 runtime failure is not the same bug as the AMD GPU crash." }, { "left": "issue:30277", "right": "issue:35545", "accept": false, - "reason": "DeepSpeed collective mismatch vs ONNX export error; unrelated." + "reason": "Different subsystems: DeepSpeed collective mismatch vs ONNX export error." }, { "left": "issue:35545", "right": "issue:42915", "accept": false, - "reason": "Different model and failure type: export to ONNX vs FineGrainedFP8 load failure." + "reason": "Different model/export problems; ModernBERT ONNX export is unrelated to Qwen3Moe FP8 failure." }, { "left": "issue:35545", "right": "issue:41720", "accept": false, - "reason": "Export-time ONNX failure vs auto device-map cuda assert; unrelated." + "reason": "Unrelated failures: ONNX export for ModernBERT vs CUDA assert with Qwen3 auto device mapping." }, { "left": "issue:44509", "right": "issue:44512", "accept": false, - "reason": "Both are docs cleanups after v5 removals, but they target different obsolete references; not one duplicate issue." + "reason": "Both are v5 docs regressions, but they reference different removed items and do not look like one concrete bug." }, { "left": "issue:35545", "right": "issue:36010", "accept": false, - "reason": "Different failures: export issue vs missing GenerationMixin import." + "reason": "Different issues: ONNX export vs GenerationMixin import path breakage." }, { "left": "issue:44805", "right": "issue:45823", "accept": false, - "reason": "Generic mask-shape mismatch vs Gemma4 PLE device mismatch; different code paths." + "reason": "Both mention shape/mismatch-type failures, but the affected models and failure modes are different." }, { "left": "issue:43638", "right": "issue:44805", "accept": false, - "reason": "ZeRO-3 Bert load crash vs generic mask indexing error; unrelated." + "reason": "Different error families: ZeRO-3 index-out-of-bounds on Bert vs mask/index shape mismatch." }, { "left": "issue:43596", "right": "issue:44805", "accept": false, - "reason": "ZeRO-3 Bert load crash vs generic mask indexing error; unrelated." + "reason": "Same general category of indexing errors, but not the same model path or concrete bug." }, { "left": "issue:44805", "right": "issue:45161", "accept": false, - "reason": "Different symptoms: mask shape mismatch vs GPT-OSS tensor-parallel failure." + "reason": "Different models and execution modes; mask-shape mismatch is unrelated to GPT-OSS tensor-parallel behavior." + }, + { + "left": "issue:44805", + "right": "issue:44918", + "accept": false, + "reason": "Different failure sites: mask/index mismatch vs Qwen3.5 embedding unpacking in TRL SFT trainer." }, { "left": "issue:34634", "right": "issue:35545", "accept": false, - "reason": "BarkProcessor voice_preset bug vs ModernBERT ONNX export issue; unrelated." + "reason": "BarkProcessor voice_preset bug is unrelated to ModernBERT ONNX export." }, { "left": "issue:30277", "right": "issue:43638", "accept": false, - "reason": "Collective mismatch during ZeRO-3 vs BertModel index-out-of-bounds on load; different bugs." + "reason": "Both involve DeepSpeed, but one is a collective mismatch and the other is a ZeRO-3 index-out-of-bounds init bug." }, { "left": "issue:43596", "right": "issue:43638", "accept": true, - "reason": "Same ZeRO-3/BertModel index-out-of-bounds load failure, just reported with nearly identical wording." + "reason": "Same ZeRO-3 index-out-of-bounds failure on BertModel; the second report is the non-pretrained Bert variant of the same underlying bug." }, { "left": "issue:43975", "right": "issue:44457", "accept": false, - "reason": "Detokenization regression vs LoRA save/load mismatch; unrelated." + "reason": "Different functionality: detokenization regression vs LoRA merge/save/reload output mismatch." }, { "left": "issue:43039", "right": "issue:45161", "accept": false, - "reason": "Liger Kernel cross-entropy routing vs GPT-OSS TP problem; different issues." + "reason": "Liger Kernel cross-entropy routing and GPT-OSS tensor-parallel failure are unrelated." }, { "left": "issue:44292", "right": "issue:44457", "accept": false, - "reason": "NVFP4 model runtime error vs LoRA merge/reload inconsistency; unrelated." + "reason": "Different areas entirely: Qwen NVFP4 runtime failure vs LoRA merge/save consistency." }, { "left": "issue:34928", "right": "issue:44805", "accept": false, - "reason": "FSDP activation-checkpointing tensor-size mismatch vs mask indexing error; different bug." + "reason": "Different distributed-training bugs: activation checkpointing/FSDP size mismatch vs mask/index mismatch." }, { "left": "issue:24643", "right": "issue:30277", "accept": false, - "reason": "DeepSpeed weight-dimension error vs collective mismatch; not the same failure." + "reason": "Both are DeepSpeed-related, but the concrete failures are different and not the same bug." }, { "left": "issue:43638", "right": "issue:45161", "accept": false, - "reason": "ZeRO-3 Bert load crash vs GPT-OSS tensor-parallel issue; unrelated." + "reason": "Bert ZeRO-3 init failure is unrelated to GPT-OSS tensor-parallel behavior." }, { "left": "issue:34928", "right": "issue:36331", "accept": false, - "reason": "Activation-checkpointing/FSDP mismatch vs trainer signature change; unrelated." + "reason": "Activation-checkpointing size mismatch and CustomTrainer signature mismatch are unrelated." }, { "left": "issue:43596", "right": "issue:45161", "accept": false, - "reason": "Different model/loading bug vs GPT-OSS TP issue." + "reason": "Different models and code paths; ZeRO-3 Bert init bug is not the GPT-OSS TP issue." }, { "left": "issue:43716", "right": "issue:45507", "accept": false, - "reason": "dtype mismatch in Mistral-3 vs invalid method call in GraniteMoEHybrid; unrelated." + "reason": "Different bugs: image preprocessor/model dtype mismatch vs invalid method call in GraniteMoEHybrid." }, { "left": "issue:39290", "right": "issue:45081", "accept": false, - "reason": "Gemma3 config attribute error vs tokenizer regex patch AttributeError; unrelated." + "reason": "Different model/config issues: Gemma3 sliding-window config vs Mistral tokenizer regex patch crash." }, { "left": "issue:43828", "right": "issue:45161", "accept": false, - "reason": "Autocast dtype mismatch vs GPT-OSS TP failure; different bugs." + "reason": "Phi-tiny-MoE autocast dtype mismatch is unrelated to GPT-OSS tensor-parallel failures." }, { - "left": "issue:43425", - "right": "issue:44292", + "left": "issue:44918", + "right": "issue:45161", "accept": false, - "reason": "Torch version incompatibility vs Qwen NVFP4 runtime error; unrelated." + "reason": "Qwen3.5 embedding unpacking in TRL is not the same bug as GPT-OSS tensor-parallel support." }, { - "left": "issue:44918", - "right": "issue:45161", + "left": "issue:43039", + "right": "issue:44918", "accept": false, - "reason": "TRL SFT trainer embedding unpacking issue vs GPT-OSS TP issue; unrelated." + "reason": "Different areas: Liger Kernel cross-entropy vs TRL SFT input embedding unpacking." }, { - "left": "issue:34928", - "right": "issue:41720", + "left": "issue:41762", + "right": "issue:43638", + "accept": false, + "reason": "Both are ZeRO-3 index errors, but they involve different model families and are not clearly the same concrete bug." + }, + { + "left": "issue:43827", + "right": "issue:44512", "accept": false, - "reason": "FSDP activation-checkpointing mismatch vs Qwen3 auto device-map cuda assert; unrelated." + "reason": "Both are docs regressions, but they refer to different removed v5 documentation surfaces." }, { "left": "issue:43716", + "right": "issue:44560", + "accept": false, + "reason": "Different multimodal models and different runtime errors; dtype mismatch is not the same as StopIteration in video embedding." + }, + { + "left": "issue:43828", "right": "issue:45685", "accept": false, - "reason": "dtype mismatch vs MPS histogram op limitation; unrelated." + "reason": "Autocast dtype mismatch and MPS histogram op missing are unrelated backend bugs." }, { - "left": "issue:43541", - "right": "issue:43716", + "left": "issue:44560", + "right": "issue:45685", "accept": false, - "reason": "Torch Dynamo grouped_mm tracing error vs Mistral-3 dtype mismatch; unrelated." + "reason": "Different model/backend issues: Qwen3-vl video StopIteration vs MPS unsupported histogram op." }, { - "left": "issue:43828", - "right": "issue:44292", + "left": "issue:43039", + "right": "issue:43828", + "accept": false, + "reason": "Different failure modes in different models; cross_entropy routing is not the same as autocast dtype mismatch." + }, + { + "left": "issue:30277", + "right": "issue:36331", "accept": false, - "reason": "Both involve model/runtime failures, but the concrete errors and code paths differ." + "reason": "DeepSpeed collective mismatch and trainer method signature mismatch are unrelated." }, { "left": "issue:34928", "right": "issue:36010", "accept": false, - "reason": "Different failures: FSDP checkpointing mismatch vs missing GenerationMixin import." + "reason": "Different problems: activation checkpointing/FSDP shape mismatch vs missing GenerationMixin import." }, { "left": "issue:43039", "right": "issue:44805", "accept": false, - "reason": "Liger Kernel cross-entropy call vs generic mask-shape mismatch; unrelated." + "reason": "Liger Kernel cross-entropy call issue is unrelated to mask/index shape mismatch." + }, + { + "left": "issue:43901", + "right": "issue:44512", + "accept": false, + "reason": "Both are docs issues, but one is about return_all_scores and the other about the removed transformers run command." }, { "left": "issue:41762", "right": "issue:43596", "accept": false, - "reason": "Both are ZeRO-3 load-time index errors, but on different model families and not clearly the same concrete bug." + "reason": "Same broad ZeRO-3 symptom, but the affected models differ enough that this is not clearly one duplicate bug." }, { - "left": "issue:43827", - "right": "issue:44512", + "left": "issue:39290", + "right": "issue:41720", "accept": false, - "reason": "Different docs removals after v5: pipeline-task docs vs transformers run-command docs." + "reason": "Gemma3 sliding_window_pattern config and Qwen3 auto device mapping CUDA assert are unrelated." }, { - "left": "issue:43716", + "left": "issue:44841", + "right": "issue:45823", + "accept": false, + "reason": "Processor failure for Voxtral is unrelated to Gemma4 PLE device mismatch in forward." + }, + { + "left": "issue:43039", "right": "issue:44560", "accept": false, - "reason": "dtype mismatch vs video StopIteration; unrelated." + "reason": "Different subsystems entirely: Liger Kernel cross-entropy vs Qwen3-vl video StopIteration." }, { - "left": "issue:43828", - "right": "issue:45685", + "left": "issue:43866", + "right": "issue:44863", "accept": false, - "reason": "Autocast dtype mismatch vs MPS histogram limitation; unrelated." + "reason": "Checkpoint loading problems, but for different models and likely different root causes." }, { - "left": "issue:44560", - "right": "issue:45685", + "left": "issue:43854", + "right": "issue:43866", "accept": false, - "reason": "Video-processing StopIteration vs MPS histogram limitation; unrelated." + "reason": "Different model-loading failures for different models; no clear shared bug." }, { - "left": "issue:43039", - "right": "issue:43828", + "left": "issue:39290", + "right": "issue:43531", "accept": false, - "reason": "Liger Kernel cross-entropy issue vs autocast dtype mismatch; different code paths." + "reason": "Related Qwen/Gemma sliding-window themes, but different model families and concrete failures." }, { - "left": "issue:30277", - "right": "issue:36331", + "left": "issue:41093", + "right": "issue:43638", "accept": false, - "reason": "Collective mismatch vs trainer kwargs regression; unrelated." + "reason": "Both are index-related errors, but the mask/tensor size mismatch is different from the ZeRO-3 empty-dimension init bug." }, { - "left": "issue:43901", - "right": "issue:44512", + "left": "issue:43425", + "right": "issue:45070", "accept": false, - "reason": "Both are docs cleanup issues, but they cover different removed/updated API references." + "reason": "Torch compatibility and pydantic model regression are unrelated." }, { - "left": "issue:39290", - "right": "issue:41720", + "left": "issue:43596", + "right": "issue:45800", "accept": false, - "reason": "Different model-specific failures: Gemma3 config attribute vs Qwen3 auto device-map assert." + "reason": "Different DeepSpeed/Bert init bug vs torch/transformers version incompatibility." }, { - "left": "issue:44841", - "right": "issue:45084", + "left": "issue:43792", + "right": "issue:45070", "accept": false, - "reason": "Processor loading failure vs template compilation error; unrelated." + "reason": "Whisper loading failure and PretrainedConfig pydantic breakage are unrelated." }, { - "left": "issue:45800", + "left": "issue:45362", + "right": "issue:45464", + "accept": false, + "reason": "Both involve Qwen3.5 and chat, but one is a transformers chat crash and the other is a streaming chat/completions API failure; not clearly the same bug." + }, + { + "left": "issue:43404", + "right": "issue:43866", + "accept": false, + "reason": "Different issue classes: tied lm_head weight bug vs corrupted checkpoint/load failure." + }, + { + "left": "issue:39290", "right": "issue:45823", "accept": false, - "reason": "Torch/transformers version incompatibility vs Gemma4 PLE device mismatch; different bugs." + "reason": "Gemma3 sliding-window config and Gemma4 device-mapping mismatch are different model issues." }, { - "left": "issue:41093", - "right": "issue:43638", + "left": "issue:45464", + "right": "issue:45685", "accept": false, - "reason": "Generic mask-shape mismatch vs ZeRO-3 Bert index-out-of-bounds; unrelated." + "reason": "Chat/completions streaming failure on Qwen3.5 is unrelated to the MPS histogram op bug." }, { - "left": "issue:43854", - "right": "issue:43866", + "left": "issue:43541", + "right": "issue:45685", "accept": false, - "reason": "GLM-4.7-Flash unit-test load failure vs Ovis2 checkpoint corruption; different issues." + "reason": "Mixtral grouped_mm tracing error and MPS histogram op missing are unrelated." }, { - "left": "issue:43404", - "right": "issue:43866", + "left": "issue:43541", + "right": "issue:43828", "accept": false, - "reason": "Untied Mistral3 lm_head vs corrupted Ovis2 checkpoint; unrelated." + "reason": "Grouped_mm tracing failure in Mixtral is unrelated to Phi-tiny-MoE autocast dtype mismatch." }, { - "left": "issue:39290", - "right": "issue:43531", + "left": "issue:36010", + "right": "issue:39290", + "accept": false, + "reason": "GenerationMixin import failure and Gemma3 sliding-window config issue are unrelated." + }, + { + "left": "issue:44841", + "right": "issue:45084", + "accept": false, + "reason": "Processor failure and template compilation TypeError are different bugs." + }, + { + "left": "issue:45800", + "right": "issue:45823", "accept": false, - "reason": "Both mention sliding_window, but they are different model families with different failures." + "reason": "Torch/transformers version incompatibility and Gemma4 device mismatch are unrelated." }, { "left": "issue:41093", "right": "issue:41762", "accept": false, - "reason": "Generic mask mismatch vs Gemma3 ZeRO-3 index error; different code paths." + "reason": "Both are index errors, but the mask/index mismatch is not the same as the ZeRO-3 empty-dimension load failure." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary stays conservative and is grounded in the packet. The single accepted duplicate family (issue:43596 / issue:43638) is well supported by the titles, and the other pair rejections do not overstate similarity. No clear evidence of a false positive or mergeability issue." + "feedback": "The summary is well grounded in the packet and stays conservative. The only accepted duplicate pair, issue:43596/issue:43638, is supported by the shared DeepSpeed ZeRO-3 + Bert index-out-of-bounds titles. The other soft-edge rejections are cautious and consistent with the visible model/backend/feature differences in the packet." }, "evaluator_used": true, "retried": false, @@ -639,8 +699,8 @@ "issue:43541|issue:45161", "issue:43828|issue:45464", "issue:44291|issue:45070", - "issue:45020|issue:45800", "issue:43366|issue:45464", + "issue:45020|issue:45800", "issue:43531|issue:45070", "issue:43541|issue:44589", "issue:41093|issue:43596", @@ -668,244 +728,244 @@ "issue:43828|issue:44928", "issue:43866|issue:44403" ], - "prepared_review_unit_hash": "0834920debe1aa402d82a41bf58e0a592c0ed319f52aafdc9d7222123a365922", + "prepared_review_unit_hash": "6e7ca954d63323d5d398617986dcb10cb3c111d7f6cc0eebe0f8990c8386375a", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 37473, - "estimated_input_tokens": 9369, - "estimated_eval_tokens": 18994 + "serialized_chars": 37468, + "estimated_input_tokens": 9367, + "estimated_eval_tokens": 18990 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is mostly a collection of unrelated bug reports with only one clear duplicate pair: the Transformers 5.x non-persistent-buffer corruption regression. Most other soft pairs share only a broad subsystem or model family, not the same concrete bug.", - "confidence": 0.89, - "canonical_issue_reason": "issue:43950 is the best canonical for the non-persistent-buffer regression cluster: it states the bug most explicitly, has the clearest regression framing, and matches the earlier report in issue:43644.", + "summary": "Most items are unrelated. The only clear duplicate cluster is the Transformers v5 non-persistent-buffer corruption regression, best anchored by issue 43950.", + "confidence": 0.92, + "canonical_issue_reason": "Issue 43950 is the most explicit statement of the underlying regression and matches the small duplicate cluster around non-persistent buffers being corrupted on load.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43950 is the most actionable and specific issue to anchor the one clear duplicate pair; it describes the underlying buffer-corruption problem directly.", + "best_issue_reason": "Issue 43950 is the best overall representative because it describes the core failure mode clearly and centrally, rather than a narrower symptom.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43296", "right": "issue:43366", "accept": false, - "reason": "Different problems: PaddleOCR-VL model loading/quantization vs GGUF support for gpt-oss." + "reason": "Different problems: PaddleOCR-VL loading under transformers/vLLM vs GGUF support for gpt-oss architecture." }, { "left": "issue:43572", "right": "issue:45070", "accept": false, - "reason": "Both are v5 config regressions, but one is missing pad_token_idx and the other is a pydantic PretrainedConfig field break; not the same bug." + "reason": "Both are v5 regressions, but one is a missing StableLm config field and the other is a pydantic PretrainedConfig field breakage; not the same bug." }, { "left": "issue:44387", "right": "issue:45005", "accept": false, - "reason": "Int4 OOM/memory regression is unrelated to tied-weights behavior in translation models." + "reason": "Different failure modes: int4 quantization OOM vs tied-weights handling in translation models." }, { "left": "issue:43541", "right": "issue:45161", "accept": false, - "reason": "Both involve MoE models, but one is a grouped_mm tracing failure and the other is a tensor-parallel support issue; different code paths." + "reason": "Both involve MoE models, but one is a grouped_mm tracing failure and the other is a tensor-parallel-only issue for GPT-OSS." }, { "left": "issue:43828", "right": "issue:45464", "accept": false, - "reason": "Autocast dtype mismatch on Phi-tiny-MoE is unrelated to Qwen3.5 streaming chat/completions failures." + "reason": "Both are dtype/runtime issues, but the model path and failure context differ: Phi-tiny-MoE autocast mismatch vs Qwen3.5 streaming inference failure." }, { "left": "issue:44291", "right": "issue:45070", "accept": false, - "reason": "Unexpected _is_hf_initialized argument during init_empty_weights is a different initialization bug than the pydantic field regression." + "reason": "Different code paths: init_empty_weights/_is_hf_initialized argument handling vs pydantic PretrainedConfig field regression." }, { - "left": "issue:45020", - "right": "issue:45800", + "left": "issue:43366", + "right": "issue:45464", "accept": false, - "reason": "Broad compatibility breakage, but one is remote_code model loading and the other is torch/transformers version incompatibility." + "reason": "Unrelated: GGUF/gpt-oss architecture support vs chat/completions streaming failure on Qwen3.5." }, { - "left": "issue:43366", - "right": "issue:45464", + "left": "issue:45020", + "right": "issue:45800", "accept": false, - "reason": "GGUF gpt-oss support and Qwen3.5 streaming inference are unrelated." + "reason": "Both are recent-version breakages, but one is remote_code model loading and the other is a torch/transformers compatibility issue." }, { "left": "issue:43531", "right": "issue:45070", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior is unrelated to the PretrainedConfig/pydantic regression." + "reason": "Different bugs: Qwen3-MoE sliding_window behavior vs PretrainedConfig field breakage in v5.4.0." }, { "left": "issue:43541", "right": "issue:44589", "accept": false, - "reason": "Grouped_mm tracing failure and missing Float8 storage are separate runtime errors." + "reason": "Both hit model execution errors, but grouped_mm tracing and missing Float8 storage are separate underlying failures." }, { "left": "issue:41093", "right": "issue:43596", "accept": false, - "reason": "Both are IndexErrors, but the mask shape mismatch and the zero-sized tensor under ZeRO-3 are different bugs." + "reason": "Both are IndexErrors, but the mask-shape mismatch and the zero-sized tensor access are different root causes." }, { "left": "issue:43054", "right": "issue:43994", "accept": false, - "reason": "Both mention SigLIP2, but one is a quality/performance complaint and the other is a model/processor mismatch in docs; not the same bug." + "reason": "Both mention siglip2, but one is degraded embedding quality and the other is nonsensical output with AutoModel/pipeline; not enough evidence of the same bug." }, { "left": "issue:45542", "right": "issue:45800", "accept": false, - "reason": "TensorBoard-without-TensorFlow backend error is unrelated to torch/transformers compatibility." + "reason": "Different regressions: missing TensorFlow backend handling vs torch 2.4.1 / transformers 5.8.0 incompatibility." }, { "left": "issue:43716", "right": "issue:44918", "accept": false, - "reason": "Mistral-3 image preprocessor dtype mismatch is unrelated to Qwen3.5 embedding unpacking in TRL SFT trainer." + "reason": "Different model families and failure paths: Mistral-3 image preprocessor dtype mismatch vs Qwen3.5 input-embedding unpacking in TRL SFT." }, { "left": "issue:44918", "right": "issue:45685", "accept": false, - "reason": "TRL embedding unpacking failure and MPS histogram error are different model/runtime issues." + "reason": "Both involve MoE-ish models, but one is TRL unpacking and the other is an MPS histogram op limitation; unrelated." }, { "left": "issue:45632", "right": "issue:45698", "accept": false, - "reason": "Cache-path collision for leaf directory names is not the same as loading the wrong custom module after save_pretrained." + "reason": "Both involve custom-module loading, but one is a trust_remote_code cache-path collision and the other is loading the wrong custom module after save_pretrained; different concrete bugs." }, { "left": "issue:43404", "right": "issue:45005", "accept": false, - "reason": "Both involve tied weights, but the affected models and failure modes differ; too broad to merge as one bug." + "reason": "Related to tied weights in broad terms, but one is Mistral3 lm_head tying and the other is translation-model tied-weight issues in v5." }, { "left": "issue:43645", "right": "issue:45399", "accept": false, - "reason": "Jupyter custom-model initialization and flash-attn2 fallback gating are unrelated." + "reason": "Notebook custom-model initialization vs flash-attn2 fallback blocking are unrelated." }, { "left": "issue:40990", "right": "issue:44960", "accept": false, - "reason": "High perplexity on gpt-oss-20b and GLM5 are different model-specific reports." + "reason": "Different symptoms and model families: high perplexity on gpt-oss-20b vs GLM5 issues." }, { "left": "issue:44530", "right": "issue:45005", "accept": false, - "reason": "PagedAttentionCache linear_attention crash is unrelated to tied-weights regressions." + "reason": "PagedAttentionCache / Qwen3.5.0 linear_attention crash is unrelated to translation tied-weight regressions." }, { "left": "issue:42915", "right": "issue:45399", "accept": false, - "reason": "Qwen3-MoE FineGrainedFP8 failure and flash-attn2 fallback checks are different issues." + "reason": "FineGrainedFP8Config failure for Qwen3-MoE is unrelated to flash-attn2 fallback blocking." }, { "left": "issue:44360", "right": "issue:44512", "accept": false, - "reason": "DSA indexer/ReLU discussion is unrelated to docs mentioning a removed run command." + "reason": "No shared underlying bug: a DSA indexer algorithm note vs stale docs about the removed run command." }, { "left": "issue:44863", "right": "issue:45005", "accept": false, - "reason": "NemotronH checkpoint loading failure is unrelated to translation-model tied weights." + "reason": "NemotronH checkpoint-loading failure is unrelated to tied weights in translation models." }, { "left": "issue:43873", "right": "issue:45005", "accept": false, - "reason": "Quantization offloading behavior is unrelated to tied-weights regressions." + "reason": "Quantization/offloading behavior and translation tied-weight handling are different problems." }, { "left": "issue:39186", "right": "issue:44918", "accept": false, - "reason": "FSDP 2-D weight error and Qwen3.5 embedding unpacking are different stack traces and code paths." + "reason": "FSDP weight dimensionality error is unrelated to Qwen3.5 embedding unpacking in TRL." }, { "left": "issue:44492", "right": "issue:44512", "accept": false, - "reason": "A typo in cache strategies is not the same as docs still mentioning a removed command." + "reason": "Different doc issues: cache-strategy typo vs removed transformers run command." }, { "left": "issue:44918", "right": "issue:45507", "accept": false, - "reason": "Embedding unpacking in TRL and GraniteMoEHybrid invalid-method calls are unrelated." + "reason": "Different failure paths: TRL input-embedding unpacking vs GraniteMoEHybrid calling an invalid method." }, { "left": "issue:43975", "right": "issue:44292", "accept": false, - "reason": "Detokenization bug and init_empty_weights _is_hf_initialized argument error are different problems." + "reason": "Different bugs: detokenization regression vs init_empty_weights constructor argument mismatch." }, { "left": "issue:43950", "right": "issue:44534", "accept": true, - "reason": "Both describe the same Transformers v5 non-persistent-buffer corruption/junk issue." + "reason": "Same underlying regression: Transformers v5 corrupts/fills non-persistent buffers during load." }, { "left": "issue:43644", "right": "issue:43950", "accept": true, - "reason": "Same regression: non-persistent buffers are corrupted/junk-filled during Transformers 5.x loading." + "reason": "Same bug cluster: non-persistent buffers being filled with junk/corrupted on from_pretrained in v5." }, { "left": "issue:44292", "right": "issue:45507", "accept": false, - "reason": "Unexpected initializer argument and GraniteMoEHybrid invalid method calls are unrelated." + "reason": "init_empty_weights argument handling and GraniteMoEHybrid invalid-method calls are unrelated." }, { "left": "issue:39692", "right": "issue:43054", "accept": false, - "reason": "SigLIP2 docs example errors and general SigLIP2 embedding quality complaint are not the same concrete bug." + "reason": "SigLIP2 docs example errors are unrelated to the later report that text embeddings are worse than SigLip." }, { "left": "issue:43828", "right": "issue:44928", "accept": false, - "reason": "Autocast dtype mismatch and Qwen3.5 gradient explosion from 3D position_ids are distinct failures." + "reason": "Both are Qwen/MoE-adjacent runtime issues, but the autocast dtype mismatch and RLHF NaN explosion have different causes." }, { "left": "issue:43866", "right": "issue:44403", "accept": false, - "reason": "Corrupted Ovis2 checkpoint and transformer loading noise are unrelated." + "reason": "Checkpoint corruption is unrelated to the generic loading-noise discussion." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet and stays conservative. The only clearly supported duplicate cluster is the non-persistent-buffer regression around issues 43644/43950/44534, and the other soft-edge rejections are appropriately cautious rather than overconfident." + "feedback": "Grounded overall and conservative in the soft-edge calls. The only clearly supported duplicate cluster in the packet is the Transformers v5 non-persistent-buffer corruption regression around issues 43950/44534/43644, and the representative choice is reasonable." }, "evaluator_used": true, "retried": false, @@ -1012,20 +1072,20 @@ "issue:43638|issue:45823", "issue:43257|issue:43866" ], - "prepared_review_unit_hash": "9504166a60dadce8b376c236e454cbf15e59cb9a8007f51f37e9c70a487f158a", + "prepared_review_unit_hash": "14f1adbe699b6788f5de374c28a2482f974afbde359bd892b81e2f1a897ea4f8", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 35, - "serialized_chars": 38121, + "serialized_chars": 38124, "estimated_input_tokens": 9531, "estimated_eval_tokens": 19318 }, @@ -1034,231 +1094,231 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of unrelated Transformers bug reports that only share a broad \u2018loading/regression\u2019 theme. The only clear duplicate soft edge is the docs pair about removed pipeline tasks; the rest are too coarse or model-specific to be safely merged.", - "confidence": 0.73, - "canonical_issue_reason": "Issue 45020 is the broadest, most representative anchor for the cluster: it captures the recurring pattern of recent-version regressions around model loading and remote_code behavior, which is the common thread across several nearby reports.", + "summary": "The packet supports two concrete duplicate pockets only: the SigLIP2 regression cluster (issue:38175, issue:43054, issue:43493) and the docs/pipeline-removal cleanup pair (issue:43827, issue:44509). The remaining soft pairs are thematic similarities, not the same underlying bug or change.", + "confidence": 0.84, + "canonical_issue_reason": "issue:43054 is the clearest global canonical issue because it sits at the center of the supported SigLIP2 pocket, linking the user-visible symptom in issue:38175 with the implementation discrepancy in issue:43493.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45020 is the best single issue to represent this cluster because it is the most general, active report about recent Transformers breakages and can subsume adjacent loading/config regressions.", + "best_issue_reason": "issue:43054 is the strongest representative issue overall: it anchors the only non-doc duplicate pocket with explicit support, while the docs pocket is a smaller cleanup pair.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:36331", "right": "issue:41762", "accept": false, - "reason": "Different failures: a CustomTrainer API mismatch vs a DeepSpeed ZeRO-3 loading IndexError." + "reason": "Different failures: a custom trainer signature mismatch vs a ZeRO-3 loading IndexError." }, { "left": "issue:41628", "right": "issue:45800", "accept": false, - "reason": "Importing AutoImageProcessor and a torch/transformers version incompatibility are unrelated problems." + "reason": "Import/export API issue vs a torch/transformers version incompatibility; not the same bug." }, { "left": "issue:43872", "right": "issue:44589", "accept": false, - "reason": "bitsandbytes Int8Params kwarg breakage and a missing Float8 storage error are different low-level incompatibilities." + "reason": "Both involve storage/quantization behavior, but the concrete errors and code paths differ." }, { "left": "issue:41553", "right": "issue:43883", "accept": false, - "reason": "Bad AutoTokenizer error messaging for Voxtral is not the same as Molmo missing all_tied_weights_keys." + "reason": "Different model-loading problems with no clear shared failure mechanism." }, { "left": "issue:38175", "right": "issue:43054", - "accept": false, - "reason": "Both mention SigLIP2, but one is zero probabilities and the other is poorer text embeddings; no clear same root bug." + "accept": true, + "reason": "Same SigLIP2 regression: one reports the bad output behavior, the other is the broader symptom cluster." }, { "left": "issue:43054", "right": "issue:45701", "accept": false, - "reason": "SigLIP2 embedding quality regression and generic tokenization changes are not the same issue." + "reason": "SigLIP2 regression vs a generic tokenization-change report; too broad to be the same bug." }, { "left": "issue:41553", "right": "issue:43881", "accept": false, - "reason": "Voxtral tokenizer messaging and GLM-4V loading failure are unrelated model-specific bugs." + "reason": "Bad tokenizer error message vs a model load failure for a different model family." }, { "left": "issue:42947", "right": "issue:44387", "accept": false, - "reason": "Gradient checkpointing/LoRA effectiveness and int4 quantization memory growth are different code paths." + "reason": "Gradient checkpointing/LoRA issue vs int4 reserved-memory/OOM regression; different concrete problems." }, { "left": "issue:44403", "right": "issue:44451", "accept": false, - "reason": "Noise during loading is not the same as failing to load vesteinn/ScandiBERT." + "reason": "Generic loading noise vs inability to load a specific model; not the same underlying defect." }, { "left": "issue:43975", "right": "issue:45520", "accept": false, - "reason": "Detokenization regression and flash_attn import/key lookup on Python 3.13 are unrelated." + "reason": "Tokenizer detokenization regression vs a flash_attn import KeyError; unrelated failures." }, { "left": "issue:43643", "right": "issue:45020", "accept": false, - "reason": "These are both remote_code-related, but 43643 is a specific missing-fields config bug while 45020 is a broad umbrella report; not the same concrete defect." + "reason": "Related in theme, but not clearly the same concrete remote_code bug." }, { "left": "issue:43054", "right": "issue:43493", - "accept": false, - "reason": "SigLIP2 text-embedding quality and SigLIP2 HF-vs-JAX discrepancy are related only by model family, not clearly the same bug." + "accept": true, + "reason": "Same SigLIP2 bug; issue:43493 describes the HF-vs-JAX discrepancy that explains issue:43054." }, { "left": "issue:43975", "right": "issue:45237", "accept": false, - "reason": "Tokenizer detokenization regression and GPT-OSS-20B AMD GPU failure are unrelated." + "reason": "Different model/use-case failures with no shared code-path evidence." }, { "left": "issue:44403", "right": "issue:44488", "accept": false, - "reason": "One is a generic loading-noise complaint, the other is a hard load failure for a specific BERT model." + "reason": "Two separate model loading failures with no concrete match." }, { "left": "issue:42947", "right": "issue:43856", "accept": false, - "reason": "LoRA gradient checkpointing and Qwen3 MoE memory usage are different performance/behavior issues." + "reason": "LoRA/checkpointing behavior vs Qwen3 MoE memory use; same area, different bug." }, { "left": "issue:43881", "right": "issue:43883", "accept": false, - "reason": "Both are model loading problems, but GLM-4V loading failure and Molmo missing tied-weights metadata are not the same bug." + "reason": "Different load-time errors for different models; not the same defect." }, { "left": "issue:41762", "right": "issue:42915", "accept": false, - "reason": "ZeRO-3 index error during load and Qwen3 MoE FineGrainedFP8 failure do not share the same concrete fault." + "reason": "Both are loading/training issues, but one is a ZeRO-3 IndexError and the other is a Qwen3 MoE FP8 config failure." }, { "left": "issue:42491", "right": "issue:44368", "accept": false, - "reason": "A LoRA compatibility regression across versions and a tied-embeddings warning in Qwen3.5 are different issues." + "reason": "LoRA/Qwen3 MoE compatibility issue vs a tie_word_embeddings warning; too indirect to merge." }, { "left": "issue:45685", "right": "issue:45823", "accept": false, - "reason": "MPS histogram support for Int and Gemma4 device mismatch are unrelated backend bugs." + "reason": "MPS histogram limitation vs Gemma4 device mismatch; different device/runtime bugs." }, { "left": "issue:43975", "right": "issue:44451", "accept": false, - "reason": "Detokenization regression and inability to load a specific model are separate problems." + "reason": "Different tokenizer/model-specific regressions." }, { "left": "issue:43824", "right": "issue:43872", "accept": false, - "reason": "Qwen2.5-VL import failure and bitsandbytes Int8Params incompatibility are unrelated." + "reason": "ImportError for Qwen2.5 VL vs bitsandbytes Int8Params kwarg incompatibility; unrelated." }, { "left": "issue:45020", "right": "issue:45823", "accept": false, - "reason": "Broad remote_code regressions and a Gemma4 PLE device_map mismatch are different failures." + "reason": "Broad remote_code breakage vs a Gemma4 PLE device-mapping issue; not the same bug." }, { "left": "issue:44991", "right": "issue:45399", "accept": false, - "reason": "Tokenizer loading for est-roberta and flash-attn fallback logic are unrelated." + "reason": "Tokenizer loading regression vs flash-attn fallback blocking; no shared concrete failure." }, { "left": "issue:44661", "right": "issue:45020", "accept": false, - "reason": "Model-registration edge case in add-new-model-like is not the same as the broader remote_code regression report." + "reason": "Tokenizer-mapping registration bug vs broader remote_code regressions; not the same underlying change." }, { "left": "issue:43828", "right": "issue:45823", "accept": false, - "reason": "autocast dtype mismatch in Phi-tiny-MoE and Gemma4 device mismatch are different runtime errors." + "reason": "Autocast dtype mismatch vs device mismatch; different execution-path errors." }, { "left": "issue:45020", "right": "issue:45356", "accept": false, - "reason": "A broad remote_code regression and a specific Kimi-K2.5 tokenizer regression are not the same underlying bug." + "reason": "Broad remote_code issue vs a specific tokenizer codec/warning regression." }, { "left": "issue:43782", "right": "issue:44960", "accept": false, - "reason": "Qwen3VL weight_only=True failure and GLM5 loading are unrelated." + "reason": "Different model-loading failures with no evidence of a shared root cause." }, { "left": "issue:45081", "right": "issue:45399", "accept": false, - "reason": "Mistral regex patch crash and flash-attn fallback blocking are separate tokenizer/dependency issues." + "reason": "Mistral tokenizer regex crash vs flash-attn fallback logic; unrelated issues." }, { "left": "issue:43827", "right": "issue:44509", "accept": true, - "reason": "Clear duplicate docs issue: both report that removed pipeline tasks are still referenced in the documentation." + "reason": "Same docs cleanup after v5 pipeline removals; both report stale pipeline references in documentation." }, { "left": "issue:43404", "right": "issue:43950", "accept": false, - "reason": "Untied lm_head weights in Mistral3 and non-persistent buffer corruption in from_pretrained are unrelated state/loading bugs." + "reason": "Weight-tying/model head issue vs non-persistent buffer corruption; different code paths." }, { "left": "issue:43792", "right": "issue:44220", "accept": false, - "reason": "Whisper runtime failure and a specific _torch_extract_fbank_features() problem may be related, but the evidence is too weak to treat them as the same defect." + "reason": "Whisper failing to run and an fbank feature-extraction issue are not clearly the same concrete bug." }, { "left": "issue:43742", "right": "issue:45399", "accept": false, - "reason": "MobileLLM load failure and flash-attn fallback gating are unrelated." + "reason": "Different loading/runtime failures for unrelated models." }, { "left": "issue:42915", "right": "issue:45020", "accept": false, - "reason": "Qwen3 MoE FineGrainedFP8 failure is model/config-specific and not the same as the broad remote_code regression report." + "reason": "Qwen3 MoE FP8 config failure vs broad remote_code regressions; not a duplicate." }, { "left": "issue:43638", "right": "issue:45823", "accept": false, - "reason": "ZeRO-3/BERT index error and Gemma4 device mismatch are different loading/runtime issues." + "reason": "ZeRO-3 IndexError vs Gemma4 device-mapping mismatch; different failures." }, { "left": "issue:43257", "right": "issue:43866", "accept": false, - "reason": "Qwen3 MoE weight-conversion failure with accelerate+deepspeed and a corrupted Ovis2 checkpoint are different model-specific loading problems." + "reason": "Qwen3 MoE weight conversion problem vs a corrupted checkpoint report; similar family, different failure." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative: it treats the cluster as mostly unrelated reports with only one clearly duplicate docs pair, which is supported by the issue titles and the soft-edge evidence. The canonical issue choice is plausible, and the pairwise rejections do not overclaim shared root causes." + "accept": false, + "feedback": "The SigLIP2 and docs/pipeline-removal pockets are grounded, but the summary overstates the evidence by claiming a separate DeepSpeed ZeRO-3 cluster that is not actually supported by the packet's soft pairs or accepted verdicts. Keep the clustering summary limited to the pockets with explicit evidence in the packet." }, "evaluator_used": true, - "retried": false, + "retried": true, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -1359,235 +1419,235 @@ "issue:43632|issue:45834", "issue:44155|issue:45584" ], - "prepared_review_unit_hash": "a18324d290a5a267307f2169a4383b3b0ed1cfa9ccf8b7d25c493c8db7c6e7fa", + "prepared_review_unit_hash": "e0e825cc8aa59d66b2a8fc5afd5352843599b5a487604b70542677d771151cec", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 32, - "serialized_chars": 37422, - "estimated_input_tokens": 9356, - "estimated_eval_tokens": 18968 + "serialized_chars": 37425, + "estimated_input_tokens": 9357, + "estimated_eval_tokens": 18970 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is mostly a mix of unrelated issue threads, with a few true duplicates: the `_is_hf_initialized`/`init_empty_weights` regression, the generic mask-shape IndexError, the Qwen3.5/Qwen3-VL video regression, the custom-model-init Transformers 5 breakage, and the SigLIP2 wrong-results report. Most other soft links are only same-model-family or same-subsystem, not the same bug.", - "confidence": 0.79, - "canonical_issue_reason": "No single canonical issue applies to the whole set because it splits into several unrelated bug families; if forced to pick one representative of a true duplicate cluster, issue #44291 is the clearest broad root-cause report (`_is_hf_initialized` / `init_empty_weights`).", + "summary": "Mostly heterogeneous cluster with a few tight duplicate subclusters around Transformers v5 init/loading regressions and a couple of near-exact symptom repeats. The SigLIP2 pair should stay separate on this evidence packet: related symptoms, but no confirmed shared root cause or duplicate-level proof.", + "confidence": 0.77, + "canonical_issue_reason": "issue:43632 is the best anchor because it is the broadest central regression here: a Transformers v5 init/loading break that overlaps with several nearby loading failures.", "canonical_pr_reason": null, - "best_issue_reason": "Issue #44291 is the strongest standalone issue to keep as a representative because it describes the underlying regression clearly and at the right abstraction level, rather than only one downstream symptom.", + "best_issue_reason": "issue:43632 is the strongest single representative of the cluster\u2019s main theme: loading/initialization regressions tied to the Transformers v5 `_is_hf_initialized` change.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43940", "right": "issue:45313", "accept": false, - "reason": "Same DeepSpeed ZeRO-3 loading symptom, but different Qwen model lines and no evidence they are the exact same code-path bug." + "reason": "Both are DeepSpeed weight-loading problems, but they affect different Qwen variants and failure modes." }, { "left": "issue:43872", "right": "issue:44291", "accept": true, - "reason": "Both point to the same `_is_hf_initialized` constructor-argument regression; one surfaces through bitsandbytes, the other through `init_empty_weights`." + "reason": "Same underlying `_is_hf_initialized` regression: a new Transformers v5 kwarg breaks constructor/loading paths in different surfaces." }, { "left": "issue:45161", "right": "issue:45823", "accept": false, - "reason": "Both mention model execution problems, but they are different models and different failures (TP/MoE load vs Gemma4 PLE device mismatch)." + "reason": "Same broad model area, but these are different bugs: TP/GPT-OSS MoE vs Gemma4 device-mismatch in forward." }, { "left": "issue:43854", "right": "issue:44863", "accept": false, - "reason": "Both are model-loading failures, but for unrelated model families and likely different loader bugs." + "reason": "Different model-loading failures for different architectures; no clear shared code path." }, { "left": "issue:41553", "right": "issue:44991", "accept": false, - "reason": "Both concern tokenizer loading, but one is a Voxtral error-message issue and the other is a concrete tokenizer-loading regression for a different model." + "reason": "Both concern tokenizers, but one is a bad error message and the other is a loading regression for a different model." }, { "left": "issue:42915", "right": "issue:43872", "accept": false, - "reason": "These are unrelated regressions: Qwen3MoE + FineGrainedFP8Config versus the `_is_hf_initialized`/bitsandbytes issue." + "reason": "Quantization/FP8 failure vs `_is_hf_initialized` constructor incompatibility; different root causes." }, { "left": "issue:44928", "right": "issue:45588", "accept": false, - "reason": "Both involve attention/video-style code paths, but the failures are different and not the same bug." + "reason": "Both mention attention/video-ish behavior, but the concrete failures and code paths are unrelated." }, { "left": "issue:43257", "right": "issue:45305", "accept": false, - "reason": "Different DeepSpeed-related problems: one is Qwen3 MoE weight conversion, the other is gradient averaging with `model_accepts_loss_kwargs=True`." + "reason": "Both involve DeepSpeed, but one is missing MOE weight conversion and the other is GAS gradient averaging." }, { "left": "issue:36331", "right": "issue:41093", "accept": false, - "reason": "Unrelated failures: custom trainer loss signature vs mask/tensor shape mismatch." + "reason": "Trainer API regression vs tensor-mask shape mismatch; unrelated bugs." }, { "left": "issue:41553", "right": "issue:44488", "accept": false, - "reason": "Both are loading-related, but the titles indicate different models and different symptoms." + "reason": "Both are loading-related, but they are different tokenizer/model issues with different failure mechanisms." }, { "left": "issue:43872", "right": "issue:45538", "accept": false, - "reason": "The `_is_hf_initialized` regression and CLIPTokenizer's `model_max_length` bug are unrelated." + "reason": "`_is_hf_initialized` constructor issue vs CLIPTokenizer max-length constant bug; unrelated." }, { "left": "issue:44530", "right": "issue:44863", "accept": false, - "reason": "Different model families and different failure modes; same broad area is not enough to merge." + "reason": "PagedAttention cache crash vs checkpoint-loading failure for NemotronH; different bugs." }, { "left": "issue:41093", "right": "issue:44805", "accept": true, - "reason": "These are effectively the same IndexError with the same mask/tensor shape mismatch symptom." + "reason": "Same concrete mask/tensor shape-mismatch error; the later report is a repeat of the same bug class." }, { "left": "issue:44479", "right": "issue:44560", - "accept": true, - "reason": "`44560` looks like a concrete Qwen3-VL video manifestation of the broader transformers 5.3.0 video-input regression reported in `44479`." + "accept": false, + "reason": "Both are Qwen video regressions, but the titles describe different concrete failures and model scopes." }, { "left": "issue:24643", "right": "issue:30064", "accept": false, - "reason": "Training with DeepSpeed and void-segmentation image processing are unrelated." + "reason": "Unrelated: DeepSpeed training error vs image processor segmentation-map bug." }, { "left": "issue:43366", "right": "issue:43828", "accept": false, - "reason": "No shared bug: GGUF architecture support versus an autocast dtype mismatch in a different model." + "reason": "GGUF support request vs autocast dtype mismatch; no shared bug." }, { "left": "issue:45663", "right": "issue:45823", "accept": false, - "reason": "Both are Gemma4-related, but one is an FSDP2 training KeyError and the other is a device-map forward mismatch." + "reason": "Same model family, but FSDP2 shared_kv_states rebuild vs device_map auto mismatch are different issues." }, { "left": "issue:43366", "right": "issue:43531", "accept": false, - "reason": "Different Qwen/GPT-OSS support topics with no shared concrete bug." + "reason": "GGUF gpt-oss support and Qwen3-MoE sliding-window handling are unrelated." }, { "left": "issue:43950", "right": "issue:45834", "accept": false, - "reason": "Non-persistent buffer corruption and Kosmos2.5 OCR index errors are unrelated." + "reason": "Non-persistent buffer corruption vs long OCR index error; unrelated." }, { "left": "issue:43550", "right": "issue:44910", "accept": false, - "reason": "Both mention attention/compile behavior, but the model, trigger, and failure mechanism differ." + "reason": "Different attention/compile failures on different models and code paths." }, { "left": "issue:43645", "right": "issue:43646", "accept": true, - "reason": "Same Transformers 5 custom-model initialization breakage; one report is just notebook-specific." + "reason": "Near-identical Transformers 5.0 custom-model initialization regression; same underlying breakage with duplicate reporting." }, { "left": "issue:43366", "right": "issue:44589", "accept": false, - "reason": "Unrelated issues: GPT-OSS GGUF support versus a missing Float8 storage type." + "reason": "GGUF support request vs missing Float8 storage type; unrelated." }, { "left": "issue:43950", "right": "issue:45325", "accept": false, - "reason": "Different regressions: buffer-loading corruption versus Qwen2.5-VL RoPE/video position scaling." + "reason": "Different regressions: buffer corruption in from_pretrained vs Qwen2.5-VL rope-index scaling." }, { "left": "issue:44560", "right": "issue:44918", "accept": false, - "reason": "Both are Qwen3.5 input-related, but one is a video regression and the other is an SFT embedding unpacking failure." + "reason": "Both are Qwen-related, but one is a video StopIteration bug and the other is an SFT embedding unpacking failure." }, { "left": "issue:42617", "right": "issue:43366", "accept": false, - "reason": "3d_parallel launch failure and GPT-OSS model-architecture support are unrelated." + "reason": "3D parallel script failure vs GGUF support request; unrelated." }, { "left": "issue:43824", "right": "issue:43957", "accept": false, - "reason": "Different loading failures for different model classes; no clear shared bug." + "reason": "Import error for a specific class vs meta-device loading regression; different issues." }, { "left": "issue:42915", "right": "issue:43957", "accept": false, - "reason": "Qwen3MoE FP8 config failure is unrelated to meta-device loading regressions." + "reason": "FP8 config failure vs meta-device loading regression; different code paths." }, { "left": "issue:44661", "right": "issue:45823", "accept": false, - "reason": "Tokenizer-mapping generation bug and Gemma4 device mismatch are unrelated." + "reason": "Tokenizer-mapping helper bug vs Gemma4 device mismatch; unrelated." }, { "left": "issue:36010", "right": "issue:42915", "accept": false, - "reason": "GenerationMixin import failure and Qwen3MoE FP8 config failure are unrelated." + "reason": "Import error for GenerationMixin vs Qwen3Moe FP8 failure; unrelated." }, { "left": "issue:38175", "right": "issue:43994", - "accept": true, - "reason": "Same SigLIP2 model producing nonsensical/zero-probability outputs; these are duplicate symptom reports of the same bug." + "accept": false, + "reason": "Both are SigLIP2 output-quality regressions, but the packet does not establish that they are duplicates or share a confirmed root cause." }, { "left": "issue:43632", "right": "issue:45834", "accept": false, - "reason": "`_is_hf_initialized`-flag regression and Kosmos2.5 OCR index error are unrelated." + "reason": "Core Transformers v5 init-flag regression vs Kosmos2.5 long-input OCR index error; unrelated." }, { "left": "issue:44155", "right": "issue:45584", "accept": false, - "reason": "AudioFlamingo3 track leakage and Whisper empty-transcription failure are different bugs in different code paths." + "reason": "AudioFlamingo3 batch leakage bug vs Whisper empty-transcription failure; unrelated." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The clustering stays grounded in the packet and is mostly conservative: the rejected links are clearly treated as same-subsystem but not same bug, while the accepted ones have strong title-level overlap or an explicit shared regression family. The overall summary does not overclaim a single canonical cluster." + "accept": false, + "feedback": "Mostly grounded, but one soft-edge verdict is too strong for the evidence packet: accepting issue:38175 \u2194 issue:43994 as the same underlying bug goes beyond what the titles support. They are both SigLIP2 output-quality regressions, but the packet does not show they are duplicates or share a confirmed root cause. The rest of the cluster framing is conservative and reasonable, but this over-acceptance means I can\u2019t approve the evaluation as written." }, "evaluator_used": true, - "retried": false, + "retried": true, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -1669,8 +1729,8 @@ "issue:42491|issue:43720", "issue:43854|issue:43931", "issue:41720|issue:42915", - "issue:42491|issue:45276", "issue:43708|issue:44075", + "issue:42491|issue:45276", "issue:43257|issue:43854", "issue:43856|issue:43873", "issue:42915|issue:44756", @@ -1686,220 +1746,214 @@ "issue:45276|issue:45335", "issue:43404|issue:43454" ], - "prepared_review_unit_hash": "3b730505d1b98aa58811f867e452f034f16a51abe72e261e085b6442a958f8eb", + "prepared_review_unit_hash": "f2a83a50b2e6438511ed909e6a28227a033cd2a5b2f37ffc668f394923db9fab", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 31, - "serialized_chars": 36415, - "estimated_input_tokens": 9104, - "estimated_eval_tokens": 18464 + "serialized_chars": 36412, + "estimated_input_tokens": 9103, + "estimated_eval_tokens": 18462 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly heterogeneous, but there are a few clear duplicate-like subgroups: a non-persistent-buffer regression pair, a tokenizer/model-loading pair, and a config-validation loading pair. Most other soft pairs look related only at a broad subsystem level and should stay separate.", - "confidence": 0.49, - "canonical_issue_reason": "issue:43644 is the clearest broad regression report and has an essentially identical duplicate in issue:44534, making it the strongest canonical issue among the accepted duplicates.", + "summary": "This cluster is mostly heterogeneous, but there are two clear duplicate families: the v5 non-persistent buffer junk regression and the resize_token_embeddings/output_embeddings regression. Most other pairs are only loosely related by subsystem or model family and should stay separate.", + "confidence": 0.74, + "canonical_issue_reason": "issue:45292 is the best umbrella issue for the resize_token_embeddings regression: it is the most generic title and cleanly covers the narrower Gemma4/T5Gemma variants.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43644 best represents the only exact duplicate subcluster here: it states the regression generically, is earlier, and is directly mirrored by issue:44534.", + "best_issue_reason": "issue:45292 is the strongest cluster anchor among the issues shown because it represents a broad, recurring code-path bug rather than a model-specific symptom.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44877", "right": "issue:45030", - "accept": true, - "reason": "Both are config-loading failures caused by stricter validation changes; a single relaxation/fix could plausibly cover both." + "accept": false, + "reason": "Both are config-validation load failures, but they affect different model families and likely different validation rules." }, { "left": "issue:44560", "right": "issue:44805", "accept": false, - "reason": "Different failures and code paths: one is a StopIteration in Qwen3-VL video embedding, the other is a mask/tensor shape mismatch." + "reason": "Different symptoms and code paths: video StopIteration in Qwen3-VL embedding versus a mask/tensor shape IndexError." }, { "left": "issue:43638", "right": "issue:44661", "accept": false, - "reason": "Unrelated bugs: ZeRO-3/Bert initialization vs add-new-model-like handling of TOKENIZER_MAPPING_NAMES." + "reason": "One is a ZeRO-3/BertModel indexing failure; the other is an add-new-model-like tokenizer mapping failure." }, { "left": "issue:43317", "right": "issue:44756", "accept": false, - "reason": "Both involve memory/offload behavior, but one is quantized model loading and the other is mmap OOM on a specific platform; not the same bug." + "reason": "GPU+CPU offload load failure is unrelated to disabling mmap to avoid OOM." }, { "left": "issue:43596", "right": "issue:44661", "accept": false, - "reason": "Different areas: DeepSpeed ZeRO-3 index error versus tokenizer-mapping logic." + "reason": "Different failure modes: ZeRO-3 with BertModel versus tokenizer mapping logic." }, { "left": "issue:44661", "right": "issue:44805", "accept": false, - "reason": "No common underlying bug; one is tokenizer-mapping behavior, the other is a mask indexing error." + "reason": "Tokenizer mapping bug and tensor mask shape mismatch are unrelated." }, { "left": "issue:38175", "right": "issue:39692", "accept": false, - "reason": "Same model family, but the reported problems differ: zero probabilities vs documentation/model-processor mismatch and quantization issues." + "reason": "Same model family, but one is zero probabilities and the other is a docs/example mismatch plus quantization failure; not the same concrete bug." }, { "left": "issue:43644", "right": "issue:44534", "accept": true, - "reason": "These describe the same v5 regression: non-persistent buffers being initialized with junk." + "reason": "These describe the same Transformers v5 regression where non-persistent buffers get filled with junk." }, { "left": "issue:45313", "right": "issue:45357", "accept": false, - "reason": "Different Qwen3.5 failures: ZeRO-3 weight loading for language_model versus incorrect visual encoder keys on save_pretrained." + "reason": "Different Qwen3.5 issues: ZeRO-3 loading of language_model weights versus incorrect visual encoder keys on save_pretrained." }, { "left": "issue:45488", "right": "issue:45812", "accept": false, - "reason": "Both are tokenizer regressions, but they target different mechanisms and model families (LlamaTokenizer pre-tokenizer override vs Granite token ID regression)." + "reason": "Both are tokenizer regressions, but they affect different model families and different tokenization failures." }, { "left": "issue:44960", "right": "issue:44977", "accept": false, - "reason": "Different models and symptoms; these are both flash-attention related only at a very high level." + "reason": "Unrelated models and failures: GLM5 versus Qwen3.5 flash-attention generation." }, { "left": "issue:44451", "right": "issue:44488", - "accept": true, - "reason": "Both report current-version model loading failures for BERT-family checkpoints, consistent with the same loader regression." + "accept": false, + "reason": "Both are model loading failures, but they involve different models and no shared concrete failure is evident." }, { "left": "issue:42491", "right": "issue:43720", "accept": false, - "reason": "Distinct model families and failure modes: LoRA compatibility across hf4.x/hf5.x versus BitNet packed-weight unpacking during accelerate loading." + "reason": "Different problems: hf4.x to hf5.x LoRA/Qwen3 MoE incompatibility versus BitNet packed-weight unpacking during accelerate load." }, { "left": "issue:43854", "right": "issue:43931", "accept": false, - "reason": "Different models and different loading errors; not the same concrete bug." + "reason": "Different models and different loading errors; no same underlying bug is evident." }, { "left": "issue:41720", "right": "issue:42915", "accept": false, - "reason": "Both are Qwen3-related, but one is a device-map/cuda assert and the other is a FineGrainedFP8Config failure." + "reason": "Both involve Qwen3, but one is device-mapping cudaErrorAssert and the other is FineGrainedFP8Config failure." }, { - "left": "issue:42491", - "right": "issue:45276", + "left": "issue:43708", + "right": "issue:44075", "accept": false, - "reason": "No clear shared code path; one is LoRA compatibility and the other is resize_token_embeddings propagation." + "reason": "Trainer resume/max_steps math is unrelated to SGD arguments being ignored." }, { - "left": "issue:43708", - "right": "issue:44075", + "left": "issue:42491", + "right": "issue:45276", "accept": false, - "reason": "Trainer resume/max_steps logic is unrelated to optimizer argument handling." + "reason": "Different model families and failure modes; one is Qwen3 MoE compatibility, the other is resize_token_embeddings propagation." }, { "left": "issue:43257", "right": "issue:43854", "accept": false, - "reason": "Qwen3 MoE weight-conversion/loading under accelerate+deepspeed is a different issue from loading GLM-4.7 in tests." + "reason": "Both concern loading, but one is Qwen3 MoE weight conversion and the other is a GLM-4.7-Flash unit-test load issue." }, { "left": "issue:43856", "right": "issue:43873", "accept": false, - "reason": "Both mention memory, but one is Qwen3 MoE training efficiency and the other is quantization/offloading behavior." + "reason": "Training memory usage and quantization offloading are only broadly related, not the same bug." }, { "left": "issue:42915", "right": "issue:44756", "accept": false, - "reason": "Different problems: FP8 config failure versus mmap OOM workaround on Strix Halo." + "reason": "Qwen3 MoE FP8 failure is unrelated to Strix Halo mmap OOM mitigation." }, { "left": "issue:39692", "right": "issue:43994", "accept": false, - "reason": "Both are SigLIP2-related, but one is a doc-example/model-processor problem and the other is nonsensical outputs with AutoModel/pipeline." + "reason": "Both mention SigLIP2, but one is a documentation/example error and the other is a model/pipeline output regression." }, { "left": "issue:42491", "right": "issue:43931", "accept": false, - "reason": "Different failure modes: LoRA model reuse across versions versus a Qwen3-VL weight-shape mismatch." + "reason": "Different Qwen3-related loading bugs with no shared concrete code path shown." }, { "left": "issue:43856", "right": "issue:44387", "accept": false, - "reason": "Qwen3 MoE training memory usage and int4 reserved-memory OOM are related only broadly, not the same bug." + "reason": "Different memory issues: MoE training memory usage versus int4 quantization reserved-memory OOM." }, { "left": "issue:43644", "right": "issue:43645", "accept": false, - "reason": "Both are v5 regressions, but one is buffer initialization and the other is notebook custom-model initialization." + "reason": "Both are Transformers v5 regressions, but one is about non-persistent buffers and the other about custom model initialization in notebooks." }, { "left": "issue:44910", "right": "issue:45588", "accept": false, - "reason": "Different flash-attention crashes with different triggers: 3D position_ids misread as packed sequence versus s_aux=None handling." + "reason": "Both touch flash-attention, but one is a packed-sequence position_ids crash and the other is an s_aux=None AttributeError." }, { "left": "issue:45362", "right": "issue:45406", "accept": false, - "reason": "Different model/processor crashes: Qwen3.5 chat generation versus Gemma4Processor missing _tokenizer in serve." + "reason": "Different serving/chat crashes and different model classes." }, { "left": "issue:44488", "right": "issue:44991", "accept": false, - "reason": "Both are tokenizer/load regressions, but they affect different tokenizer stacks and appear to have different underlying causes." - }, - { - "left": "issue:36010", - "right": "issue:39404", - "accept": false, - "reason": "Importing GenerationMixin and Whisper return_language pipeline behavior are unrelated." + "reason": "Tokenizer load failures, but for different models and likely different root causes." }, { "left": "issue:45276", "right": "issue:45335", - "accept": false, - "reason": "Same general API surface, but the model classes and reported propagation failures are different enough that this is not the same bug." + "accept": true, + "reason": "Both report resize_token_embeddings failing to update output/decoder embeddings; 45335 is the narrower T5Gemma variant of the same bug." }, { "left": "issue:43404", "right": "issue:43454", "accept": false, - "reason": "Both are weight-tying bugs, but they affect different model classes and are not clearly the same code-path failure." + "reason": "Both are weight-tying problems, but they affect different model classes and are not clearly the same concrete bug." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays mostly conservative: it identifies the strongest duplicate-like pairs without over-claiming broad cluster cohesion. The accepted soft edges are plausible from the issue titles alone, and the rejected ones mostly avoid over-merging on high-level subsystem similarity. No obvious overstatement stands out." + "feedback": "The summary is grounded in the packet and stays conservative overall. The two accepted pairs are strongly supported by the titles: issue:43644/44534 are the same non-persistent buffer junk regression, and issue:45276/45335 are the same resize_token_embeddings/output_embeddings bug family. The rest of the soft-edge verdicts are appropriately cautious and do not overclaim evidence." }, "evaluator_used": true, "retried": false, @@ -1984,8 +2038,8 @@ "issue:44291|issue:44756", "issue:44292|issue:44912", "issue:42947|issue:43381", - "issue:45684|issue:45698", "issue:45276|issue:45292", + "issue:45684|issue:45698", "issue:43381|issue:44928", "issue:44164|issue:44756", "issue:41720|issue:44155", @@ -2005,266 +2059,272 @@ "issue:44479|issue:45663", "issue:43906|issue:44530", "issue:39404|issue:45701", - "issue:43425|issue:45823", "issue:43611|issue:43645", + "issue:43425|issue:45823", "issue:43931|issue:44863" ], - "prepared_review_unit_hash": "88862b63d1dc5cdd569c79fe26be33c8a32cc5b06911122911a82215ab1ed292", + "prepared_review_unit_hash": "c33d2c8c9ee12860070b42abffa5b69b45015f4a329947c72d3b08d35119f4ed", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 38, - "serialized_chars": 39072, - "estimated_input_tokens": 9768, - "estimated_eval_tokens": 19792 + "serialized_chars": 39075, + "estimated_input_tokens": 9769, + "estimated_eval_tokens": 19794 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly heterogeneous and does not look like one duplicate set. The only clear duplicate-like subgroup is the resize_token_embeddings regression across the generic, Gemma4, and T5Gemma reports; the rest are different bugs or different code paths.", - "confidence": 0.85, - "canonical_issue_reason": "Issue 45292 is the broadest, most model-agnostic formulation of the resize_token_embeddings bug, and the Gemma4/T5Gemma reports read as variants of the same missing embedding-update behavior.", + "summary": "This cluster is mostly heterogeneous. The only convincing duplicate-style pair is the `resize_token_embeddings` regression around output embeddings; the rest are distinct bugs affecting different models/code paths.", + "confidence": 0.83, + "canonical_issue_reason": "issue:45292 is the broad, generic report for the `resize_token_embeddings` regression; issue:45276 is a model-specific Gemma4 repro of the same output-embedding problem.", "canonical_pr_reason": null, - "best_issue_reason": "45292 is the best anchor for triage because it captures the shared failure mode without model-specific noise and can plausibly subsume the closely related embedding-update reports.", + "best_issue_reason": "issue:45292 is the clearest representative issue because it states the underlying bug in the most general form and can subsume the more specific Gemma4 variant.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44164", "right": "issue:44291", "accept": false, - "reason": "Different bugs: save/from_pretrained extra_state handling vs init_empty_weights/_is_hf_initialized." + "reason": "Different bugs: save/from_pretrained extra_state handling vs `init_empty_weights` passing an unexpected `_is_hf_initialized` arg." }, { "left": "issue:44402", "right": "issue:44488", "accept": false, - "reason": "Both are load failures, but the reports point to different models and the shared root cause is not clear enough." + "reason": "Both are model-loading failures, but for different models and different root causes." }, { "left": "issue:43425", "right": "issue:43957", "accept": false, - "reason": "Both are Transformers 5 regressions, but one is Torch version incompatibility and the other is meta-device loading; different code paths." + "reason": "Torch version incompatibility is not the same as `torch.device(\"meta\")` loading breakage." }, { "left": "issue:39692", "right": "issue:44960", "accept": false, - "reason": "Unrelated models and failures: SigLIP2 docs example issues vs GLM5 loading." + "reason": "Documentation/example errors vs a GLM5 model issue; no shared underlying defect." }, { "left": "issue:42947", "right": "issue:45704", "accept": false, - "reason": "Gradient checkpointing/LoRA is unrelated to the apex RMSNorm memory leak in T5." + "reason": "Gradient checkpointing/LoRA behavior is unrelated to the apex RMSNorm memory leak." }, { "left": "issue:43335", "right": "issue:45663", "accept": false, - "reason": "Different subsystems: SwitchTransformers sparse-layer config vs Gemma4 FSDP2 shared_kv_states KeyError." + "reason": "Different model families and failures: SwitchTransformers sparse-layer config vs Gemma4 FSDP2 `KeyError`." }, { "left": "issue:43873", "right": "issue:44387", "accept": false, - "reason": "Both involve quantization/memory, but one is offloading behavior and the other is reserved-memory/OOM regression; not the same bug." + "reason": "Quantization/offloading behavior and int4 CUDA reserved-memory OOM are related only broadly, not the same bug." }, { "left": "issue:45663", "right": "issue:45759", "accept": false, - "reason": "Gemma4 FSDP2 attention state bug vs multimodal Qwen config unwrapping; unrelated." + "reason": "Gemma4 FSDP2 attention-state rebuild is unrelated to Qwen multimodal `text_config` unwrapping." }, { "left": "issue:44155", "right": "issue:45823", "accept": false, - "reason": "AudioFlamingo3 batched inference leak is unrelated to Gemma4 device-map mismatch." + "reason": "AudioFlamingo3 batch leakage and Gemma4 device-map mismatch are different code paths." }, { "left": "issue:43854", "right": "issue:45216", "accept": false, - "reason": "Different model-loading regressions on different models and releases." + "reason": "Different regressions: unit-test loading for GLM-4.7-Flash vs Qwen3.5 save_pretrained correctness." }, { "left": "issue:44387", "right": "issue:44756", "accept": false, - "reason": "Both mention memory pressure, but one is int4 quantization OOM and the other is mmap on Strix Halo; different causes." + "reason": "Int4 OOM from reserved memory growth is not the same as disabling mmap on Strix Halo." }, { "left": "issue:44291", "right": "issue:44756", "accept": false, - "reason": "init_empty_weights argument handling is unrelated to mmap/OOM on Strix Halo." + "reason": "Unexpected init arg during meta init vs mmap/OOM on Strix Halo are unrelated." }, { "left": "issue:44292", "right": "issue:44912", "accept": false, - "reason": "Generic output_embeddings resize issue vs MXFP4 quantization fallback; different bugs." + "reason": "Qwen3-VL shape mismatch and MXFP4 fallback to bf16 are different model-loading failures." }, { "left": "issue:42947", "right": "issue:43381", "accept": false, - "reason": "Gradient checkpointing in eval mode is not the same as ineffective checkpointing with PEFT LoRA." - }, - { - "left": "issue:45684", - "right": "issue:45698", - "accept": false, - "reason": "Custom-model file permission propagation is unrelated to loading the wrong custom module after save_pretrained." + "reason": "Both mention gradient checkpointing, but one is PEFT LoRA effectiveness and the other is eval-mode prohibition." }, { "left": "issue:45276", "right": "issue:45292", "accept": true, - "reason": "Same resize_token_embeddings propagation bug; 45276 is the Gemma4-specific manifestation of the broader 45292 report." + "reason": "Same underlying `resize_token_embeddings` bug: the generic output-embedding regression in 45292 matches the Gemma4-specific repro in 45276." + }, + { + "left": "issue:45684", + "right": "issue:45698", + "accept": false, + "reason": "Read-only permission propagation in save_pretrained is a different bug from loading the wrong custom module after save_pretrained." }, { "left": "issue:43381", "right": "issue:44928", "accept": false, - "reason": "Eval-mode checkpointing restriction vs RLHF NaN/SDPA math fallback issue; different failure modes." + "reason": "Eval-mode checkpointing restriction is unrelated to the RLHF NaN/SDPA fallback issue." }, { "left": "issue:44164", "right": "issue:44756", "accept": false, - "reason": "save/from_pretrained extra_state handling is unrelated to disabling mmap to avoid OOM." + "reason": "extra_state save/load handling is unrelated to Strix Halo mmap memory pressure." }, { "left": "issue:41720", "right": "issue:44155", "accept": false, - "reason": "Qwen3 auto device-map assert and AudioFlamingo3 track leakage are unrelated." + "reason": "Qwen3 device-map cuda assert and AudioFlamingo3 batch token leakage are different failures." }, { "left": "issue:45588", "right": "issue:45663", "accept": false, - "reason": "Flash-attention sinkless-model crash and Gemma4 FSDP2 KeyError are different code paths." + "reason": "Flash-attention crash on `s_aux=None` is not the same as Gemma4 FSDP2 shared-state rebuilding." }, { "left": "issue:44479", "right": "issue:44877", "accept": false, - "reason": "Video-input regression in Qwen vision models vs granite_speech config loading; not the same bug." + "reason": "Qwen video-input regression and strict-config loading for granite_speech are different issues." }, { "left": "issue:43646", "right": "issue:43708", "accept": false, - "reason": "Custom model initialization breakage vs Trainer resume/max_steps miscalculation; unrelated." + "reason": "Custom model initialization in notebooks is not the same as incorrect `max_steps` on checkpoint resume." }, { "left": "issue:43854", "right": "issue:44960", "accept": false, - "reason": "Different model families and different loading failures." + "reason": "GLM-4.7-Flash loading failure and GLM5 are different model-specific reports." }, { "left": "issue:43957", "right": "issue:44756", "accept": false, - "reason": "Meta-device loading regression is unrelated to the Strix Halo mmap OOM report." + "reason": "Meta-device loading breakage is unrelated to mmap being disabled on Strix Halo." }, { "left": "issue:45292", "right": "issue:45335", - "accept": true, - "reason": "Same underlying resize_token_embeddings bug; 45335 is the T5Gemma-specific case of the broader embedding-update failure." + "accept": false, + "reason": "Both are `resize_token_embeddings` regressions, but they hit different components (`output_embeddings` vs `decoder.embed_tokens`) and don't clearly share one exact code-path bug." }, { "left": "issue:44402", "right": "issue:44451", "accept": false, - "reason": "Similar class of model-loading complaints, but the root cause is not established as the same." + "reason": "Different model-loading failures for different checkpoints; too model-specific to merge." }, { "left": "issue:43828", "right": "issue:43957", "accept": false, - "reason": "Autocast dtype mismatch on Phi-tiny-MoE is unrelated to meta-device model loading failures." + "reason": "Dtype mismatch under autocast is unrelated to meta-device loading breakage." }, { "left": "issue:43645", "right": "issue:44164", "accept": false, - "reason": "Custom-model notebook initialization vs save/from_pretrained extra_state handling; different issues." + "reason": "Custom-model notebook initialization and extra_state save/load are different problems." }, { "left": "issue:43257", "right": "issue:44863", "accept": false, - "reason": "Qwen3 MoE conversion/loading under accelerate+deepspeed is unrelated to NemotronH checkpoint loading." + "reason": "Qwen3 MoE deepspeed conversion and NemotronH checkpoint loading are unrelated." + }, + { + "left": "issue:43425", + "right": "issue:43828", + "accept": false, + "reason": "Torch compatibility complaints and an autocast dtype mismatch are separate failures." }, { "left": "issue:44589", "right": "issue:44756", "accept": false, - "reason": "Float8 storage lookup error vs mmap/OOM on Strix Halo; different bugs." + "reason": "Float8 storage lookup error and Strix Halo mmap OOM 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." + "reason": "Compiled flex_attention on torch>=2.9 and Gemma4 PLE device mismatch are different bugs." }, { "left": "issue:44479", "right": "issue:45663", "accept": false, - "reason": "Qwen video-input regression and Gemma4 FSDP2 KeyError are unrelated." + "reason": "Video-input regression in Qwen models is unrelated to Gemma4 FSDP2 attention-state rebuilding." }, { "left": "issue:43906", "right": "issue:44530", "accept": false, - "reason": "Isolated reproduction of another issue vs Qwen3.5 PagedAttentionCache crash; not the same underlying bug." + "reason": "An isolated reproduction of a different issue is not the same as the Qwen3.5 PagedAttentionCache crash." }, { "left": "issue:39404", "right": "issue:45701", "accept": false, - "reason": "Whisper pipeline return_language regression is unrelated to tokenization changes across versions." + "reason": "Whisper `return_language` pipeline behavior and tokenizer-version drift are unrelated." }, { - "left": "issue:43425", - "right": "issue:45823", + "left": "issue:43611", + "right": "issue:43645", "accept": false, - "reason": "Torch 2.10 compatibility and Gemma4 device-map mismatch are different regressions." + "reason": "Both concern Transformers 5.0 custom-model breakage, but the failure modes are different (`base_model_prefix` loading vs notebook redefinition/init issues)." }, { - "left": "issue:43611", - "right": "issue:43645", + "left": "issue:43425", + "right": "issue:45823", "accept": false, - "reason": "Both are custom-model/Transformers 5 breakages, but one is base_model_prefix loading and the other is notebook initialization; too different to merge." + "reason": "Torch 2.10 incompatibility is not the same as the Gemma4 PLE device mismatch." }, { "left": "issue:43931", "right": "issue:44863", "accept": false, - "reason": "Qwen3-VL shape mismatch and NemotronH checkpoint loading are unrelated." + "reason": "Qwen3-VL weight-shape mismatch and NemotronH checkpoint loading are different model-loading bugs." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays mostly conservative. The only clearly supported duplicate-like subgroup is the resize_token_embeddings trio (45292, 45276, 45335), and the rest of the pairwise rejections avoid over-merging unrelated model-loading or version-specific bugs." + "feedback": "The cluster summary is grounded and appropriately conservative. The only accepted duplicate-style pair, issue:45276 with issue:45292, is well supported by the titles and high similarity around the same `resize_token_embeddings`/embedding regression. The other verdicts stay cautious by rejecting pairs that only share broad themes (e.g. model loading, custom-model breakage, or checkpointing) without clear evidence of the same root cause. " }, "evaluator_used": true, "retried": false, @@ -2357,8 +2417,8 @@ "issue:44164|issue:44479", "issue:44488|issue:45081", "issue:44661|issue:44683", - "issue:43582|issue:43632", "issue:40990|issue:43421", + "issue:43582|issue:43632", "issue:43611|issue:43644", "issue:44206|issue:44220", "issue:43716|issue:45084", @@ -2368,20 +2428,20 @@ "issue:34567|issue:40444", "issue:44811|issue:45584" ], - "prepared_review_unit_hash": "4dae3f4e8a323b9bc862b9f358cdde77eb1032c2ba4d7ab897e335c8210c6370", + "prepared_review_unit_hash": "1163698b639cdca620e89879b4c252e7b39c452a96920a2ac760163011e477e7", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 32, - "serialized_chars": 37335, + "serialized_chars": 37334, "estimated_input_tokens": 9334, "estimated_eval_tokens": 18924 }, @@ -2390,192 +2450,204 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a loose set of transformers/v5 regressions and model/tokenizer loading issues rather than a single duplicate bug. The best representative issue is the broad model-loading regression around `base_model_prefix`.", - "confidence": 0.57, - "canonical_issue_reason": "`issue:43611` is the most general, central loading regression in the set and best represents the broader v5 model-loading breakages seen across nearby reports.", + "summary": "This cluster is heterogeneous: the soft pairs span unrelated regressions, model-loading issues, audio/Whisper bugs, quantization, docs, and training behavior. I would reject all suggested duplicate edges.", + "confidence": 0.95, + "canonical_issue_reason": "No single canonical issue stands out; the items are distinct bugs/regressions across different subsystems, so collapsing them would over-merge unrelated reports.", "canonical_pr_reason": null, - "best_issue_reason": "`issue:43611` is a concise, broadly applicable duplicate target for v5 loading failures; it is more representative than the highly model-specific reports.", + "best_issue_reason": "No good global representative emerges because the cluster is not a duplicate set but a mix of separate issues.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43425", "right": "issue:43716", "accept": false, - "reason": "Different problems: Torch version compatibility vs image preprocessor/model dtype mismatch." + "reason": "Unrelated problems: torch version incompatibility vs image preprocessor/model dtype mismatch." }, { "left": "issue:44530", "right": "issue:45084", "accept": false, - "reason": "Different code paths and failures; both are crashes, but not the same underlying bug." + "reason": "Different failures: Qwen3.5 PagedAttentionCache linear_attention crash vs template compilation error." }, { "left": "issue:43901", "right": "issue:44509", "accept": false, - "reason": "Both are docs-related, but they refer to different pipeline/task documentation issues." + "reason": "Both are docs issues, but they concern different pipeline/task documentation changes." }, { "left": "issue:39404", "right": "issue:44912", "accept": false, - "reason": "Whisper pipeline language output vs MXFP4 quantization loading/fallback are unrelated." + "reason": "Whisper pipeline return_language regression is unrelated to MXFP4 quantization fallback." }, { "left": "issue:44265", "right": "issue:44898", "accept": false, - "reason": "torch.export/torch_compilable_check is unrelated to Perceiver interpolation failure." + "reason": "Different code paths: torch.export with torch_compilable_check vs Perceiver non-default resolution interpolation." }, { "left": "issue:43299", "right": "issue:43931", "accept": false, - "reason": "Same model family, but one is a loading regression and the other is a weight-shape mismatch; not the same bug." + "reason": "Both involve Qwen3-VL loading, but one is a v5 regression for MoE models and the other is a weight-shape mismatch on a specific checkpoint." }, { "left": "issue:44479", "right": "issue:44805", "accept": false, - "reason": "Video-input regression for Qwen VL models is unrelated to the mask-shape IndexError." + "reason": "Video-input regression for Qwen VL models is not the same as a generic mask/index shape mismatch." }, { "left": "issue:43425", "right": "issue:44918", "accept": false, - "reason": "Torch compatibility issue vs TRL embedding unpacking failure are different problems." + "reason": "Torch incompatibility is unrelated to TRL SFT embedding unpacking for Qwen3.5." }, { "left": "issue:44589", "right": "issue:45084", "accept": false, - "reason": "Storage/type error vs template-node compile error are different failures." + "reason": "Float8 storage lookup failure is unrelated to a template-node compilation error." }, { "left": "issue:43856", "right": "issue:44756", "accept": false, - "reason": "Different symptoms and causes: Qwen3 MoE memory use vs Apple Silicon mmap OOM." + "reason": "Both mention memory, but one is Qwen3 MoE training efficiency and the other is Strix Halo mmap OOM." }, { "left": "issue:40990", "right": "issue:42947", "accept": false, - "reason": "Perplexity anomaly on a model is unrelated to gradient checkpointing with LoRA." + "reason": "High perplexity on a model/dataset is unrelated to LoRA gradient checkpointing ineffectiveness." }, { "left": "issue:39404", "right": "issue:44671", "accept": false, - "reason": "Whisper return_language regression and CamemBERT masked-LM regression are unrelated." + "reason": "Whisper pipeline behavior and CamemBERT masked LM predictions are separate model-family bugs." }, { "left": "issue:43994", "right": "issue:44079", "accept": false, - "reason": "Nonsensical outputs on SigLIP are not the same as ModelOutput key assignment logic." + "reason": "Incorrect SigLIP outputs are not the same as ModelOutput key assignment when values were previously None." }, { "left": "issue:44164", "right": "issue:45759", "accept": false, - "reason": "extra_state save/load handling is unrelated to multimodal config unwrapping in `from_config`." + "reason": "save/from_pretrained extra_state handling is unrelated to AutoModelForCausalLM.from_config text_config unwrapping." }, { "left": "issue:36032", "right": "issue:43452", "accept": false, - "reason": "Tokenizer special-token method conflict vs gguf loading breakage are different bugs." + "reason": "Tokenizer special-token method conflict is unrelated to gguf_file loading failures." }, { "left": "issue:43596", "right": "issue:44683", "accept": false, - "reason": "Deepspeed/BertModel init failure is unrelated to flex_attention compilation on newer torch." + "reason": "DeepSpeed ZeRO-3/BertModel initialization error is unrelated to compiled flex_attention on newer torch." }, { "left": "issue:44671", "right": "issue:44991", "accept": false, - "reason": "Both concern tokenizers/model behavior, but one is prediction quality and the other is tokenizer loading." + "reason": "CamemBERT MLM prediction issues are unrelated to tokenizer loading for EMBEDDIA/est-roberta." }, { "left": "issue:44991", "right": "issue:45081", "accept": false, - "reason": "Different tokenizer-loading regressions with different root causes and models." + "reason": "Tokenizer load failure and Mistral regex patch crash are different model-specific bugs." }, { "left": "issue:32090", "right": "issue:39404", "accept": false, - "reason": "Trainer GPU broadcast TypeError is unrelated to Whisper pipeline return_language behavior." + "reason": "Trainer GPU broadcast NoneType error is unrelated to Whisper return_language behavior." }, { "left": "issue:44164", "right": "issue:44479", "accept": false, - "reason": "Save/load extra_state handling is unrelated to Qwen video-input regression." + "reason": "Checkpoint save/load extra_state handling is unrelated to Qwen video-input regressions." }, { "left": "issue:44488", "right": "issue:45081", "accept": false, - "reason": "Different tokenizer loading failures; no shared concrete bug path." + "reason": "Loading cjvt/sleng-bert and patching Mistral regex are unrelated tokenizer issues." }, { "left": "issue:44661", "right": "issue:44683", "accept": false, - "reason": "Tokenizer mapping-name failure is unrelated to flex_attention compilation." + "reason": "Model registration in TOKENIZER_MAPPING_NAMES is unrelated to compiled flex_attention on torch >= 2.9." }, { - "left": "issue:43582", - "right": "issue:43632", + "left": "issue:40990", + "right": "issue:43421", "accept": false, - "reason": "Apple Silicon warmup TypeError is unrelated to `_is_hf_initialized` flag regression." + "reason": "Perplexity regression on gpt-oss-20b is unrelated to tokenizers backend post-processor updates." }, { - "left": "issue:40990", - "right": "issue:43421", + "left": "issue:43582", + "right": "issue:43632", "accept": false, - "reason": "Perplexity issue is unrelated to runtime post-processor updates in TokenizersBackend." + "reason": "Apple Silicon caching_allocator_warmup TypeError is unrelated to the _is_hf_initialized flag regression." }, { "left": "issue:43611", "right": "issue:43644", "accept": false, - "reason": "Both are v5 regressions, but one is about base_model_prefix loading and the other about non-persistent buffers; different code paths." + "reason": "Base_model_prefix loading bug and non-persistent buffer junk are different Transformers v5 regressions." }, { "left": "issue:44206", "right": "issue:44220", "accept": false, - "reason": "Related audio area, but the reported failures are not clearly the same underlying bug." + "reason": "Unsupported center argument in LasrFeatureExtractor is not the same as a generic _torch_extract_fbank_features issue." }, { "left": "issue:43716", "right": "issue:45084", "accept": false, - "reason": "Mistral image dtype mismatch is unrelated to template-node compilation errors." + "reason": "Mistral-3 dtype mismatch and non-template-node compilation failure are unrelated." }, { "left": "issue:43883", "right": "issue:44488", "accept": false, - "reason": "Molmo tied-weights attribute error and sleng-bert tokenizer loading are unrelated." + "reason": "Missing all_tied_weights_keys on Molmo is unrelated to sleng-bert loading failure." }, { - "left": "issue:44811", - "right": "issue:45584", + "left": "issue:44488", + "right": "issue:44987", + "accept": false, + "reason": "sleng-bert tokenizer loading and physical-intelligence/fast loading failures are different model-specific issues." + }, + { + "left": "issue:43643", + "right": "issue:44530", "accept": false, - "reason": "Both involve Whisper token handling, but one is batch_decode skip_special_tokens and the other is empty-transcription generation after align_special_tokens; not the same bug." + "reason": "trust_remote_code AutoConfig field loss is unrelated to Qwen3.5 PagedAttentionCache crashes." }, { "left": "issue:34567", "right": "issue:40444", "accept": false, - "reason": "TrainerState token counting and multimodal iterable-dataset failure are unrelated." + "reason": "TrainerState num_input_tokens_seen not updating is unrelated to iterable-dataset multi-image fine-tuning failures." + }, + { + "left": "issue:44811", + "right": "issue:45584", + "accept": false, + "reason": "Ignoring skip_special_tokens in batch_decode is a different Whisper bug from empty-transcription generation after align_special_tokens." } ] }, @@ -2650,8 +2722,8 @@ "soft_pairs": [ "issue:43644|issue:43646", "issue:44451|issue:44987", - "issue:43541|issue:45084", "issue:36032|issue:42915", + "issue:43541|issue:45084", "issue:43452|issue:43482", "issue:42915|issue:44530", "issue:43317|issue:43482", @@ -2678,27 +2750,27 @@ "issue:43632|issue:44756", "issue:43296|issue:43531", "issue:42915|issue:44291", - "issue:44062|issue:44987", "issue:36032|issue:43742", + "issue:44062|issue:44987", "issue:43531|issue:44877", "issue:42915|issue:44589", "issue:43611|issue:44534", "issue:43317|issue:43856" ], - "prepared_review_unit_hash": "3f2197462aeefbe4bdd7fdb05a88df13e77d5e86666e4a0d515d680c06d0a5fb", + "prepared_review_unit_hash": "2fbb473898070b9e4ad3cfd7aff1c52f1202cd81bb17c6611d38ae3c5c2f4830", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 36, - "serialized_chars": 37913, + "serialized_chars": 37916, "estimated_input_tokens": 9479, "estimated_eval_tokens": 19214 }, @@ -2707,238 +2779,235 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is mostly a grab-bag of unrelated Transformers v5 regressions across different models and features. Only a couple of near-exact duplicate pairs stand out; most soft matches should be rejected.", - "confidence": 0.93, - "canonical_issue_reason": "issue:43644 is the clearest canonical issue because it has an almost verbatim duplicate in 44534 describing the same non-persistent-buffer regression.", + "summary": "This is a loose thematic cluster of mostly unrelated Transformers v5 regressions around loading, initialization, tokenizers, and model-specific edge cases. The similarities are largely topical rather than the same underlying bug, so none of the soft pairs should be merged as duplicates.", + "confidence": 0.91, + "canonical_issue_reason": "issue:43646 is the broadest and most representative of the recurring v5 loading/initialization breakages; it best serves as an umbrella among the set, even though the cluster is heterogeneous.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43644 is the strongest issue anchor in the set: it has the tightest duplicate match and the cleanest, model-agnostic bug description.", + "best_issue_reason": "issue:43646 is the best single issue to anchor the cluster because it is general, central, and closest in scope to many of the other v5 initialization/loading reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43644", "right": "issue:43646", "accept": false, - "reason": "Both are v5 initialization regressions, but one is about non-persistent buffers and the other about custom model init; different bugs." + "reason": "Both are v5 regressions, but one is about non-persistent buffers being initialized with junk while the other is about custom model initialization; different bugs and code paths." }, { "left": "issue:44451", "right": "issue:44987", "accept": false, - "reason": "Both are load failures, but they involve different models and likely different failure modes." + "reason": "Both are model-loading failures, but they affect different models and likely different regressions; too broad to be the same underlying bug." }, { - "left": "issue:43541", - "right": "issue:45084", + "left": "issue:36032", + "right": "issue:42915", "accept": false, - "reason": "Different code paths: torch dynamo grouped_mm tracing versus template compilation errors." + "reason": "T5 tokenizer import/load error vs Qwen3MoE FineGrainedFP8Config failure; unrelated subsystems and failure modes." }, { - "left": "issue:36032", - "right": "issue:42915", + "left": "issue:43541", + "right": "issue:45084", "accept": false, - "reason": "Tokenizer add_special_tokens conflict is unrelated to Qwen3MoE FP8 loading." + "reason": "Mixtral torch dynamo tracing error and template compilation error are distinct runtime/compiler issues, not the same bug." }, { "left": "issue:43452", "right": "issue:43482", "accept": false, - "reason": "Both mention GGUF loading, but one is a generic from_pretrained breakage and the other is a specific Qwen2.5-GGUF failure." + "reason": "Both mention GGUF/loading, but one is about gguf_file API handling and the other is a Qwen2.5-GGUF v5 load failure; not the same concrete defect." }, { "left": "issue:42915", "right": "issue:44530", "accept": false, - "reason": "Same broad Qwen/MoE area, but FP8 config failure and PagedAttentionCache linear_attention crash are different bugs." + "reason": "Qwen3MoE FineGrainedFP8Config failure and PagedAttentionCache linear_attention crash are separate Qwen-related issues with different code paths." }, { "left": "issue:43317", "right": "issue:43482", "accept": false, - "reason": "Dequantized model offload failure is not the same as GGUF loading regression." + "reason": "device_map=auto offload/dequantized-model loading is unrelated to Qwen2.5-GGUF loading failure." }, { "left": "issue:44849", "right": "issue:45325", "accept": false, - "reason": "Different models and different symptoms: hidden-states output bug versus rope index scaling bug." + "reason": "Both are Qwen3.5/Qwen2.5-VL regressions, but one is about output_hidden_states and the other about rope/position_ids scaling; different behavior." }, { "left": "issue:43883", "right": "issue:44991", "accept": false, - "reason": "Molmo tied-weights attribute error is unrelated to tokenizer loading for est-roberta." + "reason": "Molmo tied-weights attribute error is unrelated to tokenizer loading for EMBEDDIA/est-roberta." }, { "left": "issue:43404", "right": "issue:44466", "accept": false, - "reason": "Both involve tied weights, but one is a model-specific tying bug and the other is device-dependent serialization behavior." + "reason": "Both involve tied lm_head weights, but one is a missing tie in Mistral3 and the other is device-dependent serialization inconsistency; related theme, different bugs." }, { "left": "issue:36032", "right": "issue:43645", "accept": false, - "reason": "Tokenizer method conflict is unrelated to custom model initialization in notebooks." + "reason": "Tokenizer special-token conflict vs notebook custom-model initialization breakage; no same failure path." }, { "left": "issue:45084", "right": "issue:45507", "accept": false, - "reason": "Different problems entirely: template compilation versus GraniteMoEHybrid method dispatch." + "reason": "Template compilation error and GraniteMoEHybrid invalid method call are unrelated code-path failures." }, { "left": "issue:43262", "right": "issue:43377", "accept": false, - "reason": "Both are audio-related, but sampling-rate defaults and padding-mask correctness are separate issues." + "reason": "Audio processor chat-template sampling-rate default bug vs MIMI encoder padding-mask bug; both audio-adjacent but distinct." }, { "left": "issue:43646", "right": "issue:44534", "accept": false, - "reason": "Both are v5 regressions, but they affect different initialization/serialization paths." + "reason": "General custom-model initialization breakage vs non-persistent buffer junk initialization; similar release theme, not the same defect." }, { "left": "issue:44987", "right": "issue:44991", "accept": false, - "reason": "Both are loading regressions, but on different models and likely different causes." + "reason": "Different load failures affecting different artifacts/models; no evidence they share the same root cause." }, { "left": "issue:43994", "right": "issue:44610", "accept": false, - "reason": "Different vision bugs: nonsensical outputs versus processor/model input-size mismatch." + "reason": "SigLIP2 nonsensical outputs and OmDet-Turbo processor size mismatch are different model/processor bugs." }, { "left": "issue:43322", "right": "issue:43540", "accept": false, - "reason": "Different multimodal failure modes on different model families." + "reason": "Llava Next segmentation fault and Qwen3OmniMoe video ValueError are separate multimodal model failures." }, { "left": "issue:44403", "right": "issue:45081", "accept": false, - "reason": "Unnecessary loading noise is unrelated to the Mistral regex patch crash." + "reason": "Generic transformer-loading noise and Mistral regex patch crash are not the same underlying issue." }, { "left": "issue:43377", "right": "issue:43819", "accept": false, - "reason": "Both are audio/codec issues, but missing padding-mask support is unrelated to DAC.from_latents STE behavior." + "reason": "Missing padding mask in MIMI encoder and DAC latent mismatch/STE issue are unrelated audio-model bugs." }, { "left": "issue:42915", "right": "issue:45084", "accept": false, - "reason": "Qwen3MoE FP8 loading and template compilation errors are unrelated." + "reason": "FineGrainedFP8Config failure and template compilation failure do not share a concrete code path." }, { "left": "issue:43645", "right": "issue:44534", "accept": false, - "reason": "Custom model init in notebooks is not the same as the non-persistent-buffer junk regression." + "reason": "Notebook custom model initialization vs non-persistent buffer junk are different regressions." }, { "left": "issue:44451", "right": "issue:45081", "accept": false, - "reason": "Model loading failure and Mistral tokenizer regex crash are different bugs." + "reason": "Loading a specific model checkpoint and Mistral tokenizer regex patch crash are unrelated." }, { "left": "issue:45216", "right": "issue:45357", - "accept": true, - "reason": "Same Qwen3.5 save_pretrained regression; 45357 is a narrower manifestation of the same incorrect checkpoint serialization." + "accept": false, + "reason": "Both are Qwen3.5 save_pretrained regressions, but one is incorrect checkpoint content generally and the other is visual encoder key serialization specifically; not the same bug." }, { "left": "issue:44265", "right": "issue:44610", "accept": false, - "reason": "Different export/compilation bug versus processor image-size mismatch." + "reason": "torch.export/exportability issue with torch_compilable_check is unrelated to an image processor size mismatch." }, { "left": "issue:44987", "right": "issue:45081", "accept": false, - "reason": "Different model-loading regression versus tokenizer regex patch crash." + "reason": "Different model-loading regressions with different mechanisms and affected components." }, { "left": "issue:44220", "right": "issue:44683", "accept": false, - "reason": "Audio fbank extraction and compiled flex_attention failure are unrelated." + "reason": "fbank feature extraction and compiled flex_attention failures are unrelated audio vs attention/compiler bugs." }, { "left": "issue:43404", "right": "issue:45127", "accept": false, - "reason": "Tied lm_head weights in Mistral3ForConditionalGeneration is not the same as LoRA merging with extended vocabulary." + "reason": "Tied lm_head weight bug in Mistral3 and LoRA merge collapse with extended vocab are related only at a high level; different concrete failure modes." }, { "left": "issue:43632", "right": "issue:44756", "accept": false, - "reason": "_is_hf_initialized flag regression and mmap OOM on Strix Halo are unrelated." + "reason": "_is_hf_initialized flag regression and mmap/OOM on Strix Halo are unrelated issues." }, { "left": "issue:43296", "right": "issue:43531", "accept": false, - "reason": "PaddleOCR-VL load failure and Qwen3-MoE sliding_window bug are different issues." + "reason": "PaddleOCR-VL load failure in vLLM and Qwen3-MoE sliding_window issue are separate model-specific problems." }, { "left": "issue:42915", "right": "issue:44291", "accept": false, - "reason": "Both touch v5 initialization/FP8-related areas, but the specific failures are different." + "reason": "FineGrainedFP8Config failure and unexpected _is_hf_initialized argument in init_empty_weights are different v5 regressions." }, { - "left": "issue:44062", - "right": "issue:44987", + "left": "issue:36032", + "right": "issue:43742", "accept": false, - "reason": "AddedToken keyword conflict is unrelated to loading physical-intelligence/fast." + "reason": "T5 tokenizer special-token conflict and MobileLLM loading KeyError are unrelated model-loading bugs." }, { - "left": "issue:36032", - "right": "issue:43742", + "left": "issue:44062", + "right": "issue:44987", "accept": false, - "reason": "T5 tokenizer add_special_tokens conflict is unrelated to MobileLLM loading key errors." + "reason": "AddedToken special-argument duplication and a model load failure for physical-intelligence/fast are distinct tokenizer vs model issues." }, { "left": "issue:43531", "right": "issue:44877", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and granite_speech strict config are different bugs." + "reason": "Qwen3-MoE sliding_window problem and granite_speech strict config loading issue are unrelated." }, { "left": "issue:42915", "right": "issue:44589", "accept": false, - "reason": "Both mention FP8-adjacent areas, but FineGrainedFP8Config failure and missing Float8 storage are separate problems." + "reason": "FineGrainedFP8Config failure and missing Float8 storage type are different quantization/runtime issues." }, { "left": "issue:43611", "right": "issue:44534", "accept": false, - "reason": "base_model_prefix loading regression is not the same as the non-persistent-buffer junk bug." + "reason": "base_model_prefix loading breakage and non-persistent buffer junk are separate v5 regressions with different symptoms." }, { "left": "issue:43317", "right": "issue:43856", "accept": false, - "reason": "GPU+CPU offload load failure and Qwen3 MoE memory inefficiency are different issues." + "reason": "device_map=auto dequantized offload failure is unrelated to Qwen3 MoE training memory usage." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The clustering summary is grounded in the packet, and the duplicate anchor choice is well supported by the titles/body overlap (especially 43644/44534, plus 45216/45357). The soft-edge verdicts are generally conservative and do not overclaim equivalence where the packet only shows broad topical similarity." - }, - "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 }, @@ -3036,214 +3105,214 @@ "issue:44079|issue:44534", "issue:43994|issue:44898" ], - "prepared_review_unit_hash": "efca1de0091ec85d37e6c48ceb328b6c87075867653d0062d3429fbc1508635c", + "prepared_review_unit_hash": "794874f4f9a11081157d6c6881a5a5bcda21c0feb8de5341cb393de4db9c94ba", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 37017, - "estimated_input_tokens": 9255, - "estimated_eval_tokens": 18766 + "serialized_chars": 37013, + "estimated_input_tokens": 9254, + "estimated_eval_tokens": 18764 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is highly heterogeneous: most items are distinct regressions or feature requests across tokenizers, vision models, training, and backend packaging. I found no soft pair that clearly meets the strict \u201csame underlying bug/change\u201d bar, so all soft edges are rejected. The broadest representative issue is the v5 model-loading regression around `base_model_prefix` / initialization.", - "confidence": 0.27, - "canonical_issue_reason": "Issue 43611 is the broadest and most central v5 model-loading regression in the set; it generalizes better than the more model-specific tokenizer, vision, or training bugs.", + "summary": "This cluster is mostly a loose set of unrelated transformers bugs, many around v5 regressions, tokenizer loading, vision-language models, and backend/integration issues. I found no true duplicate PRs, and all soft edges should be rejected.", + "confidence": 0.42, + "canonical_issue_reason": "Issue 45020 is the best umbrella candidate: it is broad, recent, and explicitly frames a class of regressions in recent transformers versions affecting remote_code models.", "canonical_pr_reason": null, - "best_issue_reason": "43611 is the best single issue representative because it captures a general Transformers v5 loading/init breakage that is closer to several other items than the narrower reports.", + "best_issue_reason": "Issue 45020 is the strongest global representative because it is the most general open issue in the set and could plausibly anchor related v5 regression reports, even though the cluster is not a tight duplicate group.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:41553", "right": "issue:43582", "accept": false, - "reason": "Unrelated: AutoTokenizer/Voxtral error messaging vs Apple Silicon warmup TypeError." + "reason": "Different subsystems and failures: tokenizer loading error message vs Apple Silicon caching allocator TypeError." }, { "left": "issue:44479", "right": "issue:45325", "accept": false, - "reason": "Both are Qwen VL regressions, but one is a broad video-input regression and the other is a specific rope-index scaling bug; not the same concrete issue." + "reason": "Both are Qwen VL/video-related, but one is a broad video-input regression and the other is a still-image temporal position_ids scaling bug." }, { "left": "issue:40990", "right": "issue:44811", "accept": false, - "reason": "Completely different models and failure modes: perplexity on gpt-oss vs Whisper batch_decode special-token handling." + "reason": "Perplexity on GPT-OSS/WikiText and Whisper batch_decode skipping special tokens are unrelated bugs." }, { "left": "issue:43425", "right": "issue:44403", "accept": false, - "reason": "Torch compatibility/package issue vs logging/noise on transformer loading." + "reason": "Torch 2.10 compatibility and noisy transformer loading messages are different issues." }, { "left": "issue:42175", "right": "issue:45542", "accept": false, - "reason": "Both concern TF packaging, but one is missing TensorFlow in torch extras and the other is a backend-selection error when only tensorboard is installed." + "reason": "Both mention TensorFlow installation, but one is missing TF from the torch extra and the other is a backend detection/runtime error with tensorboard-only installs." }, { "left": "issue:44462", "right": "issue:45812", "accept": false, - "reason": "Both are tokenizer-loading bugs, but the symptoms and affected code paths differ: tokenizer.json selection vs wrong token IDs for Granite." + "reason": "Tokenizer.json being ignored and Granite token IDs being wrong are different tokenizer-path failures." }, { "left": "issue:43818", "right": "issue:45357", "accept": false, - "reason": "Video-LLaVA architectural bug vs Qwen3.5 save_pretrained key regression; unrelated." + "reason": "Video-LLaVA architecture/weight-sharing bug vs Qwen3.5 save_pretrained visual-key regression." }, { "left": "issue:44079", "right": "issue:45663", "accept": false, - "reason": "ModelOutput key assignment bug vs Gemma-4 FSDP2 shared_kv_states KeyError; unrelated." + "reason": "ModelOutput key assignment and Gemma-4 shared_kv_states KeyError are unrelated code paths." }, { "left": "issue:36032", "right": "issue:41720", "accept": false, - "reason": "Tokenizer method conflict vs Qwen3 device-mapping cuda assert; unrelated." + "reason": "T5 tokenizer method conflict and Qwen3 auto device mapping cuda assert are unrelated." }, { "left": "issue:43421", "right": "issue:44568", "accept": false, - "reason": "Both involve special tokens, but one is runtime post-processor updates and the other is a tokenizer-specific add_special_tokens regression." + "reason": "Runtime post-processor updates for special tokens and mdeberta add_special_tokens behavior are different tokenization bugs." }, { "left": "issue:43792", "right": "issue:44589", "accept": false, - "reason": "Whisper load failure vs Float8 storage lookup error; unrelated." + "reason": "Whisper failing to run and missing Float8 storage support are unrelated." }, { "left": "issue:43632", "right": "issue:44534", "accept": false, - "reason": "Different v5 regressions: `_is_hf_initialized` flag handling vs non-persistent buffer initialization." + "reason": "_is_hf_initialized handling and non-persistent buffer junk are separate v5 regressions." }, { "left": "issue:44403", "right": "issue:44991", "accept": false, - "reason": "Noise/warning behavior vs a hard tokenizer-loading failure for a specific model." + "reason": "Generic loading noise and est-roberta tokenizer loading failure are not the same bug." }, { "left": "issue:44683", "right": "issue:45020", "accept": false, - "reason": "Torch flex_attention compile issue vs remote_code breakage; different subsystems." + "reason": "Compiled flex_attention on torch>=2.9 and remote_code model breakage are unrelated." }, { "left": "issue:44829", "right": "issue:45479", "accept": false, - "reason": "Both are training regressions, but one is flash_attention_3-related and the other is a loss/setup bug for num_labels=1." + "reason": "Both affect training, but flash_attention_3 degeneration and single-label-classification zero-loss are distinct root causes." }, { "left": "issue:43482", "right": "issue:45538", "accept": false, - "reason": "GGUF/v5 loading failure vs CLIPTokenizer max length constant; unrelated." + "reason": "Qwen2.5-GGUF loading failure and CLIPTokenizer model_max_length are unrelated tokenizer/model-loading issues." }, { "left": "issue:33666", "right": "issue:40444", "accept": false, - "reason": "Both Qwen VL training issues, but multi-GPU training and iterable-dataset multi-image failure are distinct problems." + "reason": "Both are Qwen2 VL training reports, but multi-GPU training and iterable-dataset multi-image failure are different problems." }, { "left": "issue:43482", "right": "issue:43525", "accept": false, - "reason": "Different model/config regressions: Qwen2.5-GGUF loading vs missing `pad_token_id` on Llama4Config." + "reason": "Qwen2.5-GGUF loading and Llama4Config pad_token_id missing are unrelated." }, { "left": "issue:44291", "right": "issue:44589", "accept": false, - "reason": "Both are TypeErrors, but the root causes are unrelated: `_is_hf_initialized` argument vs missing Float8 storage object." + "reason": "init_empty_weights argument handling and Float8 storage lookup are different failures." }, { "left": "issue:40444", "right": "issue:45325", "accept": false, - "reason": "Different Qwen2.5-VL bugs: iterable dataset/multi-image failure vs still-image rope-index scaling regression." + "reason": "IterableDataset multi-image prompting and still-image rope-index scaling are separate Qwen2.5-VL bugs." }, { "left": "issue:43377", "right": "issue:45491", "accept": false, - "reason": "Both involve batching/padding, but they affect different models and code paths." + "reason": "MIMI batch-vs-single padding-mask mismatch and Gemma3 NaN embeddings with sliding windows are unrelated." }, { "left": "issue:41720", "right": "issue:45081", "accept": false, - "reason": "Qwen device-mapping crash vs Mistral regex patch tokenizer AttributeError; unrelated." + "reason": "Qwen3 auto device map cuda assert and Mistral tokenizer regex patch crash are different code paths." }, { "left": "issue:42175", "right": "issue:43296", "accept": false, - "reason": "Packaging/backend install issue vs vLLM loading failure for PaddleOCR-VL; unrelated." + "reason": "Packaging/backend install issue vs PaddleOCR-VL load failure in vLLM are unrelated." }, { "left": "issue:43531", "right": "issue:44291", "accept": false, - "reason": "Sliding-window Qwen3-MoE bug vs init_empty_weights/_is_hf_initialized TypeError; unrelated." + "reason": "Qwen3-MoE sliding_window behavior and init_empty_weights argument errors are different bugs." }, { "left": "issue:43611", "right": "issue:43646", "accept": false, - "reason": "Both are v5 model-init/load regressions, but the reported failures are different enough that I cannot treat them as the same concrete bug." + "reason": "Both are v5 init regressions, but base_model_prefix loading and custom model initialization are not the same underlying defect." }, { "left": "issue:44265", "right": "issue:45072", "accept": false, - "reason": "torch.export/compilable-check failure vs bfloat16 dtype mismatch in inference; unrelated." + "reason": "torch.export/torch_compilable_check failure and bfloat16 dtype mismatch are unrelated." }, { "left": "issue:43632", "right": "issue:45072", "accept": false, - "reason": "Different failure modes and subsystems: initialization flag regression vs dtype mismatch." + "reason": "_is_hf_initialized breakage and bfloat16 dtype mismatches are different issues." }, { "left": "issue:43454", "right": "issue:44466", "accept": false, - "reason": "Both mention tied `lm_head` weights, but one is a weight-tying bug and the other is device-dependent serialization inconsistency." + "reason": "Tied lm_head weights generation bug and device-dependent serialization of lm_head.weight are related in theme but not clearly the same concrete bug." }, { "left": "issue:44079", "right": "issue:44534", "accept": false, - "reason": "ModelOutput key bookkeeping vs buffer initialization corruption; unrelated." + "reason": "ModelOutput key assignment and non-persistent buffer initialization are unrelated." }, { "left": "issue:43994", "right": "issue:44898", "accept": false, - "reason": "Different vision-model failures on different architectures; too broad to be duplicates." + "reason": "Siglip2 nonsensical outputs and Perceiver resize/interpolation failure are separate model-specific bugs." } ] }, @@ -3325,8 +3394,8 @@ "issue:45325|issue:45381", "issue:43644|issue:44079", "issue:44387|issue:44617", - "issue:43296|issue:45800", "issue:43881|issue:44488", + "issue:43296|issue:45800", "issue:44479|issue:45823", "issue:35707|issue:40444", "issue:44291|issue:44877", @@ -3338,11 +3407,11 @@ "issue:43262|issue:43746", "issue:41720|issue:45823", "issue:43454|issue:43550", - "issue:43377|issue:44610", "issue:43526|issue:43883", + "issue:43377|issue:44610", "issue:43454|issue:45072", - "issue:43377|issue:44898", "issue:40990|issue:45584", + "issue:43377|issue:44898", "issue:43572|issue:45759", "issue:39692|issue:40444", "issue:43881|issue:44991", @@ -3350,14 +3419,14 @@ "issue:43296|issue:44991", "issue:43792|issue:44683" ], - "prepared_review_unit_hash": "622eb455780e0d698d5c31d64bb91bd771cd532e85b59521ca0288a4474aef88", + "prepared_review_unit_hash": "ba687a57a9e9a9e67ff20604f5bbc0e4c2e53f9629f401056dea2839d0373b0a", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, @@ -3372,192 +3441,186 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: most pairs are only loosely related by model family or broad subsystem, not the same concrete bug. No pull requests are present, and there is no strong single canonical duplicate across the set.", - "confidence": 0.87, - "canonical_issue_reason": null, + "summary": "The items are mostly a mix of unrelated bug reports across different models and subsystems; there is no strong duplicate cluster, so I rejected all soft matches.", + "confidence": 0.96, + "canonical_issue_reason": "No single issue cleanly serves as a canonical duplicate: the set spans unrelated loading, quantization, multimodal, tokenizer, and serialization problems.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45325 is the clearest self-contained regression report in the set, but the overall cluster is too broad for a true canonical duplicate.", + "best_issue_reason": "No issue is a clear global representative of a duplicate group; the closest overlaps are still only broad subsystem similarity, not the same concrete bug.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45325", "right": "issue:45834", "accept": false, - "reason": "Both involve vision/text position handling, but one is a Qwen2.5-VL temporal position scaling bug and the other is a Kosmos2.5 OCR length/index error." + "reason": "Both involve multimodal position/index handling, but they target different models and different failure modes (still-image temporal scaling vs long OCR index error)." }, { "left": "issue:43792", "right": "issue:43957", "accept": false, - "reason": "These are different loading failures: Whisper runtime failure vs meta-device loading breakage in unrelated models." + "reason": "Both are loading/runtime failures, but the models and root causes differ; this is too broad to be the same bug." }, { "left": "issue:42915", "right": "issue:43482", "accept": false, - "reason": "Different problems and model families: Qwen3MoE FP8 config failure vs Qwen2.5 GGUF loading in v5." + "reason": "Different model families and different failure modes: FP8 config failure vs GGUF/v5 loading failure." }, { "left": "issue:44079", "right": "issue:45072", "accept": false, - "reason": "ModelOutput key assignment and bfloat16 dtype mismatch are unrelated bugs." + "reason": "Unrelated bugs: ModelOutput key assignment vs bfloat16 dtype mismatches." }, { "left": "issue:40444", "right": "issue:45381", "accept": false, - "reason": "Multiple-image finetuning failure is not the same bug as Qwen2.5-VL video position-id regression." + "reason": "Both are Qwen2.5-VL-related, but one is training with multiple images and the other is video vision position IDs; not the same bug." }, { "left": "issue:45325", "right": "issue:45381", "accept": false, - "reason": "Same model family, but one is still-image temporal scaling and the other is video vision_position_ids; not enough evidence of one root cause." + "reason": "Related to vision position ids, but one is still-image temporal scaling and the other is video input position IDs; not concrete enough to merge." }, { "left": "issue:43644", "right": "issue:44079", "accept": false, - "reason": "Non-persistent buffer junk and ModelOutput key bookkeeping are separate serialization/data-structure issues." + "reason": "Non-persistent buffer initialization and ModelOutput key assignment are unrelated issues." }, { "left": "issue:44387", "right": "issue:44617", "accept": false, - "reason": "Both end in OOM, but one is an int4 quantization memory regression and the other is a model-specific video OOM." + "reason": "Both mention OOM, but one is int4 quantization memory regression and the other is a model-specific video OOM." }, { - "left": "issue:43296", - "right": "issue:45800", + "left": "issue:43881", + "right": "issue:44488", "accept": false, - "reason": "PaddleOCR-VL/vLLM loading failure and torch-transformers incompatibility are different versioning problems." + "reason": "Different models and likely unrelated loading failures." }, { - "left": "issue:43881", - "right": "issue:44488", + "left": "issue:43296", + "right": "issue:45800", "accept": false, - "reason": "Different models and failure modes; both are loading issues, but not the same bug." + "reason": "A vLLM/model-loading issue and a torch/transformers version incompatibility are not the same bug." }, { "left": "issue:44479", "right": "issue:45823", "accept": false, - "reason": "A Qwen video-input regression is unrelated to Gemma4 PLE device-mismatch during forward." + "reason": "Both are multimodal regressions, but video-input regression and PLE device mismatch are different code-path problems." }, { "left": "issue:35707", "right": "issue:40444", "accept": false, - "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen2.5-VL multi-image dataset finetuning." + "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen2.5-VL finetuning with multiple images." }, { "left": "issue:44291", "right": "issue:44877", "accept": false, - "reason": "Unexpected _is_hf_initialized argument and strict config rejection are different from unrelated loading paths." + "reason": "init_empty_weights/_is_hf_initialized is a loading API break; granite_speech strict config is a separate config issue." }, { "left": "issue:43317", "right": "issue:43873", "accept": false, - "reason": "Both mention offloading/quantization, but one is device_map loading of a dequantized model and the other is a broader offloading complaint." + "reason": "Both involve offloading/quantization, but one is a gpu+cpu load failure and the other is a broader offloading complaint; too imprecise to treat as one bug." }, { "left": "issue:43606", "right": "issue:44265", "accept": false, - "reason": "CPU offload device mismatch is not the same as torch.export/export compatibility with torch_compilable_check." + "reason": "CPU offload device mismatch and torch.export/export compatibility are unrelated." }, { "left": "issue:43701", "right": "issue:45325", "accept": false, - "reason": "Resume-from-checkpoint key mismatch is unrelated to Qwen2.5-VL rope-index scaling." + "reason": "Checkpoint key mismatch and Qwen2.5-VL rope/index scaling are unrelated." }, { "left": "issue:44186", "right": "issue:44265", "accept": false, - "reason": "LayoutLMv2 tokenizer padding/NER crash is a different area from torch.export failures." + "reason": "Tokenizer padding/truncation crashes and torch.export with torch_compilable_check are different subsystems." }, { "left": "issue:43606", "right": "issue:43819", "accept": false, - "reason": "Both are model execution bugs, but device-mismatch offload and DAC latent/forward mismatch are unrelated." + "reason": "Both mention model computation correctness, but one is CPU offload device mismatch and the other is DAC.from_latents/STE mismatch." }, { "left": "issue:43262", "right": "issue:43746", "accept": false, - "reason": "Audio sampling-rate defaulting and PEFT local-checkpoint loading are separate issues." + "reason": "Audio chat-template sampling rate defaults and GraniteSpeech PEFT checkpoint loading are unrelated." }, { "left": "issue:41720", "right": "issue:45823", "accept": false, - "reason": "Both involve device placement, but Qwen3 auto-map cuda assert and Gemma4 PLE mismatch are different failure modes." + "reason": "Both are device_map=auto failures, but different models and different failure types (cuda assert vs device mismatch)." }, { "left": "issue:43454", "right": "issue:43550", "accept": false, - "reason": "Weight tying corruption and torch.compile SDPA failure are unrelated bugs." - }, - { - "left": "issue:43377", - "right": "issue:44610", - "accept": false, - "reason": "Missing padding-mask support in MIMI encoder is unrelated to OmDet-Turbo processor/model resolution mismatch." + "reason": "Weight tying in AyaVision and torch.compile/SDPA in Bamba are unrelated bugs." }, { "left": "issue:43526", "right": "issue:43883", "accept": false, - "reason": "BEiT label reduction bug and missing all_tied_weights_keys attribute are unrelated." + "reason": "BeitImageProcessorFast label reduction and Molmo tied-weights attribute errors are unrelated." }, { - "left": "issue:43454", - "right": "issue:45072", + "left": "issue:43377", + "right": "issue:44610", "accept": false, - "reason": "lm_head weight tying and bfloat16 dtype mismatch are different code paths." + "reason": "Padding-mask batch consistency and processor image-size mismatch are different issues." }, { - "left": "issue:43377", - "right": "issue:44898", + "left": "issue:43454", + "right": "issue:45072", "accept": false, - "reason": "Both are input-shape related vision bugs, but the models and failure mechanisms differ." + "reason": "AyaVision weight tying and dtype mismatches in SwitchTransformers/TimmWrapperModel are unrelated." }, { "left": "issue:40990", "right": "issue:45584", "accept": false, - "reason": "Perplexity regression on gpt-oss-20b is unrelated to Whisper empty-transcription generation failure." + "reason": "High perplexity on gpt-oss-20b and Whisper empty-transcription generation failure are unrelated." }, { "left": "issue:43572", "right": "issue:45759", "accept": false, - "reason": "Missing pad_token_idx and text_config unwrapping are separate config-loading regressions." + "reason": "Missing StableLmConfig pad_token_idx and composite Qwen text_config unwrapping are different config-loading problems." }, { "left": "issue:39692", "right": "issue:40444", "accept": false, - "reason": "SigLIP2 docs issues and Qwen2.5-VL multi-image finetuning failure are unrelated." + "reason": "SigLIP2 docs errors/quantization failure and Qwen2.5-VL finetuning failure are unrelated." }, { "left": "issue:43881", "right": "issue:44991", "accept": false, - "reason": "Different models and different loading layers: model load failure vs tokenizer load failure." + "reason": "Different model/tokenizer loading failures with no shared concrete code path." }, { "left": "issue:44466", "right": "issue:45357", "accept": false, - "reason": "Serialization of tied lm_head weights and incorrect visual-encoder keys are distinct save_pretrained regressions." + "reason": "Both involve serialization, but one is lm_head weight persistence and the other is visual encoder key naming; not the same bug." }, { "left": "issue:43296", @@ -3569,7 +3632,7 @@ "left": "issue:43792", "right": "issue:44683", "accept": false, - "reason": "Whisper runtime failure and compiled flex_attention failure on torch>=2.9 are different bugs." + "reason": "Whisper runtime failure and compiled flex_attention on torch>=2.9 are unrelated." } ] }, @@ -3667,8 +3730,8 @@ "issue:36331|issue:42617", "issue:44479|issue:44918", "issue:43526|issue:43881", - "issue:42915|issue:43317", "issue:43316|issue:44913", + "issue:42915|issue:43317", "issue:43550|issue:43606", "issue:45081|issue:45812", "issue:41720|issue:43322", @@ -3676,20 +3739,20 @@ "issue:43295|issue:45701", "issue:41720|issue:45161" ], - "prepared_review_unit_hash": "d06eb94ae11446bc916794d84b68a004b98cca45c4925593f11ec524da3f845d", + "prepared_review_unit_hash": "6882e637f516a7a21e8789a1c0e940258931d5fb1773d34e055cb80a0ff0597c", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 32, - "serialized_chars": 37385, + "serialized_chars": 37387, "estimated_input_tokens": 9347, "estimated_eval_tokens": 18950 }, @@ -3698,204 +3761,204 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The proposed soft links are mostly superficial topical matches. The issues span unrelated models, APIs, and failure modes, so none should be collapsed as duplicates.", - "confidence": 0.94, - "canonical_issue_reason": "No single canonical issue stands out; the cluster is too heterogeneous across tokenization, loading/offload, serving, trainer API, and model-specific regressions.", + "summary": "This cluster is heterogeneous rather than a true duplicate set: it mixes unrelated bugs in generation, tokenization, loading/offload, training, and multimodal processors. None of the soft pairs appear to be the same concrete underlying bug or change.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue stands out; the items span unrelated code paths and model families, so there is no clear duplicate anchor.", "canonical_pr_reason": null, - "best_issue_reason": "If forced to pick one representative, issue:45701 is the broadest/open topic, but it is still too generic to serve as a true canonical for the rest.", + "best_issue_reason": "No global best issue for this cluster: the issues are thematically similar only at a very broad level, not as one shared defect.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43454", "right": "issue:43720", "accept": false, - "reason": "Different bugs: weight tying in AyaVision vs packed-weight unpacking during accelerate loading." + "reason": "Both concern model-loading problems, but one is lm_head tying and the other is packed-weight unpacking in accelerate; different bugs and code paths." }, { "left": "issue:43873", "right": "issue:45701", "accept": false, - "reason": "One is quantized offloading behavior; the other is a tokenization regression. Not the same change or failure path." + "reason": "Quantization/offload behavior versus tokenizer/version-dependent tokenization changes; too broad and unrelated." }, { "left": "issue:41720", "right": "issue:44977", "accept": false, - "reason": "CUDA assert under auto device mapping is unrelated to Qwen3.5 flash-attention generation corruption." + "reason": "Different Qwen variants and different failures: device-map cuda assert vs flash-attention generation regression." }, { "left": "issue:44008", "right": "issue:45823", "accept": false, - "reason": "Gemma3n variable-name collision in forward is a different issue from Gemma4 PLE device mismatch." + "reason": "Gemma3n audio tensor name collision is unrelated to Gemma4 PLE device mismatch during forward." }, { "left": "issue:45632", "right": "issue:45684", "accept": false, - "reason": "Cache path collision and read-only permission propagation are distinct save/load problems." + "reason": "Cache-path collision for trust_remote_code is unrelated to save_pretrained permission propagation." }, { "left": "issue:43404", "right": "issue:45357", "accept": false, - "reason": "lm_head tying in Mistral3 is unrelated to incorrect visual encoder keys in Qwen3.5 save_pretrained." + "reason": "Weight tying in Mistral3 conditional generation is a different issue from Qwen3.5 visual encoder key saving." }, { "left": "issue:34634", "right": "issue:35707", "accept": false, - "reason": "BarkProcessor voice_preset failure is not the same as progressive generation with inputs_embeds and past_key_values." + "reason": "BarkProcessor voice_preset failure is unrelated to progressive generation with inputs_embeds and past_key_values." }, { "left": "issue:43577", "right": "issue:43708", "accept": false, - "reason": "Model dtype casting during load has nothing to do with Trainer max_steps recalculation on resume." + "reason": "Blip2 dtype loading bug versus Trainer resume/max_steps calculation bug; different subsystems." }, { "left": "issue:44186", "right": "issue:45072", "accept": false, - "reason": "Tokenizer crash on NER/padding is unrelated to dtype mismatches in inference wrappers." + "reason": "Tokenizer crash on NER/padding is unrelated to dtype mismatches in SwitchTransformers/TimmWrapperModel inference." }, { "left": "issue:44423", "right": "issue:45406", "accept": false, - "reason": "Both hit serve, but one is a multimodal batching '.to' crash and the other is a missing processor tokenizer attribute." + "reason": "Both are serve/multimodal crashes, but one is continuous-batching string handling and the other is a missing _tokenizer attribute in Gemma4Processor." }, { - "left": "issue:44479", - "right": "issue:44918", + "left": "issue:35707", + "right": "issue:41720", "accept": false, - "reason": "Video-input regression and TRL SFT embedding unpacking are different code paths and symptoms." + "reason": "Both involve generation, but the concrete failures are different: cache/input-embed handling vs device-map cuda assert." }, { - "left": "issue:43550", - "right": "issue:43720", + "left": "issue:43873", + "right": "issue:44756", "accept": false, - "reason": "torch.compile with SDPA on Bamba is unrelated to BitNet packed-weight loading during accelerate." + "reason": "Quantization offload issues and disabling mmap for Strix Halo OOM are not the same underlying problem." }, { - "left": "issue:36331", - "right": "issue:42915", + "left": "issue:44877", + "right": "issue:44991", "accept": false, - "reason": "Trainer compute_loss API break and FineGrainedFP8Config failure are separate problems." + "reason": "Strict config rejection for granite_speech is unrelated to tokenizer loading failure for EMBEDDIA/est-roberta." }, { - "left": "issue:44610", - "right": "issue:44898", + "left": "issue:43749", + "right": "issue:44534", "accept": false, - "reason": "Processor output shape mismatch is not the same as Perceiver non-default-resolution handling." + "reason": "FSDP CPU RAM efficient loading is a loading-path issue; non-persistent buffer junk is a different v5 buffer initialization bug." }, { - "left": "issue:44387", - "right": "issue:45704", + "left": "issue:42617", + "right": "issue:42915", "accept": false, - "reason": "CUDA memory growth under int4 quantization is distinct from the apex RMSNorm memory leak." + "reason": "3d_parallel.py runtime failure is unrelated to Qwen3Moe failing with FineGrainedFP8Config." }, { - "left": "issue:45161", - "right": "issue:41720", + "left": "issue:43450", + "right": "issue:45701", "accept": false, - "reason": "GPT-OSS MoE tensor-parallel failure is a different model-specific parallelism bug than Qwen3 auto device mapping asserts." + "reason": "Video processor batch-shape bug is unrelated to version-dependent tokenization changes." }, { - "left": "issue:44877", - "right": "issue:44991", + "left": "issue:43550", + "right": "issue:43720", "accept": false, - "reason": "Strict config rejection and tokenizer loading failure are separate regressions." + "reason": "Bamba torch.compile/SDPA failure and BitNet packed-weight loading failure are different model/backend bugs." }, { - "left": "issue:43295", - "right": "issue:45701", + "left": "issue:36331", + "right": "issue:42915", "accept": false, - "reason": "Processor.tokenizer regression is not the same as a general tokenization-version change report." + "reason": "Trainer compute_loss signature change is unrelated to Qwen3Moe FP8 configuration loading." }, { - "left": "issue:43526", - "right": "issue:43881", + "left": "issue:44610", + "right": "issue:44898", "accept": false, - "reason": "BeitImageProcessorFast label reduction bug and glm-4v loading failure do not share a concrete root cause." + "reason": "Both are image-resolution related, but they affect different models and different preprocessing/model expectations." }, { - "left": "issue:42915", - "right": "issue:43317", + "left": "issue:44186", + "right": "issue:44610", "accept": false, - "reason": "FP8 configuration failure and dequantized model offload loading failure are different loading paths." + "reason": "LayoutLMv2 tokenizer crashes and OmDet-Turbo image-size mismatch are different processor bugs." }, { - "left": "issue:43316", - "right": "issue:44913", + "left": "issue:44387", + "right": "issue:45704", "accept": false, - "reason": "Config API discrepancy is not the same as rotary_pct not persisting on reload." + "reason": "Int4 quantization memory regression in Transformers 5.x is unrelated to apex FusedRMSNorm memory leak in T5." }, { - "left": "issue:45081", - "right": "issue:45812", + "left": "issue:36331", + "right": "issue:42617", "accept": false, - "reason": "Mistral regex patch crash and Granite tokenizer wrong IDs are different tokenizer bugs." + "reason": "CustomTrainer compute_loss API break and 3d_parallel.py failure share no concrete code path." }, { - "left": "issue:41720", - "right": "issue:43322", + "left": "issue:44479", + "right": "issue:44918", "accept": false, - "reason": "Auto device mapping assert for Qwen3 is unrelated to Llava Next segmentation fault on load." + "reason": "Video-input regression in Qwen2.5/Qwen3.5 VL is different from TRL SFT issues unpacking Qwen3.5 input embeddings." }, { - "left": "issue:40990", - "right": "issue:45701", + "left": "issue:43526", + "right": "issue:43881", "accept": false, - "reason": "High perplexity on GPT-OSS and version-dependent tokenization changes are not the same defect." + "reason": "BeitImageProcessorFast reduce_labels bug is unrelated to glm-4v-9b loading failure." }, { - "left": "issue:43873", - "right": "issue:44756", + "left": "issue:43316", + "right": "issue:44913", "accept": false, - "reason": "Quantization/offloading behavior and mmap OOM on Strix Halo are separate performance/load issues." + "reason": "Both are config reload discrepancies, but for different configs and different fields; not the same bug." }, { - "left": "issue:35707", - "right": "issue:41720", + "left": "issue:42915", + "right": "issue:43317", "accept": false, - "reason": "Progressive generation with embeddings/past KV is unrelated to Qwen3 auto device-map CUDA asserts." + "reason": "Qwen3Moe FineGrainedFP8Config failure is unrelated to dequantized model loading with gpu+cpu offload." }, { - "left": "issue:44186", - "right": "issue:44610", + "left": "issue:43550", + "right": "issue:43606", "accept": false, - "reason": "Tokenizer crash in LayoutLMv2 and video processor shape mismatch are different subsystems and bugs." + "reason": "Different failures: torch.compile/SDPA in Bamba versus CPU offload device mismatch in Bark-small." }, { - "left": "issue:44534", - "right": "issue:44918", + "left": "issue:45081", + "right": "issue:45812", "accept": false, - "reason": "Non-persistent buffer junk in v5 is unrelated to Qwen3.5 embedding unpacking in TRL SFT." + "reason": "Mistral regex patch crash on tokenizer internals is unrelated to Granite token ID regression." }, { - "left": "issue:43550", - "right": "issue:43606", + "left": "issue:41720", + "right": "issue:43322", "accept": false, - "reason": "Bamba torch.compile/SDPA failure and Bark CPU-offload device mismatch are different model/runtime issues." + "reason": "Qwen3 auto device-map cuda assert and Llava Next segfault are different model-loading failures." }, { - "left": "issue:43450", + "left": "issue:40990", "right": "issue:45701", "accept": false, - "reason": "Batched video shape bug is unrelated to a tokenization regression." + "reason": "High perplexity on gpt-oss-20b is not the same as a tokenization change regression." }, { - "left": "issue:43317", - "right": "issue:42915", + "left": "issue:43295", + "right": "issue:45701", "accept": false, - "reason": "Dequantized model offload loading failure is not the same as FP8 config failure." + "reason": "Processor.tokenizer regression with images is a different issue from generic tokenization changes across versions." }, { - "left": "issue:44610", - "right": "issue:44898", + "left": "issue:41720", + "right": "issue:45161", "accept": false, - "reason": "These are both image-processing bugs, but one is a fixed-size output shape issue and the other is non-default-resolution handling; not the same underlying defect." + "reason": "Auto device-map cuda assert on Qwen3 is unrelated to TP-only failure on GPT-OSS MoE." } ] }, @@ -3999,214 +4062,214 @@ "issue:44987|issue:45042", "issue:43531|issue:44589" ], - "prepared_review_unit_hash": "315b848b1c60a7d55d13af181e0acb6a78488833462c2f4544df44eee4815338", + "prepared_review_unit_hash": "6fc1ad4f48ef898c0c5371e76ab68c8e5fe6cbb58f62401b988e720fdd2258d3", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36374, - "estimated_input_tokens": 9094, - "estimated_eval_tokens": 18444 + "serialized_chars": 36380, + "estimated_input_tokens": 9095, + "estimated_eval_tokens": 18446 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab-bag of unrelated bug reports across tokenizers, processors, loading, quantization, and model-specific regressions. Only one soft edge is a clear duplicate; the rest are only superficially similar.", - "confidence": 0.95, - "canonical_issue_reason": "Issue 45812 is the strongest representative issue: it is a broad, silent tokenizer regression affecting all Granite models, with inbound references and a clear user-facing correctness impact.", + "summary": "Only one clear duplicate pair stands out: the two Qwen2_5_VLProcessor.apply_chat_template crash reports. All other soft links are broad similarity matches but describe different bugs, models, or code paths and should remain separate.", + "confidence": 0.89, + "canonical_issue_reason": "Issue 44545 is the better representative of the exact duplicate pair with 44514: same crash, same title/behavior, and slightly more inbound references/discussion.", "canonical_pr_reason": null, - "best_issue_reason": "45812 is the most central and impactful issue in the set because it describes a general token-id regression rather than a model-specific edge case.", + "best_issue_reason": "Issue 44545 is the strongest issue candidate overall because it is the clearest exact duplicate target in this set and has the most signal among the pair.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43531", "right": "issue:43572", "accept": false, - "reason": "Different bugs: Qwen3-MoE sliding-window behavior vs StableLm config missing pad_token_idx." + "reason": "Both mention Qwen3-MoE / config-related problems, but one is a sliding_window bug and the other is a missing pad_token_idx regression; different failures." }, { "left": "issue:44671", "right": "issue:44743", "accept": false, - "reason": "Unrelated model-specific failures: CamemBERT MLM predictions vs Qwen3.5 recurrent state reset." + "reason": "CamemBERT masked-LM prediction errors and Qwen3 modular cache state resets are unrelated bugs in different models and code paths." }, { "left": "issue:43421", "right": "issue:45812", "accept": false, - "reason": "Both tokenizer-related, but one is runtime post-processor refresh and the other is Granite-wide wrong token IDs; different code paths." + "reason": "One is about runtime post-processor updates for special tokens; the other is a silent tokenizer ID regression across Granite models. Related area, different concrete bug." }, { "left": "issue:41720", "right": "issue:42617", "accept": false, - "reason": "Different subsystems: CUDA assert during auto device mapping vs inability to run a 3d_parallel script." + "reason": "Qwen3 auto device mapping on A800 causing cudaErrorAssert is unrelated to failing to run 3d_parallel.py." }, { "left": "issue:36010", "right": "issue:42617", "accept": false, - "reason": "ImportError for GenerationMixin is unrelated to the 3d_parallel execution issue." + "reason": "An import error for GenerationMixin is a different issue from a script execution failure." }, { "left": "issue:43540", "right": "issue:43782", "accept": false, - "reason": "Different multimodal/loading problems: Qwen3OmniMoe video input processing vs Qwen3VL weight_only loading." + "reason": "Qwen3OmniMoe video-input processing and Qwen3VL weight_only loading are distinct model/load-path bugs." }, { "left": "issue:44186", "right": "issue:44361", "accept": false, - "reason": "Different tokenizer/model bugs affecting different tasks and code paths." + "reason": "Both are tokenizer crashes, but they involve different tokenizers, tasks, and failure modes; not the same underlying bug." }, { "left": "issue:45406", "right": "issue:45464", "accept": false, - "reason": "Serve crash from missing _tokenizer is unrelated to Qwen3.5 streaming inference failure." + "reason": "Gemma4Processor serve crash and Qwen3.5 streaming chat/completions failure are different processor/runtime problems." }, { "left": "issue:43421", "right": "issue:45701", "accept": false, - "reason": "A broad tokenization regression is not the same as a runtime special-token post-processor update bug." + "reason": "Runtime special-token/post-processor updates and a general tokenization-version change are not the same concrete defect." }, { "left": "issue:30064", "right": "issue:34634", "accept": false, - "reason": "Image segmentation-map processing and Bark voice preset handling are unrelated." + "reason": "Void segmentation map processing and BarkProcessor voice_preset behavior are unrelated processor bugs." }, { "left": "issue:44492", "right": "issue:44509", "accept": false, - "reason": "Cache-strategy typo/docs cleanup is unrelated to removed pipeline-task docs." + "reason": "A cache-strategy typo and stale docs for removed pipeline tasks are documentation issues of different scope." }, { "left": "issue:44779", "right": "issue:45812", "accept": false, - "reason": "Both mention tokenization regressions, but they affect different models and likely different tokenizer logic." + "reason": "Both are tokenization regressions, but for different model families and symptoms; no evidence of one shared bug." }, { "left": "issue:42915", "right": "issue:43322", "accept": false, - "reason": "Quantized Qwen3Moe failure and Llava Next segmentation fault are unrelated." + "reason": "Qwen3Moe FP8 failure and Llava Next segmentation fault are unrelated model-specific crashes." }, { "left": "issue:43606", "right": "issue:43720", "accept": false, - "reason": "CPU offload device mismatch is a different bug from packed-weight unpacking during accelerate loading." + "reason": "CPU offload device mismatch for Bark-small and packed-weights unpacking for BitNet are different accelerate/loading bugs." }, { "left": "issue:43388", "right": "issue:45059", "accept": false, - "reason": "Metric label truncation and SAM3 text/bbox behavior are unrelated." + "reason": "Mask2Former label dropping in gather_for_metrics is unrelated to SAM3 text+box prompt weirdness." }, { "left": "issue:43550", "right": "issue:43653", "accept": false, - "reason": "torch.compile SDPA failure is unrelated to tokenizer special-token registration." + "reason": "torch.compile/SDPA failure in Bamba-9B-v2 and tokenizer special-token registration in BigBirdTokenizer are unrelated." }, { "left": "issue:44514", "right": "issue:44545", "accept": true, - "reason": "Same processor, same batched padding=False crash, and effectively the same report." + "reason": "These appear to be the same exact bug report: identical title and same apply_chat_template crash on batched input with padding=False." }, { "left": "issue:43531", "right": "issue:45759", "accept": false, - "reason": "Sliding-window/recurrent-state issue is unrelated to unwrapping text_config in composite multimodal configs." + "reason": "Sliding_window behavior in Qwen3-MoE and AutoModelForCausalLM.from_config text_config unwrapping are different issues." }, { "left": "issue:43606", "right": "issue:43653", "accept": false, - "reason": "Different issues in different components: offload mismatch vs tokenizer special-token registration." + "reason": "Device mismatch under CPU offload and a tokenizer mask-token decode issue are not the same underlying problem." }, { "left": "issue:43421", "right": "issue:43927", "accept": false, - "reason": "Both involve token IDs, but one is runtime special-token updates and the other is save/load losing custom IDs." + "reason": "Special-token post-processor refresh at runtime and DiaConfig losing custom token IDs on save/load are different tokenizer/config bugs." }, { "left": "issue:34567", "right": "issue:35707", "accept": false, - "reason": "TrainerState token counting is unrelated to progressive generation cache handling." + "reason": "TrainerState token counter not updating and progressive generation with inputs_embeds/past_key_values are unrelated training vs generation issues." }, { "left": "issue:44368", "right": "issue:45357", "accept": false, - "reason": "A config warning about tie_word_embeddings is unrelated to saving incorrect visual encoder keys." + "reason": "A warning about tie_word_embeddings and incorrect save_pretrained visual encoder keys are different Qwen3.5 fine-tuning/save issues." }, { "left": "issue:43873", "right": "issue:45704", "accept": false, - "reason": "Quantization offloading and an apex RMSNorm memory leak are unrelated." + "reason": "Quantization/offloading not working and a T5 apex memory leak are different mechanisms and symptoms." }, { "left": "issue:44451", "right": "issue:44991", "accept": false, - "reason": "Both are model load failures, but for different models and likely different tokenizer issues." + "reason": "Cannot load one specific model and tokenizer loading failures for another model are separate compatibility issues." }, { "left": "issue:44898", "right": "issue:45072", "accept": false, - "reason": "Different failure modes: image interpolation on Perceiver vs dtype mismatch in SwitchTransformers/TimmWrapperModel." + "reason": "Perceiver non-default resolution handling and bfloat16 dtype mismatches in SwitchTransformers/TimmWrapperModel are different bugs." }, { "left": "issue:41720", "right": "issue:43746", "accept": false, - "reason": "CUDA assert on device mapping is unrelated to PEFT/local checkpoint loading." + "reason": "Qwen3 auto device mapping CUDA assert and GraniteSpeech PEFT local checkpoint loading are unrelated failures." }, { "left": "issue:44062", "right": "issue:44451", "accept": false, - "reason": "AddedToken constructor error and ScandiBERT load failure are unrelated." + "reason": "AddedToken special-argument collision and inability to load a model are not the same code-path." }, { "left": "issue:44987", "right": "issue:45042", "accept": false, - "reason": "Different version regressions in different subsystems: model loading vs PIL backend image processors." + "reason": "A model loading failure for physical-intelligence/fast and a PIL backend torchvision dependency regression are different issues." }, { "left": "issue:43531", "right": "issue:44589", "accept": false, - "reason": "Qwen3-MoE sliding-window bug is unrelated to the Float8 storage lookup error." + "reason": "Qwen3-MoE sliding_window and a missing Float8 storage type are unrelated bugs." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative: it identifies issue 45812 using its broad Granite-wide tokenizer regression title and inbound references, and it correctly treats the cluster as largely heterogeneous. The only accepted soft edge, 44514/44545, looks like the same Qwen2_5_VLProcessor padding=False crash with near-identical titles, so that merge is reasonable. The remaining pair verdicts are cautious and not overstated." + "feedback": "Grounded overall. The exact-duplicate call on 44514/44545 is supported by the packet: identical title, same crash description, and 44545 has slightly more inbound references/discussion. The remaining soft-edge rejections are conservative and stay within what the titles/metadata support. No PR-pair mergeability concerns apply here." }, "evaluator_used": true, "retried": false, @@ -4277,11 +4340,11 @@ "issue:45759" ], "soft_pairs": [ - "issue:33666|issue:35707", "issue:45588|issue:45759", + "issue:33666|issue:35707", "issue:43421|issue:43653", - "issue:42886|issue:44843", "issue:43381|issue:43844", + "issue:42886|issue:44843", "issue:39692|issue:44610", "issue:44918|issue:44936", "issue:43421|issue:44186", @@ -4297,10 +4360,10 @@ "issue:43596|issue:44220", "issue:43540|issue:44008", "issue:43618|issue:43697", - "issue:45183|issue:45479", "issue:44220|issue:44661", - "issue:43575|issue:45704", + "issue:45183|issue:45479", "issue:44077|issue:44625", + "issue:43575|issue:45704", "issue:43720|issue:43819", "issue:30064|issue:36010", "issue:43335|issue:43756", @@ -4308,20 +4371,20 @@ "issue:43653|issue:43720", "issue:43716|issue:44008" ], - "prepared_review_unit_hash": "0306dd0b3c2756adc7512c7b48c21669fc03a5a1a314918d84d046f0ad2322c8", + "prepared_review_unit_hash": "405c6bc498835c3765518eddf60049deee7554a9a015e55fb853e6d04f903d4b", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36814, + "serialized_chars": 36813, "estimated_input_tokens": 9204, "estimated_eval_tokens": 18664 }, @@ -4330,192 +4393,192 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly heterogeneous: many items share broad tokenizer/model/config vocabulary, but the soft pairs usually point to different concrete bugs or different code paths. I would not merge any of the proposed soft edges.", - "confidence": 0.87, - "canonical_issue_reason": "issue:44843 is the clearest representative of the tokenizer-loading/regex/offline subtheme and is a concrete bug report, but the overall cluster is too mixed to support a strong canonical beyond that.", + "summary": "This set is a heterogeneous mix of unrelated bug reports across different models, tokenizers, processors, and training/runtime paths. None of the proposed soft pairs look like the same underlying issue, so no duplicate merge is warranted.", + "confidence": 0.96, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:44843 is the best single issue anchor here because it is specific, actionable, and closest to the recurring tokenizer infrastructure issues in the set, even though most other items are not true duplicates.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:33666", - "right": "issue:35707", + "left": "issue:45588", + "right": "issue:45759", "accept": false, - "reason": "Different problems: multi-GPU training for Qwen2-VL vs generation behavior with inputs_embeds/past_key_values." + "reason": "Different subsystems and failures: flash-attention sink handling vs Qwen multimodal config unwrapping." }, { - "left": "issue:45588", - "right": "issue:45759", + "left": "issue:33666", + "right": "issue:35707", "accept": false, - "reason": "Different code paths: flash-attention sink handling vs AutoModel config unwrapping for composite Qwen configs." + "reason": "Qwen2-VL multi-GPU training is unrelated to progressive generation with inputs_embeds/past_key_values." }, { "left": "issue:43421", "right": "issue:43653", "accept": false, - "reason": "Both involve tokenizers, but one is runtime special-token post-processing and the other is BigBird mask-token registration; not the same bug." + "reason": "Both involve tokenizers, but one is runtime post-processor updates and the other is BigBird mask-token registration/decode behavior." }, { - "left": "issue:42886", - "right": "issue:44843", + "left": "issue:43381", + "right": "issue:43844", "accept": false, - "reason": "Both mention offline/cache behavior, but one is tokenizer cache loading and the other is Mistral regex patching; distinct causes." + "reason": "Gradient checkpointing eval-mode behavior is unrelated to ZeRO-3 gradient growth during training." }, { - "left": "issue:43381", - "right": "issue:43844", + "left": "issue:42886", + "right": "issue:44843", "accept": false, - "reason": "Unrelated: eval-mode checkpointing restriction vs gradient blow-up under ZeRO-3 training." + "reason": "Both mention offline/tokenizer loading, but one is generic cache loading and the other is a specific _patch_mistral_regex model_info() call." }, { "left": "issue:39692", "right": "issue:44610", "accept": false, - "reason": "Both are model/processor mismatches in docs/examples, but they affect different models and different failure modes." + "reason": "Different model-specific processor mismatches: SigLIP2 docs vs OmDet-Turbo image size expectation." }, { "left": "issue:44918", "right": "issue:44936", "accept": false, - "reason": "Both involve trainers, but one is SFT unpacking of embeddings and the other is train/evaluate lifecycle failure." + "reason": "Unpacking Qwen3.5 input embeddings with TRL SFT is not the same as trainer.evaluate() failing after train()." }, { "left": "issue:43421", "right": "issue:44186", "accept": false, - "reason": "Tokenizer-related, but runtime post-processor updates and LayoutLMv2 NER/padding crashes are different defects." + "reason": "Different tokenizer bugs: dynamic special-token post-processing vs LayoutLMv2 batching/NER crash." }, { "left": "issue:43525", "right": "issue:43531", "accept": false, - "reason": "Unrelated config attribute issue vs sliding-window behavior in Qwen3-MoE." + "reason": "Missing pad_token_id in Llama4Config is unrelated to Qwen3-MoE sliding_window behavior." }, { "left": "issue:44625", "right": "issue:45479", "accept": false, - "reason": "Both touch classification config, but one is missing num_labels propagation and the other is a zero-loss problem with problem_type=num_labels=1." + "reason": "Config propagation of num_labels is a different bug from single-label classification degenerating to zero loss." }, { "left": "issue:43782", "right": "issue:45381", "accept": false, - "reason": "Different multimodal Qwen issues: weight_only loading error vs wrong video vision_position_ids." + "reason": "Different Qwen VL issues: weight_only loading failure vs video vision_position_ids being wrong." }, { "left": "issue:45081", "right": "issue:45356", "accept": false, - "reason": "Both mention Mistral regex/tokenizer handling, but one is a crash in _patch_mistral_regex and the other is a Kimi-K2.5 regression with codec handling/warnings." + "reason": "Both touch fix_mistral_regex, but one crashes on backend_tokenizer and the other is a codec-handling regression." }, { "left": "issue:34634", "right": "issue:41720", "accept": false, - "reason": "No overlap: BarkProcessor voice_preset vs Qwen3 auto device-mapping cuda assert." + "reason": "Bark voice_preset parsing is unrelated to Qwen3 auto device mapping cudaErrorAssert." }, { "left": "issue:43994", "right": "issue:44671", "accept": false, - "reason": "Both are incorrect model outputs, but on different models and unrelated failure mechanisms." + "reason": "Different models and outputs: SigLIP2 pipeline behavior vs CamemBERT masked LM predictions." }, { "left": "issue:43381", "right": "issue:45704", "accept": false, - "reason": "Different domains: gradient checkpointing eval-mode restriction vs T5/apex RMSNorm memory leak." + "reason": "Gradient checkpointing eval-mode support is unrelated to a T5 apex.RMSNorm memory leak." }, { "left": "issue:39692", "right": "issue:43873", "accept": false, - "reason": "Doc/example mismatch vs offloading under quantization; not the same underlying bug." + "reason": "SigLIP2 example issues are unrelated to quantization offloading behavior." }, { "left": "issue:42915", "right": "issue:45759", "accept": false, - "reason": "Different issues: FineGrainedFP8Config failure in Qwen3Moe vs text_config unwrapping in AutoModelForCausalLM.from_config." + "reason": "Qwen3Moe FP8 loading failure is unrelated to Qwen multimodal config unwrapping." }, { "left": "issue:43596", "right": "issue:44220", "accept": false, - "reason": "Unrelated: DeepSpeed ZeRO-3/BertModel index error vs audio fbank feature extraction issue." + "reason": "Deepspeed ZeRO-3/BertModel index error is unrelated to _torch_extract_fbank_features()." }, { "left": "issue:43540", "right": "issue:44008", "accept": false, - "reason": "Different models and different bugs: Qwen3OmniMoe video processing vs Gemma 3n audio_mel_mask variable collision." + "reason": "Video input validation in Qwen3OmniMoe is unrelated to Gemma3n's audio_mel_mask name collision." }, { "left": "issue:43618", "right": "issue:43697", "accept": false, - "reason": "Different regressions: CLIPOutput attentions assignment vs RTDetrV2 output changes in v5." - }, - { - "left": "issue:45183", - "right": "issue:45479", - "accept": false, - "reason": "Both concern model input/config validation, but one is input_ids/inputs_embeds XOR errors and the other is classification loss degeneracy." + "reason": "CLIP attentions assignment regression is unrelated to RTDetrV2 differing outputs in v5." }, { "left": "issue:44220", "right": "issue:44661", "accept": false, - "reason": "No real overlap: fbank feature extraction vs tokenizer-model mapping logic for add-new-model-like." + "reason": "Audio feature extraction bug is unrelated to tokenizer mapping-name handling in add-new-model-like." }, { - "left": "issue:43575", - "right": "issue:45704", + "left": "issue:45183", + "right": "issue:45479", "accept": false, - "reason": "Unrelated: tensor-parallel OOM on Qwen2-57B vs T5 memory leak from apex RMSNorm." + "reason": "Incorrect input validation messaging is unrelated to the num_labels=1 zero-loss classification bug." }, { "left": "issue:44077", "right": "issue:44625", "accept": false, - "reason": "PatchTSMixer post_init policy and Qwen3.5 num_labels propagation are unrelated config issues." + "reason": "Optional post_init policy for patchtsmixer is unrelated to config propagation of num_labels." + }, + { + "left": "issue:43575", + "right": "issue:45759", + "accept": false, + "reason": "Tensor-parallel OOM on a large Qwen2 model is unrelated to multimodal text_config unwrapping." }, { "left": "issue:43720", "right": "issue:43819", "accept": false, - "reason": "Different loading/runtime failures: BitNet packed-weight unpacking vs DAC.from_latents missing STE behavior." + "reason": "Packed-weight loading in BitNet is unrelated to DAC.from_latents missing STE behavior." }, { "left": "issue:30064", "right": "issue:36010", "accept": false, - "reason": "Completely unrelated: image processor segmentation-map handling vs GenerationMixin import error." + "reason": "Image processor void-segmentation handling is unrelated to the GenerationMixin import error." }, { "left": "issue:43335", "right": "issue:43756", "accept": false, - "reason": "Different model-specific config bugs: SwitchTransformers sparse-layer creation vs Smollm3 RoPE-layer dropping." + "reason": "SwitchTransformers sparse-layer construction is unrelated to Smollm3 RoPE layer dropping." }, { "left": "issue:44756", "right": "issue:45759", "accept": false, - "reason": "No overlap: mmap OOM on Strix Halo vs composite Qwen config unwrapping." + "reason": "Strix Halo mmap OOM mitigation is unrelated to Qwen multimodal config unwrapping." }, { "left": "issue:43653", "right": "issue:43720", "accept": false, - "reason": "Different tokenizer/CI bug vs BitNet accelerate-loading bug; shared tokenized wording is superficial." + "reason": "BigBirdTokenizer special-token decode failure is unrelated to BitNet packed-weight loading." }, { "left": "issue:43716", "right": "issue:44008", "accept": false, - "reason": "Different multimodal model bugs: dtype mismatch in Mistral-3 preprocessing vs Gemma 3n attribute collision." + "reason": "Mistral-3 image dtype mismatch is unrelated to Gemma3n audio tensor attribute collision." } ] }, @@ -4588,11 +4651,11 @@ "issue:45834" ], "soft_pairs": [ - "issue:42915|issue:43746", "issue:43295|issue:45834", + "issue:42915|issue:43746", "issue:43746|issue:45381", - "issue:43746|issue:43873", "issue:43550|issue:45072", + "issue:43746|issue:43873", "issue:43550|issue:43819", "issue:44186|issue:44898", "issue:44843|issue:45081", @@ -4606,33 +4669,33 @@ "issue:44448|issue:44671", "issue:43782|issue:44977", "issue:45584|issue:45834", - "issue:44561|issue:45020", "issue:43825|issue:43994", "issue:43322|issue:45685", + "issue:44561|issue:45020", "issue:44821|issue:45084", "issue:39692|issue:45834", "issue:30064|issue:36331", - "issue:44479|issue:45834", "issue:44485|issue:44509", + "issue:44479|issue:45834", "issue:43825|issue:43827", "issue:44186|issue:44442", - "issue:43606|issue:45072", - "issue:43577|issue:43582" + "issue:43577|issue:43582", + "issue:43606|issue:45072" ], - "prepared_review_unit_hash": "b323656a634dcb658bbfe39d2286cafd6024ea7322ee89a5aaecc241fd9e915a", + "prepared_review_unit_hash": "b2e34c56af20a190fb916a286cc7ac8a0f78e9c785ded4d241f90ccb884a4fd2", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 36227, + "serialized_chars": 36226, "estimated_input_tokens": 9057, "estimated_eval_tokens": 18370 }, @@ -4641,192 +4704,192 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a noisy mix of unrelated Transformers issues (tokenizer loading, processor regressions, video/vision bugs, quantization, remote_code, and docs). The soft links are generally thematic rather than true duplicates, so I would not merge any of the candidate pairs.", - "confidence": 0.83, - "canonical_issue_reason": "issue:44561 is the most concrete regression report among the set and best captures one of the few clearly related themes (recent-version/remote_code breakage).", + "summary": "This cluster is a loose semantic bucket of mostly unrelated Transformers issues. The soft pairs share broad keywords (Qwen, tokenizer, pipeline, video, config, etc.) but not the same concrete bug or change, so none should be merged as duplicates.", + "confidence": 0.9, + "canonical_issue_reason": "No canonical issue: the similarities are superficial and span distinct models, code paths, and failure modes rather than one underlying defect.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44561 is the best representative issue here because it states a specific root cause and failure mode, while most other candidates are broader, downstream, or unrelated.", + "best_issue_reason": "No single issue is a good global representative because the set is heterogeneous and does not form a true duplicate cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:42915", - "right": "issue:43746", - "accept": false, - "reason": "Different models and failure modes: FP8 config failure vs PEFT/local checkpoint loading." - }, { "left": "issue:43295", "right": "issue:45834", "accept": false, - "reason": "Both involve processor/tokenizer behavior, but one is a regression around processor.tokenizer/images and the other is an OCR long-input index error." + "reason": "Different regressions: custom processor.tokenizer access vs OCR long-input indexing." }, { - "left": "issue:43746", - "right": "issue:45381", + "left": "issue:42915", + "right": "issue:43746", "accept": false, - "reason": "Unrelated areas: PEFT/local checkpoint loading vs Qwen2.5-VL video position IDs." + "reason": "Unrelated bugs: FineGrainedFP8Config loading vs GraniteSpeech adapter/local-checkpoint loading." }, { "left": "issue:43746", - "right": "issue:43873", + "right": "issue:45381", "accept": false, - "reason": "No shared concrete bug; local checkpoint loading is not the same as quantized offloading behavior." + "reason": "Different code paths: PEFT/local checkpoint loading vs Qwen2.5-VL video position ids." }, { "left": "issue:43550", "right": "issue:45072", "accept": false, - "reason": "Both mention dtype/SDPA-style execution issues, but the affected models and code paths differ." + "reason": "Different failures: torch.compile SDPA on Bamba vs dtype mismatch in SwitchTransformers/TimmWrapperModel." + }, + { + "left": "issue:43746", + "right": "issue:43873", + "accept": false, + "reason": "Same broad area of loading/offloading, but distinct bugs and models; not the same defect." }, { "left": "issue:43550", "right": "issue:43819", "accept": false, - "reason": "Completely different problems: torch.compile+SDPA vs DAC.from_latents/STE mismatch." + "reason": "Unrelated: SDPA compile bug vs DAC latents/STE mismatch." }, { "left": "issue:44186", "right": "issue:44898", "accept": false, - "reason": "Different tokenizers and failures: LayoutLMv2 NER/padding crash vs Perceiver image classification with interpolation." + "reason": "Different subsystems: LayoutLMv2 tokenizer padding/NER crash vs Perceiver image resolution handling." }, { "left": "issue:44843", "right": "issue:45081", "accept": false, - "reason": "Both touch tokenizer internals, but one is offline regex patching and the other is a Mistral backend_tokenizer AttributeError." + "reason": "Both touch _patch_mistral_regex, but one is offline model_info access and the other is a backend_tokenizer AttributeError; separate bugs." }, { "left": "issue:43531", "right": "issue:43643", "accept": false, - "reason": "Sliding-window Qwen3-MoE issue is unrelated to trust_remote_code config field loss." + "reason": "Different issues: Qwen3-MoE sliding_window vs AutoConfig trust_remote_code field loss." }, { "left": "issue:43643", "right": "issue:44877", "accept": false, - "reason": "Both are config/loading regressions, but one is missing fields from trust_remote_code and the other is strict config rejecting granite_speech config." + "reason": "Both involve config loading, but one loses fields and the other is strict config rejection; not the same bug." }, { "left": "issue:43421", "right": "issue:44811", "accept": false, - "reason": "Runtime special-token post-processor updates are unrelated to Whisper batch_decode skip_special_tokens handling." + "reason": "TokenizersBackend post-processor updates vs Whisper batch_decode skip_special_tokens behavior." }, { "left": "issue:43577", "right": "issue:43818", "accept": false, - "reason": "Different BLIP2 dtype bug vs Video-LLaVA temporal attention/weight-sharing issue." + "reason": "Different model bugs: BLIP2 dtype persistence vs Video-LLaVA temporal attention/weight sharing." }, { "left": "issue:44062", "right": "issue:44488", "accept": false, - "reason": "Tokenizer AddedToken argument collision is not the same as failing to load a specific model tokenizer." + "reason": "AddedToken special-arg duplication vs tokenizer load failure for cjvt/sleng-bert." }, { "left": "issue:43540", "right": "issue:45381", "accept": false, - "reason": "Different multimodal paths: Qwen3OmniMoe video processing vs Qwen2.5-VL video position IDs." + "reason": "Different video regressions in different models and failure modes." }, { "left": "issue:43653", "right": "issue:43819", "accept": false, - "reason": "Tokenizer special-token registration issue is unrelated to DAC latent decoding mismatch." + "reason": "Tokenizer special-token registration bug vs DAC forward/from_latents mismatch." }, { "left": "issue:44448", "right": "issue:44671", "accept": false, - "reason": "Pegasus output differences across v4/v5 are not the same bug as CamemBERT masked-LM prediction regressions." + "reason": "Both are output quality regressions, but on different models/tasks and likely different causes." }, { "left": "issue:43782", "right": "issue:44977", "accept": false, - "reason": "Different model families and different failures: weight_only load error vs flash-attention generation issue." + "reason": "Qwen3VL weight_only load error vs Qwen3.5 flash-attention generation issue." }, { "left": "issue:45584", "right": "issue:45834", "accept": false, - "reason": "Whisper empty-transcription/align_special_tokens issue is not the same as Kosmos2.5 long OCR input indexing." - }, - { - "left": "issue:44561", - "right": "issue:45020", - "accept": false, - "reason": "45020 is a broad umbrella report; 44561 is a specific root-cause regression, so they are not clearly the same concrete bug." + "reason": "Whisper empty-transcription generation vs Kosmos2.5 long-OCR index error." }, { "left": "issue:43825", "right": "issue:43994", "accept": false, - "reason": "One is an error-message/docs mismatch around pipeline removals, the other is a model producing bad outputs in AutoModel/pipeline." + "reason": "Error-message wording about pipeline tasks vs SigLIP2 model/pipeline output correctness." }, { "left": "issue:43322", "right": "issue:45685", "accept": false, - "reason": "Segfault on Llava Next loading is unrelated to an MPS histogram Int error in moe code." + "reason": "Segfault loading Llava Next vs MPS histogram_int dtype error in MoE; unrelated." + }, + { + "left": "issue:44561", + "right": "issue:45020", + "accept": false, + "reason": "General remote_code breakage is too broad; these are not the same concrete regression." }, { "left": "issue:44821", "right": "issue:45084", "accept": false, - "reason": "AutoImageProcessor URL loading and template-node compilation errors are different subsystems and failure modes." + "reason": "AutoImageProcessor URL loading vs template-node compilation TypeError." }, { "left": "issue:39692", "right": "issue:45834", "accept": false, - "reason": "SigLIP2 doc-example/model mismatch is unrelated to Kosmos2.5 OCR long-input indexing." + "reason": "SigLIP2 docs/example mistakes vs Kosmos2.5 OCR input indexing." }, { "left": "issue:30064", "right": "issue:36331", "accept": false, - "reason": "Image processor segmentation-map handling is unrelated to CustomTrainer.compute_loss signature breakage." + "reason": "Image segmentation-map handling vs CustomTrainer loss signature regression." }, { - "left": "issue:44479", - "right": "issue:45834", + "left": "issue:44485", + "right": "issue:44509", "accept": false, - "reason": "Qwen video-input regression across models is not the same as Kosmos2.5 OCR indexing." + "reason": "GLM-5 RoPE discussion vs docs mentioning removed pipeline tasks." }, { - "left": "issue:44485", - "right": "issue:44509", + "left": "issue:44479", + "right": "issue:45834", "accept": false, - "reason": "GLM-5 RoPE discussion is unrelated to stale v5 pipeline docs." + "reason": "Qwen video-input regression vs Kosmos2.5 OCR indexing; different models and symptoms." }, { "left": "issue:43825", "right": "issue:43827", "accept": false, - "reason": "Both concern pipeline removals, but one is an incorrect message and the other is stale documentation; not the same bug." + "reason": "Both are about v5 pipeline removals, but one is an error-message issue and the other is stale docs; not the same underlying bug." }, { "left": "issue:44186", "right": "issue:44442", "accept": false, - "reason": "LayoutLMv2 tokenizer crashes and AutoTokenizer failing to load FastSpeech2ConformerTokenizer are different tokenizer-loading problems." + "reason": "LayoutLMv2 tokenizer crash vs AutoTokenizer loading FastSpeech2ConformerTokenizer." }, { - "left": "issue:43606", - "right": "issue:45072", + "left": "issue:43577", + "right": "issue:43582", "accept": false, - "reason": "CPU offload device mismatch and bfloat16 dtype mismatch are different execution bugs." + "reason": "BLIP2 dtype retention vs Apple Silicon caching_allocator_warmup TypeError." }, { - "left": "issue:43577", - "right": "issue:43582", + "left": "issue:43606", + "right": "issue:45072", "accept": false, - "reason": "BLIP2 dtype propagation and Apple Silicon warmup TypeError are unrelated." + "reason": "CPU offload device mismatch vs bfloat16 dtype mismatch in a different model." } ] }, @@ -4903,14 +4966,14 @@ "issue:34634|issue:36010", "issue:44805|issue:44936", "issue:43452|issue:44991", - "issue:42175|issue:45081", "issue:44610|issue:45072", + "issue:42175|issue:45081", "issue:43720|issue:45072", "issue:44060|issue:44368", "issue:44671|issue:45701", "issue:45081|issue:45406", - "issue:43650|issue:43906", "issue:43404|issue:43782", + "issue:43650|issue:43906", "issue:43582|issue:44857", "issue:45459|issue:45803", "issue:43262|issue:43819", @@ -4925,199 +4988,199 @@ "issue:44589|issue:45042", "issue:43854|issue:44977", "issue:38175|issue:44610", - "issue:43526|issue:44554", - "issue:39692|issue:44912" + "issue:39692|issue:44912", + "issue:43526|issue:44554" ], - "prepared_review_unit_hash": "fdfd1b80347cf3d305808e73117e56c9c8ce7f8572f5bacfeee3f3fc70394a13", + "prepared_review_unit_hash": "4222a258e0ca48c4fce29a00cf860f2e12a19fce9c539c7d0a9e10efd42298d0", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 27, - "serialized_chars": 36045, - "estimated_input_tokens": 9012, - "estimated_eval_tokens": 18280 + "serialized_chars": 36042, + "estimated_input_tokens": 9011, + "estimated_eval_tokens": 18278 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: the items are mostly unrelated Transformers issues spanning tokenizers, processors, loading, quantization, and model-specific bugs. None of the soft-edge pairs look like true duplicates or mergeable same-fix PRs.", - "confidence": 0.62, - "canonical_issue_reason": "Issue 45020 is the broadest umbrella item here because it describes a version-wide regression affecting remote-code model loading, but the cluster is too mixed to have a true duplicate canonical.", + "summary": "Mostly unrelated issue set; the soft-similarity links look like false positives across different models, processors, loaders, and runtime failures rather than true duplicates.", + "confidence": 0.81, + "canonical_issue_reason": "No single canonical issue stands out: the items cover many distinct bugs in different subsystems and model families.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45020 is the best representative anchor for the set: it is open, cross-cutting, and closer to a general Transformers regression than the many narrow model-specific reports.", + "best_issue_reason": "If forced to pick a representative, issue:45020 is the broadest/open report, but it is still too generic to serve as a real duplicate anchor.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:34634", "right": "issue:36010", "accept": false, - "reason": "Different subsystems and failure modes: BarkProcessor voice preset behavior vs a transformers import error." + "reason": "Different failures: Bark voice preset behavior vs a GenerationMixin import error." }, { "left": "issue:44805", "right": "issue:44936", "accept": false, - "reason": "Unrelated bugs: mask/index shape mismatch in one case, trainer.evaluate() state/failure after training in the other." + "reason": "Different bugs: a mask/index shape mismatch vs trainer state breaking evaluate() after train()." }, { "left": "issue:43452", "right": "issue:44991", "accept": false, - "reason": "Both are loading-related, but one is gguf_file handling and the other is a tokenizer/version regression for a specific model." + "reason": "Both are loading issues, but they affect different tokenizer/model paths and do not appear to be the same defect." }, { - "left": "issue:42175", - "right": "issue:45081", + "left": "issue:44610", + "right": "issue:45072", "accept": false, - "reason": "Package extra/backend dependency issue vs a Mistral tokenizer regex crash; not the same bug." + "reason": "Different models and symptoms: OmDet-Turbo processor size mismatch vs dtype mismatch in SwitchTransformers/TimmWrapperModel." }, { - "left": "issue:44610", - "right": "issue:45072", + "left": "issue:42175", + "right": "issue:45081", "accept": false, - "reason": "Processor input-size mismatch and bfloat16 dtype mismatch are different code paths and symptoms." + "reason": "Unrelated: backend packaging problem vs a Mistral tokenizer regex crash." }, { "left": "issue:43720", "right": "issue:45072", "accept": false, - "reason": "Packed-weight loading failure is unrelated to the dtype mismatch reported in the other issue." + "reason": "Different problems: BitNet packed-weight loading vs bfloat16 dtype mismatches." }, { "left": "issue:44060", "right": "issue:44368", "accept": false, - "reason": "Similar warning theme, but different models and likely different tying logic; not one concrete duplicate bug." + "reason": "Related theme of tied-weight warnings, but the model families and underlying warning bug differ." }, { "left": "issue:44671", "right": "issue:45701", "accept": false, - "reason": "Masked LM prediction correctness vs generic tokenization changes; too broad and not the same underlying defect." + "reason": "CamemBERT masked-LM predictions vs a broad tokenization-regression report; not the same concrete bug." }, { "left": "issue:45081", "right": "issue:45406", "accept": false, - "reason": "One is a Mistral tokenizer patch crash, the other is Gemma4Processor missing _tokenizer in serve." + "reason": "Different processor/tokenizer crashes in different code paths and models." }, { - "left": "issue:43650", - "right": "issue:43906", + "left": "issue:43404", + "right": "issue:43782", "accept": false, - "reason": "Placeholder/non-informative issue vs a reproduction of an external issue; no shared bug." + "reason": "Different generation/loading issues: untied lm_head weights vs weight_only load error." }, { - "left": "issue:43404", - "right": "issue:43782", + "left": "issue:43650", + "right": "issue:43906", "accept": false, - "reason": "Different model families and failures: tied lm_head weights vs weight_only loading error." + "reason": "No clear overlap; one is a vague data request, the other is an isolated reproduction of a separate upstream issue." }, { "left": "issue:43582", "right": "issue:44857", "accept": false, - "reason": "Apple Silicon allocator warmup TypeError is unrelated to CUDA/AMP loss crashing." + "reason": "Different runtime environments and failures: Apple Silicon allocator warmup TypeError vs CUDA float16 AMP crash." }, { "left": "issue:45459", "right": "issue:45803", "accept": false, - "reason": "Both involve broad excepts, but in different functions and for different error classes; not the same bug." + "reason": "Both involve exception handling, but they are different functions and different error-hiding bugs." }, { "left": "issue:43262", "right": "issue:43819", "accept": false, - "reason": "Audio sampling-rate defaulting and DAC STE/forward mismatch are unrelated." + "reason": "Audio sampling-rate defaulting bug vs DAC STE mismatch; unrelated code paths." }, { "left": "issue:43526", "right": "issue:43697", "accept": false, - "reason": "Label reduction bug in BEiT fast processor vs output divergence in RTDetrV2; no shared code path." + "reason": "Different model issues: BEiT Fast label reduction vs RTDetrV2 output divergence." }, { "left": "issue:45084", "right": "issue:45538", "accept": false, - "reason": "Template compilation error and CLIPTokenizer max_length constant are different tokenizer issues." + "reason": "Template compilation error vs tokenizer max_length constant; unrelated defects." }, { "left": "issue:43644", "right": "issue:43749", "accept": false, - "reason": "Non-persistent buffer initialization junk vs FSDP CPU RAM efficient loading breakage; different problems." + "reason": "Different serialization/loading bugs: non-persistent buffer corruption vs FSDP CPU RAM efficient loading failure." }, { "left": "issue:43329", "right": "issue:45600", "accept": false, - "reason": "Multimodal token counting bug vs stale auto_mappings references; unrelated fixes." + "reason": "Different scopes: multimodal token counting video-path bug vs stale auto_mappings CI breakage." }, { "left": "issue:43653", "right": "issue:44912", "accept": false, - "reason": "Tokenizer special-token registration bug vs MXFP4 quantization loading/fallback issue." + "reason": "Tokenizer special-token registration bug vs quantization fallback to bf16; not the same issue." }, { "left": "issue:43742", "right": "issue:45020", "accept": false, - "reason": "Specific MobileLLM key error is not the same as the broader remote_code regression." + "reason": "MobileLLM loading KeyError is model-specific; the other is a broad remote_code regression report." }, { "left": "issue:43638", "right": "issue:44936", "accept": false, - "reason": "Zero3/Bert index error and trainer.evaluate() failure after train are different training/runtime issues." + "reason": "Different causes and workflows: Deepspeed ZeRO3 index error vs trainer evaluate() state regression." }, { "left": "issue:43746", "right": "issue:44811", "accept": false, - "reason": "PEFT checkpoint loading failure vs Whisper batch_decode skip_special_tokens behavior; unrelated." + "reason": "Different processor loading bug vs Whisper batch_decode skip_special_tokens handling." }, { "left": "issue:44589", "right": "issue:45042", "accept": false, - "reason": "Storage-object lookup failure is unrelated to the PIL backend torchvision dependency regression." + "reason": "Unrelated: missing Float8 storage object vs PIL image processor unexpectedly requiring torchvision." }, { "left": "issue:43854", "right": "issue:44977", "accept": false, - "reason": "Different surfaces: model load/test failure vs flash-attention generation malfunction." + "reason": "Different model-loading/test failure vs flash-attention generation bug." }, { "left": "issue:38175", "right": "issue:44610", "accept": false, - "reason": "Both are vision-model related, but zero probabilities and wrong processor image size are different bugs." + "reason": "Different symptom and subsystem: zero probabilities in SigLIP2 vs processor/model input-size mismatch." }, { - "left": "issue:43526", - "right": "issue:44554", + "left": "issue:39692", + "right": "issue:44912", "accept": false, - "reason": "Processor label reduction vs MPS attention correctness issue; no overlap beyond being model bugs." + "reason": "SigLIP2 docs/example errors are not the same as MXFP4 quantization fallback behavior." }, { - "left": "issue:39692", - "right": "issue:44912", + "left": "issue:43526", + "right": "issue:44554", "accept": false, - "reason": "Docs-example model/processor mismatch plus quantization failure is not the same as MXFP4 fallback during loading." + "reason": "Different models and failure modes; BEiT Fast label reduction is unrelated to the MPS attention correctness bug." } ] }, @@ -5192,12 +5255,12 @@ ], "soft_pairs": [ "issue:36010|issue:43645", - "issue:34634|issue:36331", "issue:43454|issue:43577", - "issue:43257|issue:43329", + "issue:34634|issue:36331", "issue:43819|issue:43927", - "issue:43720|issue:44610", + "issue:43257|issue:43329", "issue:43550|issue:44912", + "issue:43720|issue:44610", "issue:43749|issue:44291", "issue:44568|issue:45488", "issue:43531|issue:45042", @@ -5212,8 +5275,8 @@ "issue:43257|issue:43931", "issue:43697|issue:45357", "issue:44291|issue:45042", - "issue:41720|issue:43540", "issue:43452|issue:43582", + "issue:41720|issue:43540", "issue:42915|issue:43742", "issue:43317|issue:45538", "issue:33357|issue:34567", @@ -5222,239 +5285,254 @@ "issue:42915|issue:44821", "issue:44479|issue:45381", "issue:42915|issue:45538", - "issue:43540|issue:44560", "issue:43257|issue:43756", + "issue:43540|issue:44560", "issue:43540|issue:44479" ], - "prepared_review_unit_hash": "351a7f7f735394dca89791d8a98b102dec1e06190dec79778f0ec596bb26e701", + "prepared_review_unit_hash": "47732b72f8fa57bead48e94c59be81e7cbb474e04a5258a0135f932cb4d88b43", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 38061, - "estimated_input_tokens": 9516, - "estimated_eval_tokens": 19288 + "serialized_chars": 38058, + "estimated_input_tokens": 9515, + "estimated_eval_tokens": 19286 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly unrelated issues spanning loaders, tokenizers, multimodal/video, and training. The closest pair is 43329/44479, but the packet only supports topical Qwen video-regression similarity, not a confident same-root-cause duplicate. No PRs are present.", - "confidence": 0.68, - "canonical_issue_reason": "No single canonical issue cleanly covers the set. issue:43329 is the most specific Qwen video-token bug, but issue:44479 is only a broader video regression report and the packet does not establish a shared root cause.", + "summary": "This is a heterogeneous cluster of mostly unrelated Transformers v5 regressions/load/load-time bugs. The soft links are mainly superficial topic overlap (tokenization, video, loading, quantization) rather than the same underlying bug, so none should be merged as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": "No true canonical duplicate stands out; if forced to anchor the cluster, issue 45701 is the broadest umbrella for the recurring version-dependent tokenization regression theme, but it does not cover the full set.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43329 is the best representative of the closest pair because it pinpoints the concrete undefined-variable failure in the video branch; issue:44479 is broader and less actionable.", + "best_issue_reason": "Issue 45701 is the best available anchor only because it is broad and open with several inbound references; it is still too generic to represent the whole cluster cleanly.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:36010", "right": "issue:43645", "accept": false, - "reason": "Different subsystems and failure modes: GenerationMixin import regression vs custom model init in Jupyter." - }, - { - "left": "issue:34634", - "right": "issue:36331", - "accept": false, - "reason": "Unrelated bugs: BarkProcessor voice preset handling vs Trainer compute_loss argument mismatch." + "reason": "Different breakages: GenerationMixin import path vs notebook custom-model initialization under v5." }, { "left": "issue:43454", "right": "issue:43577", "accept": false, - "reason": "Different model families and issues: weight tying in AyaVision vs dtype propagation in Blip2." + "reason": "Unrelated model-specific bugs: AyaVision weight tying vs Blip2 dtype staying float32." }, { - "left": "issue:43257", - "right": "issue:43329", + "left": "issue:34634", + "right": "issue:36331", "accept": false, - "reason": "Both involve Qwen3-related code, but one is MOE weight conversion/loading and the other is a video token-counting bug; not the same change." + "reason": "Different APIs and symptoms: BarkProcessor voice_preset vs Trainer.compute_loss kwargs." }, { "left": "issue:43819", "right": "issue:43927", "accept": false, - "reason": "Different bugs: DAC latent decoding mismatch vs DiaConfig losing custom token IDs on save/load." + "reason": "Different code paths: DAC latent conversion mismatch vs DiaConfig losing custom token IDs on save/load." }, { - "left": "issue:43720", - "right": "issue:44610", + "left": "issue:43257", + "right": "issue:43329", "accept": false, - "reason": "Different code paths: BitNet packed-weight loading vs OmDet-Turbo image size mismatch." + "reason": "Qwen3 MoE weight conversion issue is distinct from multimodal video token counting undefined-variable bug." }, { "left": "issue:43550", "right": "issue:44912", "accept": false, - "reason": "Different quantization/compile issues on different models; no shared concrete bug." + "reason": "Different failures: torch.compile/SDPA on Bamba vs MXFP4 quantization fallback on gpt-oss-20b." + }, + { + "left": "issue:43720", + "right": "issue:44610", + "accept": false, + "reason": "Packed-weight loading bug in BitNet is unrelated to OmDet-Turbo processor image-size mismatch." }, { "left": "issue:43749", "right": "issue:44291", "accept": false, - "reason": "Both concern loading, but one is FSDP CPU RAM-efficient loading and the other is an init_empty_weights argument error; too different to merge." + "reason": "Both are loading-related, but one is FSDP CPU RAM efficient loading and the other is init_empty_weights/_is_hf_initialized." }, { "left": "issue:44568", "right": "issue:45488", "accept": false, - "reason": "Both are tokenizer regressions, but one is special-token insertion and the other is pre-tokenizer override for a different tokenizer family." + "reason": "Both are tokenizer regressions, but they affect different tokenizers and different implementation paths." }, { "left": "issue:43531", "right": "issue:45042", "accept": false, - "reason": "Unrelated: Qwen3-MoE sliding window behavior vs PIL image processor torchvision dependency." + "reason": "Sliding-window handling in Qwen3-MoE is unrelated to PIL backend processor torchvision dependency changes." }, { "left": "issue:43262", "right": "issue:44811", "accept": false, - "reason": "Different audio processor bugs: chat template sampling-rate default vs batch_decode special-token skipping." + "reason": "Audio chat-template sampling-rate default bug is unrelated to Whisper batch_decode skip_special_tokens handling." }, { "left": "issue:43643", "right": "issue:45759", "accept": false, - "reason": "Both touch config/model loading, but one is trust_remote_code missing fields and the other is unwrapping text_config for composite multimodal configs." + "reason": "Different config bugs: trust_remote_code field loss vs from_config not unwrapping text_config." }, { "left": "issue:44568", "right": "issue:45356", "accept": false, - "reason": "Both are tokenizer-related, but the failures are different model-specific regressions with different symptoms and likely fixes." + "reason": "Different tokenizer regressions: add_special_tokens behavior vs Kimi-K2.5 codec handling/warning issue." }, { "left": "issue:36010", "right": "issue:36331", "accept": false, - "reason": "Import path regression vs trainer API change; no shared underlying bug." + "reason": "Unrelated incompatibilities: GenerationMixin import breakage vs custom trainer loss signature change." }, { "left": "issue:44849", "right": "issue:45701", "accept": false, - "reason": "One is a Qwen3.5 hidden-states bug; the other is a tokenization-version-change report. Different issues." + "reason": "Qwen3.5 hidden-states output bug is a model forward-pass issue, not the broad tokenization regression described by 45701." }, { "left": "issue:43329", "right": "issue:43450", "accept": false, - "reason": "Both are video-related, but one is a token-counting undefined-variable bug and the other is batched video shape output; distinct defects." + "reason": "Both are video-related, but one is undefined vars in token counting and the other is batched-shape handling in video processors." }, { "left": "issue:34567", "right": "issue:40990", "accept": false, - "reason": "Trainer token accounting vs perplexity on a specific model/dataset; unrelated." + "reason": "TrainerState token-count tracking and gpt-oss perplexity are unrelated symptoms and subsystems." }, { "left": "issue:43329", "right": "issue:44479", "accept": false, - "reason": "The titles suggest related Qwen video regressions, but the packet only shows broad topical overlap; insufficient evidence to call them the same underlying bug." + "reason": "Both mention video regressions, but one is a specific undefined-variable bug and the other is a broader v5.3.0 regression across multiple models." }, { "left": "issue:43257", "right": "issue:43931", "accept": false, - "reason": "Related to Qwen model loading, but one is MOE weight conversion with accelerate/deepspeed and the other is a shape mismatch for a VL model; not the same bug." + "reason": "Different load-time failures: Qwen3 MoE weight conversion vs Qwen3-VL shape mismatch." }, { "left": "issue:43697", "right": "issue:45357", "accept": false, - "reason": "Different regressions: RTDetrV2 output differences vs Qwen3.5 save_pretrained visual encoder keys." + "reason": "RTDetrV2 output drift and Qwen3.5 save_pretrained key corruption are unrelated model-specific issues." }, { "left": "issue:44291", "right": "issue:45042", "accept": false, - "reason": "Completely different areas: init_empty_weights loading error vs image processor torchvision requirement." + "reason": "init_empty_weights constructor-arg bug is unrelated to PIL backend requiring torchvision." }, { - "left": "issue:41720", - "right": "issue:43540", + "left": "issue:43452", + "right": "issue:43582", "accept": false, - "reason": "Different model families and failures: CUDA assert on auto device map vs video-processing ValueError." + "reason": "gguf_file loading breakage and Apple Silicon caching_allocator_warmup TypeError are unrelated." }, { - "left": "issue:43452", - "right": "issue:43582", + "left": "issue:41720", + "right": "issue:43540", "accept": false, - "reason": "Different loading bugs: gguf_file tokenizer/model loading vs Apple Silicon allocator warmup TypeError." + "reason": "Different Qwen model families and different failures: cudaErrorAssert on auto device map vs video-processing ValueError." }, { "left": "issue:42915", "right": "issue:43742", "accept": false, - "reason": "Qwen3Moe FineGrainedFP8Config vs MobileLLM loading key error; unrelated." + "reason": "FineGrainedFP8Config failure on Qwen3Moe is unrelated to MobileLLM-125M key error on load." }, { "left": "issue:43317", "right": "issue:45538", "accept": false, - "reason": "Different loading/quantization issue vs tokenizer max-length constant bug." + "reason": "Dequantized model offload failure is unrelated to CLIPTokenizer model_max_length behavior." }, { "left": "issue:33357", "right": "issue:34567", "accept": false, - "reason": "MacOS bus error in CLIP model loading vs TrainerState token counter not updating; unrelated." + "reason": "MacOS CLIP model bus error and TrainerState token counter bug are unrelated." }, { "left": "issue:43927", "right": "issue:44912", "accept": false, - "reason": "Different bugs: custom token IDs lost on config save/load vs MXFP4 quantization fallback." + "reason": "DiaConfig token-ID persistence and gpt-oss quantization fallback are different bugs." }, { "left": "issue:44987", "right": "issue:45538", "accept": false, - "reason": "Model-loading failure for a specific repo vs CLIPTokenizer max_length handling; no shared code-path problem." + "reason": "Loading physical-intelligence/fast under >=5.1.0 is unrelated to CLIPTokenizer max_length." }, { "left": "issue:42915", "right": "issue:44821", "accept": false, - "reason": "Different areas: Qwen3Moe FP8 loading vs AutoImageProcessor URL loading; unrelated." + "reason": "Qwen3Moe FP8 failure is unrelated to AutoImageProcessor URL loading." }, { "left": "issue:44479", "right": "issue:45381", "accept": false, - "reason": "Both are video-input regressions, but the models and reported symptoms differ enough that they do not look like one fixable bug." + "reason": "Both are Qwen video regressions, but one is a broad multi-model regression and the other is a specific qwen2.5-vl vision_position_ids issue." + }, + { + "left": "issue:44568", + "right": "issue:45488", + "accept": false, + "reason": "Both are tokenizer-related, but they concern different tokenizers and distinct regressions." }, { "left": "issue:42915", "right": "issue:45538", "accept": false, - "reason": "Different areas: Qwen3Moe FP8 loading vs tokenizer max-length constant." + "reason": "FineGrainedFP8Config on Qwen3Moe is unrelated to CLIPTokenizer max_length." + }, + { + "left": "issue:43257", + "right": "issue:43756", + "accept": false, + "reason": "Qwen3 MoE weight conversion and Smollm3 RoPE-layer dropping are unrelated model/config bugs." }, { "left": "issue:43540", "right": "issue:44560", "accept": false, - "reason": "Both are video-input failures in Qwen-family models, but the concrete errors and affected models differ; too speculative to merge." + "reason": "Qwen3OmniMoe video-processing ValueError is unrelated to gpt-oss-20b MXFP4 loading fallback." + }, + { + "left": "issue:43540", + "right": "issue:44479", + "accept": false, + "reason": "Both are video-related, but they affect different model families and different failure modes." } ] }, - "evaluator_result": { - "accept": false, - "feedback": "The bulk of the summary is grounded, but the accepted 43329/44479 pair is stated too strongly. From the packet we only have title/metadata-level similarity (both video/Qwen regressions), which supports \u201cpossibly related\u201d but not confidently \u201csame underlying regression\u201d or \u201cclearest root-cause report.\u201d Please hedge that claim more conservatively. The rest of the soft-edge rejections look appropriately cautious." - }, - "evaluator_used": true, - "retried": true, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null @@ -5488,6 +5566,7 @@ "issue:43643", "issue:43646", "issue:43696", + "issue:43697", "issue:43742", "issue:43749", "issue:43756", @@ -5496,10 +5575,8 @@ "issue:43994", "issue:44062", "issue:44291", - "issue:44484", "issue:44488", "issue:44589", - "issue:44704", "issue:44737", "issue:44756", "issue:44792", @@ -5511,6 +5588,7 @@ "issue:45072", "issue:45081", "issue:45137", + "issue:45216", "issue:45325", "issue:45375", "issue:45381", @@ -5524,10 +5602,10 @@ "issue:43257|issue:43696", "issue:43577|issue:45072", "issue:45137|issue:45405", - "issue:38175|issue:44898", "issue:44737|issue:45588", - "issue:43742|issue:44291", + "issue:38175|issue:44898", "issue:44062|issue:44991", + "issue:43742|issue:44291", "issue:43540|issue:45823", "issue:43906|issue:45381", "issue:43646|issue:43749", @@ -5538,8 +5616,8 @@ "issue:43493|issue:43994", "issue:36010|issue:42175", "issue:42175|issue:43824", - "issue:43452|issue:44488", "issue:44291|issue:44821", + "issue:43452|issue:44488", "issue:44792|issue:44964", "issue:43526|issue:45479", "issue:28282|issue:34689", @@ -5547,205 +5625,208 @@ "issue:43756|issue:45325", "issue:43742|issue:44589", "issue:45405|issue:45600", - "issue:44484|issue:44704" + "issue:43697|issue:45216" ], - "prepared_review_unit_hash": "c2debd0c1ed6176ff43cbdc441ed3c80223e9067815bd70b6a4eb05e20ad24e9", + "prepared_review_unit_hash": "42b4363cdf6d8fbaac6f451ebce7cdf6426a41c9adefc7a485c0162880744724", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 27, - "serialized_chars": 35413, - "estimated_input_tokens": 8854, - "estimated_eval_tokens": 17964 + "serialized_chars": 35477, + "estimated_input_tokens": 8870, + "estimated_eval_tokens": 17996 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a heterogeneous mix of Transformers issues spanning different models, loaders, trainers, and runtime errors. None of the proposed soft pairs looks like the same underlying bug/change, so they should stay separate.", - "confidence": 0.79, - "canonical_issue_reason": "issue:43749 has the strongest cross-reference and discussion signal in the set, and it is a broad, high-impact regression that works best as the cluster anchor if one issue must be chosen.", + "summary": "Mostly unrelated bug reports with one plausible duplicate pair around SigLIP2 model correctness. The rest differ by model, API surface, or failure mode and should stay separate.", + "confidence": 0.81, + "canonical_issue_reason": "issue:43493 is the best canonical issue: it describes the underlying SigLIP2 implementation discrepancy at a higher level, while issue:43994 is a user-facing symptom on the same model/checkpoint.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43749 is the most representative problem in the group by activity and inbound references, making it the best single issue to surface for the cluster.", + "best_issue_reason": "issue:43493 is the strongest representative for the SigLIP2 bug because it frames the root behavior difference, not just one downstream reproduction.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43257", "right": "issue:43696", "accept": false, - "reason": "Different failures: Qwen3 MoE weight conversion under accelerate/deepspeed vs GPT-oss CUDA OOM. Same broad area, but not the same bug." + "reason": "Different problems: Qwen3 MoE weight conversion under accelerate/deepspeed vs GPT-oss CUDA OOM." }, { "left": "issue:43577", "right": "issue:45072", "accept": false, - "reason": "Both mention dtype, but they involve different models and different mismatches. Not the same concrete code-path problem." + "reason": "Both mention dtype, but they involve different models and different code paths; not the same concrete bug." }, { "left": "issue:45137", "right": "issue:45405", "accept": false, - "reason": "DeepSpeed ZeRO3 deque error and a PEFT version pin issue are unrelated." - }, - { - "left": "issue:38175", - "right": "issue:44898", - "accept": false, - "reason": "Both are vision-model accuracy issues, but the models and symptoms differ enough that they do not look like one bug." + "reason": "DeepSpeed ZeRO3 deque error is unrelated to a PEFT version bump." }, { "left": "issue:44737", "right": "issue:45588", "accept": false, - "reason": "XLNet CPU device placement bug vs flash-attention sink handling crash. Distinct code paths." + "reason": "XLNet CPU allocation bug and flash-attention crash on sink-less models are unrelated." }, { - "left": "issue:43742", - "right": "issue:44291", + "left": "issue:38175", + "right": "issue:44898", "accept": false, - "reason": "MobileLLM loading key error vs init_empty_weights argument incompatibility. Different load-time failures." + "reason": "Different models and symptoms; one is SigLIP2 zero probabilities, the other is Perceiver resolution handling." }, { "left": "issue:44062", "right": "issue:44991", "accept": false, - "reason": "Tokenizer AddedToken kwarg duplication is unrelated to EMBEDDIA/est-roberta tokenizer loading failure." + "reason": "Tokenizer AddedToken special-arg duplication is a different failure from est-roberta tokenizer loading." + }, + { + "left": "issue:43742", + "right": "issue:44291", + "accept": false, + "reason": "MobileLLM key error and init_empty_weights/_is_hf_initialized type error are different load-time issues." }, { "left": "issue:43540", "right": "issue:45823", "accept": false, - "reason": "Qwen3OmniMoe video-input validation and Gemma4 PLE device mismatch are different model-specific problems." + "reason": "Qwen3Omni video-input error and Gemma4 PLE device mismatch are distinct model-forward bugs." }, { "left": "issue:43906", "right": "issue:45381", "accept": false, - "reason": "One is a reproduction of an older issue; the other is a Qwen2.5-VL video position-id bug. No clear shared fix." + "reason": "The reproduction of issue 38071 is not the same bug as Qwen2.5-VL vision_position_ids scaling." }, { "left": "issue:43646", "right": "issue:43749", "accept": false, - "reason": "Custom model initialization breakage in Transformers 5.0.0 is unrelated to FSDP CPU RAM efficient loading." + "reason": "AutoConfig missing fields with trust_remote_code is unrelated to FSDP CPU RAM efficient loading." }, { "left": "issue:44062", "right": "issue:45081", "accept": false, - "reason": "Tokenizer AddedToken special-flag error and Mistral regex patch crash are different tokenizer internals." + "reason": "AddedToken special-arg duplication and Mistral regex patching crash are different tokenizer bugs." }, { "left": "issue:43749", "right": "issue:44756", "accept": false, - "reason": "Both concern memory/loading, but one is an FSDP regression and the other is mmap behavior on Strix Halo. Not the same bug." + "reason": "Both are memory/loading related, but the causes and fixes are different." }, { "left": "issue:43643", "right": "issue:45375", "accept": false, - "reason": "AutoConfig field loss with trust_remote_code and a missing config field in Qwen3_5MoeVisionConfig are different serialization issues." + "reason": "Both involve missing config fields, but one is generic trust_remote_code behavior and the other is a specific Qwen3_5MoeVisionConfig field being dropped by @strict." }, { "left": "issue:34567", "right": "issue:36331", "accept": false, - "reason": "TrainerState token counting and CustomTrainer.compute_loss signature mismatch are separate trainer API regressions." + "reason": "TrainerState token counting not updating is unrelated to CustomTrainer.compute_loss signature breakage." }, { "left": "issue:43493", "right": "issue:43994", - "accept": false, - "reason": "Both involve SigLIP2, but one is a HF-vs-JAX discrepancy and the other is bad outputs through AutoModel/pipeline. Not enough to merge." + "accept": true, + "reason": "Same SigLIP2 underlying model bug: both report incorrect outputs for google/siglip2-base-patch16-224, with one describing the implementation mismatch and the other the user-visible bad results." }, { "left": "issue:36010", "right": "issue:42175", "accept": false, - "reason": "GenerationMixin import failure and missing TensorFlow backend from torch install are unrelated packaging/import issues." + "reason": "GenerationMixin import error is unrelated to backend dependency packaging." }, { "left": "issue:42175", "right": "issue:43824", "accept": false, - "reason": "TensorFlow backend packaging and missing Qwen2_5_VLForConditionalGeneration symbol are different dependency/import problems." + "reason": "TensorFlow missing from a torch install extra and Qwen2_5_VL import failure are different packaging/import issues." }, { - "left": "issue:43452", - "right": "issue:44488", + "left": "issue:44291", + "right": "issue:44821", "accept": false, - "reason": "gguf_file loading failures and cjvt/sleng-bert tokenizer loading are separate loader regressions." + "reason": "Unexpected _is_hf_initialized argument and AutoImageProcessor URL loading are unrelated." }, { - "left": "issue:44291", - "right": "issue:44821", + "left": "issue:43452", + "right": "issue:44488", "accept": false, - "reason": "Unexpected _is_hf_initialized argument with init_empty_weights is unrelated to AutoImageProcessor-from-URL behavior." + "reason": "gguf_file handling for AutoTokenizer/AutoModel and a model loading failure for cjvt/sleng-bert are different problems." }, { "left": "issue:44792", "right": "issue:44964", "accept": false, - "reason": "Janus test failure and Phi-4 multimodal load failure are different model-specific issues." + "reason": "Janus image-generation test failure and Phi-4-multimodal loading failure are distinct multimodal issues." }, { "left": "issue:43526", "right": "issue:45479", "accept": false, - "reason": "BEiT Fast processor label reduction bug and single-label classification zero-loss bug are not the same code path." + "reason": "BEiT image processor label reduction bug is unrelated to sequence-classification loss degeneracy." }, { "left": "issue:28282", "right": "issue:34689", "accept": false, - "reason": "PyTorch-missing import error and Llama 3.2 Vision load breakage are different environment/model problems." + "reason": "PyTorch-missing ImportError and Llama 3.2 Vision model-loading regression are not the same bug." }, { "left": "issue:45042", "right": "issue:45081", "accept": false, - "reason": "PIL image processor torchvision requirement and Mistral regex patch crash are unrelated." + "reason": "Torchvision dependency regression in PIL image processors is unrelated to the Mistral regex tokenizer crash." }, { "left": "issue:43756", "right": "issue:45325", "accept": false, - "reason": "Smollm3 RoPE-layer configuration and Qwen2.5-VL rope-index scaling are not the same underlying defect." + "reason": "Both touch RoPE/positioning, but they affect different models and different logic." }, { "left": "issue:43742", "right": "issue:44589", "accept": false, - "reason": "MobileLLM key error and missing Float8 storage type are separate load/runtime incompatibilities." + "reason": "MobileLLM key error and missing Float8 storage object are different loading failures." }, { "left": "issue:45405", "right": "issue:45600", "accept": false, - "reason": "A not-yet-released PEFT version bump and removed Sam3LiteText auto-mapping references are different CI/dependency issues." + "reason": "Unreleased MIN_PEFT_VERSION bump and stale auto_mappings references are unrelated CI/package issues." }, { - "left": "issue:44484", - "right": "issue:44704", + "left": "issue:43697", + "right": "issue:45216", "accept": false, - "reason": "save_pretrained shard-size behavior and AutoProcessor kwargs forwarding are unrelated API concerns." + "reason": "RTDetrV2 output drift in v5 and Qwen3.5 save_pretrained corruption are different regressions." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is grounded in the packet and stays conservative. The SigLIP2 pair (43493/43994) is the only clearly plausible duplicate based on the shared model/checkpoint and overlapping symptom, while the other soft-edge pairs are reasonably rejected as different bugs, models, or failure modes. No overstatement beyond the evidence is apparent." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -5784,15 +5865,16 @@ "issue:43742", "issue:43819", "issue:43906", - "issue:43927", "issue:44361", "issue:44442", "issue:44464", "issue:44466", + "issue:44484", "issue:44514", "issue:44561", "issue:44568", "issue:44625", + "issue:44704", "issue:44743", "issue:44811", "issue:44821", @@ -5804,7 +5886,6 @@ "issue:44977", "issue:44991", "issue:45005", - "issue:45216", "issue:45290", "issue:45356", "issue:45381", @@ -5812,253 +5893,245 @@ "issue:45823" ], "soft_pairs": [ - "issue:43697|issue:45216", + "issue:44484|issue:44704", "issue:43531|issue:44821", "issue:44514|issue:45381", "issue:28282|issue:29127", "issue:43582|issue:44977", "issue:44625|issue:44877", "issue:43643|issue:44561", - "issue:42175|issue:44938", - "issue:43720|issue:44898", "issue:43452|issue:43577", - "issue:44514|issue:45290", + "issue:43720|issue:44898", + "issue:42175|issue:44938", "issue:44466|issue:45005", - "issue:43452|issue:44843", + "issue:44514|issue:45290", "issue:41762|issue:44936", + "issue:43452|issue:44843", "issue:43295|issue:45356", "issue:43819|issue:44811", "issue:43653|issue:44568", "issue:44936|issue:45823", "issue:24643|issue:29127", "issue:30064|issue:33666", - "issue:38175|issue:43295", "issue:44361|issue:44442", + "issue:38175|issue:43295", "issue:43653|issue:44361", - "issue:43577|issue:43697", "issue:44464|issue:44743", + "issue:43577|issue:43697", "issue:44843|issue:44991", "issue:41720|issue:43906", "issue:44977|issue:45356", "issue:43688|issue:45440", - "issue:43531|issue:43742", - "issue:43653|issue:43927", - "issue:44464|issue:44977" + "issue:43531|issue:43742" ], - "prepared_review_unit_hash": "04a0b0fdd500c5aa0933dd2684abfcac1d61b1737b988e6394b2ec8b927fca5d", + "prepared_review_unit_hash": "f004e77b9379f33936cc73d52812cc263920772b256022cada7903b672c5a63a", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 37447, - "estimated_input_tokens": 9362, - "estimated_eval_tokens": 18980 + "soft_pair_count": 30, + "serialized_chars": 36783, + "estimated_input_tokens": 9196, + "estimated_eval_tokens": 18648 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is not a single duplicate family; it mixes many unrelated transformer bugs spanning tokenizer loading, chat templating, generation, model serialization, device mapping, and training/runtime failures. None of the soft-similarity pairs look like the same concrete bug or change.", - "confidence": 0.98, - "canonical_issue_reason": "No single issue is a good canonical representative because the set is heterogeneous and does not center on one underlying defect.", + "summary": "This set is not a single duplicate cluster; it contains many unrelated bug reports across tokenizers, configs, processors, generation, and v5 regressions. I rejected all soft pairs because they do not appear to describe the same concrete underlying bug or change.", + "confidence": 0.93, + "canonical_issue_reason": "No true canonical issue fits the whole set; the cluster is heterogeneous. If forced to pick an anchor, issue 44466 is the most concrete recent v5 regression and the closest to another soft pair about tied weights.", "canonical_pr_reason": null, - "best_issue_reason": "No global best issue: the items cover many distinct subsystems and failure modes, so choosing one would be arbitrary.", + "best_issue_reason": "Issue 44466 is the least-bad representative anchor, but only loosely; it is about tied-weight serialization in v5 and does not duplicate the rest of the set.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43697", - "right": "issue:45216", + "left": "issue:44484", + "right": "issue:44704", "accept": false, - "reason": "Different bugs: RTDetrV2 output divergence vs Qwen3.5 save_pretrained regression." + "reason": "Different problems: shard-size/default save_pretrained behavior vs kwargs not reaching cached_file in AutoProcessor." }, { "left": "issue:43531", "right": "issue:44821", "accept": false, - "reason": "Different subsystems: Qwen3-MoE sliding_window behavior vs AutoImageProcessor URL loading." + "reason": "Unrelated: Qwen3-MoE sliding_window behavior vs loading AutoImageProcessor from URL." }, { "left": "issue:44514", "right": "issue:45381", "accept": false, - "reason": "Both touch Qwen2.5-VL, but one is batched chat template padding and the other is video vision_position_ids; distinct failures." + "reason": "Both are Qwen2.5-VL, but one is chat-template batching and the other is video vision_position_ids; distinct bugs." }, { "left": "issue:28282", "right": "issue:29127", "accept": false, - "reason": "Missing PyTorch import error is unrelated to LayoutLMv3's unclear box validation message." + "reason": "Importing PyTorch vs LayoutLMv3 box-content validation are unrelated." }, { "left": "issue:43582", "right": "issue:44977", "accept": false, - "reason": "AppleSilicon caching_allocator_warmup TypeError is unrelated to Qwen3.5 flash-attention generation." + "reason": "Apple Silicon allocator TypeError vs Qwen3.5 flash-attention generation issue; different code paths." }, { "left": "issue:44625", "right": "issue:44877", "accept": false, - "reason": "Both involve config handling, but the concrete bugs differ: num_labels propagation vs strict config rejection." + "reason": "Qwen3.5 num_labels propagation vs granite_speech config strictness are separate config-loading bugs." }, { "left": "issue:43643", "right": "issue:44561", "accept": false, - "reason": "Different regression surfaces: missing fields from AutoConfig vs v5 removal of is_torch_fx_available breaking trust_remote_code." + "reason": "Both mention trust_remote_code, but one is missing returned config fields and the other is a removed helper breaking models." }, { - "left": "issue:42175", - "right": "issue:44938", + "left": "issue:43452", + "right": "issue:43577", "accept": false, - "reason": "Packaging/backend installation issue is unrelated to a Python 3.14 import/load failure." + "reason": "gguf loading/tokenizer breakage vs Blip2 dtype not propagating; no shared underlying defect." }, { "left": "issue:43720", "right": "issue:44898", "accept": false, - "reason": "BitNet packed-weight loading bug is not the same as Perceiver non-default resolution failure." - }, - { - "left": "issue:43452", - "right": "issue:43577", - "accept": false, - "reason": "Tokenizer/model loading via gguf_file is unrelated to BLIP2 dtype propagation." + "reason": "BitNet packed-weight loading vs Perceiver interpolation failure are different model-specific bugs." }, { - "left": "issue:44514", - "right": "issue:45290", + "left": "issue:42175", + "right": "issue:44938", "accept": false, - "reason": "Same API family, but different edge cases: batched padding=False crash vs assistant tool-call messages with no content." + "reason": "Backend dependency packaging vs Python 3.14 import/load failure are unrelated." }, { "left": "issue:44466", "right": "issue:45005", "accept": false, - "reason": "Both mention tied weights, but one is device-dependent lm_head serialization and the other is translation-model tied-weight issues; not the same bug." + "reason": "Both involve tied weights, but one is a device-dependent serialization regression and the other is a broader translation-model report; not the same concrete bug." }, { - "left": "issue:43452", - "right": "issue:44843", + "left": "issue:44514", + "right": "issue:45290", "accept": false, - "reason": "Both affect loading, but one is gguf_file breakage and the other is an offline model_info call in mistral regex patching." + "reason": "Different chat-template crashes: batched padding behavior vs assistant tool-call messages with no content." }, { "left": "issue:41762", "right": "issue:44936", "accept": false, - "reason": "Gemma3 ZeRO-3 load failure is unrelated to trainer.evaluate() failing after train()." + "reason": "Gemma3 DeepSpeed ZeRO-3 loading failure vs trainer.evaluate after train; separate training/runtime issues." + }, + { + "left": "issue:43452", + "right": "issue:44843", + "accept": false, + "reason": "gguf file support vs offline model_info call in Mistral regex patch are unrelated tokenizer/loading problems." }, { "left": "issue:43295", "right": "issue:45356", "accept": false, - "reason": "Processor/tokenizer regression and Kimi codec handling are different bugs." + "reason": "Processor/tokenizer regression vs Kimi-K2.5 codec/regex warning regression; different symptoms and code paths." }, { "left": "issue:43819", "right": "issue:44811", "accept": false, - "reason": "DAC from_latents mismatch is unrelated to Whisper batch_decode skip_special_tokens handling." + "reason": "DAC latent conversion mismatch vs Whisper batch_decode skip_special_tokens handling are unrelated." }, { "left": "issue:43653", "right": "issue:44568", "accept": false, - "reason": "Both involve special tokens, but different tokenizers and behaviors; not one underlying defect." + "reason": "Both are tokenizer special-token issues, but BigBird mask-token registration and mdeberta BOS/EOS omission are different bugs." }, { "left": "issue:44936", "right": "issue:45823", "accept": false, - "reason": "Trainer lifecycle bug and Gemma4 PLE device mismatch are unrelated." + "reason": "Trainer lifecycle failure vs Gemma4 PLE device mismatch during forward are unrelated." }, { "left": "issue:24643", "right": "issue:29127", "accept": false, - "reason": "DeepSpeed training RuntimeError is unrelated to a LayoutLMv3 error-message clarity issue." + "reason": "DeepSpeed training weight-dimension error vs LayoutLMv3 box-info validation are not the same bug." }, { "left": "issue:30064", "right": "issue:33666", "accept": false, - "reason": "Void segmentation map processing and Qwen2-VL multi-GPU training are unrelated." + "reason": "Void segmentation-map processing and Qwen2-VL multi-GPU training are different image/model workflows." }, { - "left": "issue:38175", - "right": "issue:43295", + "left": "issue:44361", + "right": "issue:44442", "accept": false, - "reason": "Siglip2 zero-probabilities bug is unrelated to processor.tokenizer regression." + "reason": "MLukeTokenizer task AttributeError and AutoTokenizer failing on FastSpeech2ConformerTokenizer are separate tokenizer-load issues." }, { - "left": "issue:44361", - "right": "issue:44442", + "left": "issue:38175", + "right": "issue:43295", "accept": false, - "reason": "Both are tokenizer-related, but they affect different tokenizer classes and failure modes." + "reason": "SigLIP2 zero probabilities vs processor.tokenizer regression are unrelated." }, { "left": "issue:43653", - "right": "issue:43927", + "right": "issue:44361", "accept": false, - "reason": "BigBird special-token decode bug is unrelated to DiaConfig losing custom token IDs." + "reason": "BigBird special-token decoding bug vs MLukeTokenizer task AttributeError are distinct tokenizer issues." }, { "left": "issue:44464", "right": "issue:44743", "accept": false, - "reason": "Both are generation/state bugs, but one is compiled forward chunked generation and the other is cache state reset in a modular Qwen3.5 model." + "reason": "Chunked generation with compiled forward and Qwen3.5 recurrent-state reset are different generation/cache bugs." + }, + { + "left": "issue:43577", + "right": "issue:43697", + "accept": false, + "reason": "Blip2 dtype propagation vs RTDetrV2 output divergence are unrelated model behaviors." }, { "left": "issue:44843", "right": "issue:44991", "accept": false, - "reason": "Offline mistral regex bug and EMBEDDIA/est-roberta tokenizer load failure have different root causes." + "reason": "Offline model_info regression vs tokenizer loading failure for a specific model are not the same defect." }, { "left": "issue:41720", "right": "issue:43906", "accept": false, - "reason": "No evidence these are the same bug; one is Qwen3 auto device mapping cudaErrorAssert, the other is only an isolated reproduction of another issue." + "reason": "Qwen3 auto device mapping cudaErrorAssert vs an isolated reproduction of another issue are not duplicates." }, { "left": "issue:44977", "right": "issue:45356", "accept": false, - "reason": "Qwen3.5 flash-attention generation bug is unrelated to Kimi-K2.5 tokenizer codec handling." + "reason": "Qwen3.5 flash-attention generation bug vs Kimi tokenizer codec regression are unrelated." }, { "left": "issue:43688", "right": "issue:45440", "accept": false, - "reason": "Both mention MoE-style models, but the reported defects are different: aux-loss normalization vs DeepSeekV3 implementation divergence." + "reason": "Aux-loss normalization in MoE models vs DeepSeekV3 implementation divergence are separate MoE issues." }, { "left": "issue:43531", "right": "issue:43742", "accept": false, - "reason": "Qwen3-MoE sliding_window issue is unrelated to MobileLLM key-error loading." - }, - { - "left": "issue:43653", - "right": "issue:43927", - "accept": false, - "reason": "Tokenizer special-token registration bug is unrelated to DiaConfig custom token ID persistence." - }, - { - "left": "issue:44464", - "right": "issue:44977", - "accept": false, - "reason": "Different generation problems in different models/code paths; not mergeable as one fix." + "reason": "Qwen3-MoE sliding_window behavior and MobileLLM key error do not share a concrete bug path." } ] }, @@ -6101,11 +6174,11 @@ "issue:43493", "issue:43504", "issue:43525", - "issue:43540", "issue:43550", "issue:43572", "issue:43577", "issue:43606", + "issue:43653", "issue:43688", "issue:43720", "issue:43742", @@ -6131,6 +6204,8 @@ "issue:45812" ], "soft_pairs": [ + "issue:43653|issue:43927", + "issue:44464|issue:44977", "issue:43577|issue:45588", "issue:43550|issue:43927", "issue:43550|issue:44361", @@ -6148,8 +6223,8 @@ "issue:41720|issue:44464", "issue:33666|issue:36010", "issue:43756|issue:44913", - "issue:43688|issue:43749", "issue:43329|issue:45325", + "issue:43688|issue:43749", "issue:43493|issue:43825", "issue:43525|issue:43572", "issue:43262|issue:43295", @@ -6162,235 +6237,240 @@ "issue:39692|issue:43493", "issue:43720|issue:44361", "issue:43688|issue:45704", - "issue:43334|issue:43572", - "issue:43540|issue:44977" + "issue:43334|issue:43572" ], - "prepared_review_unit_hash": "e0d64d75d07fdbd56d7b9ec9dc896f4779c6f5fdf82c8b112cf75f026872750c", + "prepared_review_unit_hash": "e7bbd8d92b83c3a7f0361853b46666fa40d0deb54687ea18cc764990025a23c4", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 33, - "serialized_chars": 37274, - "estimated_input_tokens": 9319, - "estimated_eval_tokens": 18894 + "soft_pair_count": 34, + "serialized_chars": 37612, + "estimated_input_tokens": 9403, + "estimated_eval_tokens": 19062 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is a heterogeneous mix of Transformer bugs (tokenizer/config regressions, model-loading failures, training errors, and multimodal edge cases). I don\u2019t see any soft pair that clearly points to the same underlying bug or change, so no duplicates should be merged here.", - "confidence": 0.88, - "canonical_issue_reason": "issue:45812 is the strongest representative: it\u2019s an open, broad regression affecting all Granite models, with clear user impact and recent discussion/inbound references.", + "summary": "The cluster is broad and mostly heterogeneous, with many separate Transformers regressions across tokenization, loading, trainer, and multimodal code paths. I found no PRs, and I do not see any soft-pair that looks like the same underlying bug with enough certainty to merge.", + "confidence": 0.42, + "canonical_issue_reason": "issue:45812 is the strongest representative of the cluster\u2019s broad v4\u2192v5 tokenizer regression theme: it is open, concrete, and impacts multiple Granite models rather than a single narrow repro.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45812 is the best overall issue anchor because it describes a concrete, widespread regression rather than a narrow model-specific failure.", + "best_issue_reason": "issue:45812 is the best global issue candidate because it describes a high-impact, cross-model tokenizer regression and is phrased broadly enough to anchor related reports.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43653", + "right": "issue:43927", + "accept": false, + "reason": "Different failure classes: special-token decode behavior in BigBirdTokenizer vs config token ID persistence and generation IndexError in DiaConfig." + }, + { + "left": "issue:44464", + "right": "issue:44977", + "accept": false, + "reason": "Different code paths: compiled chunked generation inconsistency vs Qwen3.5 flash-attention generation failure." + }, { "left": "issue:43577", "right": "issue:45588", "accept": false, - "reason": "Different failures: dtype stays float32 in Blip2 loading vs a flash-attention crash on sink-less models." + "reason": "Unrelated bugs: Blip2 dtype loading stays float32 vs flash_attention.py crashing on sink-less models with s_aux=None." }, { "left": "issue:43550", "right": "issue:43927", "accept": false, - "reason": "One is a torch.compile/SDPA runtime bug; the other is a config save/load token-ID regression." + "reason": "Different areas: torch.compile/SDPA runtime failure vs custom token ID loss and generation IndexError after save/load." }, { "left": "issue:43550", "right": "issue:44361", "accept": false, - "reason": "Unrelated: Bamba compile/SDPA failure vs MLukeTokenizer task AttributeError." + "reason": "Model execution bug vs tokenizer AttributeError; they do not share the same underlying path." }, { "left": "issue:32090", "right": "issue:34567", "accept": false, - "reason": "Both involve Trainer, but the bugs differ: GPU broadcast NoneType error vs num_input_tokens_seen not updating." + "reason": "Trainer broadcast NoneType error and TrainerState token-count update bug are separate trainer issues." }, { "left": "issue:43504", "right": "issue:43550", "accept": false, - "reason": "Different model paths and root causes: legacy-field load failure vs compile/SDPA failure." + "reason": "Legacy preset loading failure for BeitForSemanticSegmentation is not the same as Bamba SDPA/compile failure." }, { "left": "issue:43504", "right": "issue:43720", "accept": false, - "reason": "Model loading bugs in different subsystems; Beit legacy config issue is unrelated to BitNet packed-weight unloading." + "reason": "Different load-time problems: legacy field handling vs BitNet packed-weight unpacking during accelerate loading." }, { "left": "issue:43122", "right": "issue:43906", "accept": false, - "reason": "The second is only a reproduction issue; no evidence it matches the tokenizer behavior regression in the first." + "reason": "Both mention tokenization, but 43906 is only an isolated reproduction of another issue and does not clearly match the same tokenizer regression." }, { "left": "issue:32090", "right": "issue:33666", "accept": false, - "reason": "Trainer error vs multi-GPU training request; not the same underlying bug." + "reason": "Trainer GPU broadcast error is unrelated to Qwen2-VL multi-GPU training behavior." }, { "left": "issue:28282", "right": "issue:35141", "accept": false, - "reason": "ImportError for missing PyTorch is unrelated to embedding reinitialization during post_init." + "reason": "ImportError for missing PyTorch is unrelated to embedding reinitialization after resize." }, { "left": "issue:43504", "right": "issue:44610", "accept": false, - "reason": "Different vision-model load/processor problems with no shared code-path evidence." + "reason": "Model preset loading and processor/model image-size mismatch are different bugs." }, { "left": "issue:43452", "right": "issue:44451", "accept": false, - "reason": "Both are load failures, but one is gguf-file handling and the other is a specific model import/load regression." + "reason": "GGUF loading breakage and inability to load a specific model are not the same code-path issue." }, { "left": "issue:44849", "right": "issue:45357", "accept": false, - "reason": "Qwen3.5 hidden-state output bug vs save_pretrained saving wrong visual encoder keys." + "reason": "Hidden-states generation bug and incorrect visual encoder keys on save are separate Qwen3.5 problems." }, { "left": "issue:44568", "right": "issue:45812", "accept": false, - "reason": "Different tokenizer regressions on different model families; no shared bug mechanism shown." + "reason": "Both are tokenizer regressions, but one is mdeberta special-token handling and the other is Granite token IDs; no clear shared underlying bug." }, { "left": "issue:43742", "right": "issue:44821", "accept": false, - "reason": "KeyError loading a text model is unrelated to loading AutoImageProcessor from a URL." + "reason": "KeyError loading MobileLLM is unrelated to AutoImageProcessor loading from URL." }, { "left": "issue:41720", "right": "issue:44464", "accept": false, - "reason": "CUDA assert during auto device mapping is not the same as inconsistent chunked generation with compiled forward." + "reason": "cudaErrorAssert under auto device mapping is a different failure than compiled-forward chunked generation inconsistency." }, { "left": "issue:33666", "right": "issue:36010", "accept": false, - "reason": "Multi-GPU training issue vs import error for GenerationMixin; different layers entirely." + "reason": "Multi-GPU training for Qwen2-VL and GenerationMixin import failure are unrelated." }, { "left": "issue:43756", "right": "issue:44913", "accept": false, - "reason": "Both touch rotary settings, but one is a dropped-layer bug and the other is config reload resetting rotary_pct." + "reason": "Smollm3 RoPE-layer mismatch and GPTNeoX rotary_pct reload defaulting are different model/config bugs." }, { - "left": "issue:43688", - "right": "issue:43749", + "left": "issue:43329", + "right": "issue:45325", "accept": false, - "reason": "Aux-loss normalization bug and FSDP CPU RAM loading bug are unrelated." + "reason": "Video multimodal token counting bug is not the same as still-image temporal position scaling in Qwen2.5-VL." }, { - "left": "issue:43329", - "right": "issue:45325", + "left": "issue:43688", + "right": "issue:43749", "accept": false, - "reason": "Multimodal token counting video-branch bug vs still-image rope index scaling bug." + "reason": "Aux-loss normalization and FSDP CPU RAM efficient loading are separate problems." }, { "left": "issue:43493", "right": "issue:43825", "accept": false, - "reason": "SigLIP2 implementation mismatch is unrelated to a pipeline error-message regression." + "reason": "SigLIP2 implementation discrepancy and pipeline error-message regression do not share the same bug." }, { "left": "issue:43525", "right": "issue:43572", "accept": false, - "reason": "Similar missing-pad-token wording, but different fields and different model configs." + "reason": "Similar theme, but different configs and missing fields: pad_token_id in Llama4Config vs pad_token_idx in StableLmConfig." }, { "left": "issue:43262", "right": "issue:43295", "accept": false, - "reason": "Audio chat-template sampling-rate default bug vs processor.tokenizer/image-passing regression." + "reason": "Audio chat-template sample-rate defaulting is unrelated to processor.tokenizer access/images-to-tokenizer regression." }, { "left": "issue:43606", "right": "issue:43927", "accept": false, - "reason": "CPU offload device mismatch is unrelated to DiaConfig token-ID persistence." + "reason": "CPU offload device mismatch is a different bug from custom token ID loss and generation failure." }, { "left": "issue:43720", "right": "issue:43927", "accept": false, - "reason": "BitNet accelerate-loading/unpack bug vs config save/load token-ID loss; different code paths." + "reason": "BitNet packed-weight loading failure and DiaConfig token persistence/generation bug are unrelated." }, { "left": "issue:28282", "right": "issue:32090", "accept": false, - "reason": "Missing PyTorch ImportError vs Trainer broadcast TypeError; no overlap beyond general training/runtime context." + "reason": "Missing PyTorch ImportError is not the same as Trainer GPU broadcast of NoneType." }, { "left": "issue:28282", "right": "issue:33357", "accept": false, - "reason": "Environment ImportError vs MacOS bus error on a CLIP model; different failure modes." + "reason": "PyTorch missing ImportError and MacOS bus error with CLIP are unrelated failures." }, { "left": "issue:43577", "right": "issue:44554", "accept": false, - "reason": "Blip2 dtype casting issue and MPS attention correctness issue are unrelated." + "reason": "Blip2 dtype loading and MPS attention correctness with differing head dims are distinct issues." }, { "left": "issue:33666", "right": "issue:36331", "accept": false, - "reason": "Multi-GPU training request vs unexpected num_items_in_batch in custom compute_loss." + "reason": "Multi-GPU training issue and unexpected compute_loss keyword argument are separate trainer-related bugs." }, { "left": "issue:39692", "right": "issue:43493", "accept": false, - "reason": "Docs example/quantization failures are not the same bug as the SigLIP2 HF-vs-JAX discrepancy." + "reason": "Documentation example errors are not the same as a SigLIP2 implementation discrepancy." }, { "left": "issue:43720", "right": "issue:44361", "accept": false, - "reason": "Packed-weight loading bug vs tokenizer task AttributeError; different subsystems." + "reason": "BitNet accelerate-loading bug and MLukeTokenizer AttributeError are different subsystems." }, { "left": "issue:43688", "right": "issue:45704", "accept": false, - "reason": "Aux-loss normalization bug is unrelated to T5 using apex.FusedRMSNorm and leaking memory." + "reason": "Aux-loss normalization and apex FusedRMSNorm memory leak are unrelated." }, { "left": "issue:43334", "right": "issue:43572", "accept": false, - "reason": "Different model families and different missing config fields." - }, - { - "left": "issue:43540", - "right": "issue:44977", - "accept": false, - "reason": "Video-input processing error in Qwen3OmniMoe is unrelated to Qwen3.5 flash-attention generation issues." + "reason": "Qwen3-VL missing pad_token_id and StableLmConfig missing pad_token_idx are different model/config regressions." } ] }, @@ -6424,6 +6504,7 @@ "issue:43334", "issue:43493", "issue:43504", + "issue:43540", "issue:43572", "issue:43575", "issue:43606", @@ -6439,7 +6520,6 @@ "issue:44291", "issue:44361", "issue:44423", - "issue:44464", "issue:44466", "issue:44479", "issue:44488", @@ -6451,19 +6531,20 @@ "issue:44792", "issue:44843", "issue:44964", + "issue:44977", "issue:45083", "issue:45200", "issue:45229", "issue:45290", "issue:45405", "issue:45520", - "issue:45584", "issue:45701", "issue:45759", "issue:45823" ], "soft_pairs": [ "issue:44964|issue:45405", + "issue:43540|issue:44977", "issue:44479|issue:44625", "issue:43493|issue:43844", "issue:43334|issue:45759", @@ -6484,210 +6565,209 @@ "issue:44423|issue:44734", "issue:43927|issue:44361", "issue:43504|issue:43653", - "issue:43575|issue:45405", "issue:44625|issue:45759", + "issue:43575|issue:45405", "issue:43606|issue:44112", "issue:44792|issue:45405", "issue:43994|issue:44493", - "issue:44521|issue:45290", - "issue:44464|issue:45584" + "issue:44521|issue:45290" ], - "prepared_review_unit_hash": "f3136336a04ae6d192630e826cfe162154084b7d4ef5cc8c158790ca4ea208a6", + "prepared_review_unit_hash": "952113fe6001b6fa7a1620e5b3cbdfe2c4b8fede404dd9e471e5233024ea3480", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 28, - "serialized_chars": 35738, - "estimated_input_tokens": 8935, - "estimated_eval_tokens": 18126 + "serialized_chars": 35705, + "estimated_input_tokens": 8927, + "estimated_eval_tokens": 18110 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These issues are mostly a heterogeneous mix of unrelated bug reports across model loading, multimodal config handling, tokenization/chat templates, serving, and training. None of the soft pairs looks like a true duplicate with the same concrete fix target.", - "confidence": 0.88, - "canonical_issue_reason": "issue:45701 is the broadest open regression here and the closest thing to a cluster anchor, but the overall set is still heterogeneous.", + "summary": "This cluster is a loose collection of unrelated Transformers bug reports, mostly spanning Qwen/Gemma/SigLIP/multimodal loading, config propagation, serving, and tokenizer regressions. There is no clear duplicate center and no PRs in the set.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue stands out; the issues cover many different models and failure modes rather than one shared bug.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45701 has the widest scope and enough discussion/inbound activity to serve as the best representative issue in this mixed cluster.", + "best_issue_reason": "Best treated as a non-duplicate cluster: the most prominent theme is broad regressions in model/config loading, but the reports are still too diverse to collapse into one issue.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44964", "right": "issue:45405", "accept": false, - "reason": "Different bugs: Phi-4 multimodal load failure vs an unreleased PEFT version constraint." + "reason": "Different problems: multimodal model loading failure vs an unreleased PEFT version bump." + }, + { + "left": "issue:43540", + "right": "issue:44977", + "accept": false, + "reason": "Both are Qwen-related, but one is a video processing ValueError and the other is a flash-attention generation issue." }, { "left": "issue:44479", "right": "issue:44625", "accept": false, - "reason": "Different regressions: Qwen video-input path vs Qwen3.5 num_labels propagation." + "reason": "Unrelated Qwen bugs: video input regression vs num_labels propagation in config." }, { "left": "issue:43493", "right": "issue:43844", "accept": false, - "reason": "Different subsystems: SigLIP2 implementation mismatch vs ZeRO-3 gradient growth." + "reason": "Different subsystems and symptoms: SigLIP2 implementation mismatch vs DeepSpeed gradient scaling anomaly." }, { "left": "issue:43334", "right": "issue:45759", "accept": false, - "reason": "Both are Qwen multimodal config-loading issues, but one is missing pad_token_id and the other is from_config not unwrapping text_config; not the same bug." + "reason": "Both involve Qwen multimodal configs, but one is a missing pad_token_id attribute and the other is from_config text_config unwrapping." }, { "left": "issue:45200", "right": "issue:45229", "accept": false, - "reason": "Gemma4 fine-tuning defaults vs multi-GPU inference OOM are distinct failures." + "reason": "Text-only Gemma4 token-type defaulting is unrelated to multi-GPU OOM." }, { "left": "issue:43756", "right": "issue:45083", "accept": false, - "reason": "Smollm3 RoPE-layer mismatch vs qwen3_omni_moe helper behavior." + "reason": "Different model families and bugs: Smollm3 RoPE layer loss vs qwen3_omni_moe helper behavior." }, { "left": "issue:44488", "right": "issue:44843", "accept": false, - "reason": "A specific model-load regression is not the same as an offline-mode network call in AutoTokenizer." + "reason": "Unrelated: model load failure for cjvt/sleng-bert vs offline tokenization patch calling model_info()." }, { "left": "issue:43257", "right": "issue:43299", "accept": false, - "reason": "Different Qwen3-MoE failures: accelerate/deepspeed conversion issue vs dev-version loading breakage." + "reason": "Both are Qwen MoE loading issues, but one is accelerate/deepspeed weight conversion and the other is a dev-load regression for Qwen3VL MoE." }, { "left": "issue:43278", "right": "issue:43844", "accept": false, - "reason": "Embedding dtype drift in evaluate vs gradient abnormality under ZeRO-3." + "reason": "Different issues: embedding dtype drift in evaluation vs abnormal gradient growth under ZeRO-3." }, { "left": "issue:43646", "right": "issue:44466", "accept": false, - "reason": "Custom model init breakage vs device-dependent serialization of tied weights." + "reason": "Custom model init breakage is not the same as lm_head.weight serialization varying by device." }, { "left": "issue:43824", "right": "issue:44589", "accept": false, - "reason": "Import/symbol exposure issue vs float8 storage deserialization error." + "reason": "Different failures: missing Qwen2_5_VL import vs missing Float8 storage type." }, { "left": "issue:45229", "right": "issue:45823", "accept": false, - "reason": "OOM during multi-GPU inference vs device mismatch in forward are different concrete bugs." + "reason": "Same model family, but OOM during multi-GPU inference is not the same bug as device mismatch in forward." }, { "left": "issue:44589", "right": "issue:45520", "accept": false, - "reason": "Float8 storage lookup/type error vs Python 3.13 flash_attn import KeyError." + "reason": "Unrelated runtime errors: Float8 storage lookup vs flash_attn KeyError on Python 3.13." }, { "left": "issue:43688", "right": "issue:45701", "accept": false, - "reason": "Aux-loss normalization bug vs tokenization changes; unrelated." + "reason": "Different topics entirely: auxiliary loss normalization vs tokenization changes across versions." }, { "left": "issue:39692", "right": "issue:43295", "accept": false, - "reason": "Both touch multimodal processor code, but one is a docs/example mismatch with quantization failure and the other is a processor.tokenizer regression." + "reason": "Both are multimodal/processor issues, but one is a doc example mismatch and the other is a regression in processor.tokenizer usage." }, { "left": "issue:33357", "right": "issue:33666", "accept": false, - "reason": "Completely different models and failure modes." + "reason": "Completely unrelated: MacOS bus error with CLIP vs Qwen2-VL multi-GPU training." }, { "left": "issue:43572", "right": "issue:44291", "accept": false, - "reason": "One is a missing config field after 5.0, the other is an init_empty_weights constructor-arg mismatch." + "reason": "Both are initialization/config regressions, but they affect different code paths and error signatures." }, { "left": "issue:43504", "right": "issue:43606", "accept": false, - "reason": "Legacy preset field issue vs CPU offload device mismatch." + "reason": "Different models and failures: legacy field in Beit loading vs CPU offload device mismatch in bark-small." }, { "left": "issue:44423", "right": "issue:44734", "accept": false, - "reason": "Both are serve crashes, but one is a multimodal .to() issue and the other is KV-cache continuation indexing." + "reason": "Both concern serve, but one crashes on multimodal input conversion and the other on KV-cache continuation indexing." }, { "left": "issue:43927", "right": "issue:44361", "accept": false, - "reason": "Config save/load token-ID bug vs tokenizer AttributeError on tasks." + "reason": "Different tokenizer/config bugs with different failure modes and models." }, { "left": "issue:43504", "right": "issue:43653", "accept": false, - "reason": "Different models and causes: legacy field load issue vs special-token registration bug." + "reason": "Beit legacy-field loading failure is unrelated to BigBirdTokenizer special-token registration." }, { - "left": "issue:43575", - "right": "issue:45405", + "left": "issue:44625", + "right": "issue:45759", "accept": false, - "reason": "TP OOM on a large model vs an unrelated PEFT version bump." + "reason": "Related Qwen config area, but num_labels propagation and text_config unwrapping are distinct bugs." }, { - "left": "issue:44625", - "right": "issue:45759", + "left": "issue:43575", + "right": "issue:45405", "accept": false, - "reason": "Both involve Qwen3.5 config handling, but num_labels propagation and text_config unwrapping are separate concrete issues." + "reason": "TP OOM on Qwen2-57B is unrelated to a version constraint bump." }, { "left": "issue:43606", "right": "issue:44112", "accept": false, - "reason": "Device mismatch in CPU offload vs a stale CI test failure." + "reason": "A real model bug vs a stale CI test; not the same underlying issue." }, { "left": "issue:44792", "right": "issue:45405", "accept": false, - "reason": "Janus image-generation test failure vs PEFT release versioning." + "reason": "Janus image-generation test failure is unrelated to PEFT version availability." }, { "left": "issue:43994", "right": "issue:44493", "accept": false, - "reason": "SigLIP2 bad outputs vs unexpected key in position IDs." + "reason": "SigLIP2 pipeline/nonsensical output issue is unrelated to unexpected position-id key warnings." }, { "left": "issue:44521", "right": "issue:45290", "accept": false, - "reason": "Both are apply_chat_template failures, but zeroed assistant masks and tool-call crashes are different code paths." - }, - { - "left": "issue:44464", - "right": "issue:45584", - "accept": false, - "reason": "Compiled-forward chunked-generation inconsistency vs Whisper empty-transcription failure." + "reason": "Both use apply_chat_template, but one is all-zero assistant masks and the other is a crash on tool-call messages with no content." } ] }, @@ -6719,7 +6799,6 @@ "issue:34567", "issue:37428", "issue:38175", - "issue:42915", "issue:43232", "issue:43278", "issue:43295", @@ -6757,33 +6836,35 @@ "issue:45412", "issue:45446", "issue:45520", + "issue:45584", "issue:45701", "issue:45823" ], "soft_pairs": [ + "issue:44464|issue:45584", "issue:29942|issue:30333", "issue:43844|issue:44964", "issue:28218|issue:34567", "issue:45084|issue:45520", "issue:44554|issue:45446", - "issue:43688|issue:43873", "issue:44568|issue:45701", - "issue:15354|issue:38175", + "issue:43688|issue:43873", "issue:43540|issue:45200", + "issue:15354|issue:38175", "issue:43232|issue:44849", "issue:45200|issue:45823", "issue:45072|issue:45198", "issue:44315|issue:44849", "issue:43278|issue:43701", - "issue:44554|issue:45356", "issue:28218|issue:33357", + "issue:44554|issue:45356", "issue:45200|issue:45325", - "issue:44466|issue:45127", "issue:43295|issue:44610", + "issue:44466|issue:45127", "issue:44898|issue:45412", - "issue:37428|issue:43957", "issue:43493|issue:43701", "issue:43572|issue:44987", + "issue:37428|issue:43957", "issue:43334|issue:43525", "issue:43572|issue:43957", "issue:43873|issue:44464", @@ -6791,246 +6872,242 @@ "issue:43844|issue:44936", "issue:43824|issue:43975", "issue:43295|issue:43873", - "issue:37428|issue:43824", "issue:38175|issue:45412", - "issue:42915|issue:43824" + "issue:37428|issue:43824" ], - "prepared_review_unit_hash": "d3c802f1aa0e5fdd02d7d5a424c552b897801f9c175b2b5e198f4e677f228af8", + "prepared_review_unit_hash": "232ad769121337cd3469cb8d9f5da0af93f68ab86d9f4133d2830354539193cc", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 33, - "serialized_chars": 37663, - "estimated_input_tokens": 9416, - "estimated_eval_tokens": 19088 + "serialized_chars": 37692, + "estimated_input_tokens": 9423, + "estimated_eval_tokens": 19102 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly heterogeneous set. The only plausible duplicate-like cluster is the pair of model-loading errors about missing `pad_token_id` on config objects; the rest are clearly unrelated bugs in different code paths.", - "confidence": 0.72, - "canonical_issue_reason": "issue:43334 is a reasonable canonical representative for the only likely duplicate pair. It clearly reports the pad_token_id config AttributeError during model loading, which matches the same failure mode described in 43525.", + "summary": "This set is highly heterogeneous: it spans tokenization, generation, multimodal processors, training state, import errors, and device/dtype issues. I found no soft pairs that look like the same underlying bug/change, so no duplicate merges are justified.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 45701 is the broadest, most umbrella-like report (tokenization regressions across versions) and is the closest thing to a cluster anchor, though the cluster is not truly one duplicate set.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43334 is the best standalone issue in this set because it is the clearest statement of the shared pad_token_id loading failure and sits in the only credible duplicate cluster.", + "best_issue_reason": "45701 is the least-bad global representative because it is broad and user-facing, but it still does not meaningfully cover the rest of the unrelated issues.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44464", + "right": "issue:45584", + "accept": false, + "reason": "Different code paths: compiled forward/chunked generation vs Whisper empty-transcription generation after special-token alignment." + }, { "left": "issue:29942", "right": "issue:30333", "accept": false, - "reason": "Different failures in different subsystems: Flash Attention 2 tests vs MLflow job-status reporting." + "reason": "Unrelated subsystems: FlashAttention2 test failures vs MLflow reporting after failed training." }, { "left": "issue:43844", "right": "issue:44964", "accept": false, - "reason": "Unrelated problems: ZeRO-3 gradient behavior vs a multimodal model-loading failure." + "reason": "Different bugs: ZeRO-3 gradient growth vs model loading failure for Phi-4 multimodal." }, { "left": "issue:28218", "right": "issue:34567", "accept": false, - "reason": "Tokenizer spacing bug and trainer token-count tracking are unrelated." + "reason": "Tokenizer spacing bug vs TrainerState token counter not updating; no shared underlying change." }, { "left": "issue:45084", "right": "issue:45520", "accept": false, - "reason": "Compilation error for template nodes is not the same as a Python 3.13 flash_attn import-key error." + "reason": "Compilation/parsing error vs import_utils flash-attn KeyError on Python 3.13." }, { "left": "issue:44554", "right": "issue:45446", "accept": false, - "reason": "MPS attention correctness issue vs flex_attention import version check; different code paths." - }, - { - "left": "issue:43688", - "right": "issue:43873", - "accept": false, - "reason": "Aux-loss normalization and quantization offloading are distinct training/runtime issues." + "reason": "MPS attention correctness issue vs a PyTorch version check/import bug in flex_attention." }, { "left": "issue:44568", "right": "issue:45701", "accept": false, - "reason": "Specific BOS/EOS omission for one tokenizer is not the same as a broad tokenization regression report." + "reason": "Specific special-token insertion regression vs a broader tokenization-version regression; not the same concrete bug." }, { - "left": "issue:15354", - "right": "issue:38175", + "left": "issue:43688", + "right": "issue:43873", "accept": false, - "reason": "TorchScript export limitation and SigLIP2 zero-probability behavior are unrelated." + "reason": "Aux-loss normalization bug vs offloading/quantization behavior; different functionality." }, { "left": "issue:43540", "right": "issue:45200", "accept": false, - "reason": "Video-input processing failure in Qwen3OmniMoe is unrelated to Gemma 4 token-type handling." + "reason": "Video input processing error in Qwen3OmniMoe vs Gemma4 multimodal token-type defaults." + }, + { + "left": "issue:15354", + "right": "issue:38175", + "accept": false, + "reason": "TorchScript GeneratorExp export failure vs SigLIP2 zero-probability inference issue." }, { "left": "issue:43232", "right": "issue:44849", "accept": false, - "reason": "Generation kwargs after sync_gpus and hidden-state output bugs do not share the same concrete failure." + "reason": "Generation kwargs update after sync_gpus vs Qwen3.5 hidden-states bug; unrelated generation paths." }, { "left": "issue:45200", "right": "issue:45823", "accept": false, - "reason": "Gemma4 token-type defaults vs device-map forward mismatch are different bugs." + "reason": "Different Gemma4 issues: missing mm_token_type_ids defaults vs device mismatch with device_map='auto'." }, { "left": "issue:45072", "right": "issue:45198", "accept": false, - "reason": "BFloat16 dtype mismatches and save/tokenization failure are not the same issue." + "reason": "BF16 dtype mismatch issues vs Wav2Vec2 save/tokenization failure; no same code path." }, { "left": "issue:44315", "right": "issue:44849", "accept": false, - "reason": "Liger Kernel application in model_init is unrelated to Qwen3.5 hidden-state output." + "reason": "Liger Kernel application in model_init vs Qwen3.5 hidden-states output bug." }, { "left": "issue:43278", "right": "issue:43701", "accept": false, - "reason": "Embedding dtype drift during eval and checkpoint key mismatch are separate training bugs." - }, - { - "left": "issue:44554", - "right": "issue:45356", - "accept": false, - "reason": "MPS attention correctness and tokenizer codec regression are unrelated." + "reason": "Training/eval dtype change vs resume_from_checkpoint key mismatch; unrelated." }, { "left": "issue:28218", "right": "issue:33357", "accept": false, - "reason": "Added-token spacing and MacOS bus error on CLIP have no common code path." + "reason": "Tokenizer added-space bug vs MacOS CLIP bus error; unrelated." }, { - "left": "issue:45200", - "right": "issue:45325", + "left": "issue:44554", + "right": "issue:45356", "accept": false, - "reason": "Both concern multimodal models, but one is token-type defaults and the other is rope-index scaling; different defects." + "reason": "Attention correctness on MPS vs Kimi-K2.5 tokenizer regression; different model paths." }, { - "left": "issue:44466", - "right": "issue:45127", + "left": "issue:45200", + "right": "issue:45325", "accept": false, - "reason": "Compiled-forward chunked generation inconsistency is not the same as LoRA merge serialization/collapse." + "reason": "Gemma4 token-type defaulting vs Qwen2.5-VL RoPE temporal-position scaling; unrelated multimodal bugs." }, { "left": "issue:43295", "right": "issue:44610", "accept": false, - "reason": "Processor API regression and wrong image size output are distinct processor bugs." + "reason": "Processor API regression vs image-size mismatch in OmDet-Turbo; not the same defect." }, { - "left": "issue:44898", - "right": "issue:45412", + "left": "issue:44466", + "right": "issue:45127", "accept": false, - "reason": "Interpolation failure at non-default resolution is unrelated to a model memory-release issue." + "reason": "Serialization/tied-weights inconsistency vs LoRA merge collapse with extended vocab; related theme but different bug." }, { - "left": "issue:37428", - "right": "issue:43957", + "left": "issue:44898", + "right": "issue:45412", "accept": false, - "reason": "Missing flash-attention helper import and meta-device loading failures are different regressions." + "reason": "Perceiver resolution handling vs RT-DETR memory release; unrelated." }, { "left": "issue:43493", "right": "issue:43701", "accept": false, - "reason": "SigLIP2 implementation discrepancy does not match a resume-from-checkpoint key mismatch." + "reason": "SigLIP2 implementation discrepancy vs checkpoint resume key mismatch; no shared bug." }, { "left": "issue:43572", "right": "issue:44987", "accept": false, - "reason": "StableLm pad-token regression and a Physical Intelligence model load failure are not the same bug." + "reason": "StableLm config attribute regression vs a model-loading failure in a different package/version path." + }, + { + "left": "issue:37428", + "right": "issue:43957", + "accept": false, + "reason": "Flash-attn import symbol error vs meta-device model loading breakage; different failures." }, { "left": "issue:43334", "right": "issue:43525", - "accept": true, - "reason": "Likely duplicate: both report model-loading failures caused by config objects missing `pad_token_id`, so they plausibly describe the same regression." + "accept": false, + "reason": "Same symptom class (missing pad_token_id) but different models/configs; insufficient evidence of one shared bug." }, { "left": "issue:43572", "right": "issue:43957", "accept": false, - "reason": "StableLmConfig pad-token regression and meta-device loading issues are unrelated." + "reason": "Config attribute missing vs meta-device loading regression; not the same underlying issue." }, { "left": "issue:43873", "right": "issue:44464", "accept": false, - "reason": "Quantization/offloading behavior is unrelated to compiled-forward generation inconsistency." + "reason": "Quantization/offloading behavior vs compiled-forward generation inconsistency; unrelated." }, { "left": "issue:43295", "right": "issue:43450", "accept": false, - "reason": "Processor tokenizer regression and batched video-shape bug are different multimodal issues." + "reason": "Processor/tokenizer regression vs batched video output shape bug; different concrete problems." }, { "left": "issue:43844", "right": "issue:44936", "accept": false, - "reason": "Training gradient blow-up with ZeRO-3 is unrelated to evaluate-after-train failure." + "reason": "DeepSpeed gradient anomaly vs evaluate-after-train failure; no common root cause evident." }, { "left": "issue:43824", "right": "issue:43975", "accept": false, - "reason": "Importing a Qwen2.5-VL class and DeepSeek detokenization are unrelated." + "reason": "Import error for a specific Qwen2.5-VL class vs DeepSeek tokenizer detokenization regression; unrelated." }, { "left": "issue:43295", "right": "issue:43873", "accept": false, - "reason": "Processor API regression and quantization/offloading behavior do not share the same bug." - }, - { - "left": "issue:37428", - "right": "issue:43824", - "accept": false, - "reason": "Flash-attention import error and Qwen2.5-VL import error are different missing-symbol problems." + "reason": "Processor API change vs quantization offloading; different subsystems." }, { "left": "issue:38175", "right": "issue:45412", "accept": false, - "reason": "SigLIP2 zero probabilities and RT-DETR memory leaks are unrelated." + "reason": "SigLIP2 inference probabilities vs RT-DETR memory cleanup; unrelated." }, { - "left": "issue:42915", + "left": "issue:37428", "right": "issue:43824", "accept": false, - "reason": "FineGrainedFP8Config failure on Qwen3Moe is unrelated to Qwen2.5-VL import exposure." + "reason": "Flash-attn import failure vs missing Qwen2_5_VL export; different import issues and code paths." } ] }, - "evaluator_result": { - "accept": false, - "feedback": "The set-level summary is mostly grounded, but the canonical-issue rationale overstates what the packet supports. From the packet alone, we can say 43334 and 43525 are likely related because both report a missing `pad_token_id` on config objects, but not that 43334 is definitively the broader or more user-visible repro. That comparison isn\u2019t evidenced in the metadata provided. The accepted soft pair should also be phrased more conservatively as a likely duplicate, not as the same underlying regression with certainty." - }, - "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 }, @@ -7050,6 +7127,7 @@ "issue:37428", "issue:38175", "issue:39692", + "issue:42915", "issue:43262", "issue:43278", "issue:43335", @@ -7063,6 +7141,7 @@ "issue:43611", "issue:43653", "issue:43819", + "issue:43824", "issue:43825", "issue:43827", "issue:43856", @@ -7080,7 +7159,6 @@ "issue:44561", "issue:44610", "issue:44743", - "issue:44779", "issue:44792", "issue:44857", "issue:45083", @@ -7093,10 +7171,10 @@ "issue:45563", "issue:45584", "issue:45588", - "issue:45593", - "issue:45823" + "issue:45593" ], "soft_pairs": [ + "issue:42915|issue:43824", "issue:43278|issue:43856", "issue:44514|issue:44521", "issue:44442|issue:45198", @@ -7117,225 +7195,217 @@ "issue:44857|issue:45593", "issue:43278|issue:43825", "issue:45290|issue:45381", - "issue:45200|issue:45588", - "issue:43606|issue:45083", "issue:43335|issue:43550", + "issue:43606|issue:45083", + "issue:45200|issue:45588", "issue:44448|issue:44743", "issue:37428|issue:44351", "issue:43611|issue:43931", "issue:39692|issue:43526", - "issue:44561|issue:44779", - "issue:44462|issue:44561", - "issue:44206|issue:45823" + "issue:44462|issue:44561" ], - "prepared_review_unit_hash": "18de9ab4f8de30859fe2971137e3925996787b23209beb56aa6c8f1cd1e554f2", + "prepared_review_unit_hash": "ef307fc5de44a06f9e98b036f2836d749c0ddd4e97e1ab241a63cb8b35a4b22d", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36685, - "estimated_input_tokens": 9172, - "estimated_eval_tokens": 18600 + "soft_pair_count": 29, + "serialized_chars": 36392, + "estimated_input_tokens": 9098, + "estimated_eval_tokens": 18452 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a loose semantic grouping of unrelated Transformers issues, with only superficial title overlap. I found no soft-edge pair that is confidently the same underlying bug/change.", - "confidence": 0.93, - "canonical_issue_reason": "issue:44521 is the most central and broadly framed issue in the set: it is open, highly discussed, and concerns a common multimodal apply_chat_template path with multiple downstream references.", + "summary": "This cluster is highly heterogeneous: the soft pairs are mostly different bugs in different models/subsystems, so none should be merged as duplicates. All soft edges are rejected.", + "confidence": 0.97, + "canonical_issue_reason": "No clear canonical issue: the set covers unrelated bugs across imports, models, processors, and docs rather than one underlying defect.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44521 is the strongest representative issue because it is recent, active, and describes a user-facing multimodal template bug with wider applicability than the more model-specific reports.", + "best_issue_reason": "issue:44521 is the strongest standalone representative: it is open, discussion-heavy, and describes a concrete multimodal apply_chat_template failure.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:42915", + "right": "issue:43824", + "accept": false, + "reason": "Different problems: Qwen3Moe FP8 config failure vs a Qwen2.5-VL import error." + }, { "left": "issue:43278", "right": "issue:43856", "accept": false, - "reason": "Different domains and failure modes: dtype drift in evaluation vs MoE training memory usage." + "reason": "Unrelated symptoms: evaluation dtype drift vs Qwen3 MoE training memory usage." }, { "left": "issue:44514", "right": "issue:44521", "accept": false, - "reason": "Both involve apply_chat_template, but one is a batched-input crash and the other is incorrect assistant masks; not the same bug." + "reason": "Both involve apply_chat_template, but one is a batched padding crash and the other is all-zero assistant masks; different bugs." }, { "left": "issue:44442", "right": "issue:45198", "accept": false, - "reason": "Different tokenizers/models and different breakages; no evidence they share a code path." + "reason": "Both are tokenizer-related, but they affect different model families with different failure modes." }, { "left": "issue:38175", "right": "issue:43493", "accept": false, - "reason": "Both mention SigLIP2, but one reports zero probabilities and the other a broader HF-vs-JAX discrepancy; too loose to merge." + "reason": "Same model family, but zero probabilities and HF-vs-JAX discrepancy are not clearly the same concrete bug." }, { "left": "issue:43827", "right": "issue:43901", "accept": false, - "reason": "Both are docs-related v5 issues, but they target different pipelines and different API behaviors." + "reason": "Both are docs issues around pipelines, but they reference different APIs and different pages." }, { "left": "issue:43653", "right": "issue:44112", "accept": false, - "reason": "Unrelated: tokenizer special-token registration vs a GraniteSpeech CI device-override failure." + "reason": "A tokenizer special-token bug and a GraniteSpeech CI device-override failure are unrelated." }, { "left": "issue:43450", "right": "issue:43493", "accept": false, - "reason": "Video batching shape bug vs SigLIP2 implementation discrepancy; different subsystems and symptoms." + "reason": "Video batch shape bug vs SigLIP2 implementation discrepancy; different code paths." }, { "left": "issue:43450", "right": "issue:43575", "accept": false, - "reason": "Shape bug in video processors is unrelated to Qwen2-57B tensor-parallel OOM." + "reason": "Incorrect video processor output shape is unrelated to Qwen2-57B TP OOM." }, { "left": "issue:44792", "right": "issue:45563", "accept": false, - "reason": "Janus image-generation test failure and a paged-generate warning are not the same defect." + "reason": "A Janus image-generation test failure and a paged generate warning are different issues." }, { "left": "issue:44360", "right": "issue:44561", "accept": false, - "reason": "No meaningful overlap: DSA indexer discussion vs removal of is_torch_fx_available breaking trust_remote_code." + "reason": "DSA indexer discussion vs trust_remote_code breakage from removed helper; no shared defect." }, { "left": "issue:45127", "right": "issue:45245", "accept": false, - "reason": "LoRA/vocab collapse and category-count limit are unrelated model/data issues." + "reason": "LoRA/tied embeddings collapse is unrelated to the category-count runtime error." }, { "left": "issue:43819", "right": "issue:45083", "accept": false, - "reason": "DAC latent/forward mismatch is unrelated to qwen3_omni_moe feature-length helper behavior." + "reason": "DAC forward/latents mismatch and qwen3_omni_moe length helper behavior are separate bugs." }, { "left": "issue:44206", "right": "issue:45584", "accept": false, - "reason": "Audio feature-extractor crash and Whisper empty-transcription generation failure are different code paths." + "reason": "Whisper empty-transcription failure and LasrFeatureExtractor center-arg crash are different audio bugs." }, { "left": "issue:43819", "right": "issue:44112", "accept": false, - "reason": "Different models, different failures, no shared underlying bug." + "reason": "DAC reconstruction mismatch is unrelated to GraniteSpeech CI device override behavior." }, { "left": "issue:43262", "right": "issue:43450", "accept": false, - "reason": "Audio chat-template sampling-rate default is unrelated to batched video shape handling." + "reason": "Audio sampling-rate defaulting and video batch-shape issues are unrelated." }, { "left": "issue:44610", "right": "issue:44857", "accept": false, - "reason": "OmDet processor size mismatch and LwDetr float16 AMP crash are distinct vision bugs." + "reason": "Processor size mismatch and AMP/CUDA loss crash are different vision/detection failures." }, { "left": "issue:37428", "right": "issue:43425", "accept": false, - "reason": "ImportError for a flash-attention helper is unrelated to a Torch 2.10 compatibility report." + "reason": "Missing internal import and Torch 2.10 incompatibility are too broad and not the same bug." }, { "left": "issue:44857", "right": "issue:45593", "accept": false, - "reason": "Both are training/runtime issues, but they affect different models and mechanisms." + "reason": "A mixed-precision loss crash and missing auxiliary losses are unrelated D-FINE issues." }, { "left": "issue:43278", "right": "issue:43825", "accept": false, - "reason": "Embedding dtype regression and a pipeline v5 error-message issue are unrelated." + "reason": "Embedding dtype drift and a stale pipeline error message are different regression types." }, { "left": "issue:45290", "right": "issue:45381", "accept": false, - "reason": "Chat-template tool-call crash vs Qwen2.5-VL video position-id bug; no shared defect." + "reason": "Chat-template tool-call crash and Qwen2.5-VL vision_position_ids error are distinct multimodal bugs." }, { - "left": "issue:45200", - "right": "issue:45588", + "left": "issue:43335", + "right": "issue:43550", "accept": false, - "reason": "Gemma4 mm_token_type_ids defaulting and flash_attention sink-less s_aux handling are unrelated." + "reason": "SwitchTransformers config construction and Bamba torch.compile/SDPA failure are unrelated." }, { "left": "issue:43606", "right": "issue:45083", "accept": false, - "reason": "CPU offload device mismatch is not the same as qwen3_omni_moe feature-length helper behavior." + "reason": "CPU offload device mismatch and qwen3_omni_moe feature-length helper behavior are unrelated." }, { - "left": "issue:43335", - "right": "issue:43550", + "left": "issue:45200", + "right": "issue:45588", "accept": false, - "reason": "SwitchTransformers config bug and Bamba torch.compile/SDPA failure are different architecture-specific issues." + "reason": "Gemma mm_token_type_ids handling and flash_attention s_aux=None are different model-path failures." }, { "left": "issue:44448", "right": "issue:44743", "accept": false, - "reason": "Pegasus v4/v5 output drift is unrelated to qwen3_5 recurrent-state reset under cache." + "reason": "Pegasus output differences across versions and qwen3_5 recurrent-state reset are unrelated." }, { "left": "issue:37428", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but they involve different symbols and modules." + "reason": "Both are import errors, but they concern different missing symbols and likely different regressions." }, { "left": "issue:43611", "right": "issue:43931", "accept": false, - "reason": "Generic base_model_prefix loading breakage is not the same as a Qwen3-VL weight-shape mismatch." + "reason": "Base_model_prefix loading breakage and Qwen3-VL weight-shape mismatch are different loading problems." }, { "left": "issue:39692", "right": "issue:43526", "accept": false, - "reason": "SigLIP2 doc/example problems and BEiT FastImageProcessor label reduction are unrelated." - }, - { - "left": "issue:44561", - "right": "issue:44779", - "accept": false, - "reason": "Removal of a helper breaking trust_remote_code is unrelated to DeepSeek tokenizer correctness in v5." + "reason": "SigLIP2 doc example issues and BeitImageProcessorFast reduce_labels behavior are unrelated." }, { "left": "issue:44462", "right": "issue:44561", "accept": false, - "reason": "Tokenizer.json loading behavior and is_torch_fx_available removal are distinct problems." - }, - { - "left": "issue:44206", - "right": "issue:45823", - "accept": false, - "reason": "Both are device-related audio/model issues, but they target different components and failure modes." + "reason": "AutoTokenizer ignoring tokenizer.json and removed is_torch_fx_available are different loading/runtime regressions." } ] }, @@ -7370,7 +7440,6 @@ "issue:40444", "issue:43335", "issue:43388", - "issue:43412", "issue:43504", "issue:43526", "issue:43550", @@ -7379,12 +7448,13 @@ "issue:43701", "issue:43720", "issue:43819", - "issue:43873", "issue:43927", "issue:44112", + "issue:44206", "issue:44292", "issue:44448", "issue:44493", + "issue:44561", "issue:44568", "issue:44779", "issue:44855", @@ -7394,7 +7464,6 @@ "issue:45083", "issue:45198", "issue:45229", - "issue:45245", "issue:45310", "issue:45372", "issue:45459", @@ -7406,13 +7475,16 @@ "issue:45593", "issue:45698", "issue:45712", - "issue:45803" + "issue:45803", + "issue:45823" ], "soft_pairs": [ + "issue:44206|issue:45823", + "issue:44561|issue:44779", "issue:43388|issue:43526", "issue:43335|issue:43653", - "issue:45310|issue:45478", "issue:43819|issue:44493", + "issue:45310|issue:45478", "issue:43550|issue:44112", "issue:44855|issue:45083", "issue:29942|issue:32090", @@ -7434,268 +7506,246 @@ "issue:44857|issue:45072", "issue:29127|issue:32090", "issue:45198|issue:45803", - "issue:43927|issue:45698", "issue:44977|issue:45563", - "issue:45372|issue:45588", + "issue:43927|issue:45698", "issue:34689|issue:35141", + "issue:45372|issue:45588", "issue:40444|issue:43526", "issue:43701|issue:45083", - "issue:43550|issue:45083", - "issue:43412|issue:45588", - "issue:45245|issue:45563", - "issue:43873|issue:44493", - "issue:29942|issue:35141" + "issue:43550|issue:45083" ], - "prepared_review_unit_hash": "69a41f040521cbef1bd405f126b90d535e0de77e453067bba7a07fcef8d83966", + "prepared_review_unit_hash": "4e47f8eb728e23f5fd36f1cf01050aa4dc0d7d9044de3b02bcdd1533f48fe9fe", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 36, - "serialized_chars": 38766, - "estimated_input_tokens": 9692, - "estimated_eval_tokens": 19640 + "soft_pair_count": 34, + "serialized_chars": 38209, + "estimated_input_tokens": 9553, + "estimated_eval_tokens": 19362 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a heterogeneous false-positive cluster: the items span unrelated bugs across tokenizers, model loading, training, CI, and generation. Only one soft pair is a true duplicate (the Qwen3.5 MoE from_pretrained regression).", - "confidence": 0.95, - "canonical_issue_reason": "issue:45310 is the clearest canonical issue because it is the earlier report of the exact duplicate pair with 45478 and has the cleaner tracker position for the shared Qwen3.5 MoE from_pretrained bug.", + "summary": "Most items are unrelated bugs across different models/subsystems. Only 45310 and 45478 look like the same underlying regression; the rest should stay separate.", + "confidence": 0.94, + "canonical_issue_reason": "If a canonical issue is needed, issue 45310 is the best anchor because 45478 is effectively the same Qwen3.5 MoE from_pretrained error with the same title and context, and 45310 is the earlier, fuller report.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45310 is the best issue representative: it is the only strong duplicate hub in this set, while the rest of the cluster is too heterogeneous to merge into it.", + "best_issue_reason": "Issue 45310 is the strongest issue representative in this set, but only for the exact duplicate pair with 45478; the broader collection is too heterogeneous to treat as one issue cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44206", + "right": "issue:45823", + "accept": false, + "reason": "Different bugs: unsupported feature arg in LasrFeatureExtractor vs Gemma4 PLE device mismatch." + }, + { + "left": "issue:44561", + "right": "issue:44779", + "accept": false, + "reason": "Different code paths: v5 fx helper removal breaking trust_remote_code vs DeepSeek tokenizer output regression." + }, { "left": "issue:43388", "right": "issue:43526", "accept": false, - "reason": "Both mention label handling, but one is about gather_for_metrics truncating labels and the other about BeitImageProcessorFast reduce_labels; different bugs." + "reason": "Both involve labels, but one is gather_for_metrics batching logic and the other is BeitImageProcessorFast label reduction." }, { "left": "issue:43335", "right": "issue:43653", "accept": false, - "reason": "Different subsystems: SwitchTransformers config creation vs BigBirdTokenizer special-token registration." - }, - { - "left": "issue:45310", - "right": "issue:45478", - "accept": true, - "reason": "Same exact Qwen3.5 MoE from_pretrained error with identical title and version scope; clear duplicate." + "reason": "Unrelated: SwitchTransformers sparse-layer config bug vs BigBirdTokenizer special-token registration issue." }, { "left": "issue:43819", "right": "issue:44493", "accept": false, - "reason": "DAC latent/STE mismatch and unexpected position-id keys are unrelated code paths and symptoms." + "reason": "Different failures: DAC latent/STE mismatch vs unexpected position-id keys across models." + }, + { + "left": "issue:45310", + "right": "issue:45478", + "accept": true, + "reason": "Same issue title and same Qwen3.5 MoE from_pretrained regression; these are duplicates." }, { "left": "issue:43550", "right": "issue:44112", "accept": false, - "reason": "Torch.compile/SDPA model failure and GraniteSpeech CI device-override test failure are unrelated." + "reason": "Different model/subsystem: Bamba SDPA torch.compile bug vs GraniteSpeech CI device override test failure." }, { "left": "issue:44855", "right": "issue:45083", "accept": false, - "reason": "Python 3.13 import syntax error in DebertaV2 vs helper-length behavior in qwen3_omni_moe; different bugs." + "reason": "Different regressions: Python 3.13 import/IndentationError vs qwen3_omni_moe feature-length helper behavior." }, { "left": "issue:29942", "right": "issue:32090", "accept": false, - "reason": "Flash Attention 2 test failures are not the same bug as Trainer _gpu_broadcast_one NoneType errors." + "reason": "Flash Attention 2 test failures are unrelated to Trainer _gpu_broadcast_one NoneType broadcast errors." }, { "left": "issue:43335", "right": "issue:43504", "accept": false, - "reason": "Both involve model/config loading, but one is sparse-layer creation and the other is a legacy preset field for BEiT; not the same issue." + "reason": "Different models and causes: SwitchTransformers config vs BEiT semantic-segmentation legacy-field loading." }, { "left": "issue:43335", "right": "issue:45083", "accept": false, - "reason": "Config bug in SwitchTransformers vs helper output-length bug in qwen3_omni_moe; no shared failure mode." + "reason": "Different bugs in different components; same broad theme of model/config issues is not enough." }, { "left": "issue:39692", "right": "issue:44493", "accept": false, - "reason": "Documentation example errors and unexpected position-id keys are different problems." + "reason": "SigLIP2 doc/example errors are unrelated to the position-id key regression." }, { "left": "issue:44292", "right": "issue:45520", "accept": false, - "reason": "Qwen-3-8B-NVFP4 runtime failure and flash_attn import KeyError on Python 3.13 are unrelated." + "reason": "Different issues: Qwen-3 NVFP4 runtime vs import_utils flash_attn KeyError on Python 3.13." }, { "left": "issue:38175", "right": "issue:44493", "accept": false, - "reason": "Zero-probability behavior in SigLIP2 is not the same as unexpected position-id keys across many models." + "reason": "siglip2 zero-probability output bug is unrelated to position-id key warnings/errors." }, { "left": "issue:44448", "right": "issue:44779", "accept": false, - "reason": "Both are version regressions, but they affect different models and likely different tokenizer/generation paths." + "reason": "Different models and failures: Pegasus v4/v5 output drift vs DeepSeek tokenizer regression." }, { "left": "issue:28218", "right": "issue:35141", "accept": false, - "reason": "Tokenizer spacing after added tokens is unrelated to embedding reinitialization after resize." + "reason": "Tokenizer spacing bug vs embedding reinitialization after resize; distinct underlying problems." }, { "left": "issue:43335", "right": "issue:45072", "accept": false, - "reason": "SwitchTransformers sparse-layer creation and dtype mismatch in bfloat16 inference are different bugs." + "reason": "Different failure modes: sparse-layer config bug vs bfloat16 dtype mismatch in inference." }, { "left": "issue:43335", "right": "issue:45593", "accept": false, - "reason": "Sparse-layer config bug and D-FINE auxiliary-loss behavior are unrelated." + "reason": "SwitchTransformers config bug and D-FINE auxiliary-loss behavior are unrelated." }, { "left": "issue:43720", "right": "issue:44112", "accept": false, - "reason": "BitNet packed-weight loading issue and GraniteSpeech CI stale override failure do not match." + "reason": "Packed-weight loading bug is unrelated to GraniteSpeech stale device-override CI failure." }, { "left": "issue:44977", "right": "issue:45229", "accept": false, - "reason": "Qwen3.5 flash-attention generation bug and Gemma4 multi-GPU OOM are unrelated." + "reason": "Different problems: flash-attention generation issue vs Gemma4 multi-GPU OOM." }, { "left": "issue:44448", "right": "issue:44977", "accept": false, - "reason": "Different models and symptoms: Pegasus output drift vs Qwen3.5 flash-attention generation failure." + "reason": "Different models and symptoms; not the same generation bug." }, { "left": "issue:44568", "right": "issue:45491", "accept": false, - "reason": "Tokenizer special-token behavior and Gemma3 NaN embeddings under sliding-window attention are different issues." + "reason": "Tokenizer BOS/EOS omission is unrelated to Gemma3 NaN embeddings from sliding-window attention." }, { "left": "issue:43606", "right": "issue:44857", "accept": false, - "reason": "CPU-offload device mismatch and float16 AMP crash are separate hardware/runtime failures." + "reason": "CPU offload device mismatch and float16 AMP CUDA crash are distinct bugs." }, { "left": "issue:45459", "right": "issue:45712", "accept": false, - "reason": "Tokenizer exception masking and leftover dummy classes in dummy_pt_objects.py are unrelated maintenance bugs." + "reason": "Tokenizer error masking is unrelated to leftover dummy class cleanup/check_repo issues." }, { "left": "issue:44857", "right": "issue:45072", "accept": false, - "reason": "Precision-related failures, but on different models/code paths and with different root causes." + "reason": "Different components and symptoms: AMP crash in LwDetrImageLoss vs dtype mismatch in SwitchTransformers/TimmWrapperModel." }, { "left": "issue:29127", "right": "issue:32090", "accept": false, - "reason": "LayoutLMv3 error messaging and Trainer broadcast TypeError are unrelated." + "reason": "LayoutLMv3 error-message clarity issue is unrelated to Trainer broadcast TypeError." }, { "left": "issue:45198", "right": "issue:45803", "accept": false, - "reason": "Wav2Vec2 save/tokenization failure and bare-except error swallowing in Fuyu are not the same bug." + "reason": "Wav2Vec2 save/tokenization failure is unrelated to Fuyu bare-except exception handling." }, { "left": "issue:43927", "right": "issue:45698", "accept": false, - "reason": "Config custom-token-ID loss and wrong custom-module loading after save_pretrained are both save/load related, but the concrete failures differ." + "reason": "Both concern save/load, but one loses custom token IDs while the other loads the wrong custom module; different root causes." }, { - "left": "issue:44977", - "right": "issue:45563", + "left": "issue:34689", + "right": "issue:35141", "accept": false, - "reason": "Paged generate stale warning is not the same as Qwen3.5 flash-attention generation failure." + "reason": "Llama 3.2 Vision model-loading regression is unrelated to embedding reinitialization after resize." }, { "left": "issue:45372", "right": "issue:45588", "accept": false, - "reason": "Gemma4 processor ImportError and flash_attention s_aux=None AttributeError are different code paths." - }, - { - "left": "issue:34689", - "right": "issue:35141", - "accept": false, - "reason": "Llama 3.2 model-loading regression and embedding reinitialization after resize are distinct problems." + "reason": "Gemma processor import dependency break and flash_attention.py s_aux=None crash are different problems." }, { "left": "issue:40444", "right": "issue:43526", "accept": false, - "reason": "Qwen2.5-VL iterable-dataset finetuning failure and BEiT label reduction bug are unrelated." + "reason": "Qwen2.5-VL iterable multi-image finetuning failure is unrelated to BEiT label-reduction behavior." }, { "left": "issue:43701", "right": "issue:45083", "accept": false, - "reason": "Checkpoint key mismatch and qwen3_omni_moe feature-length behavior are unrelated." + "reason": "resume_from_checkpoint key mismatch is unrelated to qwen3_omni_moe feature-length helper behavior." }, { "left": "issue:43550", "right": "issue:45083", "accept": false, - "reason": "Bamba torch.compile/SDPA failure and qwen3_omni_moe helper behavior are unrelated." - }, - { - "left": "issue:43412", - "right": "issue:45588", - "accept": false, - "reason": "Gemma3n export guard/erfinv problem and sink-less flash_attention AttributeError are different issues." - }, - { - "left": "issue:45245", - "right": "issue:45563", - "accept": false, - "reason": "Category-count limit error and paged-generate stale warning are unrelated." - }, - { - "left": "issue:43873", - "right": "issue:44493", - "accept": false, - "reason": "Quantization/offloading failure and unexpected position-id keys are not the same bug." - }, - { - "left": "issue:29942", - "right": "issue:35141", - "accept": false, - "reason": "Flash Attention test failures and embedding reinitialization after resize are unrelated." + "reason": "Bamba torch.compile SDPA issue is unrelated to qwen3_omni_moe output-length helper behavior." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays mostly conservative. The one accepted soft edge (45310/45478) is well supported by the identical title and overlapping bug description, and the rest of the rejects are adequately cautious. Minor phrasing like \u201ccleaner tracker position\u201d is a bit subjective, but not enough to invalidate the overall result." + "feedback": "Grounded and conservative overall. The only accepted soft pair, issue:45310 vs issue:45478, is supported by the identical title and matching Qwen3.5 MoE from_pretrained regression. The remaining pair judgments stay appropriately cautious and do not overclaim shared root causes." }, "evaluator_used": true, "retried": false, @@ -7717,12 +7767,11 @@ "cluster_id": "cluster-15354-391", "nodes": [ "issue:25251", - "issue:29127", "issue:29942", "issue:30064", "issue:32090", "issue:33357", - "issue:34567", + "issue:35141", "issue:36010", "issue:37428", "issue:39692", @@ -7732,6 +7781,7 @@ "issue:43066", "issue:43278", "issue:43381", + "issue:43412", "issue:43421", "issue:43653", "issue:43873", @@ -7746,15 +7796,14 @@ "issue:44462", "issue:44466", "issue:44479", - "issue:44610", + "issue:44493", "issue:44734", "issue:44743", "issue:44857", "issue:44871", "issue:44977", - "issue:45103", "issue:45198", - "issue:45229", + "issue:45245", "issue:45372", "issue:45406", "issue:45412", @@ -7762,19 +7811,24 @@ "issue:45459", "issue:45491", "issue:45520", + "issue:45563", "issue:45588", "issue:45823" ], "soft_pairs": [ + "issue:43412|issue:45588", + "issue:45245|issue:45563", + "issue:43873|issue:44493", + "issue:29942|issue:35141", "issue:36010|issue:37428", "issue:45372|issue:45459", "issue:43066|issue:44462", - "issue:45588|issue:45823", "issue:44857|issue:45588", + "issue:45588|issue:45823", "issue:44448|issue:44466", "issue:43278|issue:43381", - "issue:25251|issue:32090", "issue:43065|issue:45412", + "issue:25251|issue:32090", "issue:43653|issue:44743", "issue:44451|issue:45520", "issue:43976|issue:45440", @@ -7789,208 +7843,204 @@ "issue:29942|issue:30064", "issue:43065|issue:43994", "issue:44734|issue:45406", - "issue:33357|issue:39692", - "issue:29127|issue:34567", - "issue:44610|issue:45412", - "issue:45103|issue:45372", - "issue:43873|issue:45229" + "issue:33357|issue:39692" ], - "prepared_review_unit_hash": "49dfa48c96732e9701788cc4ebe9caf758b7adcce7832bbd87f70517917d15fd", + "prepared_review_unit_hash": "a87ac3cf7f88d35d752e6b6b4cd1042ee21803b528f3a621811311a5e36e601d", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36204, - "estimated_input_tokens": 9051, - "estimated_eval_tokens": 18358 + "serialized_chars": 36149, + "estimated_input_tokens": 9038, + "estimated_eval_tokens": 18332 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is highly heterogeneous. All soft pairs are only loosely similar by subsystem or error type, not the same concrete bug/change, so none should be merged as duplicates.", - "confidence": 0.94, - "canonical_issue_reason": "Issue 43873 is the most active and broadly scoped runtime/regression report in the set, so it is the best single issue to anchor the cluster if one must be chosen.", + "summary": "This cluster is heterogeneous: the soft pairs span unrelated tokenizer, export, flash-attention, quantization, generation, and model-loading issues. None look like the same underlying bug or a mergeable duplicate PR pair, so all soft edges are rejected.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue stands out because the cluster is not a coherent duplicate set; the pairings cover many unrelated failure modes across different subsystems.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43873 has the strongest discussion/reference signals and is general enough to represent the cluster better than the narrower, model-specific reports.", + "best_issue_reason": "If a representative had to be chosen, issue 45588 is the most central among the flash-attention-adjacent subset, but it is still a poor global representative for the mixed cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43412", + "right": "issue:45588", + "accept": false, + "reason": "Gemma3n export failures and a sink-less flash_attention AttributeError are different code paths and symptoms." + }, + { + "left": "issue:45245", + "right": "issue:45563", + "accept": false, + "reason": "Category-count overflow in segmentation is unrelated to a stale num_return_sequences warning in paged generate()." + }, + { + "left": "issue:43873", + "right": "issue:44493", + "accept": false, + "reason": "Quantization/offloading behavior and an unexpected position_id key are distinct bugs with different triggers." + }, + { + "left": "issue:29942", + "right": "issue:35141", + "accept": false, + "reason": "Flash Attention 2 test failures do not match the token-embedding resize/post_init reinitialization bug." + }, { "left": "issue:36010", "right": "issue:37428", "accept": false, - "reason": "Both are import errors, but they involve different missing symbols in different modules and different failure paths." + "reason": "These are separate import errors for different symbols from different modules." }, { "left": "issue:45372", "right": "issue:45459", "accept": false, - "reason": "Gemma4 processor-loading import failure vs tokenizer protobuf error masking; related area, but not the same bug." + "reason": "Gemma 4 processor import breakage from mistral_common is unrelated to protobuf masking tokenizer errors." }, { "left": "issue:43066", "right": "issue:44462", "accept": false, - "reason": "Tokenizer decoding-type mismatch and AutoTokenizer ignoring tokenizer.json are distinct loading/serialization issues." + "reason": "Both involve tokenizers, but one is a decoder-type regression and the other is AutoTokenizer ignoring tokenizer.json; not the same fix." }, { - "left": "issue:45588", - "right": "issue:45823", + "left": "issue:44857", + "right": "issue:45588", "accept": false, - "reason": "Both involve Gemma4-related inference, but one is a flash_attention crash and the other is a device-map forward mismatch." + "reason": "An AMP/CUDA loss crash and a flash_attention s_aux=None AttributeError are different failures." }, { - "left": "issue:44857", - "right": "issue:45588", + "left": "issue:45588", + "right": "issue:45823", "accept": false, - "reason": "Different components and symptoms: AMP/CUDA loss crash vs flash_attention sinkless-model AttributeError." + "reason": "flash_attention sink-less crash and Gemma4 PLE device-mismatch are unrelated model/runtime issues." }, { "left": "issue:44448", "right": "issue:44466", "accept": false, - "reason": "One is output drift between v4/v5, the other is inconsistent serialization of tied weights; not the same issue." + "reason": "Pegasus output drift and lm_head.weight serialization inconsistency are both v5 regressions, but not the same concrete code-path bug." }, { "left": "issue:43278", "right": "issue:43381", "accept": false, - "reason": "Both mention eval-mode behavior, but one is dtype drift and the other is gradient checkpointing incompatibility." + "reason": "Embedding dtype drift in eval and gradient-checkpointing in eval mode are distinct behaviors." }, { - "left": "issue:25251", - "right": "issue:32090", + "left": "issue:43065", + "right": "issue:45412", "accept": false, - "reason": "Pipeline top_k nesting bug and Trainer GPU broadcast NoneType error are unrelated." + "reason": "A dummy Conv2d in Sam3PixelDecoder is unrelated to RT-DETR memory not releasing on deletion." }, { - "left": "issue:43065", - "right": "issue:45412", + "left": "issue:25251", + "right": "issue:32090", "accept": false, - "reason": "A dummy Conv2d in Sam3PixelDecoder and an RT-DETR memory leak are different bugs in different code paths." + "reason": "Pipeline top_k nesting and Trainer _gpu_broadcast_one NoneType errors are unrelated." }, { "left": "issue:43653", "right": "issue:44743", "accept": false, - "reason": "Tokenizer special-token registration bug vs recurrent-state reset bug; no shared concrete failure." + "reason": "BigBirdTokenizer special-token registration and Qwen modular recurrent-state reset are different tokenizer/model bugs." }, { "left": "issue:44451", "right": "issue:45520", "accept": false, - "reason": "Model loading failure for ScandiBERT and Python 3.13 flash_attn import error are unrelated." + "reason": "A model-loading failure for ScandiBERT and a Python 3.13 flash_attn import KeyError are unrelated." }, { "left": "issue:43976", "right": "issue:45440", "accept": false, - "reason": "Python version compatibility issue vs DeepseekV3MoE implementation divergence; different underlying problems." + "reason": "Python version compatibility and DeepseekV3MoE behavioral divergence are different issues." }, { "left": "issue:44186", "right": "issue:45198", "accept": false, - "reason": "Both are tokenizer-related, but one crashes on NER/padding and the other fails save_pretrained/tokenization for a different model." + "reason": "LayoutLMv2Tokenizer NER/padding crashes and Wav2Vec2 save_pretrained/tokenization failures are separate tokenizer bugs." }, { "left": "issue:42915", "right": "issue:42994", "accept": false, - "reason": "Quantization failures, but one is FP8 config inference and the other is saving quantized models." + "reason": "Qwen3Moe FineGrainedFP8Config failure and quantized model saving failure are different quantization problems." }, { "left": "issue:43994", "right": "issue:45412", "accept": false, - "reason": "SigLIP2 output correctness and RT-DETR memory release are unrelated." + "reason": "SigLIP2 AutoModel/pipeline nonsensical results and RT-DETR memory leaks do not share a bug path." }, { "left": "issue:43421", "right": "issue:45491", "accept": false, - "reason": "Runtime tokenizer post-processor update bug vs Gemma3 NaN embeddings in sliding-window attention." + "reason": "Runtime special-token post-processing updates and Gemma3 NaN embeddings with mixed-length batches are unrelated." }, { "left": "issue:44361", "right": "issue:45198", "accept": false, - "reason": "Different tokenizer failures for different models and task flows; not the same bug." + "reason": "MLukeTokenizer AttributeError on tasks and Wav2Vec2 save/tokenization failure are different model-specific issues." }, { "left": "issue:44871", "right": "issue:44977", "accept": false, - "reason": "Gemma eos_token configuration mismatch and Qwen3.5 flash-attention generation failure are unrelated." + "reason": "Gemma-3 EOS config inconsistency and Qwen3.5 flash-attention generation failure are unrelated." }, { "left": "issue:44265", "right": "issue:45198", "accept": false, - "reason": "torch.export failure due to torch_compilable_check is not the same as Wav2Vec2 tokenization/save_pretrained failure." + "reason": "torch.export failure with torch_compilable_check and Wav2Vec2 save/tokenization failure are not the same defect." }, { "left": "issue:44206", "right": "issue:44479", "accept": false, - "reason": "Different regressions: unsupported center argument in image processing vs video input regression in Qwen VL models." + "reason": "An unsupported feature-extractor arg and a video-input regression in Qwen models are separate issues." }, { "left": "issue:29942", "right": "issue:30064", "accept": false, - "reason": "Flash Attention 2 test failures and void segmentation-map processing are unrelated." + "reason": "Flash Attention 2 test failures are unrelated to void segmentation map processing." }, { "left": "issue:43065", "right": "issue:43994", "accept": false, - "reason": "Sam3PixelDecoder dummy layer issue and SigLIP2 incorrect outputs are different bugs." + "reason": "Sam3PixelDecoder's dummy Conv2d and SigLIP2 model/processor mismatch are different model-family bugs." }, { "left": "issue:44734", "right": "issue:45406", "accept": false, - "reason": "Serve KV-cache indexing crash and Gemma4Processor missing _tokenizer are distinct serve failures." + "reason": "KV-cache continuation indexing in serve and Gemma4Processor missing _tokenizer are distinct failures." }, { "left": "issue:33357", "right": "issue:39692", "accept": false, - "reason": "MacOS CLIP bus error and SigLIP2 documentation example errors are not the same concrete issue." - }, - { - "left": "issue:29127", - "right": "issue:34567", - "accept": false, - "reason": "LayoutLMv3 error messaging and TrainerState token-count tracking are unrelated." - }, - { - "left": "issue:44610", - "right": "issue:45412", - "accept": false, - "reason": "Processor image-size mismatch and RT-DETR memory leak are different bugs." - }, - { - "left": "issue:45103", - "right": "issue:45372", - "accept": false, - "reason": "auto_docstring annotation parsing crash and mistral_common import error are unrelated." - }, - { - "left": "issue:43873", - "right": "issue:45229", - "accept": false, - "reason": "Both are quantization/memory-related, but one is offloading behavior and the other is a multi-GPU CUDA OOM for Gemma4." + "reason": "A MacOS bus error on a community CLIP model and documentation example errors for SigLIP2 are not the same bug." } ] }, @@ -8018,7 +8068,7 @@ "issue:29942", "issue:32090", "issue:33357", - "issue:35141", + "issue:34567", "issue:36010", "issue:41720", "issue:42915", @@ -8028,43 +8078,46 @@ "issue:43334", "issue:43381", "issue:43441", - "issue:43452", "issue:43526", "issue:43531", "issue:43575", - "issue:43646", "issue:43653", "issue:43701", "issue:43723", "issue:43756", "issue:43824", + "issue:43873", "issue:43931", - "issue:44077", "issue:44186", "issue:44206", "issue:44279", - "issue:44387", "issue:44442", "issue:44464", "issue:44554", + "issue:44610", "issue:44743", "issue:44811", "issue:44857", "issue:44977", "issue:45072", + "issue:45103", "issue:45229", "issue:45245", "issue:45372", "issue:45406", + "issue:45412", "issue:45440", "issue:45491", "issue:45588", "issue:45593", - "issue:45803", "issue:45834" ], "soft_pairs": [ + "issue:29127|issue:34567", + "issue:44610|issue:45412", + "issue:45103|issue:45372", "issue:43756|issue:45440", + "issue:43873|issue:45229", "issue:43723|issue:44554", "issue:45372|issue:45406", "issue:43278|issue:43575", @@ -8072,230 +8125,219 @@ "issue:43526|issue:45834", "issue:43653|issue:44442", "issue:41720|issue:45229", - "issue:45440|issue:45593", "issue:43278|issue:43329", + "issue:45440|issue:45593", "issue:29942|issue:36010", "issue:43334|issue:43531", "issue:32090|issue:33357", - "issue:43824|issue:44279", "issue:43526|issue:44186", "issue:44977|issue:45245", + "issue:43824|issue:44279", "issue:43526|issue:44857", - "issue:44464|issue:45491", - "issue:43441|issue:45588", "issue:42915|issue:43931", "issue:43262|issue:44206", + "issue:44464|issue:45491", + "issue:43441|issue:45588", "issue:44442|issue:45072", "issue:44743|issue:44811", - "issue:43575|issue:43701", - "issue:43646|issue:44077", - "issue:43452|issue:45588", - "issue:44442|issue:45803", - "issue:29127|issue:35141", - "issue:43575|issue:44387" + "issue:43575|issue:43701" ], - "prepared_review_unit_hash": "3aaaeda56ad4ec83e1bd0d6c09df920b692acbbed695f58c6a74f317ab0c7bca", + "prepared_review_unit_hash": "172ebcef6fa69da1d861ff450aeb74e386fa38594486b83101c931a37de3ca32", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "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": 28, + "serialized_chars": 35524, + "estimated_input_tokens": 8881, + "estimated_eval_tokens": 18018 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These issues are a mixed set of unrelated Transformers bugs and regressions; none of the soft pairs are close enough to treat as duplicates or merge into one change.", - "confidence": 0.95, - "canonical_issue_reason": null, + "summary": "This cluster is highly heterogeneous: the soft-similarity pairs share broad vocabulary at most, but they describe distinct bugs, models, and code paths. I would not merge any of the suggested pairs as duplicates.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 45229 is the best anchor only by process criteria: it is a concrete, well-described open bug with high discussion/inbound activity, while the rest of the set appears unrelated.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45372 is the clearest actionable report in the set: it is open, has concrete traceback details, and has inbound references, so it is the best single issue to keep as a representative bug report.", + "best_issue_reason": "Issue 45229 is the strongest representative issue: detailed, reproducible, and actively discussed. It is still not a duplicate of the other items, just the most substantive single issue in the mixed cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:29127", + "right": "issue:34567", + "accept": false, + "reason": "LayoutLMv3 box-content error message vs TrainerState token counter update; different components and symptoms." + }, + { + "left": "issue:44610", + "right": "issue:45412", + "accept": false, + "reason": "Processor image-size mismatch vs RT-DETR memory not released on deletion; unrelated bugs." + }, + { + "left": "issue:45103", + "right": "issue:45372", + "accept": false, + "reason": "auto_docstring annotations crash vs Gemma4 processor dependency import failure; different failure modes." + }, { "left": "issue:43756", "right": "issue:45440", "accept": false, - "reason": "Different products and failures: Smollm3 RoPE-layer count bug vs DeepSeekV3 implementation divergence." + "reason": "Smollm3 RoPE-layer bug vs DeepSeekV3MoE implementation divergence; different model logic." + }, + { + "left": "issue:43873", + "right": "issue:45229", + "accept": false, + "reason": "Quantization offloading issue vs Gemma4 multi-GPU CUDA OOM; not the same code-path problem." }, { "left": "issue:43723", "right": "issue:44554", "accept": false, - "reason": "Tokenizer import/loading regression vs MPS attention correctness issue; different code paths." + "reason": "Tokenizer loading regression vs MPS attention correctness bug; unrelated areas." }, { "left": "issue:45372", "right": "issue:45406", "accept": false, - "reason": "Both involve Gemma4, but one is a mistral_common import dependency issue and the other is a missing _tokenizer attribute in serve." + "reason": "Gemma4 loading import error vs Gemma4Processor missing _tokenizer in serve; same family, different bugs." }, { "left": "issue:43278", "right": "issue:43575", "accept": false, - "reason": "Embedding dtype flip in eval vs tensor-parallel OOM on a large model; unrelated bugs." + "reason": "Embedding dtype drift in eval vs Qwen2-57B tensor-parallel OOM; distinct issues." }, { "left": "issue:43381", "right": "issue:43701", "accept": false, - "reason": "Gradient checkpointing in eval mode and resume_from_checkpoint key mismatch are distinct trainer behaviors." + "reason": "Gradient checkpointing eval-mode error vs checkpoint resume key mismatch; different training paths." }, { "left": "issue:43526", "right": "issue:45834", "accept": false, - "reason": "BeitImageProcessorFast reduce_labels bug and Kosmos2.5 OCR index error are different components and symptoms." + "reason": "BeitImageProcessorFast reduce_labels bug vs Kosmos2.5 long-ocr index error; unrelated processors." }, { "left": "issue:43653", "right": "issue:44442", "accept": false, - "reason": "Both are tokenizer-related, but the failing models and defects differ completely." + "reason": "BigBirdTokenizer special-token decode bug vs AutoTokenizer loading failure for a speech tokenizer; different defects." }, { "left": "issue:41720", "right": "issue:45229", "accept": false, - "reason": "Qwen3 auto device mapping cuda assert vs Gemma4 multi-GPU OOM; different root causes." + "reason": "Qwen3 auto device mapping cuda assert vs Gemma4 multi-GPU OOM; different models and failures." }, { - "left": "issue:45440", - "right": "issue:45593", + "left": "issue:43278", + "right": "issue:43329", "accept": false, - "reason": "DeepSeekV3 implementation divergence and D-FINE auxiliary-loss handling are unrelated model bugs." + "reason": "BF16-to-FP32 eval dtype issue vs multimodal token counting video-branch undefined variables; unrelated code paths." }, { - "left": "issue:43278", - "right": "issue:43329", + "left": "issue:45440", + "right": "issue:45593", "accept": false, - "reason": "Embedding dtype regression vs multimodal token counting undefined variables; no shared failure mode." + "reason": "DeepSeekV3MoE divergence vs D-FINE auxiliary-loss handling; different model behavior." }, { "left": "issue:29942", "right": "issue:36010", "accept": false, - "reason": "Flash Attention 2 test failures and GenerationMixin import error are unrelated regressions." + "reason": "Flash Attention 2 test failures vs GenerationMixin import error; unrelated regression types." }, { "left": "issue:43334", "right": "issue:43531", "accept": false, - "reason": "Qwen3-VL pad_token_id loading error vs Qwen3-MoE sliding_window issue; different model families and bugs." + "reason": "Qwen3-VL pad_token_id loading error vs Qwen3-MoE sliding_window issue; different configuration bugs." }, { "left": "issue:32090", "right": "issue:33357", "accept": false, - "reason": "Trainer broadcast NoneType error and MacOS bus error on CLIP are unrelated runtime failures." - }, - { - "left": "issue:43824", - "right": "issue:44279", - "accept": false, - "reason": "Import error for a specific model class vs a broad dependency issue; not the same underlying bug." + "reason": "Trainer _gpu_broadcast_one NoneType vs MacOS CLIP bus error; unrelated runtime failures." }, { "left": "issue:43526", "right": "issue:44186", "accept": false, - "reason": "ImageProcessorFast label reduction bug vs LayoutLMv2Tokenizer padding/truncation crash; different components." + "reason": "BeitImageProcessorFast label reduction bug vs LayoutLMv2Tokenizer padding/truncation crash; different components." }, { "left": "issue:44977", "right": "issue:45245", "accept": false, - "reason": "Flash-attention generation bug vs category-count limit error; no common code path." - }, - { - "left": "issue:43526", - "right": "issue:44857", - "accept": false, - "reason": "BeitImageProcessorFast label handling and LwDetrImageLoss AMP crash are unrelated." + "reason": "Qwen3.5 flash-attention generation bug vs category-count runtime limit; no shared underlying bug." }, { - "left": "issue:44464", - "right": "issue:45491", + "left": "issue:43824", + "right": "issue:44279", "accept": false, - "reason": "Compiled-forward chunked generation inconsistency and Gemma3 sliding-window NaNs are different generation bugs." + "reason": "Specific Qwen2.5-VL import error vs vague dependency issue; not the same concrete problem." }, { - "left": "issue:43441", - "right": "issue:45588", + "left": "issue:43526", + "right": "issue:44857", "accept": false, - "reason": "Both mention FlashAttention, but they hit different failures and model setups; too broad to merge." + "reason": "BeitImageProcessorFast label reduction bug vs LwDetrImageLoss AMP/CUDA crash; different models and code paths." }, { "left": "issue:42915", "right": "issue:43931", "accept": false, - "reason": "Qwen3MoE FP8 config failure and Qwen3-VL weight-shape mismatch are separate loading problems." + "reason": "Qwen3Moe FineGrainedFP8Config failure vs Qwen3-VL weight-shape mismatch; same family only broadly, not same bug." }, { "left": "issue:43262", "right": "issue:44206", "accept": false, - "reason": "Audio chat-template sampling-rate default and feature-extractor center-arg crash are unrelated audio bugs." - }, - { - "left": "issue:44442", - "right": "issue:45072", - "accept": false, - "reason": "Tokenizer loading failure and bf16 dtype mismatch are different bugs in different subsystems." - }, - { - "left": "issue:44743", - "right": "issue:44811", - "accept": false, - "reason": "Qwen3 recurrent state reset and Whisper batch_decode skip_special_tokens handling are distinct behaviors." - }, - { - "left": "issue:43575", - "right": "issue:43701", - "accept": false, - "reason": "Tensor-parallel OOM and checkpoint key mismatch do not describe the same failure." + "reason": "Audio chat-template sampling-rate default vs unsupported center arg in LasrFeatureExtractor; unrelated audio bugs." }, { - "left": "issue:43646", - "right": "issue:44077", + "left": "issue:44464", + "right": "issue:45491", "accept": false, - "reason": "General v5 custom initialization break and patchtsmixer post_init policy are not the same concrete bug." + "reason": "Compiled-forward chunked generation inconsistency vs Gemma3 NaN embeddings from sliding-window padding; different symptoms and models." }, { - "left": "issue:43452", + "left": "issue:43441", "right": "issue:45588", "accept": false, - "reason": "gguf_file loading breakage and flash_attention s_aux=None crash are unrelated." + "reason": "Ministral-3 flash-attention failure vs sink-less flash_attention.py s_aux=None crash; both mention flash attention but not the same code-path issue." }, { "left": "issue:44442", - "right": "issue:45803", + "right": "issue:45072", "accept": false, - "reason": "FastSpeech2ConformerTokenizer loading and FuyuBatchFeature exception handling are separate issues." + "reason": "FastSpeech2ConformerTokenizer loading failure vs dtype mismatch in bfloat16 inference; unrelated issues." }, { - "left": "issue:29127", - "right": "issue:35141", + "left": "issue:44743", + "right": "issue:44811", "accept": false, - "reason": "LayoutLMv3 error-message clarity and token-embedding resize reinitialization are unrelated." + "reason": "Qwen3_5 recurrent-state reset bug vs Whisper batch_decode skip_special_tokens regression; different components." }, { "left": "issue:43575", - "right": "issue:44387", + "right": "issue:43701", "accept": false, - "reason": "Large-model tp OOM and int4 quantization reserved-memory OOM have different triggers and fixes." + "reason": "Qwen2-57B tensor-parallel OOM vs resume_from_checkpoint key mismatch; no shared underlying bug." } ] }, @@ -8326,36 +8368,33 @@ "issue:33357", "issue:34567", "issue:34689", - "issue:36010", + "issue:35141", "issue:38175", "issue:39692", "issue:41628", "issue:42907", "issue:43316", "issue:43381", + "issue:43452", "issue:43493", "issue:43525", "issue:43550", + "issue:43575", "issue:43611", - "issue:43643", - "issue:43720", + "issue:43646", "issue:43761", "issue:43827", - "issue:43856", "issue:43873", "issue:43931", "issue:44077", "issue:44079", "issue:44188", - "issue:44190", + "issue:44387", "issue:44442", "issue:44479", - "issue:44496", "issue:44568", "issue:44743", - "issue:44871", "issue:44898", - "issue:44913", "issue:44964", "issue:44977", "issue:45216", @@ -8365,10 +8404,17 @@ "issue:45479", "issue:45491", "issue:45584", + "issue:45588", "issue:45698", - "issue:45759" + "issue:45759", + "issue:45803" ], "soft_pairs": [ + "issue:43646|issue:44077", + "issue:43452|issue:45588", + "issue:44442|issue:45803", + "issue:29127|issue:35141", + "issue:43575|issue:44387", "issue:44977|issue:45584", "issue:33357|issue:38175", "issue:43316|issue:43525", @@ -8381,226 +8427,200 @@ "issue:28218|issue:32090", "issue:33290|issue:34567", "issue:39692|issue:45491", - "issue:34689|issue:43611", "issue:44442|issue:44898", + "issue:34689|issue:43611", "issue:43761|issue:44188", "issue:43381|issue:43873", "issue:43493|issue:43827", "issue:43550|issue:44743", "issue:44568|issue:45479", "issue:43316|issue:45759", - "issue:41628|issue:45372", - "issue:43643|issue:44913", - "issue:43720|issue:44442", - "issue:32090|issue:33290", - "issue:44190|issue:44496", - "issue:43873|issue:44871", - "issue:34689|issue:36010", - "issue:43381|issue:43856", - "issue:44077|issue:44479" + "issue:41628|issue:45372" ], - "prepared_review_unit_hash": "e16c65d5eac128f80bf6e3e5b6a2e3c6ebd9cc0aa560e976501f7835914b48fe", + "prepared_review_unit_hash": "2ea87565962f40d5eaec5961bf7c676cfe94c331df6f12769902b75b44f9a920", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36696, - "estimated_input_tokens": 9174, - "estimated_eval_tokens": 18604 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 26, + "serialized_chars": 35204, + "estimated_input_tokens": 8801, + "estimated_eval_tokens": 17858 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is very heterogeneous: the soft pairs mostly share only broad keywords like loading, config, tokenizer, or multimodal models, but they do not look like the same underlying bug or change. I would keep them separate; no PRs are present.", - "confidence": 0.95, - "canonical_issue_reason": "No single issue cleanly represents this set: the items span unrelated tokenizer, config, model-loading, training, and documentation problems across different models and code paths.", + "summary": "All suggested duplicate edges look like soft lexical matches rather than the same underlying bug or change. The items span unrelated tokenizer, model-loading, quantization, config, and generation regressions, so there is no single coherent canonical issue for the cluster.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue fits this set: the items cover many unrelated bugs, and the proposed duplicate pairs are not the same concrete code-path problem.", "canonical_pr_reason": null, - "best_issue_reason": "If one issue must be chosen as the broadest representative, issue:45698 is the most generally framed load/save regression, but it still does not unify the cluster.", + "best_issue_reason": "No global best issue is suitable as a cluster representative because the cluster is heterogeneous and does not collapse to one bug.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43646", + "right": "issue:44077", + "accept": false, + "reason": "Both mention initialization/post_init, but one is about custom model init in Transformers 5.0 and the other is a patchtsmixer-specific policy change; not the same bug." + }, + { + "left": "issue:43452", + "right": "issue:45588", + "accept": false, + "reason": "One is about gguf_file loading for tokenizer/model classes; the other is a flash-attention crash on s_aux=None. Different code paths and failures." + }, + { + "left": "issue:44442", + "right": "issue:45803", + "accept": false, + "reason": "Tokenizer loading failure for FastSpeech2ConformerTokenizer is unrelated to FuyuBatchFeature swallowing exceptions during tensor conversion." + }, + { + "left": "issue:29127", + "right": "issue:35141", + "accept": false, + "reason": "LayoutLMv3 error-message clarity and embedding reinitialization in post_init are unrelated issues." + }, + { + "left": "issue:43575", + "right": "issue:44387", + "accept": false, + "reason": "Both are OOM reports, but one is tensor-parallel loading for Qwen2-57B and the other is increased reserved memory under int4 quantization; different causes and fixes." + }, { "left": "issue:44977", "right": "issue:45584", "accept": false, - "reason": "Both are generation failures, but one is Qwen3.5 flash-attention behavior and the other is Whisper empty-transcription handling after chat-template alignment; different code paths." + "reason": "Qwen3.5 flash-attention generation failure and Whisper empty-transcription failure after align_special_tokens are different generation paths." }, { "left": "issue:33357", "right": "issue:38175", "accept": false, - "reason": "Both involve vision-language models, but a MacOS bus error in CLIP is not the same bug as unexpected zero probabilities in SigLIP2." + "reason": "A MacOS bus error on a community CLIP model is not the same as zero probabilities on siglip2-base-patch16-224." }, { "left": "issue:43316", "right": "issue:43525", "accept": false, - "reason": "Both are config API inconsistencies, but they concern different models and different missing fields; not the same underlying issue." + "reason": "Different config API bugs: Gemma3TextConfig parity vs missing pad_token_id on Llama4Config." }, { "left": "issue:44964", "right": "issue:45440", "accept": false, - "reason": "Different models and different failures: Phi-4 multimodal loading vs DeepSeekV3MoE implementation divergence." + "reason": "Phi-4 multimodal loading regression and DeepseekV3MoE divergence from remote implementation are distinct model-specific issues." }, { "left": "issue:43761", "right": "issue:44079", "accept": false, - "reason": "CLIPVisionModel hidden_states output bug and ModelOutput key assignment bug are distinct output-plumbing issues." + "reason": "CLIPVisionModel returning hidden_states=None is not the same bug as ModelOutput key assignment when a prior value was None." }, { "left": "issue:44479", "right": "issue:45290", "accept": false, - "reason": "A video-input regression in Qwen multimodal models is unrelated to chat-template crashes on assistant tool-call messages." + "reason": "Video-input regression for Qwen VL models and chat-template crashes on tool-call assistant messages are separate failures." }, { "left": "issue:28218", "right": "issue:29127", "accept": false, - "reason": "Tokenizer spacing after added tokens and LayoutLMv3 error-message clarity are unrelated." + "reason": "Tokenizer adds an extra space after added tokens; the other is a LayoutLMv3 error-message issue. No shared code-path." }, { "left": "issue:42907", "right": "issue:45698", "accept": false, - "reason": "Both touch save/load behavior, but dequantized save failure and wrong custom-module loading are different bugs." + "reason": "Saving dequantized models and loading the wrong custom module after save_pretrained are related to persistence generally, but not the same defect." }, { "left": "issue:43931", "right": "issue:45216", "accept": false, - "reason": "One is a shape mismatch when loading Qwen3-VL weights; the other is an incorrect checkpoint saved by save_pretrained for Qwen3.5." + "reason": "Weight-shape mismatch on Qwen3-VL loading is different from incorrect saved checkpoints in Qwen3.5 save_pretrained." }, { "left": "issue:28218", "right": "issue:32090", "accept": false, - "reason": "Tokenizer spacing and Trainer _gpu_broadcast_one NoneType errors are unrelated." + "reason": "Tokenizer spacing behavior and a Trainer NoneType GPU broadcast TypeError are unrelated." }, { "left": "issue:33290", "right": "issue:34567", "accept": false, - "reason": "AdaFactor/DeepSpeed OOM and TrainerState token-count tracking are different problems." + "reason": "Adafactor/deepspeed OOM and TrainerState.num_input_tokens_seen not updating are different trainer/runtime issues." }, { "left": "issue:39692", "right": "issue:45491", "accept": false, - "reason": "Docs/example errors for SigLIP2 are not the same as Gemma3 NaN embeddings from sliding-window attention." + "reason": "SigLIP2 documentation example problems and Gemma3 NaN embeddings under mixed-length batching are unrelated." }, { - "left": "issue:34689", - "right": "issue:43611", + "left": "issue:44442", + "right": "issue:44898", "accept": false, - "reason": "Both are load regressions, but they have different root causes and model families." + "reason": "AutoTokenizer loading failure for FastSpeech2ConformerTokenizer is not the same as Perceiver classification failing at non-default resolution." }, { - "left": "issue:44442", - "right": "issue:44898", + "left": "issue:34689", + "right": "issue:43611", "accept": false, - "reason": "Tokenizer loading failure and Perceiver interpolation failure are unrelated." + "reason": "Both are loading-related, but one is a specific Llama 3.2 Vision regression and the other is a broad base_model_prefix compatibility break." }, { "left": "issue:43761", "right": "issue:44188", "accept": false, - "reason": "Hidden_states being None in CLIPVisionModel is unrelated to attention-kernel divergence under torch.compile." + "reason": "Hidden_states output handling and attention-kernel divergence under torch.compile are different layers and symptoms." }, { "left": "issue:43381", "right": "issue:43873", "accept": false, - "reason": "Gradient-checkpointing eval-mode restriction and quantization/offloading behavior are different bugs." + "reason": "Gradient checkpointing in eval mode is unrelated to offloading failures under quantization." }, { "left": "issue:43493", "right": "issue:43827", "accept": false, - "reason": "SigLIP2 implementation mismatch is a code bug; pipeline mentions in docs after v5 removals are documentation cleanup." + "reason": "SigLIP2 implementation mismatch and docs still using pipeline() after v5 removals are not the same underlying bug." }, { "left": "issue:43550", "right": "issue:44743", "accept": false, - "reason": "Different models and different code paths: Bamba torch.compile+SDPA failure vs Qwen recurrent-state reset." + "reason": "torch.compile/SDPA failure in Bamba-9B-v2 is distinct from recurrent state reset behavior in modular_qwen3_5." }, { "left": "issue:44568", "right": "issue:45479", "accept": false, - "reason": "Tokenizer special-token addition regression and zero-loss classification behavior are unrelated." + "reason": "Missing BOS/EOS insertion for a tokenizer is unrelated to zero-loss behavior in sequence classification." }, { "left": "issue:43316", "right": "issue:45759", "accept": false, - "reason": "Config API parity for Gemma3TextConfig is not the same as AutoModelForCausalLM.from_config failing to unwrap text_config." + "reason": "Gemma3TextConfig API inconsistency and Qwen multimodal config unwrapping in from_config are separate configuration-loading issues." }, { "left": "issue:41628", "right": "issue:45372", "accept": false, - "reason": "AutoImageProcessor import failure and mistral_common ReasoningEffort import failure are unrelated import problems." - }, - { - "left": "issue:43643", - "right": "issue:44913", - "accept": false, - "reason": "Both are config reload bugs, but missing fields with trust_remote_code and rotary_pct resetting are different mechanisms." - }, - { - "left": "issue:43720", - "right": "issue:44442", - "accept": false, - "reason": "BitNet packed-weight loading and tokenizer loading for FastSpeech2ConformerTokenizer are unrelated." - }, - { - "left": "issue:32090", - "right": "issue:33290", - "accept": false, - "reason": "Trainer NoneType broadcast and AdaFactor+DeepSpeed OOM are unrelated training issues." - }, - { - "left": "issue:44190", - "right": "issue:44496", - "accept": false, - "reason": "Local dataset loading in an example script is unrelated to an unrecognized model type in config.json." - }, - { - "left": "issue:43873", - "right": "issue:44871", - "accept": false, - "reason": "Quantization/offloading behavior and Gemma3 eos_token_id mismatch are different bugs." - }, - { - "left": "issue:34689", - "right": "issue:36010", - "accept": false, - "reason": "Llama 3.2 model loading regression and missing GenerationMixin import are unrelated import/load failures." - }, - { - "left": "issue:43381", - "right": "issue:43856", - "accept": false, - "reason": "Eval-mode gradient checkpointing and MoE training memory inefficiency are different issues." - }, - { - "left": "issue:44077", - "right": "issue:44479", - "accept": false, - "reason": "PatchTSMixer post_init policy and Qwen video-input regression are unrelated." + "reason": "Missing AutoImageProcessor import and Gemma 4 processor loading failing on ReasoningEffort import are different import errors with different dependencies." } ] }, @@ -8627,11 +8647,11 @@ "issue:25251", "issue:28282", "issue:29127", + "issue:32090", "issue:33290", "issue:33357", "issue:34689", - "issue:35141", - "issue:36331", + "issue:36010", "issue:41950", "issue:42175", "issue:43065", @@ -8642,43 +8662,51 @@ "issue:43388", "issue:43404", "issue:43441", - "issue:43452", "issue:43475", "issue:43540", "issue:43550", - "issue:43575", "issue:43606", - "issue:43646", - "issue:43673", + "issue:43643", "issue:43701", + "issue:43720", "issue:43742", "issue:43761", "issue:43825", + "issue:43856", + "issue:43873", + "issue:44077", "issue:44162", + "issue:44190", + "issue:44442", "issue:44479", "issue:44496", "issue:44514", "issue:44617", - "issue:44743", "issue:44792", "issue:44855", "issue:44871", "issue:44877", "issue:44913", - "issue:44936", "issue:45127", "issue:45137", "issue:45200", "issue:45584", - "issue:45656", - "issue:45698" + "issue:45656" ], "soft_pairs": [ + "issue:43643|issue:44913", + "issue:43720|issue:44442", + "issue:32090|issue:33290", + "issue:44190|issue:44496", + "issue:43873|issue:44871", + "issue:34689|issue:36010", + "issue:43381|issue:43856", + "issue:44077|issue:44479", "issue:44855|issue:45137", "issue:33290|issue:45656", + "issue:29127|issue:33357", "issue:43316|issue:44877", "issue:43540|issue:44514", - "issue:29127|issue:33357", "issue:43381|issue:43742", "issue:42175|issue:43316", "issue:33290|issue:34689", @@ -8688,217 +8716,216 @@ "issue:43232|issue:43388", "issue:41950|issue:43475", "issue:43295|issue:43825", - "issue:45127|issue:45200", "issue:44871|issue:45584", + "issue:45127|issue:45200", "issue:43550|issue:43701", "issue:43606|issue:43701", "issue:43065|issue:44617", "issue:43295|issue:44871", - "issue:43761|issue:44479", - "issue:33290|issue:35141", - "issue:33290|issue:36331", - "issue:43673|issue:44743", - "issue:44936|issue:45698", - "issue:43452|issue:44496", - "issue:43575|issue:43742", - "issue:43646|issue:44913" + "issue:43761|issue:44479" ], - "prepared_review_unit_hash": "df25ba1578700c1e50f3655e5e3aa9c4f8ae24607e259bb8b87aadb8c0288f72", + "prepared_review_unit_hash": "47078b07b8b5ad6ca9d8e46fab6d1bce89a45b91cdbe6b208dd869a8fd16cd38", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 28, - "serialized_chars": 36114, - "estimated_input_tokens": 9029, - "estimated_eval_tokens": 18314 + "soft_pair_count": 29, + "serialized_chars": 36423, + "estimated_input_tokens": 9106, + "estimated_eval_tokens": 18468 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a heterogeneous mix of unrelated Transformers bug reports and regressions. The soft pairs only share broad subsystem vocabulary, not the same concrete failure mode, so none should be merged as duplicates.", - "confidence": 0.97, - "canonical_issue_reason": "No single canonical issue stands out: the items span unrelated bugs across loading, generation, training, configs, and vision/video pipelines.", + "summary": "No duplicate cluster here: the proposed links are all false positives across unrelated bugs/regressions in configs, loading, generation, training, and pipelines.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue: the items are heterogeneous and do not form one underlying bug cluster.", "canonical_pr_reason": null, - "best_issue_reason": "issue:41950 is the cleanest single bug report in the set, but it is not a duplicate hub for the other issues.", + "best_issue_reason": "No strong global best issue; none of these issues is a duplicate hub, so a canonical representative would be arbitrary.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44855", - "right": "issue:45137", + "left": "issue:43643", + "right": "issue:44913", "accept": false, - "reason": "Python 3.13 import parsing error vs DeepSpeed ZeRO3 deque failure; unrelated bugs." + "reason": "Both are config/loading related, but one is trust_remote_code field loss and the other is GPTNeoX rotary_pct reload; different bugs." }, { - "left": "issue:33290", - "right": "issue:45656", + "left": "issue:43720", + "right": "issue:44442", "accept": false, - "reason": "OOM in Adafactor+DeepSpeed vs optimizer step called twice; different failure modes." + "reason": "Packed-weight accelerate loading vs AutoTokenizer failing on a specific tokenizer class; unrelated failure modes." }, { - "left": "issue:43316", - "right": "issue:44877", + "left": "issue:32090", + "right": "issue:33290", "accept": false, - "reason": "Gemma3TextConfig API mismatch vs strict granite_speech config loading; different config issues." + "reason": "Trainer NoneType broadcast error vs DeepSpeed+Adafactor OOM; different subsystems and symptoms." }, { - "left": "issue:43540", - "right": "issue:44514", + "left": "issue:44190", + "right": "issue:44496", + "accept": false, + "reason": "Local dataset script issue vs unrecognized model/config loading error; unrelated." + }, + { + "left": "issue:43873", + "right": "issue:44871", + "accept": false, + "reason": "Quantization/offload behavior vs Gemma eos_token_id mismatch; different code paths." + }, + { + "left": "issue:34689", + "right": "issue:36010", + "accept": false, + "reason": "Llama 3.2 Vision loading regression vs GenerationMixin import error; not the same bug." + }, + { + "left": "issue:43381", + "right": "issue:43856", + "accept": false, + "reason": "Gradient checkpointing eval-mode error vs Qwen3 MoE memory usage; different problems." + }, + { + "left": "issue:44077", + "right": "issue:44479", + "accept": false, + "reason": "patchtsmixer post_init policy change vs video input regression in Qwen models; unrelated." + }, + { + "left": "issue:44855", + "right": "issue:45137", + "accept": false, + "reason": "Python 3.13 decorator parsing issue vs DeepSpeed ZeRO3 deque underflow; unrelated." + }, + { + "left": "issue:33290", + "right": "issue:45656", "accept": false, - "reason": "Video input ValueError in Qwen3OmniMoe vs batched chat template crash in Qwen2_5_VLProcessor." + "reason": "OOM during Adafactor+DeepSpeed vs optimizer.step being called twice; different bugs." }, { "left": "issue:29127", "right": "issue:33357", "accept": false, - "reason": "LayoutLMv3 error-message clarity issue vs CLIP model bus error on macOS; unrelated." + "reason": "LayoutLMv3 box info validation vs MacOS bus error for CLIP; unrelated." + }, + { + "left": "issue:43316", + "right": "issue:44877", + "accept": false, + "reason": "Gemma3TextConfig API discrepancy vs strict config blocking granite_speech loading; different config issues." + }, + { + "left": "issue:43540", + "right": "issue:44514", + "accept": false, + "reason": "Both are Qwen multimodal video/chat issues, but they hit different inputs and code paths; not the same underlying bug." }, { "left": "issue:43381", "right": "issue:43742", "accept": false, - "reason": "Gradient checkpointing in eval mode vs key error loading MobileLLM-125M; no shared bug." + "reason": "Gradient checkpointing eval-mode issue vs MobileLLM key error on load; unrelated." }, { "left": "issue:42175", "right": "issue:43316", "accept": false, - "reason": "Packaging/backend dependency issue vs config API discrepancy; not the same defect." + "reason": "Packaging backend omission vs Gemma3TextConfig API inconsistency; unrelated." }, { "left": "issue:33290", "right": "issue:34689", "accept": false, - "reason": "DeepSpeed OOM with Adafactor vs Llama 3.2 Vision model-loading regression; unrelated." + "reason": "DeepSpeed Adafactor OOM vs Llama 3.2 Vision model-loading regression; different bugs." }, { "left": "issue:43404", "right": "issue:43441", "accept": false, - "reason": "Weight-tying bug in Mistral3 vs FlashAttention failure in Ministral-3; different code paths." + "reason": "lm_head weight tying bug vs FlashAttention failure in Ministral-3; not the same change." }, { "left": "issue:44162", "right": "issue:44792", "accept": false, - "reason": "ESM2 model breakage vs janus image-generation test failure; no concrete overlap." + "reason": "ESM2 breakage vs Janus image-generation test failure; unrelated." }, { "left": "issue:25251", "right": "issue:28282", "accept": false, - "reason": "Pipeline top_k nesting bug vs missing PyTorch ImportError; unrelated." + "reason": "Pipeline top_k nesting bug vs missing PyTorch ImportError; completely different." }, { "left": "issue:43232", "right": "issue:43388", "accept": false, - "reason": "Generation kwargs sync issue vs gather_for_metrics label truncation; different runtime paths." + "reason": "Generation kwargs update bug vs gather_for_metrics label truncation; different code paths." }, { "left": "issue:41950", "right": "issue:43475", "accept": false, - "reason": "Pipeline uses image processor incorrectly vs missing fpn_position_embeddings on Sam3 output; different models and failures." + "reason": "Video-classification processor lookup bug vs Sam3Video missing attribute; different model/pipeline issues." }, { "left": "issue:43295", "right": "issue:43825", "accept": false, - "reason": "Processor/tokenizer regression vs misleading translation-task error message; both v5-related but not the same bug." + "reason": "Processor/tokenizer regression vs incorrect pipeline translation error message; unrelated." }, { - "left": "issue:45127", - "right": "issue:45200", + "left": "issue:44871", + "right": "issue:45584", "accept": false, - "reason": "LoRA merge collapse with tied embeddings vs Gemma 4 mm_token_type_ids defaulting; unrelated model issues." + "reason": "Gemma-3 eos_token_id mismatch vs Whisper empty-transcription handling after token alignment; different bugs." }, { - "left": "issue:44871", - "right": "issue:45584", + "left": "issue:45127", + "right": "issue:45200", "accept": false, - "reason": "Gemma-3 eos_token_id inconsistency vs Whisper empty-transcription failure after align_special_tokens; different symptoms." + "reason": "LoRA merge with extended vocab collapse vs Gemma 4 mm_token_type_ids defaulting; unrelated." }, { "left": "issue:43550", "right": "issue:43701", "accept": false, - "reason": "torch.compile/SDPA failure vs checkpoint resume key mismatch; no shared underlying defect." + "reason": "Bamba torch.compile/SDPA failure vs resume_from_checkpoint key mismatch; unrelated." }, { "left": "issue:43606", "right": "issue:43701", "accept": false, - "reason": "CPU offload device mismatch vs checkpoint key mismatch; unrelated." + "reason": "CPU offload device mismatch vs resume_from_checkpoint key mismatch; different issues." }, { "left": "issue:43065", "right": "issue:44617", "accept": false, - "reason": "Dummy Conv2d implementation vs CUDA OOM in Sam3Video; one is architectural, the other memory-related." + "reason": "Sam3PixelDecoder dummy layer concern vs Sam3Video CUDA OOM; not the same bug." }, { "left": "issue:43295", "right": "issue:44871", "accept": false, - "reason": "Processor/tokenizer regression vs eos_token_id mismatch; both tokenizer-adjacent but not duplicate." + "reason": "Processor/tokenizer regression vs Gemma-3 eos_token_id inconsistency; unrelated." }, { "left": "issue:43761", "right": "issue:44479", "accept": false, - "reason": "CLIP hidden_states regression vs Qwen VL video-input regression; different models and failure modes." - }, - { - "left": "issue:33290", - "right": "issue:35141", - "accept": false, - "reason": "DeepSpeed OOM vs embedding reinitialization after resize; unrelated training vs model-init bugs." - }, - { - "left": "issue:33290", - "right": "issue:36331", - "accept": false, - "reason": "Optimizer memory issue vs Trainer compute_loss signature breakage; different layers of the stack." - }, - { - "left": "issue:43673", - "right": "issue:44743", - "accept": false, - "reason": "Generation cache missing in v5.0.0 vs Qwen3.5 recurrent-state reset; both cache-related but not the same bug." - }, - { - "left": "issue:44936", - "right": "issue:45698", - "accept": false, - "reason": "train/evaluate lifecycle failure vs wrong custom module loaded after save/load; unrelated trainer vs serialization issues." - }, - { - "left": "issue:43452", - "right": "issue:44496", - "accept": false, - "reason": "gguf_file loading regression vs unrecognized model_type config issue; both loading-related but distinct causes." - }, - { - "left": "issue:43575", - "right": "issue:43742", - "accept": false, - "reason": "tensor-parallel OOM vs model loading key error; not the same bug." - }, - { - "left": "issue:43646", - "right": "issue:44913", - "accept": false, - "reason": "Custom model initialization breakage in v5.0.0 vs rotary_pct not persisting on reload; different config/init defects." + "reason": "CLIPVisionModel hidden_states regression vs multi-model video input regression; different models and code paths." } ] }, @@ -8922,20 +8949,22 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-391", "nodes": [ + "issue:33290", + "issue:35141", + "issue:36331", "issue:41720", "issue:42617", "issue:43065", - "issue:43335", - "issue:43540", - "issue:43550", + "issue:43452", "issue:43575", - "issue:43644", + "issue:43646", + "issue:43673", + "issue:43742", "issue:43761", "issue:43827", "issue:43873", "issue:43906", "issue:43931", - "issue:43976", "issue:43994", "issue:44060", "issue:44077", @@ -8950,12 +8979,12 @@ "issue:44617", "issue:44734", "issue:44737", + "issue:44743", "issue:44792", "issue:44855", "issue:44871", - "issue:44877", "issue:44913", - "issue:44938", + "issue:44936", "issue:44964", "issue:44977", "issue:45127", @@ -8963,14 +8992,20 @@ "issue:45290", "issue:45357", "issue:45381", - "issue:45542", "issue:45584", "issue:45588", "issue:45600", - "issue:45803", - "issue:45834" + "issue:45698", + "issue:45803" ], "soft_pairs": [ + "issue:33290|issue:35141", + "issue:33290|issue:36331", + "issue:43673|issue:44743", + "issue:43452|issue:44496", + "issue:43575|issue:43742", + "issue:44936|issue:45698", + "issue:43646|issue:44913", "issue:42617|issue:43575", "issue:43761|issue:44242", "issue:43575|issue:43873", @@ -8980,8 +9015,8 @@ "issue:44479|issue:44734", "issue:44617|issue:45584", "issue:43827|issue:43994", - "issue:43906|issue:45290", "issue:44792|issue:44977", + "issue:43906|issue:45290", "issue:43065|issue:44610", "issue:44534|issue:44913", "issue:44190|issue:44964", @@ -8990,219 +9025,198 @@ "issue:44737|issue:45803", "issue:44855|issue:45588", "issue:44521|issue:45381", - "issue:45127|issue:45357", - "issue:43335|issue:44855", - "issue:44938|issue:45542", - "issue:43976|issue:44855", - "issue:43761|issue:45834", - "issue:44077|issue:44877", - "issue:44496|issue:45803", - "issue:43550|issue:44190", - "issue:43644|issue:44913", - "issue:43540|issue:44734" + "issue:45127|issue:45357" ], - "prepared_review_unit_hash": "5f177bf48ceadca8c9ed2b40561595ca439eef213ad096551d133928e81a27a5", + "prepared_review_unit_hash": "f439e8512982ffa495ebc8ac3a126c3e45a333f2d0041656995161125d455f1c", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 29, - "serialized_chars": 35784, - "estimated_input_tokens": 8946, - "estimated_eval_tokens": 18148 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35865, + "estimated_input_tokens": 8967, + "estimated_eval_tokens": 18190 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is broad and mostly heterogeneous: the soft pairs cover different models, different failure modes, and in several cases different layers of the stack (docs, config, generation, quantization, processors, etc.). I would not merge any of the candidate edges; no clear duplicate issues or same-change PRs are present.", - "confidence": 0.94, - "canonical_issue_reason": "issue:43873 is the most central active issue in the set by discussion and inbound references, and it describes a concrete runtime problem rather than a meta/config/doc-only report.", + "summary": "These are mostly unrelated issues that only look similar on the surface. I would reject every soft edge as a duplicate match; there is no single underlying bug/change spanning the set.", + "confidence": 0.95, + "canonical_issue_reason": "No true canonical issue emerges here: the items cover unrelated bugs, regressions, and model-specific failures across different subsystems.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43873 is the strongest representative issue here: it is open, well-discussed, and has the highest external signal among the listed issues.", + "best_issue_reason": "issue:43873 is the best generic triage anchor only because it is open and has the most discussion/inbound references, but it does not subsume the other items.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:33290", + "right": "issue:35141", + "accept": false, + "reason": "OOM in Adafactor/DeepSpeed is unrelated to embedding reinitialization after resize." + }, + { + "left": "issue:33290", + "right": "issue:36331", + "accept": false, + "reason": "Different failures: optimizer OOM versus Trainer API keyword mismatch." + }, + { + "left": "issue:43673", + "right": "issue:44743", + "accept": false, + "reason": "Both mention cache, but one is a missing GenerationMixin cache in v5.0.0 and the other is recurrent state reset in Qwen3.5." + }, + { + "left": "issue:43452", + "right": "issue:44496", + "accept": false, + "reason": "Model/tokenizer loading with gguf_file is a different bug from missing model_type config resolution." + }, + { + "left": "issue:43575", + "right": "issue:43742", + "accept": false, + "reason": "OOM loading a large Qwen model is not the same as a KeyError on a small MobileLLM load." + }, + { + "left": "issue:44936", + "right": "issue:45698", + "accept": false, + "reason": "Trainer evaluate-after-train behavior is unrelated to loading the wrong custom module after save_pretrained." + }, + { + "left": "issue:43646", + "right": "issue:44913", + "accept": false, + "reason": "Custom model initialization breakage is not the same as GPTNeoX rotary_pct not persisting on reload." + }, { "left": "issue:42617", "right": "issue:43575", "accept": false, - "reason": "Both are loading/runtime failures, but one is a 3D parallel script problem and the other is tensor-parallel OOM for a specific Qwen2 model; different code paths and symptoms." + "reason": "A 3d_parallel.py runtime issue is not the same as Qwen2 TP memory exhaustion." }, { "left": "issue:43761", "right": "issue:44242", "accept": false, - "reason": "CLIPVision hidden-states regression and MoE load-balancing loss gating are unrelated bugs in different model families." + "reason": "CLIPVision hidden_states regression and MoE load-balancing loss omission are different code paths." }, { "left": "issue:43575", "right": "issue:43873", "accept": false, - "reason": "Qwen2 TP OOM and offloading-with-quantization are both memory-related, but they are distinct mechanisms and not the same concrete bug." + "reason": "Qwen2 TP OOM and quantization offloading behavior are related only broadly to memory, not the same bug." }, { "left": "issue:44077", "right": "issue:44534", "accept": false, - "reason": "PatchTST post_init enforcement is unrelated to the non-persistent-buffer junk regression." + "reason": "Optional post_init policy and junk-filled non-persistent buffers are distinct v5 issues." }, { "left": "issue:43931", "right": "issue:44060", "accept": false, - "reason": "Shape-mismatch on Qwen3-VL loading is not the same as the Qwen3-Next tied-weights warning/config issue." + "reason": "Shape mismatch when loading Qwen3-VL is not the same as an incorrect tied-weights warning in Qwen3-Next." }, { "left": "issue:44464", "right": "issue:44871", "accept": false, - "reason": "Chunked generation under compiled forward and Gemma eos_token configuration are separate problems with no shared failing code path." + "reason": "Compiled-forward generation inconsistency is unrelated to eos_token_id config mismatch." }, { "left": "issue:44479", "right": "issue:44734", "accept": false, - "reason": "A multimodal video-input regression and a serve KV-cache continuation indexing crash are different failures in different subsystems." + "reason": "Video-input regression in Qwen VL models is different from KV-cache continuation tensor indexing in serve." }, { "left": "issue:44617", "right": "issue:45584", "accept": false, - "reason": "Sam3Video CUDA OOM and Whisper empty-transcription handling after token alignment are unrelated bugs." + "reason": "Sam3Video OOM and Whisper empty-transcription generation failure are unrelated." }, { "left": "issue:43827", "right": "issue:43994", "accept": false, - "reason": "Docs still referencing pipeline() and SigLIP2 producing nonsensical outputs are not the same underlying issue." + "reason": "Docs still referencing pipeline() is not the same as SIGLIP2 producing bad outputs with pipeline." }, { - "left": "issue:43906", - "right": "issue:45290", + "left": "issue:44792", + "right": "issue:44977", "accept": false, - "reason": "The former is only an isolated reproduction reference; the latter is a specific chat-template/tool-call crash. No concrete match is shown." + "reason": "Janus image-generation test failure and Qwen3.5 flash-attention generation issues are different model-specific bugs." }, { - "left": "issue:44792", - "right": "issue:44977", + "left": "issue:43906", + "right": "issue:45290", "accept": false, - "reason": "A janus image-generation test failure and Qwen3.5 flash-attention generation bug are different model-specific issues." + "reason": "An isolated reproduction of an existing issue is not the same as apply_chat_template crashing on tool-call messages." }, { "left": "issue:43065", "right": "issue:44610", "accept": false, - "reason": "Dummy Conv2d in Sam3PixelDecoder and OmDet-Turbo input-size mismatch are separate model/processor problems." + "reason": "A dummy Conv2d placeholder in Sam3PixelDecoder is unrelated to OmDet-Turbo processor size mismatch." }, { "left": "issue:44534", "right": "issue:44913", "accept": false, - "reason": "Non-persistent buffer initialization regression and GPTNeoX rotary_pct reload defaulting are different config/runtime bugs." + "reason": "Non-persistent buffer initialization junk is not the same as rotary_pct reverting on config reload." }, { "left": "issue:44190", "right": "issue:44964", "accept": false, - "reason": "Local dataset loading in a training script is unrelated to loading a specific multimodal model checkpoint." + "reason": "Local dataset loading failure in a training script is unrelated to loading a multimodal model." }, { "left": "issue:41720", "right": "issue:45245", "accept": false, - "reason": "A Qwen3 auto-device-map cudaErrorAssert and a category-count runtime limit are unrelated failure modes." + "reason": "CUDA assert on A800 with auto device mapping is a different failure from category-count overflow." }, { "left": "issue:44792", "right": "issue:45600", "accept": false, - "reason": "Janus test failure and stale auto-mapping references to removed Sam3LiteText configs are not the same issue." + "reason": "A janus generation test failure is unrelated to stale auto_mappings references breaking CI." }, { "left": "issue:44737", "right": "issue:45803", "accept": false, - "reason": "XLNet CPU allocation regression and Fuyu bare-except exception masking are entirely different code defects." + "reason": "CPU tensor creation in XLNet and bare except swallowing errors in Fuyu are unrelated bugs." }, { "left": "issue:44855", "right": "issue:45588", "accept": false, - "reason": "Python 3.13 indentation parsing failure in DebertaV2Model is unrelated to flash_attention handling of s_aux=None." + "reason": "Python 3.13 torch.jit parsing error is unrelated to flash_attention s_aux handling." }, { "left": "issue:44521", "right": "issue:45381", "accept": false, - "reason": "All-zero assistant masks in apply_chat_template and wrong vision_position_ids in Qwen2.5-VL are different multimodal bugs." + "reason": "All-zero assistant masks and wrong video vision_position_ids are different multimodal template bugs." }, { "left": "issue:45127", "right": "issue:45357", "accept": false, - "reason": "LoRA merge collapse with extended vocabulary is unrelated to incorrect visual-encoder keys saved by Qwen3.5." - }, - { - "left": "issue:43335", - "right": "issue:44855", - "accept": false, - "reason": "SwitchTransformers sparse-layer construction and a DebertaV2 Python parsing error do not share a code path." - }, - { - "left": "issue:44938", - "right": "issue:45542", - "accept": false, - "reason": "Python 3.14 import failure and missing TensorFlow backend handling are unrelated environment/setup issues." - }, - { - "left": "issue:43976", - "right": "issue:44855", - "accept": false, - "reason": "Python-version compatibility reporting and DebertaV2 script parsing are different bugs." - }, - { - "left": "issue:43761", - "right": "issue:45834", - "accept": false, - "reason": "CLIPVision hidden_states regression and Kosmos2.5 long-OCR index error are unrelated model-specific failures." - }, - { - "left": "issue:44077", - "right": "issue:44877", - "accept": false, - "reason": "PatchTST post_init policy and granite_speech strict-config loading are separate config/validation issues." - }, - { - "left": "issue:44496", - "right": "issue:45803", - "accept": false, - "reason": "Unrecognized model-type loading and bare except swallowing interrupts are unrelated." - }, - { - "left": "issue:43550", - "right": "issue:44190", - "accept": false, - "reason": "Bamba torch.compile SDPA failure and local dataset loading in an image-classification script are unrelated." - }, - { - "left": "issue:43644", - "right": "issue:44913", - "accept": false, - "reason": "Non-persistent buffer junk and GPTNeoX rotary_pct reload behavior are different defects." - }, - { - "left": "issue:43540", - "right": "issue:44734", - "accept": false, - "reason": "Qwen3OmniMoe video processing ValueError and serve KV-cache continuation indexing are different bugs in different flows." + "reason": "LoRA merge collapse with extended vocab is unrelated to saving incorrect visual encoder keys." } ] }, @@ -9230,52 +9244,61 @@ "issue:38175", "issue:38617", "issue:41628", - "issue:41950", - "issue:42907", "issue:43065", "issue:43116", "issue:43295", "issue:43335", - "issue:43388", "issue:43404", - "issue:43441", "issue:43479", + "issue:43540", + "issue:43550", "issue:43643", + "issue:43644", "issue:43650", - "issue:43673", - "issue:43688", - "issue:43854", - "issue:44016", + "issue:43761", + "issue:43976", + "issue:44077", "issue:44112", "issue:44186", "issue:44190", "issue:44220", "issue:44246", - "issue:44336", "issue:44464", "issue:44466", "issue:44496", "issue:44561", "issue:44683", "issue:44734", + "issue:44855", + "issue:44877", "issue:44913", "issue:44933", - "issue:44964", + "issue:44938", "issue:45003", "issue:45216", "issue:45245", "issue:45276", - "issue:45335", "issue:45362", "issue:45375", - "issue:45406", "issue:45446", + "issue:45542", "issue:45563", "issue:45584", "issue:45600", - "issue:45632" + "issue:45632", + "issue:45803", + "issue:45834" ], "soft_pairs": [ + "issue:43335|issue:44855", + "issue:44938|issue:45542", + "issue:44077|issue:44877", + "issue:43550|issue:44190", + "issue:44496|issue:45803", + "issue:43761|issue:45834", + "issue:43644|issue:44913", + "issue:43976|issue:44855", + "issue:43540|issue:44734", "issue:43650|issue:44220", "issue:45276|issue:45584", "issue:44112|issue:45600", @@ -9292,237 +9315,190 @@ "issue:44734|issue:45632", "issue:44683|issue:45446", "issue:38175|issue:43065", - "issue:44561|issue:45632", "issue:44466|issue:44913", + "issue:44561|issue:45632", "issue:44246|issue:45003", "issue:36246|issue:38617", - "issue:43643|issue:44496", - "issue:45406|issue:45632", - "issue:45563|issue:45584", - "issue:41950|issue:44016", - "issue:44964|issue:45600", - "issue:43388|issue:43688", - "issue:42907|issue:43441", - "issue:44112|issue:44964", - "issue:43295|issue:45335", - "issue:43854|issue:44964", - "issue:43673|issue:44336", - "issue:44016|issue:44220" + "issue:43643|issue:44496" ], - "prepared_review_unit_hash": "e9b35b4e0c65c9f107f9c112bab4bec0fcd12f4c11c98a2398117d4306410277", + "prepared_review_unit_hash": "81a4b020da0db0c47d6d4390cd2f9deb030510fed43cc1a7b1ac558580c344df", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 37370, - "estimated_input_tokens": 9343, - "estimated_eval_tokens": 18942 + "soft_pair_count": 30, + "serialized_chars": 36697, + "estimated_input_tokens": 9175, + "estimated_eval_tokens": 18606 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is very heterogeneous: mostly unrelated bug reports spanning import regressions, model-loading issues, generation/runtime failures, and CI/test problems. I would not merge any of the proposed soft edges; they are only superficially similar by subsystem or symptom.", - "confidence": 0.34, - "canonical_issue_reason": "Issue 41628 is the best anchor because it is a direct public-API import break (`AutoImageProcessor` from `transformers`) and is broadly representative of the import/export regression subset in this cluster. That said, the cluster as a whole is not a true duplicate set.", + "summary": "The cluster is mostly heterogeneous: several unrelated import/config/regression issues plus a few loosely similar model-loading bugs. No soft pair looks like the same underlying bug/change, so none are accepted as duplicates.", + "confidence": 0.39, + "canonical_issue_reason": "Issue 43643 is the broadest representative of the cluster\u2019s config/loading regression theme, but it is still not a duplicate of the other items.", "canonical_pr_reason": null, - "best_issue_reason": "41628 is the most canonical issue to keep because it describes a clear, user-facing top-level import failure with likely broad impact. It is more general and reusable as a reference than the model-specific or test-specific reports.", + "best_issue_reason": "43643 is the best single issue anchor because it is concrete, well-scoped, and closest to the recurring config/serialization-loading failures in this set.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43335", + "right": "issue:44855", + "accept": false, + "reason": "Different models and failure modes: SwitchTransformers sparse-layer construction vs a Python 3.13 import/parsing error in DebertaV2." + }, + { + "left": "issue:44938", + "right": "issue:45542", + "accept": false, + "reason": "Both are runtime/load problems, but one is a Python 3.14 import failure and the other is a backend-missing TensorFlow/tensorboard error; not the same bug." + }, + { + "left": "issue:44077", + "right": "issue:44877", + "accept": false, + "reason": "Both involve config handling, but one is PatchTSMixer post_init validation and the other is strict loading of granite_speech config; different concrete issues." + }, + { + "left": "issue:43550", + "right": "issue:44190", + "accept": false, + "reason": "Unrelated: Bamba torch.compile/SDPA failure vs example script failing to load a local dataset." + }, + { + "left": "issue:44496", + "right": "issue:45803", + "accept": false, + "reason": "Different code paths: missing model_type in config loading vs Fuyu tensor conversion swallowing exceptions." + }, + { + "left": "issue:43761", + "right": "issue:45834", + "accept": false, + "reason": "Different multimodal models and bugs: CLIPVisionModel hidden_states regression vs Kosmos2.5 OCR index error." + }, + { + "left": "issue:43644", + "right": "issue:44913", + "accept": false, + "reason": "Both concern saved state/config behavior, but one is non-persistent buffers junk in v5 and the other is GPTNeoX rotary_pct reload behavior; not the same bug." + }, + { + "left": "issue:43976", + "right": "issue:44855", + "accept": false, + "reason": "Same Python-version area, but the root causes differ: v5.1.0 load failure vs an import-time IndentationError in DebertaV2." + }, + { + "left": "issue:43540", + "right": "issue:44734", + "accept": false, + "reason": "Different subsystems: Qwen3OmniMoe video-processing validation vs serve KV-cache continuation indexing." + }, { "left": "issue:43650", "right": "issue:44220", "accept": false, - "reason": "Completely different problems: a vague data request vs. an audio feature extraction error." + "reason": "No meaningful overlap; one issue is essentially a placeholder/spam title and the other is a feature-specific audio feature extraction bug." }, { "left": "issue:45276", "right": "issue:45584", "accept": false, - "reason": "Different models and code paths: Gemma token embedding resize vs. Whisper empty transcription generation." + "reason": "Both are generation/model issues, but one is Gemma4 embedding resizing and the other is Whisper empty transcription after align_special_tokens; not the same bug." }, { "left": "issue:44112", "right": "issue:45600", "accept": false, - "reason": "Both mention CI, but one is a stale GraniteSpeech test and the other is an auto_mappings config reference break." + "reason": "Both mention CI, but the failing behavior is different: GraniteSpeech stale device override test vs auto_mappings referencing removed Sam3LiteText configs." }, { "left": "issue:43335", "right": "issue:44190", "accept": false, - "reason": "Unrelated: SwitchTransformers config construction vs. local dataset loading in an image-classification script." + "reason": "Different domains: SwitchTransformers config bug vs image-classification example script failing to load a local dataset." }, { "left": "issue:43116", "right": "issue:44190", "accept": false, - "reason": "Different example-script failures with different causes; no shared code-path is evident." + "reason": "Both are example-script issues, but one is multi-label classification output logic and the other is local dataset loading; different bugs." }, { "left": "issue:43295", "right": "issue:45276", "accept": false, - "reason": "A processor/tokenizer regression is not the same as Gemma token embedding resize behavior." + "reason": "Separate regressions: processor/tokenizer image-passing breakage vs Gemma4 token embedding resizing not propagating." }, { "left": "issue:45245", "right": "issue:45584", "accept": false, - "reason": "No concrete overlap: category-count limit vs. Whisper generation on empty transcription." + "reason": "Completely different failures: category cardinality overflow vs Whisper empty-transcription generation." }, { "left": "issue:43479", "right": "issue:45375", "accept": false, - "reason": "Both are config-related, but they affect different model families and different missing/default fields." + "reason": "Both are config-field problems, but one is default vision/audio config initialization from None and the other is a missing deepstack_visual_indexes field dropped by strict parsing." }, { "left": "issue:36246", "right": "issue:41628", "accept": false, - "reason": "Both are import errors, but they involve different missing symbols and different modules." + "reason": "Both are import/export problems, but they are for different symbols and paths: Qwen2_5_VLImageProcessor submodule import vs top-level AutoImageProcessor import." }, { "left": "issue:44933", "right": "issue:45362", "accept": false, - "reason": "An import-from-image_utils regression is unrelated to a Qwen chat/runtime crash." + "reason": "Different issues: a missing import from image_utils vs Qwen3.5 chat runtime crash." }, { "left": "issue:44464", "right": "issue:45563", "accept": false, - "reason": "Different generation issues: compiled-forward inconsistency vs. a stale warning in paged generate." + "reason": "Different generation bugs: inconsistent compiled-forward chunked generation vs a stale warning for num_return_sequences." }, { "left": "issue:43404", "right": "issue:45216", "accept": false, - "reason": "Tied-weight/serialization bugs on different models are not the same underlying defect." + "reason": "Different model/regression targets: tied lm_head weights in Mistral3ForConditionalGeneration vs incorrect Qwen3.5 save_pretrained checkpoint output." }, { "left": "issue:44186", "right": "issue:44734", "accept": false, - "reason": "Tokenizer padding/NER crash is unrelated to serve KV-cache continuation indexing." + "reason": "Different code paths: LayoutLMv2Tokenizer NER/padding crash vs transformers serve KV-cache continuation indexing crash." }, { "left": "issue:44734", "right": "issue:45632", "accept": false, - "reason": "Serve tensor indexing and trust_remote_code cache-path collisions are unrelated." + "reason": "Different problems entirely: serve tensor indexing vs trust_remote_code cache-path collisions." }, { "left": "issue:44683", "right": "issue:45446", "accept": false, - "reason": "Both touch flex_attention, but one is a torch>=2.9 failure and the other is a version-check/import bug; not the same concrete issue." + "reason": "Related to flex_attention, but one is a compiled runtime failure on torch>=2.9 and the other is a version-check/import gating bug; not the same bug." }, { "left": "issue:38175", "right": "issue:43065", "accept": false, - "reason": "Unrelated model behavior: unexpected probabilities vs. a dummy Conv2d in Sam3PixelDecoder." - }, - { - "left": "issue:44561", - "right": "issue:45632", - "accept": false, - "reason": "Different trust_remote_code regressions: removed helper API vs. local cache-path collision." - }, - { - "left": "issue:44466", - "right": "issue:44913", - "accept": false, - "reason": "Serialization of tied weights and GPTNeoX config reload are distinct bugs." - }, - { - "left": "issue:44246", - "right": "issue:45003", - "accept": false, - "reason": "Import taking long sometimes is not the same as unsafe sys.modules access, even if both relate to import-time behavior." - }, - { - "left": "issue:36246", - "right": "issue:38617", - "accept": false, - "reason": "Both are import failures, but for different names and different modules; no shared root cause is shown." - }, - { - "left": "issue:43643", - "right": "issue:44496", - "accept": false, - "reason": "Both are config-loading issues, but one loses fields with trust_remote_code and the other lacks a model_type." - }, - { - "left": "issue:45406", - "right": "issue:45632", - "accept": false, - "reason": "Gemma4Processor missing `_tokenizer` in serve is unrelated to trust_remote_code cache collisions." - }, - { - "left": "issue:45563", - "right": "issue:45584", - "accept": false, - "reason": "A stale warning about `num_return_sequences` is not the same as Whisper failing on empty transcription." - }, - { - "left": "issue:41950", - "right": "issue:44016", - "accept": false, - "reason": "A video-classification pipeline bug is unrelated to a notebook syntax error." - }, - { - "left": "issue:44964", - "right": "issue:45600", - "accept": false, - "reason": "Different model-loading failures: Phi-4 multimodal loading vs. stale Sam3LiteText config references in CI." - }, - { - "left": "issue:43388", - "right": "issue:43688", - "accept": false, - "reason": "Different training/evaluation bugs affecting different loss or metric paths." - }, - { - "left": "issue:42907", - "right": "issue:43441", - "accept": false, - "reason": "Same model family, but dequantized save failure and FlashAttention runtime failure are distinct code-path problems." - }, - { - "left": "issue:44112", - "right": "issue:44964", - "accept": false, - "reason": "CI test flakiness in GraniteSpeech and Phi-4 multimodal load failures are unrelated." - }, - { - "left": "issue:43854", - "right": "issue:44964", - "accept": false, - "reason": "Different model-specific loading failures; no evidence of a shared underlying bug." - }, - { - "left": "issue:43673", - "right": "issue:44336", - "accept": false, - "reason": "Generation cache behavior and ANSI output formatting are unrelated." - }, - { - "left": "issue:44016", - "right": "issue:44220", - "accept": false, - "reason": "A notebook syntax error has nothing to do with `_torch_extract_fbank_features()`." + "reason": "Unrelated model bugs: siglip2 zero probabilities vs Sam3PixelDecoder dummy Conv2d behavior." } ] }, @@ -9546,22 +9522,21 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-391", "nodes": [ - "issue:32090", - "issue:35141", - "issue:38175", - "issue:39692", "issue:41950", "issue:42175", + "issue:42907", "issue:43295", "issue:43299", "issue:43381", "issue:43388", + "issue:43441", "issue:43479", "issue:43525", "issue:43526", "issue:43531", "issue:43577", - "issue:43650", + "issue:43673", + "issue:43688", "issue:43824", "issue:43827", "issue:43854", @@ -9570,15 +9545,16 @@ "issue:44016", "issue:44062", "issue:44075", + "issue:44112", + "issue:44220", "issue:44242", "issue:44279", + "issue:44336", "issue:44351", "issue:44368", "issue:44496", - "issue:44617", - "issue:44683", - "issue:44779", "issue:44933", + "issue:44964", "issue:45020", "issue:45042", "issue:45200", @@ -9587,14 +9563,26 @@ "issue:45335", "issue:45357", "issue:45405", - "issue:45542", + "issue:45406", + "issue:45563", "issue:45584", "issue:45593", "issue:45600", - "issue:45759", - "issue:45823" + "issue:45632", + "issue:45759" ], "soft_pairs": [ + "issue:45406|issue:45632", + "issue:45563|issue:45584", + "issue:41950|issue:44016", + "issue:44964|issue:45600", + "issue:43388|issue:43688", + "issue:42907|issue:43441", + "issue:44112|issue:44964", + "issue:43295|issue:45335", + "issue:43854|issue:44964", + "issue:44016|issue:44220", + "issue:43673|issue:44336", "issue:44496|issue:45042", "issue:43299|issue:45216", "issue:43531|issue:45600", @@ -9611,216 +9599,225 @@ "issue:44016|issue:44062", "issue:43388|issue:43827", "issue:41950|issue:45020", - "issue:43479|issue:43525", "issue:43976|issue:45405", "issue:43577|issue:45216", - "issue:43824|issue:44933", - "issue:41950|issue:45823", - "issue:44683|issue:45542", - "issue:43299|issue:43854", - "issue:44279|issue:45405", - "issue:38175|issue:44617", - "issue:39692|issue:43901", - "issue:32090|issue:35141", - "issue:41950|issue:43650", - "issue:43295|issue:44779" + "issue:43479|issue:43525", + "issue:43824|issue:44933" ], - "prepared_review_unit_hash": "b51853ca65a81952fcad3a3fe19a2ffcfd527528eddd3922eff915a168d38965", + "prepared_review_unit_hash": "72b52a4670ebfd826be10249859445674f1bccc6d04598e5e66344d336c0e074", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 29, - "serialized_chars": 35807, - "estimated_input_tokens": 8952, - "estimated_eval_tokens": 18160 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36992, + "estimated_input_tokens": 9248, + "estimated_eval_tokens": 18752 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is highly heterogeneous: loading regressions, config bugs, docs issues, dependency problems, and unrelated model-specific failures. Under strict duplicate triage, none of the soft pairs look like the same concrete bug/change.", - "confidence": 0.93, - "canonical_issue_reason": "No single issue cleanly subsumes the set; the items span unrelated problems rather than one duplicate cluster.", + "summary": "These issues are mostly unrelated Transformers regressions spanning model loading, processor/config handling, docs, and training/runtime behavior. No soft pair looks like the same underlying bug or change.", + "confidence": 0.92, + "canonical_issue_reason": "issue:45278 is the broadest open regression report in the set and best serves as an umbrella representative for the cluster\u2019s general post-upgrade import/load breakage theme.", "canonical_pr_reason": null, - "best_issue_reason": "No clear best issue for canonicalization because the cluster does not converge on one underlying defect.", + "best_issue_reason": "issue:45278 is the strongest cluster anchor: open, generic, and closer to the repeated import/error reports than the more model-specific bugs.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:45406", + "right": "issue:45632", + "accept": false, + "reason": "Different failures: Gemma4Processor missing _tokenizer vs trust_remote_code cache-path collision." + }, + { + "left": "issue:45563", + "right": "issue:45584", + "accept": false, + "reason": "A stale warning about num_return_sequences is unrelated to Whisper failing on empty transcription after align_special_tokens." + }, + { + "left": "issue:41950", + "right": "issue:44016", + "accept": false, + "reason": "Video-classification processor lookup is unrelated to a notebook syntax error." + }, + { + "left": "issue:44964", + "right": "issue:45600", + "accept": false, + "reason": "Phi-4 multimodal loading is a different bug than auto_mappings referencing removed Sam3LiteText configs." + }, + { + "left": "issue:43388", + "right": "issue:43688", + "accept": false, + "reason": "Batch label dropping in gather_for_metrics is unrelated to auxiliary-loss normalization in OLMoE/GPT Oss." + }, + { + "left": "issue:42907", + "right": "issue:43441", + "accept": false, + "reason": "Dequantized save failure for Ministrals/Devstrals is not the same as FlashAttention loading failure for Ministral-3." + }, + { + "left": "issue:44112", + "right": "issue:44964", + "accept": false, + "reason": "A stale GraniteSpeech CI test is unrelated to Phi-4 multimodal model loading." + }, + { + "left": "issue:43295", + "right": "issue:45335", + "accept": false, + "reason": "processor.tokenizer / image passing regression is not the same as t5gemma embedding resize not propagating to decoder.embed_tokens." + }, + { + "left": "issue:43854", + "right": "issue:44964", + "accept": false, + "reason": "GLM-4.7-Flash unit-test loading failure is a different model/config issue than Phi-4 multimodal loading." + }, + { + "left": "issue:44016", + "right": "issue:44220", + "accept": false, + "reason": "Notebook syntax error and _torch_extract_fbank_features() bug are unrelated." + }, + { + "left": "issue:43673", + "right": "issue:44336", + "accept": false, + "reason": "Generation cache missing in chunked_prefill is unrelated to ANSI codes in loading_report." + }, { "left": "issue:44496", "right": "issue:45042", "accept": false, - "reason": "Different failures: model loading/config validation vs PIL image processor dependency handling." + "reason": "Unrecognized model/config handling is unrelated to PIL image processors unnecessarily requiring torchvision." }, { "left": "issue:43299", "right": "issue:45216", "accept": false, - "reason": "Both involve Qwen, but one is a loading regression and the other is a save_pretrained checkpoint-key bug." + "reason": "Qwen3VL model loading failure and Qwen3.5 save_pretrained regression are different code paths." }, { "left": "issue:43531", "right": "issue:45600", "accept": false, - "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to stale auto_mappings references for Sam3LiteText configs." + "reason": "Qwen3-MoE sliding_window behavior is unrelated to stale Sam3LiteText auto_mappings references." }, { "left": "issue:43526", "right": "issue:43901", "accept": false, - "reason": "ImageProcessor label reduction bug vs documentation mismatch about return_all_scores." + "reason": "BeitImageProcessorFast reduce_labels bug is unrelated to a docs/behavior mismatch in TextClassificationPipeline." }, { "left": "issue:42175", "right": "issue:44279", "accept": false, - "reason": "Specific backend install omission vs a vague dependency complaint; not the same underlying bug." + "reason": "Backend packaging dependency omission is too broad and different from a generic transformers dependency issue." }, { "left": "issue:45335", "right": "issue:45584", "accept": false, - "reason": "Different code paths: token embedding resize propagation vs Whisper empty-transcription handling." + "reason": "Embedding resize propagation in t5gemma is unrelated to Whisper empty-transcription generation failure." }, { "left": "issue:44075", "right": "issue:44368", "accept": false, - "reason": "Optimizer argument usage bug is unrelated to a tie_word_embeddings warning." + "reason": "SGD args not being used is unrelated to a tie_word_embeddings warning in Qwen3.5." }, { "left": "issue:43381", "right": "issue:45593", "accept": false, - "reason": "Gradient-checkpointing eval-mode restriction is unrelated to D-FINE auxiliary-loss behavior." + "reason": "Gradient checkpointing in eval mode and D-FINE auxiliary-loss handling are different bugs." }, { "left": "issue:43299", "right": "issue:45357", "accept": false, - "reason": "Different Qwen3.5 regressions: model loading vs incorrect visual-encoder keys on save." + "reason": "Both touch Qwen3.5, but one is a loading bug and the other is a save_pretrained key regression; not the same fix." }, { "left": "issue:43901", "right": "issue:44496", "accept": false, - "reason": "Docs note about pipeline output options is unrelated to an unrecognized-model loading error." + "reason": "A docs mismatch about return_all_scores is unrelated to unrecognized model config loading." }, { "left": "issue:44496", "right": "issue:45759", "accept": false, - "reason": "Unrecognized model/config registration issue vs from_config text_config unwrapping for composite multimodal configs." + "reason": "Unrecognized model/config errors are not the same as AutoModelForCausalLM.from_config failing to unwrap text_config for composite multimodal configs." }, { "left": "issue:44351", "right": "issue:45278", "accept": false, - "reason": "Specific missing HybridCache export vs broad post-upgrade import failures; too umbrella-like to be the same duplicate." + "reason": "Missing HybridCache export is a specific import bug, while 45278 is a broad report of many import errors; not the same root cause." }, { "left": "issue:44242", "right": "issue:45200", "accept": false, - "reason": "Missing load-balancing loss and mm_token_type_ids defaulting are different Mixture-of-Experts/multimodal problems." + "reason": "Router-logit auxiliary-loss behavior is unrelated to mm_token_type_ids defaults in Gemma 4." }, { "left": "issue:44016", "right": "issue:44062", "accept": false, - "reason": "Notebook syntax error is unrelated to tokenizers.AddedToken keyword duplication." + "reason": "Notebook syntax error and AddedToken special-argument TypeError are unrelated." }, { "left": "issue:43388", "right": "issue:43827", "accept": false, - "reason": "Metric gathering bug vs documentation still mentioning removed pipeline APIs." + "reason": "Metric label truncation is unrelated to docs still mentioning pipeline()." }, { "left": "issue:41950", "right": "issue:45020", "accept": false, - "reason": "Video-classification processor lookup bug is not the same as remote_code compatibility regressions." - }, - { - "left": "issue:43479", - "right": "issue:43525", - "accept": false, - "reason": "Multimodal config default initialization vs missing pad_token_id on Llama4Config are distinct issues." + "reason": "Image-processor lookup in video classification is not the same as the broad remote_code breakage report." }, { "left": "issue:43976", "right": "issue:45405", "accept": false, - "reason": "Python-version compatibility break vs unreleased PEFT version bump." + "reason": "Python version compatibility and unreleased MIN_PEFT_VERSION are unrelated packaging/versioning issues." }, { "left": "issue:43577", "right": "issue:45216", "accept": false, - "reason": "BLIP2 dtype propagation bug is unrelated to Qwen3.5 save_pretrained checkpoint corruption." + "reason": "Blip2 dtype propagation and Qwen3.5 checkpoint save regression are different model-specific bugs." }, { - "left": "issue:43824", - "right": "issue:44933", + "left": "issue:43479", + "right": "issue:43525", "accept": false, - "reason": "Both are import errors, but they are for different missing symbols and different root causes." + "reason": "Default multimodal config initialization is unrelated to Llama4Config missing pad_token_id." }, { - "left": "issue:44683", - "right": "issue:45542", + "left": "issue:43824", + "right": "issue:44933", "accept": false, - "reason": "Torch flex_attention compilation failure is unrelated to TensorBoard/TensorFlow backend installation." - }, - { - "left": "issue:43299", - "right": "issue:43854", - "accept": false, - "reason": "Qwen3VL loading regression and GLM-4.7-Flash test-loading failure are different model-specific bugs." - }, - { - "left": "issue:44279", - "right": "issue:45405", - "accept": false, - "reason": "Generic dependency issue is not the same as the specific MIN_PEFT_VERSION release problem." - }, - { - "left": "issue:38175", - "right": "issue:44617", - "accept": false, - "reason": "Unexpected zero probabilities in SigLIP2 is unrelated to Sam3Video CUDA out-of-memory." - }, - { - "left": "issue:39692", - "right": "issue:43901", - "accept": false, - "reason": "SigLIP2 example errors are unrelated to the TextClassificationPipeline docs mismatch." - }, - { - "left": "issue:32090", - "right": "issue:35141", - "accept": false, - "reason": "Trainer GPU broadcast NoneType error is unrelated to embedding reinitialization after resize_token_embeddings." - }, - { - "left": "issue:41950", - "right": "issue:43650", - "accept": false, - "reason": "Video-classification processor bug is unrelated to the placeholder 'ADD THE DATA' issue." - }, - { - "left": "issue:43295", - "right": "issue:44779", - "accept": false, - "reason": "Custom-model processor/tokenizer regression and Deepseek tokenizer regression affect different subsystems and code paths." + "reason": "Missing Qwen2_5_VL import and a nonexistent image_utils import are different import/export issues." } ] }, @@ -9844,35 +9841,35 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-391", "nodes": [ + "issue:32090", + "issue:35141", + "issue:38175", + "issue:39692", + "issue:41950", "issue:42994", "issue:43065", - "issue:43122", - "issue:43262", + "issue:43295", + "issue:43299", "issue:43317", "issue:43441", "issue:43450", "issue:43650", "issue:43653", - "issue:43673", "issue:43723", - "issue:43761", "issue:43784", + "issue:43854", "issue:43901", "issue:43976", - "issue:43994", "issue:44016", "issue:44162", - "issue:44188", "issue:44261", "issue:44263", "issue:44279", - "issue:44291", "issue:44297", - "issue:44336", - "issue:44361", "issue:44462", "issue:44466", "issue:44521", + "issue:44617", "issue:44623", "issue:44671", "issue:44683", @@ -9885,6 +9882,7 @@ "issue:45290", "issue:45325", "issue:45397", + "issue:45405", "issue:45542", "issue:45593", "issue:45600", @@ -9893,13 +9891,22 @@ "issue:45834" ], "soft_pairs": [ + "issue:41950|issue:45823", + "issue:44683|issue:45542", + "issue:43299|issue:43854", + "issue:44279|issue:45405", + "issue:38175|issue:44617", + "issue:39692|issue:43901", + "issue:32090|issue:35141", + "issue:41950|issue:43650", + "issue:43295|issue:44779", "issue:44462|issue:45081", "issue:44297|issue:44462", "issue:43976|issue:44279", "issue:45600|issue:45712", "issue:43976|issue:44987", - "issue:43441|issue:45834", "issue:44521|issue:45325", + "issue:43441|issue:45834", "issue:44462|issue:44779", "issue:43450|issue:44279", "issue:42994|issue:43317", @@ -9910,238 +9917,233 @@ "issue:43650|issue:45290", "issue:43723|issue:44162", "issue:44623|issue:44749", - "issue:44623|issue:45397", "issue:43653|issue:45593", + "issue:44623|issue:45397", "issue:44671|issue:44779", "issue:43784|issue:45542", - "issue:44016|issue:45823", - "issue:44987|issue:45542", - "issue:44361|issue:45593", - "issue:43901|issue:43994", - "issue:43122|issue:43650", - "issue:44016|issue:44291", - "issue:43262|issue:44521", - "issue:43673|issue:43761", - "issue:44188|issue:44336" + "issue:44016|issue:45823" ], - "prepared_review_unit_hash": "029288061b46e3b1f78f3cabf12aadd9131a49059ddb958092dcd2db90422d8d", + "prepared_review_unit_hash": "dea1b6d1ffcf15dce0bf29c24243a125d7c563f957024d87ba1fc4162f977c2e", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 30, - "serialized_chars": 36182, - "estimated_input_tokens": 9046, - "estimated_eval_tokens": 18348 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36943, + "estimated_input_tokens": 9236, + "estimated_eval_tokens": 18728 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a mixed issue set with one plausible tokenizer-loader duplicate pair; the rest are separate bugs across unrelated subsystems. No PRs were provided.", - "confidence": 0.74, - "canonical_issue_reason": "Issue 44462 is the best anchor: it states a concrete tokenizer-loading regression in v5 and is broad enough to cover downstream incorrect-tokenization reports like model-specific failures.", + "summary": "This is a heterogeneous cluster: mostly tokenizer/processor/model-loading regressions in Transformers v5, plus several unrelated vision, quantization, video, and runtime issues. None of the listed soft pairs look like true duplicates or the same code change, so all soft edges should be rejected. No PRs are present.", + "confidence": 0.55, + "canonical_issue_reason": "issue:44462 is the clearest representative of the tokenizer-loading regression subset: it describes a concrete AutoTokenizer v5 failure with repository-provided tokenizer.json and broad user impact.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44462 is the strongest representative because it names a likely root cause, is actionable, and is more general than the model-specific symptom reports.", + "best_issue_reason": "Best overall issue candidate because it is specific, reproducible, and broad enough to anchor the tokenizer-loading subgroup, even though the overall cluster is not a tight duplicate set.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:41950", + "right": "issue:45823", + "accept": false, + "reason": "Different bugs: video-classification processor lookup vs Gemma4 PLE device-mismatch during forward." + }, + { + "left": "issue:44683", + "right": "issue:45542", + "accept": false, + "reason": "Unrelated failures: compiled flex_attention on torch>=2.9 vs tensorboard/TensorFlow backend installation issue." + }, + { + "left": "issue:43299", + "right": "issue:43854", + "accept": false, + "reason": "Both are model-loading problems, but for different models and different failure modes." + }, + { + "left": "issue:44279", + "right": "issue:45405", + "accept": false, + "reason": "Dependency/version complaints, but one is a generic transformers dependency issue and the other is a PEFT minimum-version bump." + }, + { + "left": "issue:38175", + "right": "issue:44617", + "accept": false, + "reason": "Different model/runtime symptoms: SIGLIP2 zero probabilities vs Sam3Video CUDA OOM." + }, + { + "left": "issue:39692", + "right": "issue:43901", + "accept": false, + "reason": "Documentation/example errors versus a pipeline docs behavior mismatch; not the same bug." + }, + { + "left": "issue:32090", + "right": "issue:35141", + "accept": false, + "reason": "Training GPU broadcast TypeError is unrelated to embedding reinitialization after resize_token_embeddings." + }, + { + "left": "issue:41950", + "right": "issue:43650", + "accept": false, + "reason": "Video-classification processor bug is unrelated to a placeholder issue title with no technical overlap." + }, + { + "left": "issue:43295", + "right": "issue:44779", + "accept": false, + "reason": "Different tokenizer regressions: processor.tokenizer/image handling vs Deepseek tokenizer output correctness." + }, { "left": "issue:44462", "right": "issue:45081", "accept": false, - "reason": "Both are tokenizer-related, but one is a general loading regression and the other is a Mistral regex patch crash; different code paths and failure modes." + "reason": "Both involve tokenizer loading, but one is AutoTokenizer ignoring tokenizer.json and the other is a Mistral regex patch crash." }, { "left": "issue:44297", "right": "issue:44462", "accept": false, - "reason": "Serialization of tokenizer_class metadata is distinct from AutoTokenizer ignoring tokenizer.json; related area, not the same bug." + "reason": "Serialization mismatch in tokenizer_config.json is a different path from AutoTokenizer ignoring tokenizer.json." }, { "left": "issue:43976", "right": "issue:44279", "accept": false, - "reason": "One is a Python version compatibility issue, the other is a generic dependency complaint; too broad and not the same underlying problem." + "reason": "Python version support issue vs generic dependency issue; not the same underlying bug." }, { "left": "issue:45600", "right": "issue:45712", "accept": false, - "reason": "Both involve repo/CI hygiene, but one is stale auto_mappings references and the other is leftover dummy classes; separate fixes." + "reason": "Both are repo maintenance/CI-adjacent, but one is auto_mappings references and the other is dummy_pt_objects cleanup." }, { "left": "issue:43976", "right": "issue:44987", "accept": false, - "reason": "Different failures: interpreter support vs a specific model-loading regression." + "reason": "Different version breakages affecting different model loads; no shared code-path evidence." }, { - "left": "issue:43441", - "right": "issue:45834", + "left": "issue:44521", + "right": "issue:45325", "accept": false, - "reason": "Different model-specific bugs in unrelated paths; no shared underlying code defect is evident." + "reason": "apply_chat_template assistant mask bug is unrelated to Qwen2.5-VL rope index scaling." }, { - "left": "issue:44521", - "right": "issue:45325", + "left": "issue:43441", + "right": "issue:45834", "accept": false, - "reason": "Both touch multimodal/chat-template behavior, but one is assistant masks and the other is temporal RoPE indexing; not the same bug." + "reason": "FlashAttention failure for Ministral-3 is unrelated to Kosmos2.5 OCR index error." }, { "left": "issue:44462", "right": "issue:44779", - "accept": true, - "reason": "This looks like the same tokenizer-loading regression manifesting on Deepseek: wrong tokenizer selection/ignoring tokenizer.json can plausibly explain incorrect tokenization." + "accept": false, + "reason": "Both are tokenizer-related, but one is repository tokenizer loading and the other is incorrect Deepseek tokenization results." }, { "left": "issue:43450", "right": "issue:44279", "accept": false, - "reason": "Video processor batching shape bug is unrelated to a dependency issue." + "reason": "Video processor batched-shape bug is unrelated to a generic dependency issue." }, { "left": "issue:42994", "right": "issue:43317", "accept": false, - "reason": "Both concern quantization/dequantization, but saving failure and offloaded loading failure are distinct code paths." + "reason": "Quantized model saving failure is a different problem from dequantized model loading with device_map=auto." }, { "left": "issue:44261", "right": "issue:44263", "accept": false, - "reason": "Different model internals and symptoms; not a duplicate pair." + "reason": "Precision/error discussion for MLA q_a_layernorm is unrelated to torch.split return values in GlmMoeDsaIndexer." }, { "left": "issue:44466", "right": "issue:45276", "accept": false, - "reason": "Serialization of tied weights and Gemma4 resize_token_embeddings behavior are separate issues." + "reason": "Both concern resizing/tied weights behavior, but one is serialization of lm_head.weight and the other is Gemma4 embed/output modules not updating." }, { "left": "issue:43065", "right": "issue:44898", "accept": false, - "reason": "Unrelated vision-model bugs; same broad domain only." + "reason": "Dummy Conv2d in Sam3PixelDecoder and Perceiver non-default resolution failure are different model-specific bugs." }, { "left": "issue:43441", "right": "issue:44683", "accept": false, - "reason": "Both are attention-kernel failures, but they hit different kernels and different triggering conditions." + "reason": "Different attention-path failures: Ministral-3 FlashAttention vs compiled flex_attention on newer torch." }, { "left": "issue:43650", "right": "issue:45290", "accept": false, - "reason": "Completely unrelated issues: one is a vague data request, the other is a chat-template crash." + "reason": "Placeholder issue title is unrelated to apply_chat_template crashing on tool-call assistant messages." }, { "left": "issue:43723", "right": "issue:44162", "accept": false, - "reason": "A tokenizer-loading regression and an ESM2 model breakage are not the same underlying issue." + "reason": "Tokenizer loading regression in v5 is unrelated to the ESM2 model breakage." }, { "left": "issue:44623", "right": "issue:44749", "accept": false, - "reason": "Processor save_pretrained missing files is unrelated to a performance regression report." + "reason": "processor.save_pretrained missing files is unrelated to a Chinese report about slower filtering after upgrade." }, { - "left": "issue:44623", - "right": "issue:45397", + "left": "issue:43653", + "right": "issue:45593", "accept": false, - "reason": "Different subsystems and failures; no duplicate evidence." + "reason": "BigBirdTokenizer special-token registration bug is unrelated to D-FINE auxiliary-loss behavior." }, { - "left": "issue:43653", - "right": "issue:45593", + "left": "issue:44623", + "right": "issue:45397", "accept": false, - "reason": "Tokenizer special-token registration and D-FINE auxiliary-loss behavior are unrelated." + "reason": "Both mention saving/loading-ish flows, but one is processor saving and the other is Gemma-4 zero3 from_pretrained." }, { "left": "issue:44671", "right": "issue:44779", "accept": false, - "reason": "Both are v5 regressions, but one is masked LM predictions and the other is tokenizer output; too different to merge." + "reason": "CamemBERT masked LM prediction regression is not the same as Deepseek tokenizer output regression." }, { "left": "issue:43784", "right": "issue:45542", "accept": false, - "reason": "Both are import/dependency flavored, but they concern different backends and different failure points." + "reason": "Missing nn import in sentence-transformers is unrelated to the TensorFlow backend/tensorboard installation issue." }, { "left": "issue:44016", "right": "issue:45823", "accept": false, "reason": "Notebook syntax error and Gemma4 device mismatch are unrelated." - }, - { - "left": "issue:44987", - "right": "issue:45542", - "accept": false, - "reason": "Specific model-loading regression vs TensorFlow backend error; different problems." - }, - { - "left": "issue:44361", - "right": "issue:45593", - "accept": false, - "reason": "Different model-specific bugs with no shared code path evident." - }, - { - "left": "issue:43901", - "right": "issue:43994", - "accept": false, - "reason": "Docs mismatch for a pipeline argument is not the same as a model producing bad results." - }, - { - "left": "issue:43122", - "right": "issue:43650", - "accept": false, - "reason": "Different topics entirely." - }, - { - "left": "issue:44016", - "right": "issue:44291", - "accept": false, - "reason": "Notebook syntax error and init_empty_weights TypeError are unrelated." - }, - { - "left": "issue:43262", - "right": "issue:44521", - "accept": false, - "reason": "Both involve apply_chat_template, but audio sampling-rate defaulting and multimodal assistant masks are distinct behaviors." - }, - { - "left": "issue:43673", - "right": "issue:43761", - "accept": false, - "reason": "Generation cache handling and CLIPVisionModel hidden_states are unrelated." - }, - { - "left": "issue:44188", - "right": "issue:44336", - "accept": false, - "reason": "Attention-kernel branching bug and ANSI-color logging are unrelated." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The clustering is mostly conservative and grounded in the packet. The only accepted soft edge, issue:44462 with issue:44779, is a plausible tokenizer-regression match, and the other rejections are appropriately cautious. No PR pair mergeability issues apply since no PRs were provided." - }, - "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 }, @@ -10165,61 +10167,69 @@ "issue:41950", "issue:42491", "issue:42757", + "issue:43122", + "issue:43262", "issue:43299", "issue:43352", "issue:43450", "issue:43525", "issue:43643", + "issue:43650", "issue:43673", + "issue:43761", "issue:43824", "issue:43881", "issue:43901", "issue:43976", - "issue:44162", + "issue:43994", + "issue:44016", + "issue:44188", "issue:44220", "issue:44242", "issue:44246", + "issue:44291", "issue:44295", "issue:44297", "issue:44336", "issue:44351", + "issue:44361", "issue:44373", "issue:44485", "issue:44496", + "issue:44521", "issue:44561", "issue:44589", "issue:44617", - "issue:44749", - "issue:44779", - "issue:44857", "issue:44871", - "issue:44877", "issue:44964", - "issue:45200", + "issue:44987", "issue:45250", - "issue:45292", "issue:45362", - "issue:45375", "issue:45397", - "issue:45561", - "issue:45563", - "issue:45584", - "issue:45593", - "issue:45701" + "issue:45542", + "issue:45593" ], "soft_pairs": [ + "issue:44361|issue:45593", + "issue:44987|issue:45542", + "issue:43901|issue:43994", + "issue:43122|issue:43650", + "issue:44016|issue:44291", + "issue:43262|issue:44521", + "issue:43673|issue:43761", + "issue:44188|issue:44336", "issue:43525|issue:43881", "issue:43525|issue:43643", "issue:44297|issue:45397", "issue:43450|issue:44617", "issue:42491|issue:43299", "issue:44589|issue:45362", - "issue:43901|issue:44246", "issue:41628|issue:43976", + "issue:43901|issue:44246", "issue:44336|issue:44373", "issue:44242|issue:45593", - "issue:43352|issue:45250", "issue:16998|issue:22355", + "issue:43352|issue:45250", "issue:44295|issue:45397", "issue:44246|issue:44496", "issue:41950|issue:44220", @@ -10229,231 +10239,216 @@ "issue:43673|issue:44561", "issue:43824|issue:43976", "issue:43976|issue:44964", - "issue:44246|issue:44871", - "issue:44242|issue:44857", - "issue:45362|issue:45375", - "issue:44162|issue:45200", - "issue:44295|issue:44749", - "issue:44779|issue:45292", - "issue:44877|issue:45375", - "issue:45292|issue:45584", - "issue:45292|issue:45701", - "issue:45561|issue:45563" + "issue:44246|issue:44871" ], - "prepared_review_unit_hash": "ca213bd53244f9822ccd87d2f95ee4ee3ac802655f67a84f7d197b630493580c", + "prepared_review_unit_hash": "b7d8353c3fc15026bda0ec89449862d647de1f74630d7e3d9fc27b30743de44b", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 36541, - "estimated_input_tokens": 9136, - "estimated_eval_tokens": 18528 + "soft_pair_count": 30, + "serialized_chars": 36378, + "estimated_input_tokens": 9095, + "estimated_eval_tokens": 18446 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is mostly a grab bag of unrelated transformer regressions and import/config issues. I don\u2019t see any strong duplicate PR candidates, and the soft issue pairs are mostly superficial overlaps (same subsystem or same model family, but different concrete failures), so I reject them all.", - "confidence": 0.61, - "canonical_issue_reason": "issue:43824 is the clearest, high-signal bug report in the set: a concrete, reproducible import failure with multiple inbound references and broad user impact.", + "summary": "The set is heterogeneous: mostly unrelated import/model-loading regressions plus a few docs/perf/UI issues. None of the soft edges is strong enough to merge, so there is no reliable duplicate cluster here.", + "confidence": 0.81, + "canonical_issue_reason": "No single issue cleanly subsumes the set; if one representative is needed, 44521 is the strongest standalone bug report because it is concrete, reproducible, and has the most discussion/inbound references among the active issues.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43824 is the best anchor for duplicate triage because it is specific, actionable, and appears to have attracted the most downstream attention.", + "best_issue_reason": "44521 is the best overall issue candidate: it has a clear failure mode, active discussion, and enough traction to serve as a representative bug report, even though it does not duplicate the others.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43525", - "right": "issue:43881", + "left": "issue:44361", + "right": "issue:45593", "accept": false, - "reason": "Both are loading/import failures, but they hit different models and likely different code paths (Llama4 config vs GLM-4V load failure)." + "reason": "Different bugs: MLukeTokenizer task-time AttributeError vs D-FINE missing auxiliary losses when denoising is off." }, { - "left": "issue:43525", - "right": "issue:43643", + "left": "issue:44987", + "right": "issue:45542", "accept": false, - "reason": "One is a missing config attribute on Llama4; the other is a trust_remote_code config field-loss bug. Similar theme, not the same bug." + "reason": "Unrelated failures: a transformers>=5.1.0 model-loading regression vs a TensorFlow backend/import issue." }, { - "left": "issue:44297", - "right": "issue:45397", + "left": "issue:43901", + "right": "issue:43994", "accept": false, - "reason": "Tokenizer config mismatch and ZeRO3 from_pretrained are unrelated failures." + "reason": "Docs mismatch for return_all_scores is unrelated to SigLIP2 nonsensical outputs." }, { - "left": "issue:43450", - "right": "issue:44617", + "left": "issue:43122", + "right": "issue:43650", "accept": false, - "reason": "Both mention video models, but one is a batched-shape bug and the other is CUDA OOM; different problems." + "reason": "Tokenization regression and a vague unrelated report; no shared code-path bug." }, { - "left": "issue:42491", - "right": "issue:43299", + "left": "issue:44016", + "right": "issue:44291", "accept": false, - "reason": "Both involve Qwen MoE variants, but one is LoRA compatibility across versions and the other is a model-loading regression for Qwen3VL MoE." + "reason": "Notebook syntax error and init_empty_weights argument mismatch are unrelated." }, { - "left": "issue:44589", - "right": "issue:45362", - "accept": false, - "reason": "A Float8 storage import/storage error is not the same as a Qwen3.5 chat crash." - }, - { - "left": "issue:43901", - "right": "issue:44246", + "left": "issue:43262", + "right": "issue:44521", "accept": false, - "reason": "Docs mismatch for return_all_scores and intermittent import slowness are unrelated." + "reason": "Both mention apply_chat_template, but one is audio sampling-rate defaulting and the other is multimodal assistant_mask generation; different bugs." }, { - "left": "issue:41628", - "right": "issue:43976", + "left": "issue:43673", + "right": "issue:43761", "accept": false, - "reason": "AutoImageProcessor import failure and Python-version compatibility regression are different issues." + "reason": "Generation cache regression and CLIP hidden_states regression are different model/code paths." }, { - "left": "issue:44336", - "right": "issue:44373", + "left": "issue:44188", + "right": "issue:44336", "accept": false, - "reason": "ANSI escape handling in loading_report and a wrong docstring for position_ids are not the same bug." + "reason": "Torch.compile attention branching bug vs ANSI codes in loading_report; unrelated." }, { - "left": "issue:44242", - "right": "issue:45593", + "left": "issue:43525", + "right": "issue:43881", "accept": false, - "reason": "Both concern missing losses, but they affect different models/features and different loss terms." + "reason": "Both are model-loading errors, but they affect different models and have different failure causes." }, { - "left": "issue:43352", - "right": "issue:45250", + "left": "issue:43525", + "right": "issue:43643", "accept": false, - "reason": "Model-specific Flash Attention support error is not the same as a generic Flash Attention issue." + "reason": "Missing pad_token_id on Llama4Config is not the same as trust_remote_code returning missing fields." }, { - "left": "issue:16998", - "right": "issue:22355", + "left": "issue:44297", + "right": "issue:45397", "accept": false, - "reason": "Model max length behavior and a missing transformers.onnx module are unrelated." + "reason": "Tokenizer save metadata mismatch is unrelated to gemma-4 zero3 from_pretrained loading failure." }, { - "left": "issue:44295", - "right": "issue:45397", + "left": "issue:43450", + "right": "issue:44617", "accept": false, - "reason": "Buffer registration/position_ids error and ZeRO3 loading regression are different failures." + "reason": "Batched video processor shape bug and CUDA OOM are not the same underlying problem." }, { - "left": "issue:44246", - "right": "issue:44496", + "left": "issue:42491", + "right": "issue:43299", "accept": false, - "reason": "Import slowness and unrecognized model/config errors do not share the same underlying cause." + "reason": "Both involve Qwen3/MoE and v5-era breakage, but one is LoRA compatibility and the other is base Qwen3VL model loading; not clearly the same fixable bug." }, { - "left": "issue:41950", - "right": "issue:44220", + "left": "issue:44589", + "right": "issue:45362", "accept": false, - "reason": "Video-classification processor selection and audio feature extraction are unrelated." + "reason": "Float8 storage import error and Qwen3.5-35B chat crash are unrelated." }, { - "left": "issue:36246", - "right": "issue:42757", + "left": "issue:41628", + "right": "issue:43976", "accept": false, - "reason": "These are different missing-symbol import errors from different packages/modules." + "reason": "AutoImageProcessor import error and Python-version compatibility regression are different issues." }, { - "left": "issue:36246", - "right": "issue:44351", + "left": "issue:43901", + "right": "issue:44246", "accept": false, - "reason": "Both are import errors, but they expose different missing symbols and likely different export regressions." + "reason": "Documentation mismatch vs import-time slowness; no overlap beyond being transformers issues." }, { "left": "issue:44336", - "right": "issue:44485", + "right": "issue:44373", "accept": false, - "reason": "Loading-report ANSI formatting and GLM-5 RoPE implementation are unrelated." + "reason": "ANSI output in loading_report and a wrong docstring are unrelated." }, { - "left": "issue:43673", - "right": "issue:44561", + "left": "issue:44242", + "right": "issue:45593", "accept": false, - "reason": "Generation cache missing during chunked prefill and removal of is_torch_fx_available are distinct internal API breakages." + "reason": "Both mention losses, but load-balancing loss in MoE and D-FINE auxiliary losses are distinct mechanisms." }, { - "left": "issue:43824", - "right": "issue:43976", + "left": "issue:16998", + "right": "issue:22355", "accept": false, - "reason": "A Qwen2.5-VL import error and a Python-version support issue are not the same underlying problem." + "reason": "A model_max_length question and a missing transformers.onnx module are unrelated." }, { - "left": "issue:43976", - "right": "issue:44964", + "left": "issue:43352", + "right": "issue:45250", "accept": false, - "reason": "Different symptoms: one is version/platform incompatibility, the other is a specific multimodal model load failure." + "reason": "45250 is too generic to be the same concrete bug as NemotronH not supporting Flash Attention 2.0 yet." }, { - "left": "issue:44246", - "right": "issue:44871", + "left": "issue:44295", + "right": "issue:45397", "accept": false, - "reason": "Import performance and Gemma eos_token_id inconsistency are unrelated." + "reason": "Reading a buffered position_ids tensor and gemma-4 zero3 loading are unrelated." }, { - "left": "issue:44242", - "right": "issue:44857", + "left": "issue:44246", + "right": "issue:44496", "accept": false, - "reason": "Load-balancing loss omission and LwDETR AMP crash are different model/training failures." + "reason": "Import latency and an unrecognized model/config error are different problems." }, { - "left": "issue:45362", - "right": "issue:45375", + "left": "issue:41950", + "right": "issue:44220", "accept": false, - "reason": "Same Qwen3.5 family, but one is a chat crash and the other is a config-field/schema issue; too far apart to merge." + "reason": "Video-classification looking for image processors and audio fbank extraction are different code paths." }, { - "left": "issue:44162", - "right": "issue:45200", + "left": "issue:36246", + "right": "issue:42757", "accept": false, - "reason": "ESM2 workflow breakage and Gemma 4 token-type defaults are unrelated." + "reason": "Different missing imports from different packages; same symptom class, not the same bug." }, { - "left": "issue:44295", - "right": "issue:44749", + "left": "issue:36246", + "right": "issue:44351", "accept": false, - "reason": "Position_ids buffer access and tokenizer slowdown after upgrade are unrelated." + "reason": "Both are import errors, but for different missing symbols and likely different export fixes." }, { - "left": "issue:44779", - "right": "issue:45292", + "left": "issue:44336", + "right": "issue:44485", "accept": false, - "reason": "Deepseek tokenizer regression and resize_token_embeddings not updating output embeddings are different bugs." + "reason": "Terminal ANSI output bug and GLM-5 RoPE discussion are unrelated." }, { - "left": "issue:44877", - "right": "issue:45375", + "left": "issue:43673", + "right": "issue:44561", "accept": false, - "reason": "Both touch strict config handling, but they affect different model configs and different failure modes; not enough to treat as the same bug." + "reason": "Generation cache issue and removal of is_torch_fx_available are different regression sources." }, { - "left": "issue:45292", - "right": "issue:45584", + "left": "issue:43824", + "right": "issue:43976", "accept": false, - "reason": "Embedding resize behavior and Whisper empty-transcription generation are unrelated." + "reason": "Qwen2.5-VL import error and Python-version support regression are not the same issue." }, { - "left": "issue:45292", - "right": "issue:45701", + "left": "issue:43976", + "right": "issue:44964", "accept": false, - "reason": "Tokenizer-version changes and embedding resize behavior do not share a concrete code-path bug." + "reason": "Broad packaging/version-compatibility failure vs a specific model-loading problem; different scope and fix." }, { - "left": "issue:45561", - "right": "issue:45563", + "left": "issue:44246", + "right": "issue:44871", "accept": false, - "reason": "A pytest-xdist race in testing utils and a stale generation warning are different issues entirely." + "reason": "Import slowness and eos_token_id config inconsistency are unrelated." } ] }, @@ -10488,24 +10483,22 @@ "issue:43526", "issue:43540", "issue:43596", - "issue:43646", "issue:43653", - "issue:43673", "issue:43901", - "issue:43931", "issue:44060", "issue:44077", "issue:44162", + "issue:44242", "issue:44246", - "issue:44297", + "issue:44295", "issue:44336", "issue:44368", "issue:44466", "issue:44496", - "issue:44623", - "issue:44734", - "issue:44869", - "issue:44933", + "issue:44749", + "issue:44779", + "issue:44857", + "issue:44877", "issue:44977", "issue:44987", "issue:44995", @@ -10513,21 +10506,32 @@ "issue:45137", "issue:45200", "issue:45229", + "issue:45292", "issue:45335", - "issue:45356", + "issue:45362", "issue:45375", "issue:45397", "issue:45459", "issue:45561", "issue:45563", + "issue:45584", "issue:45588", - "issue:45600", "issue:45656", + "issue:45701", "issue:45759" ], "soft_pairs": [ - "issue:44987|issue:45459", + "issue:44242|issue:44857", + "issue:45362|issue:45375", + "issue:44162|issue:45200", + "issue:44779|issue:45292", + "issue:44295|issue:44749", + "issue:44877|issue:45375", + "issue:45292|issue:45584", + "issue:45292|issue:45701", + "issue:45561|issue:45563", "issue:43901|issue:45003", + "issue:44987|issue:45459", "issue:45375|issue:45759", "issue:43232|issue:44336", "issue:36246|issue:43596", @@ -10545,219 +10549,210 @@ "issue:44368|issue:44977", "issue:44368|issue:45563", "issue:45397|issue:45656", - "issue:44060|issue:44077", - "issue:43673|issue:44734", - "issue:45003|issue:45561", - "issue:45137|issue:45600", - "issue:43931|issue:44368", - "issue:44869|issue:45356", - "issue:44933|issue:45459", - "issue:44297|issue:44623", - "issue:43646|issue:44162", - "issue:43232|issue:43673" + "issue:44060|issue:44077" ], - "prepared_review_unit_hash": "35a6b2bc05def2631ac5555e102f0398e95c70966fb2f877988d115500acdedf", + "prepared_review_unit_hash": "f78480df2eb971d0dd7e3f6ff3e0bffd1f0070bc89c79e2362119cb9dd08cce3", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 29, - "serialized_chars": 36059, - "estimated_input_tokens": 9015, - "estimated_eval_tokens": 18286 + "serialized_chars": 35948, + "estimated_input_tokens": 8987, + "estimated_eval_tokens": 18230 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These are not duplicates overall: the soft pairs span unrelated bugs across tokenization, generation, model loading, SAM/Qwen, DeepSpeed, and docs/utilities. None of the candidate edges look like the same concrete underlying issue.", - "confidence": 0.98, - "canonical_issue_reason": null, + "summary": "The set is mostly heterogeneous: several issues are superficially similar around configs, embeddings, tokenizers, and generation, but they do not describe the same underlying bug. No soft duplicate pair is convincing enough to merge.", + "confidence": 0.91, + "canonical_issue_reason": "issue:45292 is the clearest concrete bug report in the set for a model-agnostic embedding-resize failure, and it best serves as an anchor for the smaller related resize/embed reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45003 is the most central node in the weak similarity graph, but it is still too generic to be a true duplicate anchor.", + "best_issue_reason": "issue:45292 is the best overall issue to represent the cluster because it describes a specific, reproducible code-path problem with broad applicability, unlike the many model-specific or orthogonal reports here.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44987", - "right": "issue:45459", + "left": "issue:44242", + "right": "issue:44857", + "accept": false, + "reason": "Different bugs: MoE load-balancing loss gating vs AMP/CUDA loss crash." + }, + { + "left": "issue:45362", + "right": "issue:45375", + "accept": false, + "reason": "Both involve Qwen models, but one is a chat-generation crash and the other is a config field being dropped; not the same fix." + }, + { + "left": "issue:44162", + "right": "issue:45200", + "accept": false, + "reason": "ESM2 breakage and Gemma mm_token_type_ids defaulting are unrelated model families and code paths." + }, + { + "left": "issue:44779", + "right": "issue:45292", "accept": false, - "reason": "Different failure modes: model import/config mismatch vs tokenizer error masking when protobuf is absent." + "reason": "Tokenizer correctness regression vs embedding-resize propagation bug; different subsystems." + }, + { + "left": "issue:44295", + "right": "issue:44749", + "accept": false, + "reason": "Position_ids buffer access error is unrelated to the transformers upgrade performance slowdown." + }, + { + "left": "issue:44877", + "right": "issue:45375", + "accept": false, + "reason": "Both are strict-config complaints, but they concern different model configs and different missing fields." + }, + { + "left": "issue:45292", + "right": "issue:45584", + "accept": false, + "reason": "Resize-token-embeddings/output-head mismatch is unrelated to Whisper's empty-transcription failure after align_special_tokens." + }, + { + "left": "issue:45292", + "right": "issue:45701", + "accept": false, + "reason": "Embedding resize propagation and version-dependent tokenization are different problems." + }, + { + "left": "issue:45561", + "right": "issue:45563", + "accept": false, + "reason": "pytest-xdist file race vs a stale generation warning; no shared failure mode." }, { "left": "issue:43901", "right": "issue:45003", "accept": false, - "reason": "Docs mismatch for `return_all_scores` is unrelated to unsafe `sys.modules` access in modeling internals." + "reason": "Docs mismatch about return_all_scores vs unsafe sys.modules access in modeling_utils." + }, + { + "left": "issue:44987", + "right": "issue:45459", + "accept": false, + "reason": "Loading a package under transformers>=5.1.0 and protobuf decode error handling are different tokenizer-loading issues." }, { "left": "issue:45375", "right": "issue:45759", "accept": false, - "reason": "Both involve Qwen multimodal configs, but one is a missing config field and the other is `from_config` unwrapping; different bugs." + "reason": "Both are Qwen multimodal config issues, but one is a missing config field and the other is from_config not unwrapping text_config." }, { "left": "issue:43232", "right": "issue:44336", "accept": false, - "reason": "Generation kwargs update after `sync_gpus` is unrelated to ANSI codes emitted by loading report." + "reason": "Generation kwargs sync_gpus bug and ANSI-code logging output are unrelated." }, { "left": "issue:36246", "right": "issue:43596", "accept": false, - "reason": "ImportError for a missing processor class and a DeepSpeed/BERT index error are unrelated." + "reason": "ImportError for a missing processor export is unrelated to the DeepSpeed/BertModel index error." }, { "left": "issue:43526", "right": "issue:45003", "accept": false, - "reason": "BEiT image processor label reduction bug is unrelated to `sys.modules` access in `modeling_utils`." + "reason": "Beit label reduction bug and sys.modules access bug do not share the same code path." }, { "left": "issue:44496", "right": "issue:45003", "accept": false, - "reason": "Unrecognized model/config loading failure is unrelated to the `sys.modules` access bug." + "reason": "Unrecognized-model config loading failure is unrelated to sys.modules indexing in modeling_utils." }, { "left": "issue:43257", "right": "issue:45137", "accept": false, - "reason": "Both mention DeepSpeed, but one is Qwen3 MOE weight conversion and the other is a ZeRO3 deque underflow." + "reason": "Qwen3 MoE weight-conversion under accelerate+deepspeed and ZeRO3 deque underflow are different distributed-loading failures." }, { "left": "issue:43065", "right": "issue:43408", "accept": false, - "reason": "SAM3 pixel decoder dummy conv and sam3_video/sam3_tracker warning are different configuration/model-class issues." + "reason": "Sam3PixelDecoder dummy layer issue and sam3_video/sam3_tracker warning are different SAM3 model problems." }, { "left": "issue:44995", "right": "issue:45200", "accept": false, - "reason": "GlmMoeDsa stale cache crash is unrelated to Gemma 4 `mm_token_type_ids` defaulting." + "reason": "GlmMoeDsa stale indexer cache and Gemma 4 mm_token_type_ids defaulting are unrelated." }, { "left": "issue:43262", "right": "issue:44246", "accept": false, - "reason": "Audio chat-template sampling rate bug is unrelated to intermittent import latency." + "reason": "Audio chat-template sample-rate defaulting and import-time slowness are not the same bug." }, { "left": "issue:42886", "right": "issue:45229", "accept": false, - "reason": "Offline tokenizer cache loading and Gemma4 multi-GPU OOM are unrelated." + "reason": "Offline tokenizer cache loading failure and Gemma multi-GPU OOM are unrelated." }, { "left": "issue:43475", "right": "issue:43540", "accept": false, - "reason": "SAM3 vision encoder output attribute error and Qwen3OmniMoe video processing error are different bugs." + "reason": "SAM3 vision encoder missing attribute vs Qwen3OmniMoe video-input ValueError; different failures." }, { "left": "issue:44466", "right": "issue:45335", "accept": false, - "reason": "Tied-weight serialization inconsistency and `resize_token_embeddings` not updating decoder embeddings are not the same code path." + "reason": "Tied-weights serialization inconsistency and resize_token_embeddings not updating decoder.embed_tokens are related by theme but not the same concrete bug." }, { "left": "issue:44995", "right": "issue:45588", "accept": false, - "reason": "Different failures: stale indexer cache on second forward vs flash-attention crash on `s_aux=None`." + "reason": "Stale indexer cache and flash_attention s_aux=None crash are different runtime errors." }, { "left": "issue:43653", "right": "issue:44162", "accept": false, - "reason": "BigBirdTokenizer special-token decode bug and ESM2 breakage are unrelated." + "reason": "BigBirdTokenizer special-token registration and ESM2 breakage are unrelated." }, { "left": "issue:44368", "right": "issue:44977", "accept": false, - "reason": "Tie-word-embeddings warning and Qwen3.5 flash-attention generation failure are different bugs." + "reason": "Tie-word-embeddings warning in Qwen3.5 and flash-attention generation failure are distinct issues." }, { "left": "issue:44368", "right": "issue:45563", "accept": false, - "reason": "Stale `num_return_sequences` warning in paged generate is unrelated to the tie-word-embeddings warning." + "reason": "Qwen tied-weights warning and stale num_return_sequences warning are unrelated generation issues." }, { "left": "issue:45397", "right": "issue:45656", "accept": false, - "reason": "Gemma-4 ZeRO3 loading and duplicate optimizer steps under DeepSpeed are different problems." + "reason": "Gemma zero3 from_pretrained and double optimizer.step under deepspeed are different distributed-training bugs." }, { "left": "issue:44060", "right": "issue:44077", "accept": false, - "reason": "Qwen3-Next tied-weights warning and patchtsmixer `post_init` policy are unrelated." - }, - { - "left": "issue:43673", - "right": "issue:44734", - "accept": false, - "reason": "Chunked-prefill cache handling and KV-cache continuation in `transformers serve` are different concrete paths." - }, - { - "left": "issue:45003", - "right": "issue:45561", - "accept": false, - "reason": "`sys.modules` access bug is unrelated to a pytest-xdist race in captured_info handling." - }, - { - "left": "issue:45137", - "right": "issue:45600", - "accept": false, - "reason": "ZeRO3 deque underflow and removed Sam3LiteText auto-mapping references are unrelated." - }, - { - "left": "issue:43931", - "right": "issue:44368", - "accept": false, - "reason": "Qwen3-VL weight-shape mismatch and a tie-word-embeddings warning are not the same issue." - }, - { - "left": "issue:44869", - "right": "issue:45356", - "accept": false, - "reason": "Whisper trailing-replacement decode crash and Kimi-K2.5 tokenizer regression are different tokenizer bugs." - }, - { - "left": "issue:44933", - "right": "issue:45459", - "accept": false, - "reason": "Missing image_utils import and protobuf-related tokenizer error hiding are unrelated." - }, - { - "left": "issue:44297", - "right": "issue:44623", - "accept": false, - "reason": "Tokenizer save metadata mismatch and processor save missing files are similar area but different serialization bugs." - }, - { - "left": "issue:43646", - "right": "issue:44162", - "accept": false, - "reason": "Transformers 5.0 custom model init regression and ESM2 breakage are too broad to be the same concrete bug." - }, - { - "left": "issue:43232", - "right": "issue:43673", - "accept": false, - "reason": "Generation kwargs update after `sync_gpus` and missing generation cache in v5.0.0 are distinct generation regressions." + "reason": "Incorrect tied-weights warning in Qwen3-Next and patchtsmixer post_init support are not the same underlying change." } ] }, @@ -10783,55 +10778,63 @@ "nodes": [ "issue:28282", "issue:30990", - "issue:36010", "issue:41628", - "issue:42175", - "issue:42222", "issue:42371", "issue:42673", "issue:42831", "issue:42994", - "issue:43377", + "issue:43232", "issue:43452", "issue:43475", "issue:43482", "issue:43525", "issue:43643", + "issue:43646", "issue:43673", "issue:43704", - "issue:43824", "issue:43827", - "issue:43901", + "issue:43931", "issue:44038", + "issue:44162", "issue:44222", - "issue:44242", + "issue:44297", "issue:44351", "issue:44361", - "issue:44496", + "issue:44368", "issue:44623", "issue:44661", + "issue:44734", "issue:44749", + "issue:44869", "issue:44933", - "issue:44964", "issue:45003", "issue:45081", "issue:45092", "issue:45103", "issue:45137", - "issue:45200", "issue:45230", "issue:45310", + "issue:45356", "issue:45375", "issue:45381", "issue:45397", + "issue:45459", "issue:45478", "issue:45561", "issue:45600", - "issue:45656", - "issue:45712" + "issue:45656" ], "soft_pairs": [ + "issue:43673|issue:44734", + "issue:45003|issue:45561", + "issue:45137|issue:45600", + "issue:43931|issue:44368", + "issue:44869|issue:45356", + "issue:44933|issue:45459", + "issue:44297|issue:44623", + "issue:43646|issue:44162", "issue:44361|issue:44749", + "issue:43232|issue:43673", "issue:43673|issue:43827", "issue:43452|issue:45003", "issue:28282|issue:30990", @@ -10839,8 +10842,8 @@ "issue:42994|issue:43482", "issue:43475|issue:43643", "issue:44749|issue:45310", - "issue:42371|issue:45092", "issue:44749|issue:45656", + "issue:42371|issue:45092", "issue:43475|issue:45381", "issue:44038|issue:44623", "issue:45230|issue:45397", @@ -10851,252 +10854,229 @@ "issue:42673|issue:43704", "issue:43475|issue:44661", "issue:45310|issue:45375", - "issue:41628|issue:44351", "issue:42831|issue:44222", - "issue:42222|issue:45200", - "issue:44222|issue:45310", - "issue:42994|issue:43824", - "issue:42175|issue:45712", - "issue:36010|issue:41628", - "issue:43377|issue:44242", - "issue:44038|issue:45310", - "issue:44496|issue:45561", - "issue:45561|issue:45600", - "issue:44964|issue:45137", - "issue:43901|issue:45561" + "issue:41628|issue:44351" ], - "prepared_review_unit_hash": "fe0054fe36a6787e4e6633af71d8d48153e11fb779b90b5dfcc389e1d3ce0587", + "prepared_review_unit_hash": "216b54cc7c38d55e3bc044214ef41e861aa802f7e6c67ea5894cf81f0f2fad85", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 33, - "serialized_chars": 37395, - "estimated_input_tokens": 9349, - "estimated_eval_tokens": 18954 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36249, + "estimated_input_tokens": 9063, + "estimated_eval_tokens": 18382 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: a few import/export regressions and a couple of exact duplicate Qwen3.5-Moe issues, but most pairs are unrelated model-specific bugs, docs, or performance reports.", - "confidence": 0.41, - "canonical_issue_reason": "Issue 41628 is the broadest, most representative import-regression report in the set; several nearby issues are also import/export breakages. The rest of the cluster is too mixed for a stronger canonical choice.", + "summary": "Mostly heterogeneous issue set. The only clear duplicate cluster is the Qwen3.5-Moe v5 from_pretrained/config regression, centered on the missing-field report; there is also one exact duplicate pair for the Qwen3ForCausalLM VRAM leak.", + "confidence": 0.74, + "canonical_issue_reason": "issue:45375 is the most root-cause-like and central report in the Qwen3.5-Moe cluster: it names the missing `deepstack_visual_indexes` field being dropped by `@strict`, which plausibly explains the related from_pretrained failures.", "canonical_pr_reason": null, - "best_issue_reason": "41628 is the best single issue to anchor the cluster, but only loosely: it captures the general transformers import breakage theme better than the more model-specific reports.", + "best_issue_reason": "issue:45375 is the strongest issue anchor overall because it is specific, technically actionable, and sits at the center of the only credible multi-issue duplicate cluster in this set.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43673", + "right": "issue:44734", + "accept": false, + "reason": "Both involve generation/cache failures, but they describe different APIs and failure modes (`chunked_prefill` cache lookup vs `/v1/responses` tensor indexing)." + }, + { + "left": "issue:45003", + "right": "issue:45561", + "accept": false, + "reason": "`sys.modules` access safety and an xdist `captured_info.txt` race are unrelated bugs in different code paths." + }, + { + "left": "issue:45137", + "right": "issue:45600", + "accept": false, + "reason": "DeepSpeed ZeRO3 deque underflow and a removed-Sam3LiteText config reference are unrelated regressions." + }, + { + "left": "issue:43931", + "right": "issue:44368", + "accept": false, + "reason": "A Qwen3-VL weight-shape mismatch is not the same bug as a Qwen3.5 LoRA warning about `tie_word_embeddings`." + }, + { + "left": "issue:44869", + "right": "issue:45356", + "accept": false, + "reason": "Whisper timestamp decoding and Kimi-K2.5 tokenizer codec handling are different tokenizer bugs." + }, + { + "left": "issue:44933", + "right": "issue:45459", + "accept": false, + "reason": "Missing import from `image_utils` and protobuf-related tokenizer error masking are different failures." + }, + { + "left": "issue:44297", + "right": "issue:44623", + "accept": false, + "reason": "Both are save_pretrained-related, but one is tokenizer metadata mismatch while the other is processor file omission; not the same concrete bug." + }, + { + "left": "issue:43646", + "right": "issue:44162", + "accept": false, + "reason": "A broad custom-model init regression and an ESM2-specific break are too different to treat as the same underlying bug." + }, { "left": "issue:44361", "right": "issue:44749", "accept": false, - "reason": "Tokenizer attribute error vs. a throughput regression; different failure modes and code paths." + "reason": "`MLukeTokenizer` task-attribute errors and a general slowdown after upgrading Transformers are unrelated." + }, + { + "left": "issue:43232", + "right": "issue:43673", + "accept": false, + "reason": "These are both generation-related, but one is a post-`sync_gpus` kwargs update issue and the other is a missing generation cache in chunked prefill." }, { "left": "issue:43673", "right": "issue:43827", "accept": false, - "reason": "Runtime cache bug vs. documentation referencing removed pipeline APIs; not the same underlying change." + "reason": "A runtime cache bug and docs still mentioning removed `pipeline()` APIs are not the same issue." }, { "left": "issue:43452", "right": "issue:45003", "accept": false, - "reason": "gguf loading failure vs. unsafe sys.modules access; related area, but not the same concrete bug." + "reason": "`gguf_file`/from_pretrained loading failure is unrelated to unsafe `sys.modules` access in modeling_utils." }, { "left": "issue:28282", "right": "issue:30990", "accept": false, - "reason": "PyTorch missing dependency error vs. sentence-transformers hanging on load; different symptoms and causes." + "reason": "PyTorch-missing import errors and Sentence Transformers hanging during load are different problem reports." }, { "left": "issue:44623", "right": "issue:45656", "accept": false, - "reason": "Processor save/export problem vs. deepspeed optimizer double-step; unrelated." + "reason": "Processor save_pretrained file output and a deepspeed double-step optimizer bug are unrelated." }, { "left": "issue:42994", "right": "issue:43482", "accept": false, - "reason": "Quantized model saving failure vs. GGUF loading failure in v5; both serialization/loading related but not one bug." + "reason": "Quantized model saving and Qwen2.5-GGUF loading under v5 are different IO paths and failure modes." }, { "left": "issue:43475", "right": "issue:43643", "accept": false, - "reason": "SAM 3 video output attribute mismatch vs. missing fields from trust_remote_code config loading; different defects." + "reason": "SAM 3 video output attribute errors and missing fields from `trust_remote_code=True` config loading are unrelated." }, { "left": "issue:44749", "right": "issue:45310", "accept": false, - "reason": "Performance slowdown after upgrade vs. Qwen3.5-Moe from_pretrained error; not the same issue." + "reason": "A performance regression is not the same concrete bug as a Qwen3.5-Moe from_pretrained failure." }, { - "left": "issue:42371", - "right": "issue:45092", + "left": "issue:44749", + "right": "issue:45656", "accept": false, - "reason": "TF32 API guidance/docs issue vs. remote-code meta-init incompatibility; unrelated." + "reason": "Performance slowdown and optimizer stepping twice under deepspeed are different issues." }, { - "left": "issue:44749", - "right": "issue:45656", + "left": "issue:42371", + "right": "issue:45092", "accept": false, - "reason": "Data filtering slowdown vs. deepspeed optimizer stepping twice; different subsystems." + "reason": "TF32 settings docs are unrelated to old InternVL2 remote-code checkpoints failing meta initialization." }, { "left": "issue:43475", "right": "issue:45381", "accept": false, - "reason": "Different multimodal model bugs: missing attribute vs. wrong video position ids." + "reason": "These are distinct model-specific vision bugs with different missing/incorrect fields." }, { "left": "issue:44038", "right": "issue:44623", "accept": false, - "reason": "Qwen3-VL-Moe upgrade bug vs. processor save_pretrained missing files; no concrete overlap." + "reason": "Qwen3-VL-Moe model loading and processor.save_pretrained file omission are unrelated." }, { "left": "issue:45230", "right": "issue:45397", "accept": false, - "reason": "Generic bug report vs. Gemma-4 zero3 from_pretrained failure; too little evidence of sameness." + "reason": "A placeholder bug report is not the same as the Gemma-4 ZeRO3 from_pretrained failure." }, { "left": "issue:41628", "right": "issue:44933", "accept": false, - "reason": "Both are import-related, but they refer to different missing symbols and different modules." + "reason": "Different import problems: `AutoImageProcessor` missing vs a nonexistent import from `image_utils`." }, { "left": "issue:45081", "right": "issue:45103", "accept": false, - "reason": "Tokenizer regex patch crash vs. auto-docstring annotation handling crash; different code paths." + "reason": "A Mistral regex/tokenizer backend crash and an auto-docstring annotations crash are unrelated." }, { "left": "issue:42994", "right": "issue:43525", "accept": false, - "reason": "Quantized save failure vs. missing pad_token_id on Llama4Config; not the same bug." + "reason": "Quantized model saving failure and missing `pad_token_id` on Llama4Config are different bugs." }, { "left": "issue:45375", "right": "issue:45478", - "accept": false, - "reason": "Related Qwen3.5-Moe family, but one is a missing config field and the other is a broader from_pretrained error." + "accept": true, + "reason": "Both describe the same Qwen3.5-Moe v5 regression: a config-field handling problem that surfaces as from_pretrained failure." }, { "left": "issue:42673", "right": "issue:43704", "accept": true, - "reason": "Same title, same model, same VRAM leak in multiple dataloader threads; clear duplicate." + "reason": "Exact duplicate title and same VRAM leak in Qwen3ForCausalLM with multiple dataloader threads." }, { "left": "issue:43475", "right": "issue:44661", "accept": false, - "reason": "SAM 3 video encoder attribute issue vs. tokenizer mapping add-new-model-like failure; unrelated." + "reason": "SAM3 vision output attribute errors and `TOKENIZER_MAPPING_NAMES`-related add-new-model-like failure are unrelated." }, { "left": "issue:45310", "right": "issue:45375", - "accept": false, - "reason": "Broad Qwen3.5-Moe load error vs. missing deepstack_visual_indexes field; related family, but not the same defect." - }, - { - "left": "issue:41628", - "right": "issue:44351", - "accept": false, - "reason": "Both import errors, but different missing exports and likely different fixes." + "accept": true, + "reason": "Same Qwen3.5-Moe v5 loading regression; 45375 appears to capture the underlying missing-field cause behind the from_pretrained error in 45310." }, { "left": "issue:42831", "right": "issue:44222", "accept": false, - "reason": "FP8 accuracy issue vs. FP8 save_pretrained failure; different symptom and scope." - }, - { - "left": "issue:42222", - "right": "issue:45200", - "accept": false, - "reason": "VitPose breakage vs. Gemma 4 token-type defaulting; unrelated models and behavior." - }, - { - "left": "issue:44222", - "right": "issue:45310", - "accept": false, - "reason": "FP8 save_pretrained moe vs. Qwen3.5-Moe from_pretrained error; not the same underlying bug." - }, - { - "left": "issue:42994", - "right": "issue:43824", - "accept": false, - "reason": "Quantized model saving failure vs. missing Qwen2_5_VL symbol import; different issues." - }, - { - "left": "issue:42175", - "right": "issue:45712", - "accept": false, - "reason": "TensorFlow backend packaging issue vs. leftover dummy class cleanup; unrelated." - }, - { - "left": "issue:36010", - "right": "issue:41628", - "accept": false, - "reason": "Both import-name failures, but for different symbols and regression surfaces." - }, - { - "left": "issue:43377", - "right": "issue:44242", - "accept": false, - "reason": "Padding-mask correctness bug vs. router-logits auxiliary-loss bug; distinct model behaviors." - }, - { - "left": "issue:44038", - "right": "issue:45310", - "accept": false, - "reason": "Both involve Qwen MOE variants, but the reported failures are different and not mergeable as one fix." - }, - { - "left": "issue:44496", - "right": "issue:45561", - "accept": false, - "reason": "Model config recognition failure vs. xdist file race; unrelated." - }, - { - "left": "issue:45561", - "right": "issue:45600", - "accept": false, - "reason": "Test-worker race condition vs. stale auto_mappings references; different root causes." - }, - { - "left": "issue:44964", - "right": "issue:45137", - "accept": false, - "reason": "Model loading failure vs. DeepSpeed ZeRO3 deque underflow; unrelated." + "reason": "FineGrainedFP8 accuracy loss and FP8 moe save_pretrained failure are different bug classes." }, { - "left": "issue:43901", - "right": "issue:45561", + "left": "issue:41628", + "right": "issue:44351", "accept": false, - "reason": "Docs mismatch vs. test-worker race; no shared bug." + "reason": "Different import targets and causes (`AutoImageProcessor` vs `HybridCache`)." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet: it correctly characterizes the cluster as mixed, notes the exact duplicate Qwen3.5-Moe pair, and uses 41628 as a broad import-regression anchor without overstating it. The soft-edge verdicts are mostly conservative false negatives, with the one true duplicate (42673/43704) supported by identical titles." + "feedback": "Grounded overall. The summary stays conservative, and the two accepted clusters are well supported by the packet: the Qwen3.5-Moe v5 from_pretrained/config regression centered on issue:45375, and the exact duplicate Qwen3ForCausalLM VRAM-leak pair. The rejected soft pairs are also framed cautiously and do not overclaim duplication." }, "evaluator_used": true, "retried": false, @@ -11120,53 +11100,64 @@ "issue:30333", "issue:30990", "issue:33290", - "issue:34689", "issue:35141", - "issue:36331", + "issue:36010", "issue:41628", "issue:42175", - "issue:42757", - "issue:42831", + "issue:42222", "issue:42898", + "issue:42994", "issue:43232", "issue:43278", + "issue:43377", "issue:43479", "issue:43723", "issue:43784", "issue:43824", "issue:43874", - "issue:43976", + "issue:43901", + "issue:44038", "issue:44060", "issue:44079", "issue:44188", - "issue:44206", "issue:44222", + "issue:44242", "issue:44295", - "issue:44297", "issue:44373", "issue:44448", + "issue:44496", "issue:44568", "issue:44623", "issue:44734", "issue:44779", "issue:44792", - "issue:44869", - "issue:44933", "issue:44964", - "issue:44977", "issue:45092", "issue:45103", "issue:45125", "issue:45137", + "issue:45200", "issue:45310", "issue:45356", "issue:45405", + "issue:45561", + "issue:45600", "issue:45656", "issue:45712", - "issue:45812", - "issue:45834" + "issue:45812" ], "soft_pairs": [ + "issue:42222|issue:45200", + "issue:44222|issue:45310", + "issue:42994|issue:43824", + "issue:43377|issue:44242", + "issue:36010|issue:41628", + "issue:42175|issue:45712", + "issue:44038|issue:45310", + "issue:44496|issue:45561", + "issue:44964|issue:45137", + "issue:45561|issue:45600", + "issue:43901|issue:45561", "issue:30990|issue:35141", "issue:42898|issue:44568", "issue:43232|issue:44792", @@ -11187,260 +11178,228 @@ "issue:42898|issue:44448", "issue:44188|issue:44964", "issue:43479|issue:44079", - "issue:44295|issue:45656", - "issue:44933|issue:45712", - "issue:42175|issue:44933", - "issue:44222|issue:44297", - "issue:43976|issue:44933", - "issue:43479|issue:44869", - "issue:44297|issue:45656", - "issue:44206|issue:45103", - "issue:42831|issue:44623", - "issue:44222|issue:44623", - "issue:30333|issue:36331", - "issue:41628|issue:42757", - "issue:43232|issue:45834", - "issue:30333|issue:34689", - "issue:43723|issue:44977" + "issue:44295|issue:45656" ], - "prepared_review_unit_hash": "9296a468d961cd9fa12ee9f115f4e981bfe2b8a52c64ba32d186e983a5290e6b", + "prepared_review_unit_hash": "fdc3a090b1e51a48529e1ff51ebda9742a0f215b08dcfe4f95bec49c9b290962", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 35, - "serialized_chars": 38195, - "estimated_input_tokens": 9549, - "estimated_eval_tokens": 19354 + "soft_pair_count": 32, + "serialized_chars": 37368, + "estimated_input_tokens": 9342, + "estimated_eval_tokens": 18940 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is broadly heterogeneous: mostly unrelated regressions around tokenizers, model loading, generation, imports, and DeepSpeed. No soft pair looks like a true duplicate, so there is no strong merge candidate among the listed artifacts.", - "confidence": 0.83, - "canonical_issue_reason": "issue:42898 is the most representative concrete bug in the set: a clear v5 tokenizer-behavior regression with broader relevance than the model-specific reports around it.", + "summary": "This cluster is highly heterogeneous: most items are unrelated regressions across tokenizers, multimodal configs, generation, packaging, DeepSpeed, and docs. The soft pairs are mostly superficial keyword matches and should not be merged as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": "No safe canonical issue: the items do not describe one underlying bug or change, so choosing a single issue as the cluster representative would be misleading.", "canonical_pr_reason": null, - "best_issue_reason": "issue:42898 is the best anchor issue because it describes a specific, user-visible regression and is broad enough to stand as a representative tokenizer bug, even though the cluster does not contain true duplicates.", + "best_issue_reason": "No single issue is a good global representative; even the closest textual matches point to different code paths, models, or failure modes.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:30990", - "right": "issue:35141", + "left": "issue:42222 - All vitpose model were brokentransformers/models/vitpose_", + "right": "issue:45200 - [Gemma 4] mm_token_type_ids required for text-only fine-tuning - should default to zeros", "accept": false, - "reason": "Different failures: loading hangs for Sentence Transformers vs embedding reinitialization after resize. No shared code-path bug." + "reason": "Both are multimodal/model issues, but one is a broken VitPose model path and the other is Gemma 4 text-only mm_token_type_ids handling; different bugs and fixes." }, { - "left": "issue:42898", - "right": "issue:44568", + "left": "issue:44222 - [Bug] FP8 save_pretrained moe", + "right": "issue:45310 - [BUG] transformers>=5.4.0, Qwen3.5 Moe from_pretrained error", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different behaviors and models: cleanup spacing vs BOS/EOS insertion." + "reason": "Save-pretrained FP8/MoE serialization is a different failure path from a Qwen3.5 MoE load-time from_pretrained error." }, { - "left": "issue:43232", - "right": "issue:44792", + "left": "issue:42994 - quantized model saving failed", + "right": "issue:43824 - ImportError: cannot import name 'Qwen2_5_VLForConditionalGeneration' from 'transformers'", "accept": false, - "reason": "Generation state update after sync_gpus is unrelated to a Janus image-generation test failure." + "reason": "Quantized save failures and missing class imports are unrelated problems in different subsystems." }, { - "left": "issue:43232", - "right": "issue:44188", + "left": "issue:43377 - [BUG] MIMI Encoder produces different outputs for batched vs single inputs due to missing padding mask support", + "right": "issue:44242 - Load balancing loss not added when output_router_logits=False", "accept": false, - "reason": "Different generation/attention paths; one is sync_gpus kwargs handling, the other is torch.compile branch divergence in attention kernels." + "reason": "One is an encoder batching/padding-mask correctness bug; the other is MoE loss computation gated by router logits." }, { - "left": "issue:42898", - "right": "issue:45812", + "left": "issue:36010 - ImportError: cannot import name 'GenerationMixin' from 'transformers.generation'", + "right": "issue:41628 - Cannot import name 'AutoImageProcessor' from 'transformers'", "accept": false, - "reason": "Both concern tokenization regressions, but the concrete bug differs: cleanup spacing vs Granite token IDs across all models." + "reason": "Both are import errors, but they affect different symbols and likely different export/re-export fixes." }, { - "left": "issue:44188", - "right": "issue:44373", + "left": "issue:42175 - Tensorflow not include in the backend when using pip install '.[torch]'", + "right": "issue:45712 - Six leftover dummy classes in `dummy_pt_objects.py` fail `check_repo.py` and leak into `dir(transformers)` without torch", "accept": false, - "reason": "Attention-kernel branching and a docstring typo are unrelated." + "reason": "Packaging dependency selection and leftover dummy class cleanup are different repository hygiene issues." }, { - "left": "issue:43784", - "right": "issue:43824", + "left": "issue:44038 - [bug] transformers 5.0 & Qwen3-VL-Moe", + "right": "issue:45310 - [BUG] transformers>=5.4.0, Qwen3.5 Moe from_pretrained error", "accept": false, - "reason": "Different import errors in different packages/models; not the same underlying bug." + "reason": "Same broad model family, but different versions/models and different load/runtime failures; too coarse to be one duplicate." }, { - "left": "issue:43278", - "right": "issue:45137", + "left": "issue:44496 - ValueError: Unrecognized model in allenai/Olmo-Hybrid-Instruct-SFT-7B. Should have a `model_type` key in its config.json.", + "right": "issue:45561 - [Bug] pytest-xdist workers race on captured_info.txt in patched testing utils", "accept": false, - "reason": "BF16-to-FP32 dtype change in evaluation is unrelated to DeepSpeed ZeRO3 deque underflow." + "reason": "A model config loading error is unrelated to a parallel test-file race." }, { - "left": "issue:44623", - "right": "issue:45310", + "left": "issue:44964 - Cannot load `microsoft/Phi-4-multimodal-instruct` model with latest transformers.", + "right": "issue:45137 - IndexError: pop from an empty deque with DeepSpeed ZeRO3", "accept": false, - "reason": "Processor save_pretrained file omission is unrelated to Qwen3.5 MoE from_pretrained failure." + "reason": "Multimodal model loading and DeepSpeed ZeRO3 deque underflow are unrelated failures." }, { - "left": "issue:44734", - "right": "issue:45103", + "left": "issue:45561 - [Bug] pytest-xdist workers race on captured_info.txt in patched testing utils", + "right": "issue:45600 - auto_mappings.py references removed Sam3LiteText configs, breaking CI", "accept": false, - "reason": "KV-cache indexing crash in serve and auto_docstring future-annotations crash are different subsystems and bugs." + "reason": "A pytest-xdist file race and stale auto-mapping references are different CI problems." }, { - "left": "issue:30333", - "right": "issue:33290", + "left": "issue:43901 - TextClassificationPipeline docs still mention return_all_scores, but behavior differs", + "right": "issue:45561 - [Bug] pytest-xdist workers race on captured_info.txt in patched testing utils", "accept": false, - "reason": "MLFlow job status reporting and an Adafactor/DeepSpeed OOM are unrelated." + "reason": "Documentation mismatch for a pipeline option is unrelated to a test utility race." }, { - "left": "issue:43723", - "right": "issue:44568", + "left": "issue:30990 - Sentence Transformers Gets Stuck loading", + "right": "issue:35141 - resizing token embeddings causes output embedding to be reinitialized in `post_init` when `tie_word_embedding` is False", "accept": false, - "reason": "Tokenizer loading in v5 and add_special_tokens behavior for mdeberta are different issues." + "reason": "Loading hangs and embedding reinitialization are different code paths and different bugs." }, { - "left": "issue:44060", - "right": "issue:45125", + "left": "issue:42898 - `clean_up_tokenization_spaces` behavior changes in v5", + "right": "issue:44568 - [BUG] add_special_tokens=True doesn't add BOS/EOS tokens for microsoft/mdeberta-v3-base tokenizer in transformers >=5.0", "accept": false, - "reason": "Tied-weights warning in Qwen3-Next is unrelated to missing _tp_plan in Qwen3.5 MoE." + "reason": "Both touch tokenizer behavior, but one is cleanup-space semantics and the other is special-token insertion for a specific tokenizer." }, { - "left": "issue:42898", - "right": "issue:45356", + "left": "issue:43232 - _update_model_kwargs_for_generation after sync_gpus when generation", + "right": "issue:44792 - Failed test case `test_model_generate_images` for janus model", "accept": false, - "reason": "Both are tokenizer regressions, but they target different models and failures (spacing behavior vs codec handling/warning)." + "reason": "Generation kwargs update after sync_gpus is not the same as a Janus image-generation test failure." }, { - "left": "issue:43874", - "right": "issue:45092", + "left": "issue:43232 - _update_model_kwargs_for_generation after sync_gpus when generation", + "right": "issue:44188 - Diverging attention kernels due to `allow_is_bidirectional_skip` branching on torch.compile", "accept": false, - "reason": "Both are multimodal loading issues, but they involve different models and mechanisms; not the same bug." + "reason": "Generation bookkeeping and attention-kernel divergence are different runtime bugs." }, { - "left": "issue:42898", - "right": "issue:44779", + "left": "issue:42898 - `clean_up_tokenization_spaces` behavior changes in v5", + "right": "issue:45812 - `AutoTokenizer` produces wrong token IDs for all Granite models (silent v4\u2192v5 regression)", "accept": false, - "reason": "Different tokenizer regressions for different model families; no shared concrete failure mode." + "reason": "Both are tokenizer regressions, but one is a space-cleanup behavior change and the other is model-specific token ID generation." }, { - "left": "issue:45405", - "right": "issue:45712", + "left": "issue:44188 - Diverging attention kernels due to `allow_is_bidirectional_skip` branching on torch.compile", + "right": "issue:44373 - Wrong docstring for position_ids", "accept": false, - "reason": "Package version bump problem and leftover dummy class cleanup are unrelated." + "reason": "A runtime kernel divergence bug does not match a documentation-only position_ids issue." }, { - "left": "issue:42898", - "right": "issue:44448", + "left": "issue:43784 - NameError: name 'nn' is not defined when importing sentence-transformers with latest transformers", + "right": "issue:43824 - ImportError: cannot import name 'Qwen2_5_VLForConditionalGeneration' from 'transformers'", "accept": false, - "reason": "Pegasus output differences are a separate model regression from clean_up_tokenization_spaces behavior changes." + "reason": "Both are import failures, but they are for different packages/symbols and likely different missing exports." }, { - "left": "issue:44188", - "right": "issue:44964", + "left": "issue:43278 - Embedding layer dtype changed from BF16 in training to FP32 in evaluate.", + "right": "issue:45137 - IndexError: pop from an empty deque with DeepSpeed ZeRO3", "accept": false, - "reason": "Attention kernel divergence and Phi-4 multimodal loading are unrelated." + "reason": "A dtype mismatch between train/eval is unrelated to a DeepSpeed ZeRO3 deque error." }, { - "left": "issue:43479", - "right": "issue:44079", + "left": "issue:44623 - [BUG] processor.save_pretrained(...) missing files", + "right": "issue:45310 - [BUG] transformers>=5.4.0, Qwen3.5 Moe from_pretrained error", "accept": false, - "reason": "Both mention None handling, but one is config initialization and the other is ModelOutput key assignment; different code paths." + "reason": "Save-pretrained missing files and load-time from_pretrained errors are opposite sides of the API but not the same bug." }, { - "left": "issue:44295", - "right": "issue:45656", + "left": "issue:44734 - `transformers serve`: /v1/responses crashes on KV cache continuation due to wrong tensor indexing", + "right": "issue:45103 - [auto_docstring] _process_kwargs_parameters crashes with AttributeError when module uses from __future__ import annotations", "accept": false, - "reason": "Position_ids buffer read error and double optimizer step under DeepSpeed are unrelated." + "reason": "Serving KV-cache tensor indexing and auto-docstring annotation handling are unrelated failures." }, { - "left": "issue:44933", - "right": "issue:45712", + "left": "issue:30333 - If a training job job failed MLFlow will not be reported and MLFlow shows job still running", + "right": "issue:33290 - oom when using adafactor optimizer in deepspeed", "accept": false, - "reason": "Missing image_utils import and dummy PT object cleanup are not the same issue." + "reason": "MLflow job-state reporting and optimizer OOM in DeepSpeed are unrelated issues." }, { - "left": "issue:42175", - "right": "issue:44933", + "left": "issue:43723 - Issue loading tokenizer using AutoTokenizer.from_pretrained in v5", + "right": "issue:44568 - [BUG] add_special_tokens=True doesn't add BOS/EOS tokens for microsoft/mdeberta-v3-base tokenizer in transformers >=5.0", "accept": false, - "reason": "TensorFlow extra/backend packaging and image_utils import exposure are unrelated." + "reason": "A generic tokenizer loading issue is not the same as a specific BOS/EOS insertion regression." }, { - "left": "issue:44222", - "right": "issue:44297", + "left": "issue:44060 - Qwen3-Next: Incorrect tied weights warning ties embed_tokens.weight to linear_attn.dt_bias across all layers", + "right": "issue:45125 - Qwen3_5MoeForConditionalGeneration missing _tp_plan for tensor parallelism", "accept": false, - "reason": "Both involve save_pretrained, but one is FP8/MoE serialization and the other is tokenizer config class metadata; different bugs." + "reason": "These are different Qwen model classes with different bugs: tied-weights warning vs missing tensor-parallel plan." }, { - "left": "issue:43976", - "right": "issue:44933", + "left": "issue:42898 - `clean_up_tokenization_spaces` behavior changes in v5", + "right": "issue:45356 - Regression in Kimi-K2.5 tokenizer from 5.3.0 to 5.4.0: incorrect codec handling and misleading fix_mistral_regex warning", "accept": false, - "reason": "Python version compatibility and a missing import are unrelated." + "reason": "Both are tokenizer regressions, but the root causes and affected models differ." }, { - "left": "issue:43479", - "right": "issue:44869", + "left": "issue:43874 - [GLM46V] Glm46VImageProcessorFast missing get_number_of_image_patches breaks _get_num_multimodal_tokens (AttributeError)", + "right": "issue:45092 - [Bug] Old InternVL2 remote-code checkpoints are incompatible with Transformers v5 meta initialization", "accept": false, - "reason": "Phi4Multimodal config defaults and Whisper timestamp decoding are unrelated." + "reason": "Both are multimodal-model issues, but one is a missing image-processor method and the other is v5 meta-init incompatibility for a different model family." }, { - "left": "issue:44297", - "right": "issue:45656", + "left": "issue:42898 - `clean_up_tokenization_spaces` behavior changes in v5", + "right": "issue:44779 - Deepseek tokenizer produces incorrect results as of v5 (works in v4)", "accept": false, - "reason": "Tokenizer save_pretrained metadata mismatch is unrelated to DeepSpeed double-step behavior." + "reason": "Same area, but different tokenizer regressions with different model-specific causes." }, { - "left": "issue:44206", - "right": "issue:45103", + "left": "issue:45405 - `MIN_PEFT_VERSION` bumped to 0.18.2 which is not yet released on PyPI", + "right": "issue:45712 - Six leftover dummy classes in `dummy_pt_objects.py` fail `check_repo.py` and leak into `dir(transformers)` without torch", "accept": false, - "reason": "Feature-extractor argument incompatibility and auto_docstring future-annotation crash are different bugs." + "reason": "A dependency version bump is unrelated to dummy class cleanup and symbol leakage." }, { - "left": "issue:42831", - "right": "issue:44623", + "left": "issue:42898 - `clean_up_tokenization_spaces` behavior changes in v5", + "right": "issue:44448 - [BUG] Different output for google/pegasus-cnn_dailymail between Transformers v4 and v5", "accept": false, - "reason": "FP8 accuracy concerns and processor.save_pretrained missing files are unrelated." + "reason": "Both mention v4\u2192v5 behavior changes, but one is a tokenizer cleanup flag and the other is model output drift." }, { - "left": "issue:44222", - "right": "issue:44623", + "left": "issue:44188 - Diverging attention kernels due to `allow_is_bidirectional_skip` branching on torch.compile", + "right": "issue:44964 - Cannot load `microsoft/Phi-4-multimodal-instruct` model with latest transformers.", "accept": false, - "reason": "Different save_pretrained failures in different components; not mergeable as one fix." + "reason": "Attention-kernel divergence and Phi-4 model loading are different problems." }, { - "left": "issue:30333", - "right": "issue:36331", + "left": "issue:43479 - Phi4MultimodalConfig incorrectly initializes default vision/audio configs when passed as None", + "right": "issue:44079 - `ModelOutput` keys aren't correctly assigned if key was previously None", "accept": false, - "reason": "MLFlow reporting failure and CustomTrainer compute_loss signature breakage are unrelated." + "reason": "A multimodal config defaulting bug is unrelated to ModelOutput key assignment semantics." }, { - "left": "issue:41628", - "right": "issue:42757", + "left": "issue:44295 - An error occurs when reading position_ids after registering it as a buffer.", + "right": "issue:45656 - Optimizer step being called 2 times when using deepspeed", "accept": false, - "reason": "Two separate import errors from different dependencies; no common code-path." - }, - { - "left": "issue:43232", - "right": "issue:45834", - "accept": false, - "reason": "Generation kwargs handling and Kosmos2.5 OCR long-input indexing are unrelated." - }, - { - "left": "issue:30333", - "right": "issue:34689", - "accept": false, - "reason": "Training-job MLFlow reporting and Llama 3.2 vision model loading are unrelated." - }, - { - "left": "issue:43723", - "right": "issue:44977", - "accept": false, - "reason": "Tokenizer loading regression and Qwen3.5 flash-attention generation bug are different problems." + "reason": "Position_ids buffer access and double optimizer stepping in DeepSpeed are different runtime bugs." } ] }, @@ -11468,25 +11427,24 @@ "issue:28282", "issue:30333", "issue:30990", + "issue:34689", "issue:35141", + "issue:36331", "issue:38175", - "issue:39401", "issue:41628", - "issue:42371", + "issue:42175", "issue:42757", "issue:42831", "issue:42886", "issue:42994", - "issue:43066", "issue:43232", "issue:43479", - "issue:43576", "issue:43644", "issue:43723", - "issue:43784", "issue:43824", "issue:43873", - "issue:44038", + "issue:43976", + "issue:44206", "issue:44222", "issue:44261", "issue:44263", @@ -11498,27 +11456,43 @@ "issue:44484", "issue:44485", "issue:44514", - "issue:44568", "issue:44623", + "issue:44869", "issue:44871", "issue:44913", + "issue:44933", "issue:44945", + "issue:44977", "issue:45092", "issue:45103", - "issue:45278", "issue:45310", "issue:45357", "issue:45406", - "issue:45593", - "issue:45753" + "issue:45656", + "issue:45712", + "issue:45834" ], "soft_pairs": [ + "issue:44933|issue:45712", + "issue:42175|issue:44933", + "issue:44222|issue:44297", + "issue:43976|issue:44933", + "issue:43479|issue:44869", + "issue:44297|issue:45656", + "issue:44206|issue:45103", + "issue:42831|issue:44623", + "issue:44222|issue:44623", + "issue:30333|issue:36331", + "issue:41628|issue:42757", + "issue:43232|issue:45834", + "issue:30333|issue:34689", + "issue:43723|issue:44977", "issue:43873|issue:44945", "issue:44222|issue:44484", "issue:44297|issue:45310", "issue:28282|issue:30333", - "issue:44263|issue:45092", "issue:30333|issue:35141", + "issue:44263|issue:45092", "issue:44483|issue:45406", "issue:44263|issue:44360", "issue:44373|issue:44485", @@ -11532,226 +11506,240 @@ "issue:43479|issue:44913", "issue:42886|issue:42994", "issue:41628|issue:43824", - "issue:22355|issue:28282", - "issue:42831|issue:45310", - "issue:39401|issue:44568", - "issue:44263|issue:44373", - "issue:44945|issue:45753", - "issue:42831|issue:44038", - "issue:42757|issue:45278", - "issue:43576|issue:43784", - "issue:42886|issue:43066", - "issue:43232|issue:45593", - "issue:42371|issue:44263" + "issue:22355|issue:28282" ], - "prepared_review_unit_hash": "30be542d78a69878b91f4c1c4562e210f801a3b64d0a079daa04bd5fc8de812c", + "prepared_review_unit_hash": "f15cab0f109701929c1a9eeb009949df7d9f0b155f2b0c99ebb25a7e2b45aaee", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 30, - "serialized_chars": 35689, - "estimated_input_tokens": 8923, - "estimated_eval_tokens": 18102 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37581, + "estimated_input_tokens": 9396, + "estimated_eval_tokens": 19048 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: most pairs are superficially related by broad subsystem or v5 regression themes, but they do not describe the same concrete bug/change. I found no true duplicate pair to merge, though issue 45278 is the broadest umbrella candidate among the import-error reports.", - "confidence": 0.84, - "canonical_issue_reason": "No single issue cleanly represents the whole set; the cluster is not a true duplicate group. If an umbrella must be chosen, issue 45278 is the broadest because it covers multiple import failures after a 4.57.0\u21925.5.0 upgrade.", + "summary": "These soft pairs are mostly false positives: they mix unrelated import errors, save/load regressions, config persistence bugs, training issues, and serving/runtime crashes. I found no pair that clearly reflects the same underlying bug or change, so there is no real duplicate cluster here.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue is appropriate; the items do not converge on one underlying bug and the strongest similarities are only broad subsystem overlap.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the best global anchor because it is the most general, still-open report and plausibly subsumes several import-related regressions, but it is not a precise duplicate target for the rest of the cluster.", + "best_issue_reason": "No issue stands out as a global best representative because the set is heterogeneous and the soft links are not true duplicates.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44933", + "right": "issue:45712", + "accept": false, + "reason": "Both involve imports/symbol exposure, but one is about image_utils and the other about dummy PT objects leaking into dir(transformers); different code paths and fixes." + }, + { + "left": "issue:42175", + "right": "issue:44933", + "accept": false, + "reason": "Backend-install packaging problem vs missing image_utils import; unrelated symptoms and remediation." + }, + { + "left": "issue:44222", + "right": "issue:44297", + "accept": false, + "reason": "Both mention save_pretrained, but one is FP8 MoE saving and the other is tokenizer metadata mismatch; not the same bug." + }, + { + "left": "issue:43976", + "right": "issue:44933", + "accept": false, + "reason": "Python version compatibility regression vs import exposure issue; no shared underlying failure." + }, + { + "left": "issue:43479", + "right": "issue:44869", + "accept": false, + "reason": "Config default initialization bug vs Whisper timestamp decode crash; completely different subsystems." + }, + { + "left": "issue:44297", + "right": "issue:45656", + "accept": false, + "reason": "Tokenizer save metadata issue vs deepspeed optimizer double-step; unrelated." + }, + { + "left": "issue:44206", + "right": "issue:45103", + "accept": false, + "reason": "Feature extractor argument regression vs auto-docstring crash from future annotations; different layers." + }, + { + "left": "issue:42831", + "right": "issue:44623", + "accept": false, + "reason": "FP8 accuracy issue vs processor.save_pretrained missing files; not the same code path." + }, + { + "left": "issue:44222", + "right": "issue:44623", + "accept": false, + "reason": "Both touch saving, but model FP8 saving failure and processor artifact omission are distinct bugs." + }, + { + "left": "issue:30333", + "right": "issue:36331", + "accept": false, + "reason": "MLflow reporting on failed training vs CustomTrainer signature change; different behaviors and fixes." + }, + { + "left": "issue:41628", + "right": "issue:42757", + "accept": false, + "reason": "Different missing imports from different packages and likely different dependency/version causes." + }, + { + "left": "issue:43232", + "right": "issue:45834", + "accept": false, + "reason": "Generation kwarg update bug vs long OCR input index error; unrelated model paths." + }, + { + "left": "issue:30333", + "right": "issue:34689", + "accept": false, + "reason": "Training failure reporting vs model loading regression for a specific vision model; not duplicates." + }, + { + "left": "issue:43723", + "right": "issue:44977", + "accept": false, + "reason": "Tokenizer loading regression vs flash-attention generation issue; different failure modes." + }, { "left": "issue:43873", "right": "issue:44945", "accept": false, - "reason": "Different problems: quantization/offloading behavior vs incorrect output under pipeline parallelism." + "reason": "Quantization offloading behavior vs pipeline parallelism output correctness; too different to merge." }, { "left": "issue:44222", "right": "issue:44484", "accept": false, - "reason": "One is about FP8 MoE save_pretrained behavior; the other is about max_shard_size documentation/behavior. Not the same bug." + "reason": "FP8 save_pretrained failure vs a question about shard size defaults; not the same bug." }, { "left": "issue:44297", "right": "issue:45310", "accept": false, - "reason": "Tokenizer config mismatch on save vs Qwen3.5 MoE from_pretrained failure; different code paths and symptoms." + "reason": "Tokenizer config saved incorrectly vs Qwen3.5 MoE from_pretrained error; related broadly to load/save, but not the same underlying defect." }, { "left": "issue:28282", "right": "issue:30333", "accept": false, - "reason": "PyTorch missing import error vs MLflow reporting after failed training. No shared underlying defect." + "reason": "Missing PyTorch import dependency vs failed-job MLflow reporting; unrelated." }, { - "left": "issue:44263", - "right": "issue:45092", + "left": "issue:30333", + "right": "issue:35141", "accept": false, - "reason": "DSA indexer torch.split issue vs remote-code checkpoint/meta-init incompatibility. Unrelated failures." + "reason": "Training-job reporting bug vs embedding reinitialization in post_init; different subsystems." }, { - "left": "issue:30333", - "right": "issue:35141", + "left": "issue:44263", + "right": "issue:45092", "accept": false, - "reason": "MLflow failure reporting vs embedding resize reinitialization in post_init. Different subsystems and causes." + "reason": "DSA indexer split behavior vs remote-code checkpoint meta-init incompatibility; distinct bugs." }, { "left": "issue:44483", "right": "issue:45406", "accept": false, - "reason": "API request rejection vs Gemma4Processor missing _tokenizer. Not the same failure mode." + "reason": "Chat completions request handling vs Gemma4Processor attribute error in serve; different code paths." }, { "left": "issue:44263", "right": "issue:44360", "accept": false, - "reason": "Both mention the DSA indexer, but one is about torch.split return values and the other about a missing ReLU. Different bugs." + "reason": "Both mention the DSA indexer, but one is about split outputs and the other about missing ReLU; not clearly the same defect." }, { "left": "issue:44373", "right": "issue:44485", "accept": false, - "reason": "Docstring mismatch for position_ids vs RoPE implementation bug. One is documentation, the other is code behavior." + "reason": "Docstring mismatch vs RoPE implementation discussion; one is documentation, the other model behavior." }, { "left": "issue:44514", "right": "issue:45103", "accept": false, - "reason": "Batched chat-template crash vs auto-docstring AttributeError with future annotations. Not the same code path." + "reason": "Processor batched chat-template crash vs auto-docstring annotation crash; unrelated." }, { "left": "issue:43479", "right": "issue:43644", "accept": false, - "reason": "Config default initialization bug vs non-persistent buffer corruption. Different model lifecycle issues." + "reason": "Specific multimodal config defaulting bug vs general non-persistent buffer initialization regression; only broad init-theme overlap." }, { "left": "issue:44295", "right": "issue:44623", "accept": false, - "reason": "Buffer registration/read error for position_ids vs processor save_pretrained missing files. Not a duplicate." + "reason": "Reading position_ids after buffer registration vs processor.save_pretrained missing files; different issues." }, { "left": "issue:44261", "right": "issue:45092", "accept": false, - "reason": "RMS norm epsilon precision issue in MLA vs meta initialization incompatibility. Different underlying defects." + "reason": "Missing rms_norm_eps precision issue vs v5 meta-initialization incompatibility for old checkpoints; not the same root cause." }, { "left": "issue:45092", "right": "issue:45357", "accept": false, - "reason": "Meta-init incompatibility for old checkpoints vs incorrect visual encoder keys on save_pretrained. Not the same bug." + "reason": "Both concern Qwen/remote-code model handling, but one is InternVL2 meta-init compatibility and the other is incorrect visual encoder keys on save_pretrained." }, { "left": "issue:43723", "right": "issue:44871", "accept": false, - "reason": "AutoTokenizer load issue in v5 vs eos_token_id inconsistency for Gemma-3. Different tokenizer problems." + "reason": "Tokenizer loading regression vs eos_token_id config mismatch; related to model/tokenizer config but not duplicate bugs." }, { "left": "issue:30990", "right": "issue:38175", "accept": false, - "reason": "Stuck loading Sentence Transformers vs zero probabilities in a specific SigLIP2 model. No shared concrete defect." + "reason": "Sentence Transformers loading stall vs zero-probability outputs in a vision model; unrelated symptoms." }, { "left": "issue:43479", "right": "issue:44913", "accept": false, - "reason": "Default config initialization on None vs rotary_pct not persisting on reload. Separate config bugs." + "reason": "Different config persistence bugs in different model classes; too broad to treat as duplicates." }, { "left": "issue:42886", "right": "issue:42994", "accept": false, - "reason": "Offline cache load failure for tokenizers vs quantized model saving failure. Different features and failure modes." + "reason": "Offline tokenizer cache loading vs quantized model saving failure; unrelated." }, { "left": "issue:41628", "right": "issue:43824", "accept": false, - "reason": "Both are import errors, but for different missing symbols and likely different breakages. Too broad to merge." + "reason": "Both are import errors, but for different exported classes and likely different fixes." }, { "left": "issue:22355", "right": "issue:28282", "accept": false, - "reason": "Missing transformers.onnx module vs AutoModel requiring PyTorch. Different imports and causes." - }, - { - "left": "issue:42831", - "right": "issue:45310", - "accept": false, - "reason": "FineGrainedFP8 accuracy regression vs Qwen3.5 MoE from_pretrained error. Not the same issue." - }, - { - "left": "issue:39401", - "right": "issue:44568", - "accept": false, - "reason": "Qwen3 offset_mapping bug vs mdeberta add_special_tokens regression. Both tokenizer-related, but not the same concrete bug." - }, - { - "left": "issue:44263", - "right": "issue:44373", - "accept": false, - "reason": "DSA indexer runtime bug vs a wrong docstring. No shared underlying change." - }, - { - "left": "issue:44945", - "right": "issue:45753", - "accept": false, - "reason": "Pipeline-parallelism output bug vs Qwen3.5 infinite loop on a specific image. Different model behaviors and triggers." - }, - { - "left": "issue:42831", - "right": "issue:44038", - "accept": false, - "reason": "FineGrainedFP8 accuracy issue vs Qwen3-VL-Moe bug in v5. Not the same concrete defect." - }, - { - "left": "issue:42757", - "right": "issue:45278", - "accept": false, - "reason": "Specific missing import from huggingface_hub vs a broad 'many import errors after upgrade' report. Related theme, but not the same bug." - }, - { - "left": "issue:43576", - "right": "issue:43784", - "accept": false, - "reason": "Broken transformers env command vs sentence-transformers import NameError for nn. Different failure surfaces." - }, - { - "left": "issue:42886", - "right": "issue:43066", - "accept": false, - "reason": "Offline tokenizer cache loading bug vs wrong tokenizer decoder type in v5. Not a duplicate." - }, - { - "left": "issue:43232", - "right": "issue:45593", - "accept": false, - "reason": "Generation kwargs/sync_gpus issue vs D-FINE auxiliary-loss behavior. Completely different functionality." - }, - { - "left": "issue:42371", - "right": "issue:44263", - "accept": false, - "reason": "TF32 settings warning/regression vs DSA indexer torch.split issue. Unrelated." + "reason": "Missing submodule import vs missing backend library dependency; different causes and scopes." } ] }, @@ -11778,53 +11766,63 @@ "issue:29127", "issue:30990", "issue:33357", - "issue:36296", "issue:39401", "issue:41628", + "issue:42371", "issue:42757", + "issue:42831", + "issue:42886", "issue:42907", "issue:43066", "issue:43116", "issue:43122", "issue:43232", "issue:43408", - "issue:43452", "issue:43475", + "issue:43576", "issue:43723", "issue:43756", "issue:43784", - "issue:43824", "issue:43937", "issue:44038", "issue:44060", "issue:44117", - "issue:44222", "issue:44261", + "issue:44263", "issue:44295", "issue:44297", - "issue:44351", "issue:44360", "issue:44373", "issue:44442", "issue:44485", - "issue:44488", "issue:44514", "issue:44521", + "issue:44568", "issue:44625", "issue:44704", "issue:44779", "issue:44829", "issue:44945", - "issue:44964", "issue:45200", - "issue:45216", "issue:45245", "issue:45278", "issue:45290", - "issue:45478", - "issue:45600" + "issue:45310", + "issue:45593", + "issue:45600", + "issue:45753" ], "soft_pairs": [ + "issue:39401|issue:44568", + "issue:42831|issue:45310", + "issue:44263|issue:44373", + "issue:44945|issue:45753", + "issue:42831|issue:44038", + "issue:42757|issue:45278", + "issue:43576|issue:43784", + "issue:42886|issue:43066", + "issue:43232|issue:45593", + "issue:42371|issue:44263", "issue:43756|issue:44060", "issue:44117|issue:44704", "issue:42907|issue:44945", @@ -11834,257 +11832,226 @@ "issue:43232|issue:45245", "issue:44295|issue:44297", "issue:43784|issue:45600", - "issue:41628|issue:45278", "issue:42907|issue:43937", "issue:43723|issue:44779", + "issue:41628|issue:45278", "issue:43723|issue:44442", - "issue:43408|issue:43475", "issue:44261|issue:44373", + "issue:43408|issue:43475", "issue:44360|issue:44485", "issue:43122|issue:44521", "issue:44625|issue:44829", "issue:43116|issue:45200", - "issue:29127|issue:30990", - "issue:42907|issue:44964", - "issue:43122|issue:44779", - "issue:44038|issue:44297", - "issue:42907|issue:43452", - "issue:39401|issue:44488", - "issue:44038|issue:44295", - "issue:43824|issue:44351", - "issue:36296|issue:43116", - "issue:44038|issue:44222", - "issue:39401|issue:44521", - "issue:45216|issue:45478", - "issue:42757|issue:44351", - "issue:43066|issue:44779" + "issue:29127|issue:30990" ], - "prepared_review_unit_hash": "908d3496c4ea28fc67f2799ce0c85f982aed720ff5f565764f3798b4717cea55", + "prepared_review_unit_hash": "e2dc9b6a0d0a29e3f5b8bac9ba6ded16e4c1156fc4b350741a151b5e5a50290b", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 33, - "serialized_chars": 37265, - "estimated_input_tokens": 9317, - "estimated_eval_tokens": 18890 + "soft_pair_count": 30, + "serialized_chars": 36340, + "estimated_input_tokens": 9085, + "estimated_eval_tokens": 18426 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is highly heterogeneous: most items are distinct transformers/model/tokenizer/import regressions rather than duplicates. None of the soft pairs looks like the same concrete bug or change, so all soft edges are rejected.", - "confidence": 0.84, - "canonical_issue_reason": "Issue 45278 is the broadest umbrella report in the set (multiple import errors after an upgrade) and is the closest thing to a cluster representative.", + "summary": "This is a mixed cluster of distinct Transformers issues, mostly around v5 import/tokenizer/processor regressions, plus several unrelated model-specific bugs. No soft pair looks like the same underlying change, so none should be merged as duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue 45278 has the widest scope and is the most plausible representative for the overall upgrade/import-breakage theme, even though the set is not a tight duplicate cluster.", + "best_issue_reason": "issue:45278 is the broadest umbrella for the repeated v5 import/loading regressions, but the cluster is too heterogeneous for a true duplicate canonical.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43756", - "right": "issue:44060", + "left": "issue:39401", + "right": "issue:44568", "accept": false, - "reason": "Different model families and different bug shapes: RoPE-layer configuration vs tied-weights warning." + "reason": "Different tokenizer bugs: Qwen3 offset mapping vs mdeberta-v3 BOS/EOS handling." }, { - "left": "issue:44117", - "right": "issue:44704", + "left": "issue:42831", + "right": "issue:45310", "accept": false, - "reason": "Tokenizer mapping fallback bug vs AutoProcessor/cached_file kwarg forwarding bug." + "reason": "Different failures: FineGrainedFP8 accuracy vs Qwen3.5-Moe from_pretrained error." }, { - "left": "issue:42907", - "right": "issue:44945", + "left": "issue:44263", + "right": "issue:44373", "accept": false, - "reason": "Saving dequantized models is unrelated to pipeline-parallelism output corruption." + "reason": "One is a runtime torch.split issue, the other is a docstring problem." }, { - "left": "issue:30990", - "right": "issue:33357", - "accept": false, - "reason": "Both involve loading, but one is a hang and the other is a MacOS bus error on a specific model." - }, - { - "left": "issue:44704", - "right": "issue:45290", + "left": "issue:44945", + "right": "issue:45753", "accept": false, - "reason": "Processor kwargs forwarding and chat-template tool-call handling are different code paths." + "reason": "Both are output problems, but one is pipeline parallelism and the other is an image-specific infinite loop." }, { - "left": "issue:43122", - "right": "issue:44514", + "left": "issue:42831", + "right": "issue:44038", "accept": false, - "reason": "Generic tokenizer regression vs batched chat-template crash with padding=False." + "reason": "Unrelated model bugs: FineGrainedFP8 accuracy vs Qwen3-VL-Moe compatibility." }, { - "left": "issue:43232", - "right": "issue:45245", + "left": "issue:42757", + "right": "issue:45278", "accept": false, - "reason": "Generation kwargs update bug and category-count runtime limit are unrelated." + "reason": "Specific import failure for one symbol vs a broad bundle of import errors; not the same bug." }, { - "left": "issue:44295", - "right": "issue:44297", + "left": "issue:43576", + "right": "issue:43784", "accept": false, - "reason": "Reading a registered buffer is not the same bug as tokenizer_class mismatch on save_pretrained." + "reason": "Different code paths: broken env command vs NameError importing sentence-transformers." }, { - "left": "issue:43784", - "right": "issue:45600", + "left": "issue:42886", + "right": "issue:43066", "accept": false, - "reason": "Import NameError in sentence-transformers vs stale auto_mappings references; different failures and fixes." + "reason": "Offline cache loading failure is distinct from wrong tokenizer decoder type." }, { - "left": "issue:41628", - "right": "issue:45278", + "left": "issue:43232", + "right": "issue:45593", "accept": false, - "reason": "Specific missing AutoImageProcessor symbol vs broad post-upgrade import errors; not the same concrete bug." + "reason": "Generation kwargs bug vs missing auxiliary losses in D-FINE training." }, { - "left": "issue:42907", - "right": "issue:43937", + "left": "issue:42371", + "right": "issue:44263", "accept": false, - "reason": "Dequantized save failure and invalid GenerationConfig are unrelated." + "reason": "TF32 settings guidance is unrelated to the torch.split indexer bug." }, { - "left": "issue:43723", - "right": "issue:44779", + "left": "issue:43756", + "right": "issue:44060", "accept": false, - "reason": "AutoTokenizer load failure vs incorrect Deepseek tokenization; different symptoms and likely different causes." + "reason": "Different model-specific regressions: Smollm3 RoPE layers vs Qwen3-Next tied-weights warning." }, { - "left": "issue:43723", - "right": "issue:44442", + "left": "issue:44117", + "right": "issue:44704", "accept": false, - "reason": "Both affect AutoTokenizer, but they target different tokenizer classes and the titles do not indicate the same root cause." + "reason": "Tokenizer mapping None handling and AutoProcessor kwargs passing are separate loading issues." }, { - "left": "issue:43408", - "right": "issue:43475", + "left": "issue:42907", + "right": "issue:44945", "accept": false, - "reason": "SAM3 model-type mismatch warning is separate from missing fpn_position_embeddings on a vision output object." + "reason": "Saving dequantized checkpoints is unrelated to pipeline-parallel inference output correctness." }, { - "left": "issue:44261", - "right": "issue:44373", + "left": "issue:30990", + "right": "issue:33357", "accept": false, - "reason": "Layernorm epsilon precision bug vs docstring typo for position_ids." + "reason": "Loading hang in Sentence Transformers vs MacOS bus error on a CLIP model." }, { - "left": "issue:44360", - "right": "issue:44485", + "left": "issue:44704", + "right": "issue:45290", "accept": false, - "reason": "DSA indexer activation issue and GLM-5 RoPE implementation are unrelated." + "reason": "AutoProcessor file loading and chat-template batching crashes are different code paths." }, { "left": "issue:43122", - "right": "issue:44521", + "right": "issue:44514", "accept": false, - "reason": "Different chat/tokenization regressions: tokenization changes vs all-zero assistant masks for multimodal inputs." + "reason": "Different tokenizer/processor regressions: tokenization changes vs batched apply_chat_template crash." }, { - "left": "issue:44625", - "right": "issue:44829", + "left": "issue:43232", + "right": "issue:45245", "accept": false, - "reason": "Config propagation bug for num_labels is unrelated to flash_attention_3 training degeneration." + "reason": "Generation state update bug vs a categorical-cardinality runtime error." }, { - "left": "issue:43116", - "right": "issue:45200", + "left": "issue:44295", + "right": "issue:44297", "accept": false, - "reason": "Example-script empty outputs and Gemma 4 mm_token_type_ids fine-tuning behavior are different issues." + "reason": "Position_ids buffer access and tokenizer_class mismatch are unrelated serialization issues." }, { - "left": "issue:29127", - "right": "issue:30990", + "left": "issue:43784", + "right": "issue:45600", "accept": false, - "reason": "Error-message clarity for LayoutLMv3 boxes is unrelated to Sentence Transformers loading hangs." + "reason": "Importing sentence-transformers failure is unrelated to auto_mappings references to removed configs." }, { "left": "issue:42907", - "right": "issue:44964", + "right": "issue:43937", "accept": false, - "reason": "Saving dequantized Ministrals/Devstrals is unrelated to loading Phi-4 multimodal models." + "reason": "Dequantized save failure vs invalid GenerationConfig are separate bugs." }, { - "left": "issue:43122", + "left": "issue:43723", "right": "issue:44779", "accept": false, - "reason": "Both mention tokenizer changes in v5, but one is a broad regression report and the other is Deepseek-specific incorrect output." - }, - { - "left": "issue:44038", - "right": "issue:44297", - "accept": false, - "reason": "Qwen3-VL-Moe bug and tokenizer_class mismatch on save_pretrained are different failure modes." - }, - { - "left": "issue:42907", - "right": "issue:43452", - "accept": false, - "reason": "Dequantized save failure does not match gguf_file loading failures for tokenizer/model classes." + "reason": "Tokenizer loading failure and incorrect Deepseek tokenization are not the same underlying defect." }, { - "left": "issue:39401", - "right": "issue:44488", + "left": "issue:41628", + "right": "issue:45278", "accept": false, - "reason": "Wrong offset_mapping in Qwen3 tokenizer is unrelated to failure loading cjvt/sleng-bert." + "reason": "One specific missing export vs many import errors overall; too broad to treat as the same bug." }, { - "left": "issue:44038", - "right": "issue:44295", + "left": "issue:43723", + "right": "issue:44442", "accept": false, - "reason": "Qwen3-VL-Moe behavior and position_ids buffer access are unrelated." + "reason": "FastSpeech2ConformerTokenizer loading failure is distinct from Deepseek tokenization correctness." }, { - "left": "issue:43824", - "right": "issue:44351", + "left": "issue:44261", + "right": "issue:44373", "accept": false, - "reason": "Different missing imports from different modules; no shared concrete code path is evident." + "reason": "MLA layernorm precision/config issue is unrelated to a position_ids docstring." }, { - "left": "issue:36296", - "right": "issue:43116", + "left": "issue:43408", + "right": "issue:43475", "accept": false, - "reason": "Tensor-parallel training bug and multi-label classification example output are unrelated." + "reason": "Same SAM 3 family, but different concrete bugs: model-type warning vs missing attribute crash." }, { - "left": "issue:44038", - "right": "issue:44222", + "left": "issue:44360", + "right": "issue:44485", "accept": false, - "reason": "Qwen3-VL-Moe bug and FP8 save_pretrained moe issue are separate model-path problems." + "reason": "DSA indexer ReLU discussion and GLM-5 RoPE implementation are unrelated." }, { - "left": "issue:39401", + "left": "issue:43122", "right": "issue:44521", "accept": false, - "reason": "Wrong offset_mapping and zero assistant masks are different tokenizer/chat-template bugs." + "reason": "Both touch templating/tokenization, but one is tokenization drift and the other is multimodal assistant masks." }, { - "left": "issue:45216", - "right": "issue:45478", + "left": "issue:44625", + "right": "issue:44829", "accept": false, - "reason": "Both involve Qwen3.5 and 5.4.0+, but one is checkpoint saving corruption and the other is a from_pretrained error; not the same concrete bug." + "reason": "Qwen3.5 num_labels propagation and flash_attention_3 degenerate training are unrelated." }, { - "left": "issue:42757", - "right": "issue:44351", + "left": "issue:43116", + "right": "issue:45200", "accept": false, - "reason": "Importing from huggingface_hub and importing from transformers are different libraries and different missing symbols." + "reason": "Example-script multi-label empty outputs and Gemma 4 mm_token_type_ids defaults are different bugs." }, { - "left": "issue:43066", - "right": "issue:44779", + "left": "issue:29127", + "right": "issue:30990", "accept": false, - "reason": "Tokenizer decoder type regression and Deepseek tokenization correctness are different tokenizer issues." + "reason": "LayoutLMv3 error messaging and Sentence Transformers loading hang are unrelated." } ] }, @@ -12112,63 +12079,76 @@ "issue:28218", "issue:29127", "issue:33453", + "issue:36296", "issue:38617", "issue:39401", "issue:42222", "issue:42371", - "issue:42673", + "issue:42757", "issue:42831", "issue:42907", "issue:43064", + "issue:43066", "issue:43116", "issue:43122", - "issue:43381", - "issue:43421", "issue:43441", - "issue:43450", + "issue:43452", "issue:43493", "issue:43673", "issue:43784", + "issue:43824", "issue:43825", - "issue:43827", "issue:43874", "issue:43937", "issue:43976", - "issue:43986", "issue:43992", "issue:44038", "issue:44075", - "issue:44117", "issue:44222", + "issue:44295", + "issue:44297", + "issue:44351", "issue:44485", + "issue:44488", + "issue:44521", "issue:44704", "issue:44779", "issue:44936", - "issue:44945", + "issue:44964", "issue:44977", - "issue:45072", - "issue:45081", "issue:45083", + "issue:45216", "issue:45278", - "issue:45405", "issue:45446", "issue:45478", "issue:45563", - "issue:45588", "issue:45753" ], "soft_pairs": [ + "issue:42907|issue:44964", + "issue:43122|issue:44779", + "issue:44038|issue:44297", + "issue:42907|issue:43452", + "issue:39401|issue:44488", + "issue:44038|issue:44295", + "issue:36296|issue:43116", + "issue:43824|issue:44351", + "issue:44038|issue:44222", + "issue:39401|issue:44521", + "issue:45216|issue:45478", + "issue:42757|issue:44351", + "issue:43066|issue:44779", "issue:22355|issue:28218", "issue:42907|issue:44222", - "issue:43992|issue:44704", "issue:43122|issue:44977", + "issue:43992|issue:44704", "issue:43784|issue:45446", "issue:43937|issue:44485", "issue:42222|issue:43441", "issue:42371|issue:43874", "issue:43976|issue:45446", - "issue:43493|issue:45753", "issue:42831|issue:45753", + "issue:43493|issue:45753", "issue:44075|issue:44936", "issue:43116|issue:43825", "issue:44038|issue:45478", @@ -12180,237 +12160,252 @@ "issue:38617|issue:45278", "issue:39401|issue:44779", "issue:33453|issue:39401", - "issue:44117|issue:45588", - "issue:22355|issue:29127", - "issue:43381|issue:44945", - "issue:43827|issue:45446", - "issue:43937|issue:44977", - "issue:45405|issue:45446", - "issue:43450|issue:43986", - "issue:42673|issue:45072", - "issue:44117|issue:45081", - "issue:39401|issue:43421" + "issue:22355|issue:29127" ], - "prepared_review_unit_hash": "a37e36c4fc4b0b4ffc5370fcabffdd67896c0e0881021198f527a5281d470803", + "prepared_review_unit_hash": "34b30488a9487adc13de1a0557d21f9670f726578b7780ee1eeac3d6b00d9d86", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 36913, - "estimated_input_tokens": 9229, - "estimated_eval_tokens": 18714 + "soft_pair_count": 36, + "serialized_chars": 37929, + "estimated_input_tokens": 9483, + "estimated_eval_tokens": 19222 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These issue pairs are mostly soft-similarity false positives: they span unrelated model-specific regressions, tokenizer behavior changes, import errors, and training/runtime bugs rather than the same underlying defect.", - "confidence": 0.32, - "canonical_issue_reason": "issue:45278 is the broadest umbrella report here: an update-triggered wave of import errors, making it the closest general duplicate target among the available issues.", + "summary": "This set is highly heterogeneous: tokenizer regressions, import errors, training bugs, multimodal issues, and model-specific failures are all mixed together. None of the soft pairs look like the same underlying bug/change, so I would not collapse this cluster.", + "confidence": 0.94, + "canonical_issue_reason": "No single canonical issue stands out because the items are mostly unrelated bugs across different code paths and models; this does not look like a true duplicate cluster.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the most suitable issue to anchor this cluster because it is open, broad, and can plausibly subsume nearby import-regression reports better than the highly specific model or tokenizer bugs.", + "best_issue_reason": "issue:44779 is the most self-contained concrete bug report, but only as a standalone issue; it does not represent the rest of the mixed cluster well.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:22355", - "right": "issue:28218", + "left": "issue:42907", + "right": "issue:44964", + "accept": false, + "reason": "Both involve model loading/saving, but they are different models and different failure modes: dequantized save vs multimodal load." + }, + { + "left": "issue:43122", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different models and symptoms; not the same concrete bug." + }, + { + "left": "issue:44038", + "right": "issue:44297", "accept": false, - "reason": "Different symptoms and code paths: missing module import vs tokenizer spacing after added tokens." + "reason": "Qwen3-VL-Moe bug vs tokenizer save_pretrained metadata mismatch; unrelated code paths." }, { "left": "issue:42907", + "right": "issue:43452", + "accept": false, + "reason": "Different operations and paths: saving dequantized models versus loading GGUF through AutoTokenizer/AutoModel." + }, + { + "left": "issue:39401", + "right": "issue:44488", + "accept": false, + "reason": "Qwen3 tokenizer offset mapping bug and sleng-bert load failure are unrelated." + }, + { + "left": "issue:44038", + "right": "issue:44295", + "accept": false, + "reason": "Different model bug versus a position_ids buffer read error; no shared concrete failure." + }, + { + "left": "issue:36296", + "right": "issue:43116", + "accept": false, + "reason": "Tensor-parallel training bug and example-script multi-label output bug are unrelated." + }, + { + "left": "issue:43824", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but for different symbols and likely different fixes." + }, + { + "left": "issue:44038", "right": "issue:44222", "accept": false, - "reason": "Both involve saving, but one is dequantized model saving and the other is FP8 MoE save_pretrained; not the same bug." + "reason": "Both mention MoE, but one is a Qwen3-VL-Moe issue and the other is FP8 save_pretrained; not the same bug." }, { - "left": "issue:43992", - "right": "issue:44704", + "left": "issue:39401", + "right": "issue:44521", + "accept": false, + "reason": "Tokenizer offset mapping and multimodal assistant mask generation are different behaviors and code paths." + }, + { + "left": "issue:45216", + "right": "issue:45478", + "accept": false, + "reason": "Both mention Qwen3.5 and v5.4+, but one is save_pretrained checkpoint correctness and the other is from_pretrained loading failure." + }, + { + "left": "issue:42757", + "right": "issue:44351", + "accept": false, + "reason": "Different missing imports from different packages; these are separate dependency/export issues." + }, + { + "left": "issue:43066", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer regressions in v5, but they concern different tokenizers and different symptoms." + }, + { + "left": "issue:22355", + "right": "issue:28218", + "accept": false, + "reason": "No module transformers.onnx and extra-space tokenization are unrelated." + }, + { + "left": "issue:42907", + "right": "issue:44222", "accept": false, - "reason": "Loading missing weights vs AutoProcessor kwargs forwarding are unrelated failure modes." + "reason": "Both are save-related, but dequantized save failure and FP8 save_pretrained failure are not the same concrete problem." }, { "left": "issue:43122", "right": "issue:44977", "accept": false, - "reason": "Tokenizer regression and flash-attention generation failure are different issues on different code paths." + "reason": "Tokenizer regression versus flash-attention generation failure; unrelated bugs." + }, + { + "left": "issue:43992", + "right": "issue:44704", + "accept": false, + "reason": "Missing embed_tokens.weight in from_pretrained and kwargs not forwarded to cached_file are different load-path issues." }, { "left": "issue:43784", "right": "issue:45446", "accept": false, - "reason": "Sentence-transformers import NameError vs an AuxRequest/PyTorch version check bug are unrelated." + "reason": "Sentence-transformers import NameError and flex_attention version-check bug are unrelated." }, { "left": "issue:43937", "right": "issue:44485", "accept": false, - "reason": "GLM-5 config validation and GLM-5 RoPE implementation are distinct problems." + "reason": "Both touch GLM-5, but generation config validation and RoPE implementation are distinct issues." }, { "left": "issue:42222", "right": "issue:43441", "accept": false, - "reason": "VitPose model breakage and Ministral flash-attention failure are unrelated." + "reason": "Different models and different failures: vitpose model breakage versus Ministral-3 FlashAttention failure." }, { "left": "issue:42371", "right": "issue:43874", "accept": false, - "reason": "TF32 API warning and missing video-patch helper are different subsystems and bugs." + "reason": "TF32 API messaging and a missing image-patch helper are unrelated." }, { "left": "issue:43976", "right": "issue:45446", "accept": false, - "reason": "Python support/version compatibility issue vs incorrect PyTorch version gating for a specific import." + "reason": "Python-version compatibility and flex_attention import version checks are separate issues." }, { - "left": "issue:43493", + "left": "issue:42831", "right": "issue:45753", "accept": false, - "reason": "SigLIP2 implementation discrepancy and Qwen3.5 image looping are not the same underlying bug." + "reason": "FP8 accuracy issue and Qwen3_5 infinite-loop-on-image are different bugs." }, { - "left": "issue:42831", + "left": "issue:43493", "right": "issue:45753", "accept": false, - "reason": "FP8 accuracy regression and infinite-loop generation bug are unrelated." + "reason": "SigLIP2 implementation discrepancy and Qwen3_5 generation loop are unrelated." }, { "left": "issue:44075", "right": "issue:44936", "accept": false, - "reason": "Optimizer argument handling and trainer evaluate-after-train failure are different problems." + "reason": "SGD args ignored and train/evaluate interaction failure are different training bugs." }, { "left": "issue:43116", "right": "issue:43825", "accept": false, - "reason": "Example-script multi-label output bug and pipeline deprecation/error-message issue are unrelated." + "reason": "Example script output bug and pipeline error-message bug are unrelated." }, { "left": "issue:44038", "right": "issue:45478", "accept": false, - "reason": "Different Qwen variants and different failure modes: VL-Moe behavior vs Qwen3.5 MoE from_pretrained error." + "reason": "Different Qwen3.5/Qwen3-VL-Moe failures: one is a model-specific bug, the other is a from_pretrained error." }, { "left": "issue:44222", "right": "issue:45478", "accept": false, - "reason": "FP8 save_pretrained on MoE is not the same as a from_pretrained loading error." + "reason": "FP8 save_pretrained and Qwen3.5 Moe from_pretrained are distinct code paths." }, { "left": "issue:43116", "right": "issue:43493", "accept": false, - "reason": "Classification example failure and SigLIP2 implementation discrepancy are unrelated." + "reason": "Multi-label classification example bug and SigLIP2 discrepancy are unrelated." }, { "left": "issue:43937", "right": "issue:45563", "accept": false, - "reason": "Invalid GenerationConfig and a stale warning about num_return_sequences are different issues." + "reason": "GenerationConfig validation and stale num_return_sequences warning are different generation-related issues." }, { "left": "issue:43064", "right": "issue:43673", "accept": false, - "reason": "FSDP/PEFT optimizer-state corruption and missing GenerationMixin cache are unrelated training vs generation bugs." + "reason": "Wrong optimizer states in FSDP2/PEFT training and missing GenerationMixin cache in chunked_prefill are unrelated." }, { "left": "issue:45083", "right": "issue:45446", "accept": false, - "reason": "Feature-extraction length behavior in qwen3_omni_moe vs a PyTorch version-check import bug are unrelated." + "reason": "Qwen3_omni_moe helper behavior and flex_attention import version check are unrelated." }, { "left": "issue:38617", "right": "issue:45278", "accept": false, - "reason": "45278 is broad, but this specific missing-symbol ImportError is not clearly the same underlying bug without stronger evidence." + "reason": "Specific missing layer_type_validation import and a broader upgrade-related import-error report are not the same bug." }, { "left": "issue:39401", "right": "issue:44779", "accept": false, - "reason": "Different tokenizer regressions affecting different models and outputs." + "reason": "Different tokenizers and different manifestations; not a duplicate." }, { "left": "issue:33453", "right": "issue:39401", "accept": false, - "reason": "Tokenizer loading regression and wrong offset_mapping are different symptoms." - }, - { - "left": "issue:44117", - "right": "issue:45588", - "accept": false, - "reason": "Tokenizer mapping fallback bug and flash_attention crash on s_aux=None are unrelated." + "reason": "Generic tokenizer loading regression and Qwen3 offset_mapping bug are too different to merge." }, { "left": "issue:22355", "right": "issue:29127", "accept": false, - "reason": "Missing transformers.onnx import and LayoutLMV3 box-validation messaging are unrelated." - }, - { - "left": "issue:43381", - "right": "issue:44945", - "accept": false, - "reason": "Gradient checkpointing in eval mode and incorrect output under pipeline parallelism are not the same bug." - }, - { - "left": "issue:43827", - "right": "issue:45446", - "accept": false, - "reason": "Documentation still referencing pipeline() and a flex_attention version-check bug are unrelated." - }, - { - "left": "issue:43937", - "right": "issue:44977", - "accept": false, - "reason": "GenerationConfig validation and flash-attention generation failure are different issues." - }, - { - "left": "issue:45405", - "right": "issue:45446", - "accept": false, - "reason": "PEFT minimum-version bump and a PyTorch import version check are unrelated dependency problems." - }, - { - "left": "issue:43450", - "right": "issue:43986", - "accept": false, - "reason": "Batched video shape bug and missing torchvision crash are different video-processing failures." - }, - { - "left": "issue:42673", - "right": "issue:45072", - "accept": false, - "reason": "VRAM leak across dataloader threads and dtype mismatch in bfloat16 inference are unrelated." - }, - { - "left": "issue:44117", - "right": "issue:45081", - "accept": false, - "reason": "Tokenizer mapping None handling and Mistral regex patch crashing on backend_tokenizer are different code paths." - }, - { - "left": "issue:39401", - "right": "issue:43421", - "accept": false, - "reason": "Wrong offset_mapping and runtime special-token post-processor update behavior are different tokenizer bugs." + "reason": "ONNX import failure and LayoutLMv3 error-message clarity are unrelated." } ] }, @@ -12435,55 +12430,63 @@ "cluster_id": "cluster-15354-391", "nodes": [ "issue:38617", + "issue:39401", "issue:41628", "issue:42175", "issue:42222", "issue:42371", "issue:42673", - "issue:42831", "issue:43012", "issue:43066", - "issue:43097", "issue:43122", - "issue:43335", + "issue:43381", "issue:43408", - "issue:43441", - "issue:43452", + "issue:43421", + "issue:43450", "issue:43577", "issue:43742", + "issue:43827", "issue:43874", - "issue:43906", - "issue:43931", "issue:43937", "issue:43986", - "issue:44038", "issue:44075", "issue:44117", "issue:44220", "issue:44261", - "issue:44263", "issue:44291", "issue:44410", "issue:44483", "issue:44484", - "issue:44485", - "issue:44493", "issue:44496", "issue:44829", + "issue:44945", "issue:44964", "issue:44977", - "issue:44987", "issue:45071", + "issue:45072", + "issue:45081", "issue:45216", "issue:45278", "issue:45310", + "issue:45405", + "issue:45446", "issue:45459", "issue:45464", "issue:45478", "issue:45542", + "issue:45588", "issue:45656" ], "soft_pairs": [ + "issue:44117|issue:45588", + "issue:43381|issue:44945", + "issue:43827|issue:45446", + "issue:43937|issue:44977", + "issue:45405|issue:45446", + "issue:43450|issue:43986", + "issue:42673|issue:45072", + "issue:44117|issue:45081", + "issue:39401|issue:43421", "issue:38617|issue:41628", "issue:44829|issue:45216", "issue:44410|issue:45310", @@ -12499,265 +12502,224 @@ "issue:43012|issue:43577", "issue:43066|issue:43122", "issue:42371|issue:44261", - "issue:43874|issue:44261", "issue:38617|issue:44220", "issue:44410|issue:45478", + "issue:43874|issue:44261", "issue:44484|issue:45478", - "issue:42222|issue:45459", - "issue:43097|issue:44493", - "issue:44263|issue:44485", - "issue:43441|issue:44829", - "issue:43012|issue:43408", - "issue:41628|issue:45071", - "issue:45071|issue:45542", - "issue:42673|issue:44977", - "issue:42371|issue:42831", - "issue:42673|issue:43906", - "issue:43931|issue:44038", - "issue:43986|issue:44484", - "issue:43335|issue:45071", - "issue:44117|issue:44987", - "issue:43097|issue:43452" + "issue:42222|issue:45459" ], - "prepared_review_unit_hash": "d021d25f88e8640c114bc78a6a38f277552c8ddb23e508d2f1265283b66111fe", + "prepared_review_unit_hash": "cfce38af58a126a060bf3e50d1efb1d5d3eeb96eeae4a4206584ff85da56303a", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 34, - "serialized_chars": 37711, - "estimated_input_tokens": 9428, - "estimated_eval_tokens": 19112 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35813, + "estimated_input_tokens": 8954, + "estimated_eval_tokens": 18164 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a heterogeneous cluster, not one true duplicate set. Most soft edges are false positives across unrelated bugs; the only clear duplicate-like pair is the TensorFlow backend packaging regression for the torch extra.", - "confidence": 0.91, - "canonical_issue_reason": "issue:45278 is the broadest current umbrella report about version-related import breakage after upgrading to 5.5.0, so it is the best representative issue for this mixed regression cluster.", + "summary": "This cluster is mostly unrelated transformer bug reports, with only one plausible duplicate-style tokenizer regression pair. The broadest representative issue is the upgrade-related regression report 45278.", + "confidence": 0.42, + "canonical_issue_reason": "45278 is the broadest umbrella report in the set: a general breakage after upgrading to 5.5.0, which makes it the best single anchor for this mixed cluster.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the most umbrella-like and upstream-facing issue here; it best represents the cluster even though the cluster is too mixed to treat as one duplicate family.", + "best_issue_reason": "45278 best represents the cluster because it captures the widest user-facing regression scope and can subsume narrower follow-up reports.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44117", + "right": "issue:45588", + "accept": false, + "reason": "Different subsystems: tokenizer mapping fallback vs flash-attention sinkless-model crash." + }, + { + "left": "issue:43381", + "right": "issue:44945", + "accept": false, + "reason": "Unrelated bugs: grad checkpointing in eval mode vs wrong outputs under pipeline parallelism." + }, + { + "left": "issue:43827", + "right": "issue:45446", + "accept": false, + "reason": "Docs cleanup about pipeline() vs a flex_attention import/version check bug." + }, + { + "left": "issue:43937", + "right": "issue:44977", + "accept": false, + "reason": "Different model-specific generation failures with no clear shared code path." + }, + { + "left": "issue:45405", + "right": "issue:45446", + "accept": false, + "reason": "PEFT version pinning issue is unrelated to the flex_attention import check." + }, + { + "left": "issue:43450", + "right": "issue:43986", + "accept": false, + "reason": "Batched video processor shape bug vs missing torchvision crash during AutoProcessor loading." + }, + { + "left": "issue:42673", + "right": "issue:45072", + "accept": false, + "reason": "VRAM leak in Qwen3ForCausalLM threads is unrelated to dtype mismatch in inference." + }, + { + "left": "issue:44117", + "right": "issue:45081", + "accept": false, + "reason": "Both involve tokenizer loading, but the concrete failures are different code paths and symptoms." + }, + { + "left": "issue:39401", + "right": "issue:43421", + "accept": false, + "reason": "Offset_mapping regression and runtime special-token post-processing are related only broadly, not the same bug." + }, { "left": "issue:38617", "right": "issue:41628", "accept": false, - "reason": "Both are import failures, but they involve different missing symbols and different code paths; not the same bug." + "reason": "Both are import errors, but they involve different missing symbols and different breakages." }, { "left": "issue:44829", "right": "issue:45216", "accept": false, - "reason": "One is a FlashAttention 3 training degeneration issue; the other is a save_pretrained checkpoint regression. Different failures." + "reason": "Flash-attention training degeneration is unrelated to save_pretrained checkpoint corruption." }, { "left": "issue:44410", "right": "issue:45310", "accept": false, - "reason": "Different Qwen3-related failures: missing projections in a model layer versus a from_pretrained error." + "reason": "Different Qwen model families and different loading failures; not the same underlying defect." }, { "left": "issue:44291", "right": "issue:45071", "accept": false, - "reason": "One is init_empty_weights argument handling; the other is PreTrainedConfig type checking. Different regressions." + "reason": "init_empty_weights argument handling and PretrainedConfig type checking are separate loading regressions." }, { "left": "issue:42222", "right": "issue:45542", "accept": false, - "reason": "Different model/package problems; vitpose breakage is unrelated to TensorFlow backend installation." + "reason": "VitPose model breakage is unrelated to the tensorboard/TensorFlow backend import issue." }, { "left": "issue:44484", - "right": "issue:45310", + "right": "issue:45464", "accept": false, - "reason": "max_shard_size behavior has no concrete overlap with the Qwen3.5 MoE from_pretrained regression." + "reason": "Chat/completions request validation and streaming inference failure are different API problems." }, { "left": "issue:42175", "right": "issue:42222", "accept": false, - "reason": "TensorFlow backend packaging issue versus a vitpose model breakage; unrelated." + "reason": "Package dependency/backend installation issue vs a broken model implementation." }, { "left": "issue:43937", "right": "issue:44496", "accept": false, - "reason": "GenerationConfig validation failure and unrecognized model/config.json failure are different loading errors." + "reason": "GenerationConfig validation in GLM-5 is unrelated to unrecognized-model config loading." }, { "left": "issue:43408", "right": "issue:43742", "accept": false, - "reason": "A model-type warning for sam3_tracker is unrelated to the MobileLLM key error." + "reason": "Model-type warning for sam3_tracker is not the same as MobileLLM config loading failure." }, { "left": "issue:44483", "right": "issue:45464", "accept": false, - "reason": "Both involve chat/completions-style failures, but one is an API acceptance issue and the other is streaming inference on a specific model; not the same concrete bug." + "reason": "Both mention chat/completions, but one is endpoint request acceptance and the other is streaming generation failure on a specific model." }, { "left": "issue:44075", "right": "issue:45656", "accept": false, - "reason": "Both concern optimizers, but one is ignored SGD args while the other is double optimizer stepping with deepspeed." + "reason": "SGD argument handling and double optimizer steps under deepspeed are different optimizer bugs." }, { "left": "issue:44964", "right": "issue:45278", "accept": false, - "reason": "A specific multimodal model load failure is not the same as a broad report of import errors after upgrade." + "reason": "A specific model load failure is not the same as the broad import-error regression report." }, { "left": "issue:43012", "right": "issue:43577", "accept": false, - "reason": "One is a PyTorch compile precision warning; the other is a dtype propagation bug on BLIP2. Different issues." + "reason": "Both touch dtype/precision, but one is a compile-time warning and the other is incorrect loaded dtypes." }, { "left": "issue:43066", "right": "issue:43122", - "accept": false, - "reason": "Both are tokenizer regressions, but the reports describe different symptoms and no shared concrete failure mode is established." + "accept": true, + "reason": "These look like the same tokenizer v5 regression: a wrong decoder type can directly explain changed tokenization behavior across versions." }, { "left": "issue:42371", "right": "issue:44261", "accept": false, - "reason": "TF32 configuration warnings and MLA layernorm epsilon precision errors are unrelated." - }, - { - "left": "issue:43874", - "right": "issue:44261", - "accept": false, - "reason": "Missing image-patch method in GLM46V is unrelated to the MLA q_a_layernorm epsilon issue." + "reason": "TF32 behavior API guidance is unrelated to the missing rms_norm_eps precision issue." }, { "left": "issue:38617", "right": "issue:44220", "accept": false, - "reason": "An import error for configuration_utils versus an audio feature extraction bug are unrelated." + "reason": "ImportError for a config helper and an audio feature-extraction issue are unrelated." }, { "left": "issue:44410", "right": "issue:45478", "accept": false, - "reason": "Different Qwen3-family loading bugs; missing projections is not the same as a from_pretrained error." + "reason": "Different Qwen model-loading failures; no evidence they share the same fix." + }, + { + "left": "issue:43874", + "right": "issue:44261", + "accept": false, + "reason": "Multimodal token counting bug vs MLA layer norm epsilon precision issue." }, { "left": "issue:44484", "right": "issue:45478", "accept": false, - "reason": "max_shard_size behavior and Qwen3.5 MoE loading errors are unrelated." + "reason": "Request acceptance at chat/completions and Qwen3.5 MoE loading are separate bugs." }, { "left": "issue:42222", "right": "issue:45459", "accept": false, - "reason": "vitpose model breakage is unrelated to tokenizer protobuf error masking." - }, - { - "left": "issue:43097", - "right": "issue:44493", - "accept": false, - "reason": "Removed tie_embeddings_and_encoder_decoder API and unexpected position-id keys are different regression classes." - }, - { - "left": "issue:44263", - "right": "issue:44485", - "accept": false, - "reason": "torch.split return values in an indexer and a RoPE implementation issue are unrelated." - }, - { - "left": "issue:43441", - "right": "issue:44829", - "accept": false, - "reason": "Both mention FlashAttention, but they are different model-specific failures and not the same code-path bug." - }, - { - "left": "issue:43012", - "right": "issue:43408", - "accept": false, - "reason": "A dtype/compile warning and a sam3 model-type warning are unrelated." - }, - { - "left": "issue:41628", - "right": "issue:45071", - "accept": false, - "reason": "AutoImageProcessor import failure and PreTrainedConfig type-check regression are unrelated." - }, - { - "left": "issue:45071", - "right": "issue:45542", - "accept": false, - "reason": "PreTrainedConfig type checking and TensorFlow backend installation are different problems." - }, - { - "left": "issue:42673", - "right": "issue:44977", - "accept": false, - "reason": "VRAM leak in Qwen3ForCausalLM under multithreaded dataloaders is unrelated to Qwen3.5 flash-attention generation issues." - }, - { - "left": "issue:42371", - "right": "issue:42831", - "accept": false, - "reason": "TF32 settings and FineGrainedFP8 accuracy are different numerical issues." - }, - { - "left": "issue:42673", - "right": "issue:43906", - "accept": false, - "reason": "A Qwen3ForCausalLM VRAM leak and a reproduction of another issue are not the same bug." - }, - { - "left": "issue:43931", - "right": "issue:44038", - "accept": false, - "reason": "Weight-shape mismatch for Qwen3-VL-30B and a Qwen3-VL-Moe bug are different model-loading failures." - }, - { - "left": "issue:43986", - "right": "issue:44484", - "accept": false, - "reason": "AutoProcessor without torchvision and max_shard_size have no overlap." - }, - { - "left": "issue:43335", - "right": "issue:45071", - "accept": false, - "reason": "SwitchTransformers sparse-layer creation and PreTrainedConfig type checking are unrelated." - }, - { - "left": "issue:44117", - "right": "issue:44987", - "accept": false, - "reason": "TOKENIZER_MAPPING_NAMES returning None and a model-loading failure for physical-intelligence/fast are different bugs." - }, - { - "left": "issue:43097", - "right": "issue:43452", - "accept": false, - "reason": "Removed tie_embeddings API and gguf_file breaking AutoTokenizer/AutoModel loading are unrelated." + "reason": "VitPose model breakage is unrelated to tokenizer protobuf decode-error masking." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative about the mostly unrelated set of issues. The chosen anchor issue 45278 is a reasonable broad representative for the upgrade/regression subset, and the only accepted soft pair (43066/43122) is plausibly the same tokenizer regression." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -12775,22 +12737,23 @@ "cluster_id": "cluster-15354-391", "nodes": [ "issue:38617", - "issue:42617", + "issue:41628", + "issue:42371", "issue:42673", "issue:42757", + "issue:42831", "issue:43012", "issue:43064", "issue:43066", "issue:43097", - "issue:43208", + "issue:43335", "issue:43408", + "issue:43441", + "issue:43452", "issue:43502", - "issue:43550", "issue:43577", - "issue:43825", "issue:43873", "issue:43874", - "issue:43901", "issue:43906", "issue:43931", "issue:43976", @@ -12807,276 +12770,296 @@ "issue:44410", "issue:44484", "issue:44485", - "issue:44655", + "issue:44493", "issue:44704", "issue:44829", "issue:44945", "issue:44964", + "issue:44977", + "issue:44987", "issue:45071", "issue:45103", "issue:45245", - "issue:45290", "issue:45341", - "issue:45372", "issue:45375", - "issue:45561", + "issue:45542", "issue:45588" ], "soft_pairs": [ + "issue:43097|issue:44493", + "issue:44263|issue:44485", + "issue:43441|issue:44829", + "issue:43012|issue:43408", + "issue:41628|issue:45071", + "issue:45071|issue:45542", + "issue:42673|issue:44977", + "issue:42371|issue:42831", + "issue:42673|issue:43906", + "issue:43931|issue:44038", + "issue:43986|issue:44484", + "issue:43335|issue:45071", + "issue:44117|issue:44987", + "issue:43097|issue:43452", "issue:43976|issue:44945", - "issue:43874|issue:44263", "issue:43408|issue:44368", + "issue:43874|issue:44263", "issue:44829|issue:45245", "issue:38617|issue:42757", "issue:43064|issue:44315", "issue:44410|issue:45375", "issue:38617|issue:44351", - "issue:44062|issue:44117", - "issue:42673|issue:43577", "issue:43097|issue:45588", + "issue:44062|issue:44117", "issue:44704|issue:45103", "issue:44038|issue:44410", + "issue:42673|issue:43577", "issue:44964|issue:45071", "issue:43931|issue:44410", - "issue:44964|issue:45341", "issue:44261|issue:44485", + "issue:44964|issue:45341", "issue:43502|issue:43992", "issue:43986|issue:44117", "issue:43097|issue:43873", - "issue:43066|issue:43906", - "issue:43986|issue:45290", - "issue:45341|issue:45561", - "issue:43986|issue:45372", - "issue:43986|issue:45341", - "issue:43097|issue:43901", - "issue:42617|issue:45341", - "issue:43825|issue:44655", - "issue:43992|issue:44484", - "issue:45341|issue:45588", - "issue:43012|issue:43550", - "issue:43208|issue:43550", - "issue:43012|issue:44368" + "issue:43066|issue:43906" ], - "prepared_review_unit_hash": "e022333c89b9288531bc3b04aa4d01da2c711ed4d99203527dc0ff9bd63461e8", + "prepared_review_unit_hash": "d6bf840f7d22994a4c9df8ce1ec1ee71e76a279fbf855087293664fefc274f73", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 33, - "serialized_chars": 37020, - "estimated_input_tokens": 9255, - "estimated_eval_tokens": 18766 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 38130, + "estimated_input_tokens": 9533, + "estimated_eval_tokens": 19322 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: the soft pairs are mostly superficial title similarity across different subsystems (imports, model loading, training, quantization, docs, and testing). I would not merge any of the proposed soft edges as duplicates or same-change PRs.", - "confidence": 0.94, - "canonical_issue_reason": null, + "summary": "This cluster is mostly false-positive similarity across disparate Transformers issues. The items share broad v5/API/model-regression language, but none of the soft pairs look like the same underlying bug or change.", + "confidence": 0.23, + "canonical_issue_reason": "Issue 43097 is the cleanest representative of the cluster\u2019s broad v5 compatibility-breakage theme: it is concrete, user-facing, and about a specific API removal. That said, the set does not contain a strong true-duplicate group.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43986 is the clearest self-contained regression report, but the set is too diverse for a true canonical duplicate representative.", + "best_issue_reason": "43097 is the best issue to anchor on if this cluster must be collapsed, because it is the most general and actionable regression report among the candidates. The rest are mostly model-specific, import-specific, or unrelated behavioral bugs.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43976", - "right": "issue:44945", + "left": "issue:43097", + "right": "issue:44493", "accept": false, - "reason": "Both are runtime problems, but one is Python-version compatibility and the other is pipeline-parallel inference correctness; different bugs." + "reason": "One is a removed embedding-tie API complaint; the other is an unexpected position-id key warning. Different failure modes and code paths." }, { - "left": "issue:43874", - "right": "issue:44263", + "left": "issue:44263", + "right": "issue:44485", "accept": false, - "reason": "Same broad GLM area, but one is a missing processor method and the other is a torch.split indexing issue; not the same code-path." + "reason": "Torch split return-value issue in a MoE indexer is not the same as a RoPE implementation discussion." }, { - "left": "issue:43408", - "right": "issue:44368", + "left": "issue:43441", + "right": "issue:44829", "accept": false, - "reason": "Both mention warnings, but one is a model-type mismatch and the other is a tie-word-embeddings config warning; distinct issues." + "reason": "Both mention flash attention, but one is a Ministral-3 load failure and the other is degenerate training with FA3. Different bugs." }, { - "left": "issue:44829", - "right": "issue:45245", + "left": "issue:43012", + "right": "issue:43408", "accept": false, - "reason": "Training degeneration with flash_attention_3 is unrelated to the 2^24 category limit runtime error." + "reason": "bfloat16 compilation warning is unrelated to the SAM3 model-type warning during AutoProcessor/model loading." }, { - "left": "issue:38617", - "right": "issue:42757", + "left": "issue:41628", + "right": "issue:45071", "accept": false, - "reason": "Both are import errors, but for different missing symbols in different modules; not duplicates." + "reason": "AutoImageProcessor import failure and PretrainedConfig type-checking breakage are unrelated API regressions." }, { - "left": "issue:43064", - "right": "issue:44315", + "left": "issue:45071", + "right": "issue:45542", "accept": false, - "reason": "Different training stack failures: wrong optimizer state on FSDP2/PEFT vs Liger Kernel not applied with model_init." + "reason": "Type-checking breakage in PretrainedConfig is not the same as the TensorFlow-backend detection error." }, { - "left": "issue:44410", - "right": "issue:45375", + "left": "issue:42673", + "right": "issue:44977", "accept": false, - "reason": "Both touch Qwen config/model structure, but one is missing attention projections and the other is a dropped config field." + "reason": "VRAM leak in multithreaded Qwen3ForCausalLM and flash-attention generation corruption are different problems." }, { - "left": "issue:38617", - "right": "issue:44351", + "left": "issue:42371", + "right": "issue:42831", "accept": false, - "reason": "ImportError in unrelated symbols/modules; no shared underlying bug." + "reason": "TF32 settings warning is not the same as FineGrainedFP8 accuracy issues." }, { - "left": "issue:44062", - "right": "issue:44117", + "left": "issue:42673", + "right": "issue:43906", "accept": false, - "reason": "Tokenizer construction TypeError vs mapping lookup assuming non-None; different failure points." + "reason": "A Qwen3 VRAM leak is not the same as an isolated reproduction of an unrelated earlier issue." }, { - "left": "issue:42673", - "right": "issue:43577", + "left": "issue:43931", + "right": "issue:44038", "accept": false, - "reason": "VRAM leak under dataloader threads is unrelated to BLIP2 dtype staying float32." + "reason": "Weight-shape mismatch on one Qwen3-VL model is distinct from a broader transformers-5/Qwen3-VL-Moe failure report." }, { - "left": "issue:43097", - "right": "issue:45588", + "left": "issue:43986", + "right": "issue:44484", "accept": false, - "reason": "A removed config attribute warning is unrelated to the flash_attention crash on s_aux=None." + "reason": "Missing torchvision crash for video models is unrelated to max_shard_size behavior." }, { - "left": "issue:44704", - "right": "issue:45103", + "left": "issue:43335", + "right": "issue:45071", "accept": false, - "reason": "One is kwargs not forwarded in AutoProcessor, the other is an auto_docstring annotations crash; different code paths." + "reason": "SwitchTransformers sparse-layer construction bug is unrelated to PretrainedConfig type checking." }, { - "left": "issue:44038", - "right": "issue:44410", + "left": "issue:44117", + "right": "issue:44987", "accept": false, - "reason": "Different Qwen families and different breakages: model loading regression vs missing layer projections." + "reason": "Tokenizer mapping None-handling is not the same as a model-loading failure for physical-intelligence/fast." }, { - "left": "issue:44964", - "right": "issue:45071", + "left": "issue:43097", + "right": "issue:43452", + "accept": false, + "reason": "Removed tie-embeddings API and gguf_file failing in tokenizer/model loading are different issues." + }, + { + "left": "issue:43976", + "right": "issue:44945", + "accept": false, + "reason": "Python version compatibility problem and incorrect output under pipeline parallelism are unrelated." + }, + { + "left": "issue:43408", + "right": "issue:44368", + "accept": false, + "reason": "SAM3 model-type warning and tie_word_embeddings warning are different configuration warnings with different causes." + }, + { + "left": "issue:43874", + "right": "issue:44263", "accept": false, - "reason": "Model loading mismatch and PretrainedConfig type-checking are separate regressions." + "reason": "Missing image-patch method in a processor is not the same as a torch.split return-value issue." }, { - "left": "issue:43931", - "right": "issue:44410", + "left": "issue:44829", + "right": "issue:45245", "accept": false, - "reason": "Weight-shape mismatch for Qwen3-VL is not the same as missing qwen3next projections." + "reason": "FlashAttention-3 training degeneration is unrelated to the 2^24 category limit runtime error." }, { - "left": "issue:44964", - "right": "issue:45341", + "left": "issue:38617", + "right": "issue:42757", "accept": false, - "reason": "A multimodal model load failure and a testing_utils bug are unrelated." + "reason": "These are different import errors from different modules and symbols." }, { - "left": "issue:44261", - "right": "issue:44485", + "left": "issue:43064", + "right": "issue:44315", "accept": false, - "reason": "Precision/eps handling in MLA is unrelated to GLM-5 RoPE implementation." + "reason": "Wrong optimizer states under FSDP2/PEFT and Liger kernel not applied with model_init are different training-stack bugs." }, { - "left": "issue:43502", - "right": "issue:43992", + "left": "issue:44410", + "right": "issue:45375", "accept": false, - "reason": "Local-files-only network leak and missing embed_tokens.weight are different loading bugs." + "reason": "Missing qkv/gate projections in qwen3next is unrelated to a missing vision config field." }, { - "left": "issue:43986", - "right": "issue:44117", + "left": "issue:38617", + "right": "issue:44351", "accept": false, - "reason": "Video processor dependency crash vs tokenizer mapping None assumption; unrelated." + "reason": "Both are import errors, but they concern different missing names and likely different compatibility fixes." }, { "left": "issue:43097", - "right": "issue:43873", + "right": "issue:45588", "accept": false, - "reason": "Config deprecation/warning issue is unrelated to offloading/quantization behavior." + "reason": "Embedding-tie API removal is unrelated to a flash_attention.py crash on s_aux=None." }, { - "left": "issue:43066", - "right": "issue:43906", + "left": "issue:44062", + "right": "issue:44117", "accept": false, - "reason": "Wrong tokenizer decoder type and the linked reproduction issue do not describe the same bug." + "reason": "AddedToken constructor argument collision is not the same as tokenizer-mapping None handling." }, { - "left": "issue:43986", - "right": "issue:45290", + "left": "issue:44704", + "right": "issue:45103", "accept": false, - "reason": "One is a video-model torchvision dependency crash; the other is a chat-template tool-call crash." + "reason": "AutoProcessor kwargs forwarding and auto_docstring crashes with future annotations are unrelated." }, { - "left": "issue:45341", - "right": "issue:45561", + "left": "issue:44038", + "right": "issue:44410", "accept": false, - "reason": "Both mention testing utils, but one is a generic bug report and the other is a pytest-xdist file race; not enough evidence of the same bug." + "reason": "A Qwen3-VL-Moe loading bug is different from a qwen3next layer-structure bug." }, { - "left": "issue:43986", - "right": "issue:45372", + "left": "issue:42673", + "right": "issue:43577", "accept": false, - "reason": "Video model loading without torchvision is unrelated to the mistral_common ReasoningEffort import failure." + "reason": "VRAM leak under dataloader threads is unrelated to BLIP2 dtype propagation problems." }, { - "left": "issue:43986", - "right": "issue:45341", + "left": "issue:44964", + "right": "issue:45071", "accept": false, - "reason": "Completely different areas: video-model loading vs testing_utils." + "reason": "Phi-4 multimodal load failure and PretrainedConfig type-checking breakage are unrelated." }, { - "left": "issue:42617", - "right": "issue:45341", + "left": "issue:43931", + "right": "issue:44410", "accept": false, - "reason": "3d_parallel.py execution failure is unrelated to the testing_utils bug." + "reason": "Weight-shape mismatch in Qwen3-VL is not the same as missing projections in qwen3next." }, { - "left": "issue:43825", - "right": "issue:44655", + "left": "issue:44261", + "right": "issue:44485", "accept": false, - "reason": "Pipeline translation-task messaging and save_pretrained Pipeline serialization are different issues." + "reason": "Missing rms_norm_eps in MLA q_a_layernorm and GLM-5 RoPE implementation are separate model-specific issues." }, { - "left": "issue:43992", - "right": "issue:44484", + "left": "issue:44964", + "right": "issue:45341", "accept": false, - "reason": "Missing encoder weights on load is unrelated to the max_shard_size default question." + "reason": "Model loading failure and a testing_utils bug are unrelated." }, { - "left": "issue:45341", - "right": "issue:45588", + "left": "issue:43502", + "right": "issue:43992", "accept": false, - "reason": "Testing utils bug and flash_attention s_aux=None crash are unrelated." + "reason": "local_files_only still making API requests is unrelated to missing embed_tokens weights on UMT5Encoder load." }, { - "left": "issue:43012", - "right": "issue:43550", + "left": "issue:43986", + "right": "issue:44117", "accept": false, - "reason": "A PyTorch warning about float32 precision is not the same as a torch.compile SDPA failure." + "reason": "Video-model crashes without torchvision and tokenizer mapping None assumptions are unrelated." }, { - "left": "issue:43208", - "right": "issue:43550", + "left": "issue:43097", + "right": "issue:43873", "accept": false, - "reason": "xLSTM training blockers and Bamba SDPA compile failure are distinct problems." + "reason": "Removed tie-embeddings API and offloading/quantization behavior are different concerns." }, { - "left": "issue:43012", - "right": "issue:44368", + "left": "issue:43066", + "right": "issue:43906", "accept": false, - "reason": "Different warning classes and different subsystems; not the same bug." + "reason": "Wrong tokenizer decoder type in v5 and an isolated reproduction issue are not the same bug." } ] }, @@ -13103,6 +13086,7 @@ "issue:36010", "issue:36683", "issue:38617", + "issue:42617", "issue:42757", "issue:42886", "issue:42907", @@ -13110,45 +13094,57 @@ "issue:43012", "issue:43064", "issue:43066", + "issue:43097", "issue:43208", "issue:43299", "issue:43352", "issue:43381", - "issue:43441", "issue:43475", "issue:43502", - "issue:43525", "issue:43526", "issue:43531", + "issue:43550", "issue:43618", - "issue:43756", "issue:43761", "issue:43824", + "issue:43825", "issue:43867", - "issue:43881", "issue:43901", - "issue:43906", "issue:43937", - "issue:43976", + "issue:43986", + "issue:43992", "issue:44038", "issue:44220", "issue:44230", - "issue:44242", - "issue:44263", + "issue:44368", "issue:44380", "issue:44483", - "issue:44485", + "issue:44484", "issue:44514", "issue:44655", "issue:44683", "issue:45216", "issue:45290", "issue:45341", + "issue:45372", "issue:45381", - "issue:45440", + "issue:45561", + "issue:45588", "issue:45753" ], "soft_pairs": [ + "issue:43986|issue:45290", + "issue:45341|issue:45561", + "issue:43986|issue:45372", + "issue:43097|issue:43901", + "issue:42617|issue:45341", + "issue:43986|issue:45341", + "issue:43825|issue:44655", + "issue:45341|issue:45588", + "issue:43992|issue:44484", + "issue:43012|issue:43550", + "issue:43208|issue:43550", + "issue:43012|issue:44368", "issue:42886|issue:43381", "issue:42907|issue:45753", "issue:42907|issue:44655", @@ -13158,12 +13154,12 @@ "issue:43012|issue:44683", "issue:42757|issue:43824", "issue:44038|issue:44230", - "issue:43618|issue:43761", "issue:38617|issue:43502", + "issue:43618|issue:43761", "issue:44483|issue:45290", - "issue:44483|issue:45381", - "issue:43299|issue:43867", "issue:43064|issue:44380", + "issue:43299|issue:43867", + "issue:44483|issue:45381", "issue:43901|issue:44655", "issue:36683|issue:43475", "issue:43867|issue:45216", @@ -13171,251 +13167,256 @@ "issue:43352|issue:43937", "issue:43066|issue:45381", "issue:36010|issue:42757", - "issue:44220|issue:44655", - "issue:43756|issue:44485", - "issue:44263|issue:45440", - "issue:36683|issue:43525", - "issue:43618|issue:45216", - "issue:43906|issue:45753", - "issue:36683|issue:44242", - "issue:43299|issue:45753", - "issue:36683|issue:43881", - "issue:43352|issue:43976", - "issue:43441|issue:44380" + "issue:44220|issue:44655" ], - "prepared_review_unit_hash": "fedaa05fdf25d9f19964b46a3279392bc6b2654d50058a96d6a5499b8932ba89", + "prepared_review_unit_hash": "0ca3a5f689b20593d0ab6ff1bd7f9c259a6555fed6fecca380470a70314df04d", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 33, - "serialized_chars": 36487, - "estimated_input_tokens": 9122, - "estimated_eval_tokens": 18500 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 37884, + "estimated_input_tokens": 9471, + "estimated_eval_tokens": 19198 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is not a duplicate set: it spans many unrelated Transformers regressions and API changes (imports, model config attributes, saving/loading, chat templating, attention backends, MoE, and offline behavior). All soft-similarity edges look superficial rather than the same underlying bug.", - "confidence": 0.96, - "canonical_issue_reason": "No single canonical issue: the items describe distinct bugs across different subsystems and models, so merging the cluster around one issue would be misleading.", + "summary": "This cluster is mostly heterogeneous Transformers bugs that only superficially overlap by subsystem or version. I found one likely true duplicate pair in CLIP vision outputs; the rest should stay separate.", + "confidence": 0.71, + "canonical_issue_reason": "issue:43761 is the best anchor because it clearly describes the CLIPVisionModel.forward regression and is broad enough to cover the paired CLIP attentions-output omission in issue:43618.", "canonical_pr_reason": null, - "best_issue_reason": "No global best issue stands out; even the broadest reports are model- or subsystem-specific and do not represent the rest of the cluster well.", + "best_issue_reason": "issue:43761 is the strongest standalone report: it has a concrete symptom, a regression framing, and a specific code path/model component, making it the most useful issue to keep as the representative bug.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43986", + "right": "issue:45290", + "accept": false, + "reason": "Different failures: AutoProcessor loading without torchvision vs chat template tokenization on assistant tool-call messages." + }, + { + "left": "issue:45341", + "right": "issue:45561", + "accept": false, + "reason": "Both involve testing utilities, but one is a small utility bug and the other is a pytest-xdist race; not the same underlying defect." + }, + { + "left": "issue:43986", + "right": "issue:45372", + "accept": false, + "reason": "Unrelated crashes in different areas: video processor loading vs mistral_common ReasoningEffort import failure." + }, + { + "left": "issue:43097", + "right": "issue:43901", + "accept": false, + "reason": "One is a removed API without notice; the other is outdated docs for TextClassificationPipeline. Similar theme, different bug." + }, + { + "left": "issue:42617", + "right": "issue:45341", + "accept": false, + "reason": "3d_parallel.py runtime failure is unrelated to a testing_utils issue." + }, + { + "left": "issue:43986", + "right": "issue:45341", + "accept": false, + "reason": "Different components and symptoms; no shared code-path evidence." + }, + { + "left": "issue:43825", + "right": "issue:44655", + "accept": false, + "reason": "Pipeline error-message wording vs saving Pipeline objects are separate issues." + }, + { + "left": "issue:45341", + "right": "issue:45588", + "accept": false, + "reason": "Testing utils bug vs flash_attention crash on sink-less models are different code paths." + }, + { + "left": "issue:43992", + "right": "issue:44484", + "accept": false, + "reason": "Model weight-loading omission vs save_pretrained shard-size question; not the same bug." + }, + { + "left": "issue:43012", + "right": "issue:43550", + "accept": false, + "reason": "A precision warning during compile is not the same as a torch.compile SDPA failure." + }, + { + "left": "issue:43208", + "right": "issue:43550", + "accept": false, + "reason": "xLSTM training bugs and Bamba SDPA compile failure are unrelated." + }, + { + "left": "issue:43012", + "right": "issue:44368", + "accept": false, + "reason": "Both are warnings, but they concern different behaviors and different configs." + }, { "left": "issue:42886", "right": "issue:43381", "accept": false, - "reason": "Offline tokenizer cache loading and gradient-checkpointing eval-mode behavior are unrelated bugs." + "reason": "Offline-cache loading and gradient checkpointing in eval mode are separate regressions." }, { "left": "issue:42907", "right": "issue:45753", "accept": false, - "reason": "Both are model-specific failures, but dequantized model saving and a Qwen3.5 image loop are different problems." + "reason": "Saving dequantized models vs an infinite-loop generation bug are unrelated." }, { "left": "issue:42907", "right": "issue:44655", "accept": false, - "reason": "Saving dequantized models and saving Pipeline objects touch the same verb but not the same code path or bug." + "reason": "Both touch saving, but one is quantized model serialization and the other is Pipeline.save_pretrained." }, { "left": "issue:43531", "right": "issue:45341", "accept": false, - "reason": "Qwen3-MoE sliding-window handling and a testing_utils bug are unrelated." + "reason": "Qwen3-MoE sliding_window behavior is unrelated to a testing_utils bug." }, { "left": "issue:43208", "right": "issue:43526", "accept": false, - "reason": "xLSTM training bugs and BEiTImageProcessorFast label reduction are separate issues in different components." + "reason": "xLSTM training issues and BeitImageProcessorFast reduce_labels behavior do not share a concrete bug." }, { "left": "issue:44483", "right": "issue:44514", "accept": false, - "reason": "Chat-completions request validation and batched apply_chat_template padding crashes are different layers and failures." + "reason": "Chat/completions API rejection and apply_chat_template batching crash are different layers and different failures." }, { "left": "issue:43012", "right": "issue:44683", "accept": false, - "reason": "A PyTorch warning during bfloat16 compile is not the same as compiled flex_attention failing on torch>=2.9." + "reason": "Precision warning during bfloat16 compile is not the same as flex_attention compile failure on torch>=2.9." }, { "left": "issue:42757", "right": "issue:43824", "accept": false, - "reason": "These are both import errors, but they concern different missing symbols from different modules." + "reason": "Both are import errors, but for different symbols, different packages, and different breakages." }, { "left": "issue:44038", "right": "issue:44230", "accept": false, - "reason": "Qwen3-VL-MoE breakage and fp8 support for Qwen3-VL/Qwen3.5-MoE are not the same concrete bug." - }, - { - "left": "issue:43618", - "right": "issue:43761", - "accept": false, - "reason": "Both involve CLIP, but one drops attentions and the other returns hidden_states=None; different regressions." + "reason": "Related model family, but one is a generic Qwen3-VL-Moe regression and the other is fp8 inference support; not the same issue." }, { "left": "issue:38617", "right": "issue:43502", "accept": false, - "reason": "An import error in configuration_utils and unexpected network calls despite local_files_only=True are unrelated." + "reason": "Importing layer_type_validation and ignoring local_files_only are unrelated bugs." + }, + { + "left": "issue:43618", + "right": "issue:43761", + "accept": true, + "reason": "Both describe the same CLIP vision output regression in the forward/output path: missing attentions and missing hidden_states under v5-style behavior." }, { "left": "issue:44483", "right": "issue:45290", "accept": false, - "reason": "Both are chat-related, but request acceptance and tool-call templating crashes are different code paths." + "reason": "An API endpoint compatibility issue is not the same as a chat-template tokenization crash." }, { - "left": "issue:44483", - "right": "issue:45381", + "left": "issue:43064", + "right": "issue:44380", "accept": false, - "reason": "Chat request handling and Qwen2.5-VL video position ids are unrelated bugs." + "reason": "FSDP2/PEFT optimizer-state corruption and GPT2 attention-scaling being ignored are different training/runtime bugs." }, { "left": "issue:43299", "right": "issue:43867", "accept": false, - "reason": "Both are loading failures, but Qwen3-VL-MoE loading and sorted-state_dict load errors are different root causes." + "reason": "Qwen3VL MoE loading breakage and sorted state_dict load errors are distinct model-loading problems." }, { - "left": "issue:43064", - "right": "issue:44380", + "left": "issue:44483", + "right": "issue:45381", "accept": false, - "reason": "FSDP2/PEFT optimizer-state corruption and GPT2 attention scaling under SDPA/FlashAttention are unrelated." + "reason": "Chat completions API behavior and Qwen2.5-VL video vision_position_ids are unrelated." }, { "left": "issue:43901", "right": "issue:44655", "accept": false, - "reason": "Outdated docs for return_all_scores and inability to save Pipeline objects are not the same bug." + "reason": "Documentation drift for a pipeline flag is not the same as inability to save Pipeline objects." }, { "left": "issue:36683", "right": "issue:43475", "accept": false, - "reason": "Missing vocab_size on Gemma3Config and missing fpn_position_embeddings on Sam3VisionEncoderOutput are different attribute errors." + "reason": "Gemma3Config missing vocab_size and SAM3 output missing fpn_position_embeddings are unrelated attribute errors." }, { "left": "issue:43867", "right": "issue:45216", "accept": false, - "reason": "A load error with sorted state_dict is not the same as a save_pretrained regression for Qwen3.5 checkpoints." + "reason": "Both concern loading/saving state, but one is a load error and the other is a save_pretrained regression for Qwen3.5 checkpoints." }, { "left": "issue:42994", "right": "issue:44655", "accept": false, - "reason": "Quantized model saving failure and Pipeline save_pretrained failure are different save paths and failures." + "reason": "Quantized model saving failure and Pipeline.save_pretrained failure are different serialization paths." }, { "left": "issue:43352", "right": "issue:43937", "accept": false, - "reason": "FlashAttention support for Nemotron and invalid GenerationConfig for GLM-5 are unrelated." + "reason": "Flash Attention support error for Nemotron differs from GLM generation-config validation failure." }, { "left": "issue:43066", "right": "issue:45381", "accept": false, - "reason": "Tokenizer decoder type mismatch and Qwen2.5-VL video position-id errors are unrelated." + "reason": "Tokenizer decoder type in v5 and Qwen2.5-VL video position IDs are unrelated regressions." }, { "left": "issue:36010", "right": "issue:42757", "accept": false, - "reason": "Both are import errors, but they involve different public symbols and modules with no shared bug." + "reason": "Different missing imports from different modules; no common code-path bug." }, { "left": "issue:44220", "right": "issue:44655", "accept": false, - "reason": "Audio feature extraction and Pipeline serialization are unrelated subsystems." - }, - { - "left": "issue:43756", - "right": "issue:44485", - "accept": false, - "reason": "RoPE-layer dropping in Smollm3 and GLM-5 RoPE implementation concerns are different model-specific issues." - }, - { - "left": "issue:44263", - "right": "issue:45440", - "accept": false, - "reason": "A torch.split result issue in one MoE indexer and DeepSeekV3 implementation divergence are not the same defect." - }, - { - "left": "issue:36683", - "right": "issue:43525", - "accept": false, - "reason": "Different model configs missing different attributes; no shared underlying bug." - }, - { - "left": "issue:43618", - "right": "issue:45216", - "accept": false, - "reason": "CLIP attentions regression and Qwen3.5 save_pretrained corruption are unrelated." - }, - { - "left": "issue:43906", - "right": "issue:45753", - "accept": false, - "reason": "An isolated reproduction of an earlier issue and a specific Qwen3.5 image loop are not duplicates." - }, - { - "left": "issue:36683", - "right": "issue:44242", - "accept": false, - "reason": "Gemma3Config attribute loss and missing load-balancing loss for MoE are different problems." - }, - { - "left": "issue:43299", - "right": "issue:45753", - "accept": false, - "reason": "Qwen3-VL-MoE loading failure and a Qwen3.5 infinite loop on one image are different bugs." - }, - { - "left": "issue:36683", - "right": "issue:43881", - "accept": false, - "reason": "Gemma3Config missing vocab_size and glm-4v-9b loading failure are unrelated." - }, - { - "left": "issue:43352", - "right": "issue:43976", - "accept": false, - "reason": "FlashAttention support for Nemotron and Python-version incompatibility are unrelated." - }, - { - "left": "issue:43441", - "right": "issue:44380", - "accept": false, - "reason": "Ministral FlashAttention failure and GPT2 attention scaling being ignored are different attention-backend bugs." + "reason": "Audio feature extraction and Pipeline.save_pretrained are unrelated." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and appropriately conservative overall. The only accepted duplicate pair (issue:43618 \u2194 issue:43761) is supported by the packet as the same CLIP vision output regression, and the other soft-edge verdicts stay on the cautious side without overclaiming shared causes." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -13437,31 +13438,32 @@ "issue:36296", "issue:36683", "issue:41669", - "issue:42548", "issue:42890", - "issue:43010", "issue:43064", "issue:43097", + "issue:43299", "issue:43352", - "issue:43502", + "issue:43441", + "issue:43525", "issue:43575", "issue:43618", + "issue:43756", "issue:43867", + "issue:43881", "issue:43901", + "issue:43906", "issue:43976", "issue:43992", - "issue:44062", - "issue:44075", "issue:44077", - "issue:44112", "issue:44117", - "issue:44222", + "issue:44242", "issue:44246", - "issue:44351", + "issue:44263", "issue:44355", "issue:44360", "issue:44380", "issue:44484", + "issue:44485", "issue:44492", "issue:44655", "issue:44855", @@ -13469,9 +13471,8 @@ "issue:44938", "issue:44964", "issue:44977", - "issue:45125", + "issue:45216", "issue:45310", - "issue:45341", "issue:45357", "issue:45375", "issue:45440", @@ -13479,17 +13480,27 @@ "issue:45479", "issue:45561", "issue:45588", - "issue:45600" + "issue:45753" ], "soft_pairs": [ - "issue:43992|issue:45310", + "issue:43756|issue:44485", + "issue:44263|issue:45440", + "issue:36683|issue:43525", + "issue:43618|issue:45216", + "issue:43906|issue:45753", + "issue:36683|issue:44242", + "issue:43299|issue:45753", + "issue:36683|issue:43881", + "issue:43352|issue:43976", + "issue:43441|issue:44380", "issue:43352|issue:44380", + "issue:43992|issue:45310", "issue:44380|issue:45479", "issue:43867|issue:45375", "issue:41669|issue:44246", - "issue:43618|issue:44380", "issue:43064|issue:44492", "issue:43352|issue:45478", + "issue:43618|issue:44380", "issue:44355|issue:44855", "issue:36683|issue:44117", "issue:42890|issue:45561", @@ -13506,253 +13517,267 @@ "issue:43618|issue:44857", "issue:43618|issue:45588", "issue:43618|issue:44977", - "issue:44655|issue:44938", "issue:43901|issue:44938", - "issue:42890|issue:44964", - "issue:42890|issue:44112", - "issue:42890|issue:45600", - "issue:43502|issue:44062", - "issue:43010|issue:44075", - "issue:42890|issue:45341", - "issue:44222|issue:45125", - "issue:42548|issue:44351" + "issue:44655|issue:44938", + "issue:42890|issue:44964" ], - "prepared_review_unit_hash": "8d3f8611329a49b7bb89e2710bfd628a0524a4a9b8aa38fbf9092f0445b4eb49", + "prepared_review_unit_hash": "7413898c4948b771711287688f34d6e1e1af7664dc13b0d950935fb75eb255df", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 34, - "serialized_chars": 37453, - "estimated_input_tokens": 9364, - "estimated_eval_tokens": 18984 + "soft_pair_count": 37, + "serialized_chars": 38412, + "estimated_input_tokens": 9603, + "estimated_eval_tokens": 19462 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Only one pair looks like the same underlying bug: slow `transformers` import due to model-module `import *` usage. The rest are distinct issues across unrelated models, loading paths, or test flakes. No PRs were provided.", - "confidence": 0.74, - "canonical_issue_reason": "issue:41669 is the better canonical issue because it names the likely root cause of the import-time slowdown and covers the broader symptom reported by issue:44246.", + "summary": "This is a noisy cluster of mostly unrelated Transformers regression reports (model config/loading, flash-attention, save/load, Python-version issues). I only found one convincing duplicate-like pair: import-time slowdown reports in issue:41669 and issue:44246. All other soft pairs look like different bugs that merely share a broad subsystem or symptom class.", + "confidence": 0.77, + "canonical_issue_reason": "issue:41669 is the best canonical issue because it most concretely describes the underlying import-time regression and cleanly subsumes issue:44246's vaguer 'import transformers takes long sometimes' report.", "canonical_pr_reason": null, - "best_issue_reason": "issue:41669 is the strongest representative issue in this set: it is root-cause oriented, broader than the symptom-only report, and the only clear duplicate target.", + "best_issue_reason": "issue:41669 is the strongest representative of the only clear duplicate-like topic in this set and is specific enough to serve as the anchor for deduping the import slowdown reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43992", - "right": "issue:45310", + "left": "issue:43756", + "right": "issue:44485", + "accept": false, + "reason": "Different model-specific bugs: Smollm3 RoPE-layer count vs GLM-5 RoPE implementation." + }, + { + "left": "issue:44263", + "right": "issue:45440", + "accept": false, + "reason": "Unrelated failures: a torch.split DSA indexer issue vs DeepseekV3MoE divergence from remote implementation." + }, + { + "left": "issue:36683", + "right": "issue:43525", + "accept": false, + "reason": "Similar shape (missing config attribute) but different models and missing fields; not the same bug." + }, + { + "left": "issue:43618", + "right": "issue:45216", + "accept": false, + "reason": "CLIPOutput attentions regression is unrelated to Qwen3.5 save_pretrained checkpoint corruption." + }, + { + "left": "issue:43906", + "right": "issue:45753", + "accept": false, + "reason": "A reproduction of an old issue vs a Qwen3_5 image-specific generation loop; no clear shared code-path." + }, + { + "left": "issue:36683", + "right": "issue:44242", + "accept": false, + "reason": "Both are model/config related, but one is missing vocab_size and the other is load-balancing loss logic." + }, + { + "left": "issue:43299", + "right": "issue:45753", + "accept": false, + "reason": "Qwen3VL MoE loading breakage is unrelated to a Qwen3_5 infinite-loop generation bug." + }, + { + "left": "issue:36683", + "right": "issue:43881", + "accept": false, + "reason": "Different model-loading failures with different missing attributes/behaviors." + }, + { + "left": "issue:43352", + "right": "issue:43976", + "accept": false, + "reason": "FlashAttention support error for a model vs Python version compatibility regression; not the same underlying issue." + }, + { + "left": "issue:43441", + "right": "issue:44380", "accept": false, - "reason": "Both are `from_pretrained` failures, but they involve different models and different failure modes (`embed_tokens.weight` vs Qwen3.5 MoE load error)." + "reason": "Both involve attention backends, but one is a specific model unsupported by FlashAttention 2.0 while the other is GPT-2 scaling being ignored." }, { "left": "issue:43352", "right": "issue:44380", "accept": false, - "reason": "Different concrete problems: unsupported Flash Attention for one model versus GPT-2 attention scaling being ignored in backends." + "reason": "Different attention-backend bugs affecting different models and code paths." + }, + { + "left": "issue:43992", + "right": "issue:45310", + "accept": false, + "reason": "Different loading failures: missing embed_tokens.weight for UMT5Encoder vs Qwen3.5 MoE from_pretrained error." }, { "left": "issue:44380", "right": "issue:45479", "accept": false, - "reason": "Unrelated bugs: attention backend config handling versus a sequence-classification loss bug." + "reason": "Attention-backend scaling bug vs a classification-loss degenerate-zero-loss bug." }, { "left": "issue:43867", "right": "issue:45375", "accept": false, - "reason": "Different code paths: sorted `state_dict` load error versus a Qwen3.5 vision config field being dropped by `@strict`." + "reason": "Both are loading-related, but one is state_dict sorting and the other is a missing config field silently dropped by strict parsing." }, { "left": "issue:41669", "right": "issue:44246", "accept": true, - "reason": "Same underlying issue: import-time slowdown in `transformers`, with 41669 pointing to the likely root cause and 44246 reporting the symptom." - }, - { - "left": "issue:43618", - "right": "issue:44380", - "accept": false, - "reason": "Different models and failures; one is a missing attentions assignment, the other is GPT-2 attention-scaling behavior." + "reason": "Both report the same underlying import-time slowdown in transformers; one is the more specific root-cause report and the other is the user-facing symptom." }, { "left": "issue:43064", "right": "issue:44492", "accept": false, - "reason": "Completely unrelated: distributed optimizer state corruption versus a cache-strategy typo." + "reason": "FSDP/PEFT optimizer-state corruption is unrelated to a cache-strategy typo." }, { "left": "issue:43352", "right": "issue:45478", "accept": false, - "reason": "Same broad area only; the specific bugs differ (`Flash Attention 2.0` unsupported vs Qwen3.5 MoE `from_pretrained` error)." + "reason": "Similar FlashAttention/version wording, but the affected models and failure modes are different." + }, + { + "left": "issue:43618", + "right": "issue:44380", + "accept": false, + "reason": "Different attention-related regressions with different symptoms and models." }, { "left": "issue:44355", "right": "issue:44855", "accept": false, - "reason": "Both mention import/runtime errors, but one is about compiled Python files generally and the other is a specific `torch.jit.script` parsing issue in DebertaV2Model." + "reason": "Compiled Python file import errors are not the same as a torch.jit.script parsing/IndentationError regression." }, { "left": "issue:36683", "right": "issue:44117", "accept": false, - "reason": "Different loading bugs: missing `vocab_size` on one config versus tokenizer mapping returning `None`." + "reason": "Both involve missing metadata, but one is a config attribute bug and the other is tokenizer mapping returning None." }, { "left": "issue:42890", "right": "issue:45561", "accept": false, - "reason": "Test flakiness in a model integration test is not the same as an xdist race writing `captured_info.txt`." + "reason": "Test flakiness from missing set_seed is unrelated to xdist races on captured_info.txt." }, { "left": "issue:43097", "right": "issue:43352", "accept": false, - "reason": "Different topics: a removed config method versus Flash Attention support for Nemotron." + "reason": "Removed tie_embeddings behavior and FlashAttention support are unrelated." }, { "left": "issue:43867", "right": "issue:45357", "accept": false, - "reason": "Both concern loading/saving, but one is a `state_dict` load failure and the other is incorrect visual encoder keys on save." + "reason": "State-dict sorting load failure vs incorrect visual encoder keys in save_pretrained; different serialization paths." }, { "left": "issue:43976", "right": "issue:44938", "accept": false, - "reason": "Both are Python-version compatibility reports, but they are different versions and different failure causes." + "reason": "Both are Python-version compatibility reports, but they concern different versions and failure mechanisms." }, { "left": "issue:43992", "right": "issue:45478", "accept": false, - "reason": "Same family of `from_pretrained` issues, but different models and different missing/invalid fields." + "reason": "Different Qwen3.5 loading bugs with different missing/mis-generated weights." }, { "left": "issue:44360", "right": "issue:45440", "accept": false, - "reason": "Unrelated subsystems: a DSA-indexer activation question versus a DeepSeekV3 implementation divergence." + "reason": "DSA indexer ReLU discussion is unrelated to DeepseekV3MoE implementation divergence." }, { "left": "issue:43064", "right": "issue:44655", "accept": false, - "reason": "Different areas: FSDP/PEFT optimizer state handling versus saving `Pipeline` objects." + "reason": "Training optimizer-state corruption is unrelated to saving Pipeline objects." }, { "left": "issue:36296", "right": "issue:43575", "accept": false, - "reason": "Both mention tensor parallelism, but one is a training bug and the other is an OOM while loading a specific model." + "reason": "Tensor-parallel training bug is not the same as Qwen2-57B-A14B-Instruct OOM on load." }, { "left": "issue:44077", "right": "issue:44380", "accept": false, - "reason": "Different model/config behavior: `patchtsmixer` post-init policy versus GPT-2 attention scaling under SDPA/FlashAttention." + "reason": "Patchtsmixer optional post_init policy change is unrelated to GPT-2 attention scaling under SDPA/FlashAttention." }, { "left": "issue:43867", "right": "issue:44484", "accept": false, - "reason": "Load failure versus a question about the default `max_shard_size`; not the same bug." + "reason": "A load-model error from sorted state_dict is unrelated to the 50GB max_shard_size question." }, { "left": "issue:28282", "right": "issue:33453", "accept": false, - "reason": "Different symptoms and subsystems: missing PyTorch dependency versus tokenizer-loading regression." + "reason": "Generic import/config errors, but one is an AutoModel PyTorch missing-dependency ImportError and the other is tokenizer loading regression." }, { "left": "issue:43618", "right": "issue:44857", "accept": false, - "reason": "Different model failures: missing CLIP attentions assignment versus an AMP/CUDA crash in LwDetrImageLoss." + "reason": "CLIP attentions assignment regression is unrelated to LwDetrImageLoss AMP/CUDA crash." }, { "left": "issue:43618", "right": "issue:45588", "accept": false, - "reason": "Both touch attention-related code, but the failures are different (`CLIPOutput.attentions` vs `s_aux=None` in flash_attention.py)." + "reason": "Missing CLIPOutput attentions and flash_attention.py s_aux=None crash are different bugs." }, { "left": "issue:43618", "right": "issue:44977", "accept": false, - "reason": "Different concrete issues: CLIP attentions output assignment versus Qwen3.5 generation failing with flash-attention." + "reason": "CLIP attentions regression is unrelated to Qwen3.5 flash-attention generation failure." }, { - "left": "issue:44655", + "left": "issue:43901", "right": "issue:44938", "accept": false, - "reason": "Different problems: saving `Pipeline` objects versus Python 3.14 import compatibility." + "reason": "Docs mismatch about return_all_scores is unrelated to Python 3.14 import/load failure." }, { - "left": "issue:43901", + "left": "issue:44655", "right": "issue:44938", "accept": false, - "reason": "Documentation mismatch in a pipeline API versus a runtime failure on Python 3.14." + "reason": "Saving Pipeline objects and loading transformers on Python 3.14 are unrelated." }, { "left": "issue:42890", "right": "issue:44964", "accept": false, - "reason": "Different model-specific failures; one is a flaky test, the other is a latest-transformers load regression for Phi-4 multimodal." - }, - { - "left": "issue:42890", - "right": "issue:44112", - "accept": false, - "reason": "Both are CI/test-related, but they describe different failures and causes." - }, - { - "left": "issue:42890", - "right": "issue:45600", - "accept": false, - "reason": "Different issues in test/auto-mapping code; no evidence they share the same root cause." - }, - { - "left": "issue:43502", - "right": "issue:44062", - "accept": false, - "reason": "Different bug classes: offline loading still making API requests versus tokenizer `AddedToken` argument duplication." - }, - { - "left": "issue:43010", - "right": "issue:44075", - "accept": false, - "reason": "Cache update decoration and SGD optimizer arguments are unrelated." - }, - { - "left": "issue:42890", - "right": "issue:45341", - "accept": false, - "reason": "The titles suggest unrelated test issues; no clear shared bug." - }, - { - "left": "issue:44222", - "right": "issue:45125", - "accept": false, - "reason": "Different Qwen3.5 MoE problems: FP8 save-pretrained behavior versus missing `_tp_plan` for tensor parallelism." - }, - { - "left": "issue:42548", - "right": "issue:44351", - "accept": false, - "reason": "Both are import-related, but one is a `PreTrainedModel` export issue and the other is a missing `HybridCache` import." + "reason": "Test seeding flakiness is unrelated to Phi-4 multimodal loading failure." } ] }, - "evaluator_result": null, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The 41669/44246 pair is plausibly the only clearly duplicate-like import-slowdown cluster, and the rest of the soft-edge rejections are cautious rather than overstated." + }, "evaluator_used": true, "retried": false, "accepted_nontrivial_soft_edge": true, @@ -13783,22 +13808,23 @@ "issue:42548", "issue:42757", "issue:42831", + "issue:42890", "issue:43010", "issue:43299", "issue:43502", - "issue:43526", "issue:43576", "issue:43723", "issue:43824", "issue:43827", "issue:43874", - "issue:43901", "issue:43976", + "issue:44062", + "issue:44075", + "issue:44112", "issue:44222", "issue:44230", - "issue:44261", - "issue:44273", "issue:44315", + "issue:44351", "issue:44355", "issue:44360", "issue:44368", @@ -13807,21 +13833,27 @@ "issue:44556", "issue:44655", "issue:44749", - "issue:44908", "issue:44938", "issue:44961", - "issue:45003", "issue:45071", "issue:45125", "issue:45278", + "issue:45341", "issue:45375", "issue:45468", "issue:45478", - "issue:45542", + "issue:45600", "issue:45656", "issue:45850" ], "soft_pairs": [ + "issue:42890|issue:44112", + "issue:42890|issue:45600", + "issue:43502|issue:44062", + "issue:43010|issue:44075", + "issue:42890|issue:45341", + "issue:44222|issue:45125", + "issue:42548|issue:44351", "issue:43502|issue:44485", "issue:44355|issue:44749", "issue:45125|issue:45478", @@ -13838,9 +13870,9 @@ "issue:43502|issue:44961", "issue:42831|issue:43874", "issue:30333|issue:36296", - "issue:44355|issue:45071", - "issue:44222|issue:44230", "issue:44355|issue:44655", + "issue:44222|issue:44230", + "issue:44355|issue:45071", "issue:43827|issue:44961", "issue:33453|issue:34689", "issue:42757|issue:43576", @@ -13851,290 +13883,286 @@ "issue:43576|issue:43824", "issue:44961|issue:45468", "issue:31795|issue:38617", - "issue:38617|issue:43576", "issue:41628|issue:42548", + "issue:38617|issue:43576", "issue:43010|issue:43976", - "issue:43576|issue:44938", - "issue:44908|issue:44961", - "issue:36296|issue:43901", - "issue:41628|issue:44355", - "issue:36296|issue:43526", - "issue:44261|issue:44360", - "issue:44273|issue:45003", - "issue:44273|issue:45542" + "issue:43576|issue:44938" ], - "prepared_review_unit_hash": "f48612d951c09736a9ad9293a490860ae7e78622e09b3ba41ee4d5cbf97ca822", + "prepared_review_unit_hash": "657b76dce04fcace7e111f8fb932c0e3b9715a0cb0a2754f7230cc0a059087a6", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 40, - "serialized_chars": 39145, - "estimated_input_tokens": 9787, - "estimated_eval_tokens": 19830 + "serialized_chars": 39096, + "estimated_input_tokens": 9774, + "estimated_eval_tokens": 19804 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a loose collection of Transformers v5 regression reports, centered on import/loading/model-compatibility breakages rather than one single duplicate bug. There is one plausible lazy-loading subcluster, but most soft links are too broad or too model-specific to merge.", - "confidence": 0.67, - "canonical_issue_reason": "45278 is the best umbrella issue: it is broad enough to represent the recurring post-v5 import/loading breakage theme without being as vague as the most generic complaints.", + "summary": "This cluster is mostly a loose collection of Transformers v5 regression reports, especially import/loading breakages, but many items are about different symbols, models, or unrelated test/feature bugs. I would not merge the soft pairs; they are too diverse to be true duplicates.", + "confidence": 0.83, + "canonical_issue_reason": "issue:45278 is the best umbrella issue: it broadly captures the recurring v5 import/load regressions that overlap with several other reports, without being as vague as the most generic complaints.", "canonical_pr_reason": null, - "best_issue_reason": "45278 is the most representative issue for the cluster overall because it captures the shared upgrade-regression pattern seen across many of the other reports.", + "best_issue_reason": "issue:45278 is the strongest representative for triage because it summarizes the dominant cluster theme (post-upgrade import failures) and is specific enough to anchor related reports.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:42890", + "right": "issue:44112", + "accept": false, + "reason": "Both mention CI/test flakiness, but one is a seed-related SamHQ integration issue and the other is a GraniteSpeech stale device override test; different failing paths." + }, + { + "left": "issue:42890", + "right": "issue:45600", + "accept": false, + "reason": "Both are test/CI-related, but the bugs are unrelated: missing set_seed in SamHQ tests vs broken auto_mappings references for removed configs." + }, + { + "left": "issue:43502", + "right": "issue:44062", + "accept": false, + "reason": "One is an offline/network regression in from_pretrained; the other is an AddedToken constructor TypeError. No shared concrete code path." + }, + { + "left": "issue:43010", + "right": "issue:44075", + "accept": false, + "reason": "Cache update no_grad and SGD optimizer args are separate runtime behaviors in different subsystems." + }, + { + "left": "issue:42890", + "right": "issue:45341", + "accept": false, + "reason": "Both touch testing utilities, but one is a model test needing seed setup and the other is a generic bug in testing_utils.py; too little evidence of the same defect." + }, + { + "left": "issue:44222", + "right": "issue:45125", + "accept": false, + "reason": "Both involve Qwen3.5/MoE/FP8, but one is FP8 save_pretrained and the other is tensor-parallel _tp_plan loading; not the same bug." + }, + { + "left": "issue:42548", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but for different symbols (PreTrainedModel vs HybridCache) and likely different missing-export paths." + }, { "left": "issue:43502", "right": "issue:44485", "accept": false, - "reason": "Different bugs: local_files_only network leakage vs GLM-5 RoPE implementation." + "reason": "local_files_only network access and GLM-5 RoPE implementation are unrelated." }, { "left": "issue:44355", "right": "issue:44749", "accept": false, - "reason": "Unrelated failures: compiled Python file handling vs a Chinese report about slower filtering after upgrade." + "reason": "Compiled Python file errors and a slowdown after upgrading to v5.3 are not the same underlying issue." }, { "left": "issue:45125", "right": "issue:45478", "accept": false, - "reason": "Both involve Qwen3.5 MoE, but one is tensor-parallel plan support and the other is from_pretrained loading; not the same concrete bug." + "reason": "Same model family, but one reports missing _tp_plan for tensor parallelism while the other is a broader from_pretrained error; insufficient evidence for one concrete fix." }, { "left": "issue:30990", "right": "issue:31795", "accept": false, - "reason": "Sentence-Transformers loading hang and documentation wording are unrelated." + "reason": "Loading gets stuck vs documentation confusion; these are not duplicate bugs." }, { "left": "issue:45125", "right": "issue:45375", "accept": false, - "reason": "Same model family, but missing _tp_plan and missing deepstack_visual_indexes are distinct configuration/support issues." + "reason": "MoE tensor-parallel loading and a missing config field silently dropped by @strict are unrelated." }, { "left": "issue:44387", "right": "issue:44556", "accept": false, - "reason": "Both are upgrade regressions, but one is int4 memory/OOM behavior and the other is checkpoint reload compatibility." + "reason": "Both are version-regression complaints, but one is int4 memory/OOM behavior and the other is checkpoint reload compatibility." }, { "left": "issue:30333", "right": "issue:33453", "accept": false, - "reason": "MLflow reporting after failed training is unrelated to tokenizer loading regressions." + "reason": "MLflow reporting on training failure and tokenizer loading regression are unrelated." }, { "left": "issue:43010", "right": "issue:45278", "accept": false, - "reason": "Decorator/no_grad cache-update issue is unrelated to broad import failures." + "reason": "No overlap: no_grad cache update vs broad import errors after upgrade." }, { "left": "issue:43723", "right": "issue:44556", "accept": false, - "reason": "Tokenizer loading in v5 and checkpoint reload failures are different code paths." + "reason": "Tokenizer loading and checkpoint reload after upgrade are both load-related, but not the same bug or code path." }, { "left": "issue:43502", "right": "issue:44222", "accept": false, - "reason": "local_files_only API leakage and FP8 save_pretrained issues are unrelated." + "reason": "local_files_only network leakage and FP8 MoE save_pretrained are unrelated." }, { "left": "issue:43010", "right": "issue:44368", "accept": false, - "reason": "Cache update no_grad behavior and tie_word_embeddings warning are different bugs." + "reason": "Cache update semantics and a tie_word_embeddings warning in Qwen3.5 LoRA are unrelated." }, { "left": "issue:44315", "right": "issue:44360", "accept": false, - "reason": "Liger kernel application and DSA indexer activation are unrelated." + "reason": "Liger kernel application and a DSA indexer ReLU discussion have nothing in common." }, { "left": "issue:43010", "right": "issue:45850", "accept": false, - "reason": "One is a specific cache-update decorator bug; the other is a generic complaint about many models breaking." + "reason": "A specific no_grad cache bug and a generic 'too many models break' complaint are not duplicates." }, { "left": "issue:43502", "right": "issue:44961", "accept": false, - "reason": "No overlap beyond being general issues; 'racoon' is not a loading regression." + "reason": "No relation between offline file loading and the racoon issue." }, { "left": "issue:42831", "right": "issue:43874", "accept": false, - "reason": "FineGrainedFP8 accuracy and GLM46V image processor token counting are unrelated." + "reason": "FineGrainedFP8 accuracy and a missing image-patch-count method in GLM46V are different model bugs." }, { "left": "issue:30333", "right": "issue:36296", "accept": false, - "reason": "MLflow reporting and tensor-parallel training are different failures." + "reason": "MLflow failure reporting and tensor-parallel training bug are unrelated." }, { "left": "issue:44355", - "right": "issue:45071", + "right": "issue:44655", "accept": false, - "reason": "Compiled-file errors and PretrainedConfig type-checking are unrelated." + "reason": "Compiled Python file handling and save_pretrained for Pipeline objects are different features." }, { "left": "issue:44222", "right": "issue:44230", "accept": false, - "reason": "Both mention FP8 and Moe, but save_pretrained failure and inference support are different changes." + "reason": "FP8 save_pretrained for MoE and FP8 inference support for Qwen3-VL/Qwen3.5 are related thematically but not the same concrete change." }, { "left": "issue:44355", - "right": "issue:44655", + "right": "issue:45071", "accept": false, - "reason": "Compiled Python file errors and Pipeline.save_pretrained are not the same bug." + "reason": "Compiled-file errors and PretrainedConfig type-checking are separate regressions." }, { "left": "issue:43827", "right": "issue:44961", "accept": false, - "reason": "Docs still mentioning pipeline() is unrelated to 'racoon'." + "reason": "Docs still referencing pipeline() and the racoon issue are unrelated." }, { "left": "issue:33453", "right": "issue:34689", "accept": false, - "reason": "Tokenizer loading regression and Llama 3.2 Vision model loading are different symptoms and code paths." + "reason": "Tokenizer loading regression and Llama 3.2 vision model loading breakage are distinct failures." }, { "left": "issue:42757", "right": "issue:43576", "accept": false, - "reason": "Missing is_offline_mode import and the transformers env command are not the same issue." + "reason": "Missing is_offline_mode import and broken transformers env command are different v5 breakages." }, { "left": "issue:43299", "right": "issue:44556", "accept": false, - "reason": "Both are v5 upgrade loading regressions, but on different model/loading paths; too broad to merge confidently." + "reason": "Qwen3VL MoE loading and checkpoint reload incompatibility are both loading issues, but not the same bug." }, { "left": "issue:43010", - "right": "issue:45656", - "accept": false, - "reason": "Cache update no_grad and duplicate optimizer.step under deepspeed are unrelated." - }, - { - "left": "issue:43576", - "right": "issue:45071", - "accept": false, - "reason": "Broken env command and PretrainedConfig type checking are unrelated." - }, - { - "left": "issue:43576", - "right": "issue:43824", - "accept": false, - "reason": "Env command breakage and Qwen2.5-VL import failure are different issues." - }, - { - "left": "issue:44961", - "right": "issue:45468", - "accept": false, - "reason": "No shared bug; 'racoon' is unrelated to Gemma4AudioRelPositionalEncoding." - }, - { - "left": "issue:31795", - "right": "issue:38617", - "accept": false, - "reason": "Documentation wording and an ImportError for layer_type_validation are unrelated." - }, - { - "left": "issue:38617", - "right": "issue:43576", + "right": "issue:45656", "accept": false, - "reason": "Different failure modes: missing import vs env command breakage." + "reason": "No_grad cache updates and optimizer step called twice with deepspeed are separate optimizer/runtime behaviors." }, { - "left": "issue:41628", - "right": "issue:42548", + "left": "issue:33453", + "right": "issue:36296", "accept": false, - "reason": "Both are top-level import errors, but they concern different missing symbols and are not clearly the same concrete bug." + "reason": "Tokenizer loading regression and tensor-parallel training bug are unrelated." }, { - "left": "issue:43010", - "right": "issue:43976", + "left": "issue:43576", + "right": "issue:45071", "accept": false, - "reason": "No overlap between cache-update behavior and Python version support." + "reason": "Broken env command and PretrainedConfig type checking are not the same issue." }, { "left": "issue:43576", - "right": "issue:44938", + "right": "issue:43824", "accept": false, - "reason": "Broken env command and Python 3.14 loading failure are different problems." + "reason": "CLI/env command breakage and missing Qwen2.5-VL export are distinct import/command problems." }, { - "left": "issue:44908", - "right": "issue:44961", + "left": "issue:44961", + "right": "issue:45468", "accept": false, - "reason": "Scheduler kwargs handling is unrelated to 'racoon'." + "reason": "No meaningful overlap between the racoon issue and Gemma-4 audio positional encoding." }, { - "left": "issue:36296", - "right": "issue:43901", + "left": "issue:31795", + "right": "issue:38617", "accept": false, - "reason": "Tensor-parallel training bug and pipeline docs mentioning return_all_scores are unrelated." + "reason": "Documentation confusion and missing layer_type_validation export are different kinds of issues." }, { "left": "issue:41628", - "right": "issue:44355", + "right": "issue:42548", "accept": false, - "reason": "AutoImageProcessor import failure and compiled Python file errors are different bugs." + "reason": "Both are import errors, but for different missing names and likely different modules/exports; too broad to merge." }, { - "left": "issue:36296", - "right": "issue:43526", + "left": "issue:38617", + "right": "issue:43576", "accept": false, - "reason": "Tensor parallel training and BeitImageProcessorFast reduce_labels are unrelated." + "reason": "An import name error and a broken env command are unrelated." }, { - "left": "issue:44261", - "right": "issue:44360", + "left": "issue:43010", + "right": "issue:43976", "accept": false, - "reason": "MLA q_a_layernorm precision/config issue and DSA indexer activation issue are unrelated." - }, - { - "left": "issue:44273", - "right": "issue:45003", - "accept": true, - "reason": "These look like the same lazy-loading bug: a generic report that lazy loading is broken and a more specific root-cause report about unsafe sys.modules access in modeling_utils." + "reason": "Cache update no_grad and Python 3.9/3.10 compatibility are separate bugs." }, { - "left": "issue:44273", - "right": "issue:45542", + "left": "issue:43576", + "right": "issue:44938", "accept": false, - "reason": "Lazy loading problems and a TensorFlow/tensorboard backend error are unrelated." + "reason": "Broken env command and Python 3.14 load failure are both compatibility issues, but not the same defect." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet: the cluster is indeed dominated by v5-era import/loading/compatibility regressions, and the lone accepted soft edge (44273\u219445003) is plausibly the same lazy-loading bug. The other pair judgments are conservative and mostly rest on clearly different titles or code paths. No overstatement stands out." - }, - "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 }, @@ -14151,68 +14179,75 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-391", "nodes": [ - "issue:31795", - "issue:33453", + "issue:36296", + "issue:41628", "issue:41669", - "issue:42371", "issue:42890", "issue:43352", "issue:43408", - "issue:43519", "issue:43525", + "issue:43526", "issue:43572", "issue:43576", "issue:43673", "issue:43827", "issue:43874", + "issue:43901", "issue:43976", "issue:44038", "issue:44077", "issue:44246", + "issue:44261", "issue:44273", "issue:44295", "issue:44297", "issue:44315", "issue:44351", + "issue:44355", "issue:44360", "issue:44373", "issue:44393", - "issue:44485", "issue:44492", "issue:44556", "issue:44623", - "issue:44655", "issue:44704", "issue:44829", "issue:44861", "issue:44908", "issue:44961", + "issue:45003", "issue:45071", "issue:45092", "issue:45125", "issue:45230", "issue:45278", - "issue:45310", "issue:45341", "issue:45362", - "issue:45468", "issue:45478", + "issue:45542", "issue:45753", "issue:45850" ], "soft_pairs": [ + "issue:44908|issue:44961", + "issue:36296|issue:43901", + "issue:41628|issue:44355", + "issue:36296|issue:43526", + "issue:44261|issue:44360", + "issue:44273|issue:45003", + "issue:44273|issue:45542", "issue:44623|issue:45230", "issue:44393|issue:45753", - "issue:44038|issue:45125", "issue:43576|issue:44556", + "issue:44038|issue:45125", "issue:44393|issue:45071", "issue:43525|issue:45850", "issue:41669|issue:43673", "issue:44556|issue:45092", "issue:44373|issue:44961", "issue:43572|issue:45850", - "issue:44704|issue:44908", "issue:45362|issue:45850", + "issue:44704|issue:44908", "issue:42890|issue:44908", "issue:44246|issue:44273", "issue:43408|issue:44315", @@ -14228,301 +14263,279 @@ "issue:45278|issue:45850", "issue:44393|issue:45478", "issue:43874|issue:45850", - "issue:41669|issue:44492", "issue:44315|issue:44829", + "issue:41669|issue:44492", "issue:43827|issue:44908", - "issue:44077|issue:44861", - "issue:31795|issue:33453", - "issue:41669|issue:44908", - "issue:42371|issue:44492", - "issue:43576|issue:44273", - "issue:44655|issue:44861", - "issue:44360|issue:45468", - "issue:44492|issue:45850", - "issue:44273|issue:44908", - "issue:45230|issue:45310", - "issue:43519|issue:44485" + "issue:44077|issue:44861" ], - "prepared_review_unit_hash": "a80c074130ef56dfddd1772ff873e8a65195c21fa8bfba5f4afbbb19093cb9fa", + "prepared_review_unit_hash": "2082ef0bb15b724250fda3cd5be71d8f6fc34d4fa67aedd2d90e9c7b0845ffda", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 41, - "serialized_chars": 39212, - "estimated_input_tokens": 9803, - "estimated_eval_tokens": 19862 + "soft_pair_count": 38, + "serialized_chars": 38444, + "estimated_input_tokens": 9611, + "estimated_eval_tokens": 19478 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab bag of Transformers v5 regression reports, with a few import/lazy-loading issues and many unrelated model-specific breakages. Only one pair looks like the same underlying import/lazy-loading problem; the rest are too distinct to merge.", - "confidence": 0.43, - "canonical_issue_reason": "issue:45850 is the broadest umbrella for the cluster\u2019s recurring \u201cv5 breaks things\u201d reports, even though it is vague.", + "summary": "This cluster is mostly a mixed bag of unrelated Transformers issues, with one clear duplicate-like pair around import/lazy-loading slowness. Most other pairs share only broad theme words (v5 regressions, import problems, specific model bugs) rather than the same concrete defect.", + "confidence": 0.73, + "canonical_issue_reason": "issue:44246 is the best anchor for the only clear duplicate subgroup: it is the broadest and most central report about `import transformers` being slow, and it aligns closely with the more specific lazy-loading complaint.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45850 is the best single representative of the cluster because it captures the general v5 regression theme across many of the items.", + "best_issue_reason": "issue:44246 is the strongest representative issue in this set because it captures the import/lazy-loading regression at a higher level and has the clearest overlap with the only accepted soft duplicate.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44908", + "right": "issue:44961", + "accept": false, + "reason": "Different topics entirely: scheduler kwargs vs an unrelated one-word issue." + }, + { + "left": "issue:36296", + "right": "issue:43901", + "accept": false, + "reason": "Tensor-parallel training bug and a docs mismatch for text classification are unrelated." + }, + { + "left": "issue:41628", + "right": "issue:44355", + "accept": false, + "reason": "Both are errors, but one is a missing top-level import and the other is a runtime problem with compiled Python files." + }, + { + "left": "issue:36296", + "right": "issue:43526", + "accept": false, + "reason": "Tensor parallelism vs a BeitImageProcessorFast `reduce_labels` bug; no shared code-path." + }, + { + "left": "issue:44261", + "right": "issue:44360", + "accept": false, + "reason": "Different model families and different failures; not the same underlying bug." + }, + { + "left": "issue:44273", + "right": "issue:45003", + "accept": false, + "reason": "Both touch loading behavior, but one is lazy loading not working and the other is unsafe `sys.modules` access; too different to merge." + }, + { + "left": "issue:44273", + "right": "issue:45542", + "accept": false, + "reason": "Lazy loading/import behavior is unrelated to the TensorBoard-without-TF backend error." + }, { "left": "issue:44623", "right": "issue:45230", "accept": false, - "reason": "One is a concrete `processor.save_pretrained` bug; the other is just a generic bug report." + "reason": "A specific `save_pretrained` file-missing bug and a generic bug report are not duplicates." }, { "left": "issue:44393", "right": "issue:45753", "accept": false, - "reason": "Different Qwen3-VL failure modes: 2D bbox output corruption vs an infinite loop on a specific image." + "reason": "Both involve Qwen3-VL/Qwen3_5 vision behavior, but the symptoms and concrete failures differ too much." }, { - "left": "issue:44038", - "right": "issue:45125", + "left": "issue:43576", + "right": "issue:44556", "accept": false, - "reason": "Both mention Qwen3.5-MoE, but one is a Flash Attention support/error issue and the other is missing tensor-parallel metadata." + "reason": "Both are v5 regressions, but one is the `transformers env` command and the other is checkpoint reload incompatibility." }, { - "left": "issue:43576", - "right": "issue:44556", + "left": "issue:44038", + "right": "issue:45125", "accept": false, - "reason": "Both are v5 upgrade complaints, but one is about the `transformers env` command and the other about checkpoint reload compatibility." + "reason": "General Qwen3-VL-Moe bug vs missing `_tp_plan` for tensor parallelism; different problems." }, { "left": "issue:44393", "right": "issue:45071", "accept": false, - "reason": "Unrelated problems: Qwen3-VL output errors vs `PretrainedConfig` type-checking breakage." + "reason": "Vision hallucination/bounding-box failure is not the same as a `PretrainedConfig` type-checking regression." }, { "left": "issue:43525", "right": "issue:45850", "accept": false, - "reason": "Specific `Llama4Config` attribute error versus a broad, unspecific v5 breakage complaint." + "reason": "Specific missing config attribute in Llama4 vs a broad complaint that v5 breaks many models." }, { "left": "issue:41669", "right": "issue:43673", "accept": false, - "reason": "Import-time slowdown from `import *` usage is not the same bug as missing generation cache support during chunked prefill." + "reason": "Import-star slowdown and missing generation cache in chunked prefill are different bugs." }, { "left": "issue:44556", "right": "issue:45092", "accept": false, - "reason": "Both concern upgrade/reload incompatibility, but they describe different code paths and model situations." + "reason": "Both are upgrade compatibility issues, but they affect different model families and failure mechanisms." }, { "left": "issue:44373", "right": "issue:44961", "accept": false, - "reason": "Docstring typo vs an unrelated placeholder issue." + "reason": "A docstring bug and an unrelated issue title are not duplicates." }, { "left": "issue:43572", "right": "issue:45850", "accept": false, - "reason": "A missing `pad_token_idx` in `StableLmConfig` is a model-specific config regression, not the same as the generic v5 complaint." + "reason": "Missing `pad_token_idx` in StableLmConfig is too specific to be the same as a vague global v5 complaint." }, { - "left": "issue:44704", - "right": "issue:44908", + "left": "issue:45362", + "right": "issue:45850", "accept": false, - "reason": "Different APIs and failures: `AutoProcessor.from_pretrained` kwargs forwarding vs scheduler kwargs handling." + "reason": "Qwen3.5 chat crash is a concrete model bug, not the same as a broad regression report." }, { - "left": "issue:45362", + "left": "issue:45278", + "right": "issue:45850", + "accept": false, + "reason": "Both are about v5 breakage, but one is specifically many import errors after upgrade and the other is a generic complaint; not the same bug." + }, + { + "left": "issue:43874", "right": "issue:45850", "accept": false, - "reason": "Qwen3.5 chat crash is a specific model/runtime bug, not the same as a broad v5 breakage report." + "reason": "Specific GLM46V multimodal token bug vs a broad v5 breakage report." + }, + { + "left": "issue:44704", + "right": "issue:44908", + "accept": false, + "reason": "`AutoProcessor.from_pretrained` kwargs forwarding and scheduler kwargs handling are unrelated." }, { "left": "issue:42890", "right": "issue:44908", "accept": false, - "reason": "Test flakiness due to missing seeds is unrelated to scheduler kwarg handling." + "reason": "Seed-related integration flakiness and scheduler kwargs are different problems." }, { "left": "issue:44246", "right": "issue:44273", "accept": true, - "reason": "Both point to the same lazy-loading/import regression around `import transformers` being unexpectedly slow or broken." + "reason": "These describe the same import/lazy-loading regression: one says imports are intermittently slow, the other says lazy loading is not working properly." }, { "left": "issue:43408", "right": "issue:44315", "accept": false, - "reason": "Model-type warning during instantiation is unrelated to Liger Kernel application with `model_init`." + "reason": "Model-type warning and Liger Kernel application are different code paths and symptoms." }, { "left": "issue:43976", "right": "issue:44393", "accept": false, - "reason": "Python version compatibility and Qwen3-VL output errors are unrelated." + "reason": "Python version compatibility is unrelated to the Qwen3-VL bounding-box issue." }, { "left": "issue:43352", "right": "issue:44393", "accept": false, - "reason": "Flash Attention 2 support gating is unrelated to Qwen3-VL bounding-box hallucinations/errors." + "reason": "Flash Attention support limitations are a different issue from the Qwen3-VL output problem." }, { "left": "issue:43576", "right": "issue:44908", "accept": false, - "reason": "One is a broken CLI command; the other is a scheduler configuration bug." - }, - { - "left": "issue:44908", - "right": "issue:45071", - "accept": false, - "reason": "Scheduler kwargs being ignored is unrelated to `PretrainedConfig` type-checking failures." + "reason": "The env command regression is unrelated to scheduler kwargs." }, { "left": "issue:44297", "right": "issue:45230", "accept": false, - "reason": "Tokenizer save metadata mismatch is a specific serialization bug; the other issue is generic." + "reason": "Tokenizer class metadata mismatch in `save_pretrained` is not the same as a generic bug report." }, { "left": "issue:41669", "right": "issue:44351", "accept": false, - "reason": "Import performance regression from lazy-loading changes is not the same as missing `HybridCache` export/import failure." + "reason": "Import-star performance cost and missing `HybridCache` import are distinct import-related bugs." }, { "left": "issue:44295", "right": "issue:45230", "accept": false, - "reason": "Buffer access error for `position_ids` is unrelated to the generic bug report." + "reason": "Position-ids buffer reading error and a generic bug report do not match." }, { "left": "issue:43673", "right": "issue:44492", "accept": false, - "reason": "Generation cache missing and a typo in cache strategies are different problems." + "reason": "Chunked-prefill cache regression and a typo in cache strategies are unrelated." }, { "left": "issue:41669", "right": "issue:45341", "accept": false, - "reason": "Import-time slowdown and a testing_utils bug are unrelated." + "reason": "Import performance regression and a testing utility bug are not the same defect." }, { "left": "issue:45278", "right": "issue:45850", "accept": false, - "reason": "Both are broad v5 complaints, but one is specifically about import errors after upgrade while the other is a generic model-breakage report." + "reason": "Duplicate theme only at a very high level; the specific failure mode differs and should not be merged." }, { "left": "issue:44393", "right": "issue:45478", "accept": false, - "reason": "Different Qwen families and different failure modes; not the same underlying bug." + "reason": "Different Qwen3.5/Qwen3-VL failure modes with no evidence of a shared code-path." }, { "left": "issue:43874", "right": "issue:45850", "accept": false, - "reason": "A missing multimodal-token helper is a concrete API regression, not the same as a generic breakage report." - }, - { - "left": "issue:41669", - "right": "issue:44492", - "accept": false, - "reason": "Import performance regression is unrelated to a cache-strategy typo." + "reason": "Specific GLM46V multimodal-token bug vs a generic v5 complaint." }, { "left": "issue:44315", "right": "issue:44829", "accept": false, - "reason": "`model_init`/Liger-Kernel application and Flash Attention 3 training degeneration are different issues." - }, - { - "left": "issue:43827", - "right": "issue:44908", - "accept": false, - "reason": "Docs still referencing `pipeline()` is unrelated to scheduler kwargs handling." - }, - { - "left": "issue:44077", - "right": "issue:44861", - "accept": false, - "reason": "A model config validation change and a tied-weights helper crash are different code paths." - }, - { - "left": "issue:31795", - "right": "issue:33453", - "accept": false, - "reason": "Doc confusion about `input_ids`/`past_key_values` is unrelated to tokenizer loading regression." + "reason": "Liger Kernel not applied during model init and Flash Attention 3 degenerate training are different issues." }, { "left": "issue:41669", - "right": "issue:44908", - "accept": false, - "reason": "Import/lazy-loading performance issues are unrelated to scheduler configuration." - }, - { - "left": "issue:42371", "right": "issue:44492", "accept": false, - "reason": "TF32 settings guidance is unrelated to a cache-strategy typo." - }, - { - "left": "issue:43576", - "right": "issue:44273", - "accept": false, - "reason": "Broken `transformers env` command and lazy-loading/import slowness are different symptoms and code paths." - }, - { - "left": "issue:44655", - "right": "issue:44861", - "accept": false, - "reason": "Saving Pipeline objects and tied-weights helper crashes are unrelated." - }, - { - "left": "issue:44360", - "right": "issue:45468", - "accept": false, - "reason": "DSA indexer/ReLU discussion is unrelated to Gemma-4 audio positional encoding." - }, - { - "left": "issue:44492", - "right": "issue:45850", - "accept": false, - "reason": "A typo in cache strategy docs/code is not the same as broad v5 breakage." + "reason": "Import slowdown vs a typo in cache strategy text are unrelated." }, { - "left": "issue:44273", + "left": "issue:43827", "right": "issue:44908", "accept": false, - "reason": "Lazy loading problems are 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." + "reason": "Docs for removed pipeline APIs and scheduler kwargs handling are unrelated." }, { - "left": "issue:43519", - "right": "issue:44485", + "left": "issue:44077", + "right": "issue:44861", "accept": false, - "reason": "Qwen3VL timestamp math and GLM-5 RoPE implementation are different model-specific bugs." + "reason": "PatchTSMixer post-init policy and tied-weight key handling are different model internals, not the same bug." } ] }, "evaluator_result": { "accept": true, - "feedback": "The clustering summary is grounded in the packet: most items are indeed v5-era regression reports, with a few import/lazy-loading items and many clearly model- or API-specific bugs. The single accepted soft-edge pair, issue:44246 and issue:44273, is a conservative match on the same import/lazy-loading theme. No pair is overstated, and the broad canonical issue choice is reasonable as an umbrella for the vague v5-breakage reports." + "feedback": "The summary stays grounded in the packet and is appropriately conservative: it identifies only one plausible duplicate-like pair (44246/44273) and treats the rest as largely unrelated. I don\u2019t see evidence of overstatement in the soft-edge verdicts." }, "evaluator_used": true, "retried": false, @@ -14560,25 +14573,41 @@ "issue:43519", "issue:43525", "issue:43572", + "issue:43576", "issue:43606", "issue:43704", "issue:44230", "issue:44261", "issue:44263", + "issue:44273", "issue:44315", "issue:44360", "issue:44485", + "issue:44492", "issue:44623", + "issue:44655", "issue:44861", "issue:44908", "issue:44998", "issue:45092", + "issue:45230", "issue:45290", "issue:45310", "issue:45341", - "issue:45468" + "issue:45468", + "issue:45850" ], "soft_pairs": [ + "issue:31795|issue:33453", + "issue:41669|issue:44908", + "issue:42371|issue:44492", + "issue:43576|issue:44273", + "issue:44360|issue:45468", + "issue:44655|issue:44861", + "issue:44492|issue:45850", + "issue:44273|issue:44908", + "issue:45230|issue:45310", + "issue:43519|issue:44485", "issue:41669|issue:43704", "issue:43519|issue:44908", "issue:31795|issue:34689", @@ -14612,226 +14641,286 @@ "issue:31515|issue:33453", "issue:43298|issue:44230" ], - "prepared_review_unit_hash": "e59629260307b2507cba2e2f9bf42cc3b41d5f6a6e6fbc0ab0137b6a9ff7140e", + "prepared_review_unit_hash": "74e0136bdbba713ac92fdbe20e7a436c69b67af734cf1a3c48b6088bda4b7351", "component_budget": { "node_count": 391, "item_count": 391, "soft_pair_count": 1476, - "serialized_chars": 759364, - "estimated_input_tokens": 189841, - "estimated_eval_tokens": 379938 + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 }, "budget": { - "node_count": 34, - "item_count": 34, - "soft_pair_count": 32, - "serialized_chars": 28733, - "estimated_input_tokens": 7184, - "estimated_eval_tokens": 14624 + "node_count": 40, + "item_count": 40, + "soft_pair_count": 42, + "serialized_chars": 34927, + "estimated_input_tokens": 8732, + "estimated_eval_tokens": 17720 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is mostly heterogeneous with no true duplicate pair; the closest thematic overlap is around tokenizer/model-loading regressions, but the soft matches are still too coarse to merge.", - "confidence": 0.38, - "canonical_issue_reason": "Issue 33453 is the strongest representative bug report here: it is concrete, user-facing, and has substantial discussion around a loading regression, making it the best stand-in for the noisy loading-related subset.", + "summary": "This cluster is mostly a loose semantic grouping of unrelated Transformers issues. The only clear common thread is broad v5 regression fallout, but the individual soft pairs do not look like the same underlying bug/change, so I would not merge any of them.", + "confidence": 0.73, + "canonical_issue_reason": "Issue 45850 is the broadest umbrella report here: it explicitly frames Transformers v5 as breaking many models, so it can serve as the top-level tracker for related regression reports.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 33453 is the most suitable issue to anchor this set because it is a clear, substantive regression report rather than a doc nit, feature request, or highly model-specific edge case.", + "best_issue_reason": "Issue 45850 is the best representative of the cluster because it is broad, current, and could subsume multiple v5-related complaints, even though it is not a precise root-cause report.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:31795", + "right": "issue:33453", + "accept": false, + "reason": "Different problem types: documentation wording vs tokenizer loading regression." + }, + { + "left": "issue:41669", + "right": "issue:44908", + "accept": false, + "reason": "Import-star cleanup/perf issue is unrelated to inverse_sqrt scheduler kwargs handling." + }, + { + "left": "issue:42371", + "right": "issue:44492", + "accept": false, + "reason": "TF32 settings behavior and a cache-strategy typo are unrelated." + }, + { + "left": "issue:43576", + "right": "issue:44273", + "accept": false, + "reason": "Broken env command is not the same code path as lazy loading behavior." + }, + { + "left": "issue:44360", + "right": "issue:45468", + "accept": false, + "reason": "DSA indexer/ReLU complaint is unrelated to Gemma-4 audio positional encoding." + }, + { + "left": "issue:44655", + "right": "issue:44861", + "accept": false, + "reason": "Saving pipelines and tied-weight-key crashes are different save-path bugs." + }, + { + "left": "issue:44492", + "right": "issue:45850", + "accept": false, + "reason": "A typo report is not the same as the broad v5 breakage umbrella." + }, + { + "left": "issue:44273", + "right": "issue:44908", + "accept": false, + "reason": "Lazy loading and scheduler kwargs are unrelated subsystems." + }, + { + "left": "issue:45230", + "right": "issue:45310", + "accept": false, + "reason": "The first is too generic to establish the same concrete from_pretrained failure as the Qwen3.5 MoE report." + }, + { + "left": "issue:43519", + "right": "issue:44485", + "accept": false, + "reason": "Different models and different logic: timestamp calculation vs RoPE implementation." + }, { "left": "issue:41669", "right": "issue:43704", "accept": false, - "reason": "Different problems: import-time wildcard usage/perf vs a Qwen3 VRAM leak in multithreaded dataloaders." + "reason": "Import cleanup/perf work is unrelated to a VRAM leak in dataloader threads." }, { "left": "issue:43519", "right": "issue:44908", "accept": false, - "reason": "Unrelated subsystems: Qwen3VL timestamp calculation vs inverse_sqrt scheduler kwargs handling." + "reason": "Timestamp math in Qwen3VL and scheduler keyword propagation are unrelated." }, { "left": "issue:31795", "right": "issue:34689", "accept": false, - "reason": "Documentation confusion is not the same as a model-loading regression for Llama 3.2." + "reason": "Documentation confusion and Llama 3.2 model loading regression are not the same bug." }, { "left": "issue:31795", "right": "issue:37428", "accept": false, - "reason": "Doc wording issue vs an ImportError from flash attention utils; no shared bug." + "reason": "Docs issue vs flash-attention import error; no shared code-path bug." }, { "left": "issue:44315", "right": "issue:45092", "accept": false, - "reason": "Both concern loading paths, but one is Liger kernel application in model_init and the other is remote-code/meta-init incompatibility." + "reason": "Both mention loading/init, but they target different models and different failures." }, { "left": "issue:43704", "right": "issue:44485", "accept": false, - "reason": "VRAM leak in dataloader threads is unrelated to GLM-5 RoPE implementation." + "reason": "VRAM leak under threading is unrelated to GLM-5 RoPE math." }, { "left": "issue:44485", "right": "issue:45468", "accept": false, - "reason": "Both are model-specific encoding bugs, but they affect different models and different code paths." + "reason": "RoPE implementation and audio positional encoding are different model-specific issues." }, { "left": "issue:42371", "right": "issue:43704", "accept": false, - "reason": "TF32 configuration guidance is unrelated to a Qwen3 memory leak." + "reason": "TF32 configuration and VRAM leakage are unrelated." }, { "left": "issue:43502", "right": "issue:43519", "accept": false, - "reason": "local_files_only network leakage is unrelated to timestamp computation in a processor." + "reason": "local_files_only network leakage and timestamp calculation are unrelated." }, { "left": "issue:43519", "right": "issue:44623", "accept": false, - "reason": "Different processor bugs: timestamp math vs missing files on save_pretrained." + "reason": "Timestamp calculation bug is unrelated to missing processor save files." }, { "left": "issue:43519", "right": "issue:43704", "accept": false, - "reason": "Processor timestamp bug is unrelated to a VRAM leak." + "reason": "Different failures: timestamp math vs multi-thread VRAM leak." }, { "left": "issue:44263", "right": "issue:45468", "accept": false, - "reason": "Different model internals: torch.split return behavior vs Gemma-4 audio positional encoding." + "reason": "torch.split return handling and Gemma audio encoding are unrelated." }, { "left": "issue:43606", "right": "issue:45468", "accept": false, - "reason": "CPU offload device mismatch is not the same bug as Gemma-4 audio positional encoding." + "reason": "CPU offload device mismatch is a different bug than Gemma audio positional encoding." }, { "left": "issue:44261", "right": "issue:45468", "accept": false, - "reason": "MLA q_a_layernorm precision/config issue is unrelated to Gemma-4 audio positional encoding." + "reason": "Missing rms_norm_eps in MLA and Gemma audio positional encoding are different implementation issues." }, { "left": "issue:43704", "right": "issue:45341", "accept": false, - "reason": "A VRAM leak and a small testing_utils bug are not the same underlying change." + "reason": "VRAM leak and a testing_utils bug are unrelated." }, { "left": "issue:42913", "right": "issue:43066", "accept": false, - "reason": "Both involve tokenizer v5 behavior, but one is a broad regression report and the other is a specific decoder-type mismatch; not the same concrete bug." + "reason": "Both are tokenizer-v5 complaints, but they describe different concrete failures and code paths." }, { "left": "issue:44908", "right": "issue:44998", "accept": false, - "reason": "Scheduler kwargs handling has nothing to do with an unrelated low-signal issue." + "reason": "Scheduler kwargs bug is unrelated to the placeholder 'Unemployment' issue." }, { "left": "issue:44861", "right": "issue:45290", "accept": false, - "reason": "Tied-weights key handling is unrelated to apply_chat_template crashes on tool-call messages." + "reason": "Tied-weight-key crash and apply_chat_template tool-call crash are different code paths." }, { "left": "issue:43704", "right": "issue:44998", "accept": false, - "reason": "A real Qwen3 memory leak is unrelated to the unrelated low-content issue." + "reason": "VRAM leak in threading and the unrelated placeholder issue are not duplicates." }, { "left": "issue:41669", "right": "issue:44998", "accept": false, - "reason": "Import-star performance regression is unrelated to the unrelated low-content issue." + "reason": "Import-star cleanup/perf issue does not match the unrelated placeholder issue." }, { "left": "issue:43525", "right": "issue:44861", "accept": false, - "reason": "pad_token_id missing in Llama4Config is not the same as tied-weight key crashes." + "reason": "Missing pad_token_id and tied_weight_key crashes are unrelated." }, { "left": "issue:43066", "right": "issue:44998", "accept": false, - "reason": "Tokenizer v5 decoder-type bug is unrelated to the unrelated low-content issue." + "reason": "Wrong tokenizer decoder type and the unrelated placeholder issue are not the same bug." }, { "left": "issue:43572", "right": "issue:44861", "accept": false, - "reason": "StableLmConfig pad_token_idx regression is unrelated to tied-weight key handling." + "reason": "Missing pad_token_idx and tied-weight-key crashes are separate compatibility bugs." }, { "left": "issue:31515", "right": "issue:38617", "accept": false, - "reason": "Checkpoint loading slowness and an ImportError from configuration_utils are distinct issues." + "reason": "Both are loading-related, but one is slow checkpoint loading and the other is an ImportError." }, { "left": "issue:44360", "right": "issue:44998", "accept": false, - "reason": "DSA indexer missing ReLU is unrelated to the unrelated low-content issue." + "reason": "DSA/ReLU discussion is unrelated to the placeholder issue." }, { "left": "issue:44230", "right": "issue:44623", "accept": false, - "reason": "fp8 inference support is not the same bug as missing processor save files." + "reason": "FP8 inference support for Qwen models is unrelated to processor.save_pretrained missing files." }, { "left": "issue:30333", "right": "issue:31515", "accept": false, - "reason": "MLFlow status reporting failure is unrelated to slow from_pretrained checkpoint loading." + "reason": "MLflow failure reporting and slow checkpoint loading are unrelated." }, { "left": "issue:31515", "right": "issue:36296", "accept": false, - "reason": "Slow checkpoint loading is unrelated to a tensor-parallel training bug." + "reason": "Slow checkpoint loading and tensor-parallel training bugs are different issues." }, { "left": "issue:31515", "right": "issue:31795", "accept": false, - "reason": "Checkpoint loading performance and documentation confusion are different classes of issues." + "reason": "Checkpoint loading performance and documentation confusion are unrelated." }, { "left": "issue:44230", "right": "issue:45310", "accept": false, - "reason": "fp8 support/inference and a Qwen3.5 MoE from_pretrained error are not the same concrete bug." + "reason": "Both involve Qwen3.5, but FP8 inference support and a from_pretrained error are not the same concrete bug." }, { "left": "issue:31515", "right": "issue:33453", "accept": false, - "reason": "Both are loading-related, but one is slowness and the other is a tokenizer loading regression; not the same failure mode." + "reason": "Checkpoint load slowness and tokenizer-loading regression are not the same failure." }, { "left": "issue:43298", "right": "issue:44230", "accept": false, - "reason": "Audio continuous batching support and Qwen3 fp8 support are different feature areas." + "reason": "Continuous batching audio-model support and Qwen fp8 inference support are different feature gaps." } ] }, @@ -14861,40 +14950,40 @@ "soft_pairs": [ "issue:26350|issue:33260" ], - "prepared_review_unit_hash": "58c47de5b36a4bf733d1b52246dcb4a7c369e3678ea1926eefb811244a533dee", + "prepared_review_unit_hash": "519d87cc20a230a2a3f40bc380aef1c2475c503c0ce9f47bc203d14abd88334a", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1588, - "estimated_input_tokens": 397, - "estimated_eval_tokens": 1050 + "serialized_chars": 1590, + "estimated_input_tokens": 398, + "estimated_eval_tokens": 1052 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1588, - "estimated_input_tokens": 397, - "estimated_eval_tokens": 1050 + "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 similar in form but not in substance: one requests Flash Attention 2 support for more architectures, the other requests GGUF support for more architectures. They are related feature-expansion ideas, not the same underlying change.", - "confidence": 0.97, + "summary": "These are not duplicates: one requests Flash Attention 2 support for more architectures, the other requests GGUF support for more architectures. They concern different features and code paths, so the soft link should be rejected.", + "confidence": 0.98, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "No canonical issue applies because the requests are for different capabilities; if one must be chosen as the cluster representative, issue 26350 is older and has more discussion, but it is not a duplicate of 33260.", + "best_issue_reason": "Issue #33260 is the more active and recent enhancement request, with more inbound references; it is the better representative issue in this cluster, though it is not a duplicate of #26350.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:26350", "right": "issue:33260", "accept": false, - "reason": "Different feature requests touching different formats/subsystems (Flash Attention 2 vs GGUF); they do not describe the same underlying bug or change." + "reason": "Different underlying feature requests: Flash Attention 2 support vs GGUF support. They target separate functionality and are not the same bug or change." } ] }, @@ -14911,7 +15000,7 @@ "component_count": 16, "review_unit_index": 1, "review_unit_count": 1, - "status": "error", + "status": "reviewed", "reason": null, "source": "llm", "cache_hit": false, @@ -14931,12 +15020,12 @@ "issue:43834|issue:44365", "issue:44365|issue:44370" ], - "prepared_review_unit_hash": "880a630ad9de1d4aeee9caa49170291aa4fe7528be674c7c1f672099d415ee62", + "prepared_review_unit_hash": "57e9692166a6300ac44fc31a952b9144d3960b11c7d5cb8da932fc36bb01b983", "component_budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 4453, + "serialized_chars": 4454, "estimated_input_tokens": 1114, "estimated_eval_tokens": 2484 }, @@ -14944,7 +15033,7 @@ "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 4453, + "serialized_chars": 4454, "estimated_input_tokens": 1114, "estimated_eval_tokens": 2484 }, @@ -14952,7 +15041,52 @@ "trimmed": false, "aggressively_trimmed": false, "split": false, - "analyst_result": null, + "analyst_result": { + "summary": "These are all docs translation request issues with highly similar template titles, but they do not provide enough evidence to treat them as the same underlying request. The Turkish-specific issue is concrete; the others look like placeholder/template variants that may refer to different languages.", + "confidence": 0.94, + "canonical_issue_reason": "issue:27088 is the most concrete and specific item: it names Turkish explicitly, while the other issues use generic placeholder titles and do not show a shared target language or identical request.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:27088 is the best representative issue because it is specific, actionable, and not just a template placeholder.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:27088", + "right": "issue:44370", + "accept": false, + "reason": "Same broad i18n-docs theme, but no evidence they target the same language or the same request; one is Turkish-specific and the other is a generic template issue." + }, + { + "left": "issue:27088", + "right": "issue:44365", + "accept": false, + "reason": "The titles are superficially similar, but the generic placeholder issue does not show it is the same Turkish translation request." + }, + { + "left": "issue:27088", + "right": "issue:43834", + "accept": false, + "reason": "No proof these are the same underlying issue; 27088 is explicitly Turkish, while 43834 is a generic language template." + }, + { + "left": "issue:43834", + "right": "issue:44370", + "accept": false, + "reason": "Both are generic translation-template issues, but there is no concrete language target or identical request evidence to merge them." + }, + { + "left": "issue:43834", + "right": "issue:44365", + "accept": false, + "reason": "The near-identical template titles are not enough to confirm the same issue; they could be separate translation requests for different languages." + }, + { + "left": "issue:44365", + "right": "issue:44370", + "accept": false, + "reason": "They look templated and similar, but without explicit language targets or shared unique details, they should not be treated as duplicates." + } + ] + }, "evaluator_result": null, "evaluator_used": false, "retried": false, @@ -15026,250 +15160,244 @@ "issue:37015|issue:45027", "issue:43837|issue:45177" ], - "prepared_review_unit_hash": "fef203f7476cd19461d08af90d8a191b596aa6f0e3181f4eea411af0f6bd9663", + "prepared_review_unit_hash": "774ef35ef344ffc8d750041b305d6553f672888c3f09c1fb512514bd140073a8", "component_budget": { "node_count": 15, "item_count": 15, "soft_pair_count": 35, - "serialized_chars": 20710, - "estimated_input_tokens": 5178, - "estimated_eval_tokens": 10612 + "serialized_chars": 20705, + "estimated_input_tokens": 5177, + "estimated_eval_tokens": 10610 }, "budget": { "node_count": 15, "item_count": 15, "soft_pair_count": 35, - "serialized_chars": 20710, - "estimated_input_tokens": 5178, - "estimated_eval_tokens": 10612 + "serialized_chars": 20705, + "estimated_input_tokens": 5177, + "estimated_eval_tokens": 10610 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are all model-support request issues, but they refer to different models, architectures, and sometimes different modalities (text, ASR, TTS, OCR, vision). I don\u2019t see true duplicates or a single underlying bug/change; the cluster is only loosely similar by topic.", - "confidence": 0.96, - "canonical_issue_reason": "issue:42738 is the broadest and most general request in the set ('BERT-like models with RoPE'), so it is the most reasonable anchor if one issue must represent the cluster.", + "summary": "These are superficially similar model-request issues, but they target different architectures and products (vision, audio, OCR, BERT-like, etc.). They are not duplicates of the same underlying change.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue stands out: the cluster contains distinct requests for different models, not one shared bug or feature.", "canonical_pr_reason": null, - "best_issue_reason": "issue:42738 is the best representative issue because it is broader than the other model-specific requests and fits the shared theme without overfitting to one exact model.", + "best_issue_reason": "No global best issue for deduping; the items are separate model-support requests and should remain independent.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:36817", "right": "issue:41211", "accept": false, - "reason": "Different model requests: EuroBert config support vs DEIMv2 architecture support." + "reason": "Different requested models (EuroBert vs DEIMv2); only the generic 'add model' framing matches." }, { "left": "issue:36817", "right": "issue:43671", "accept": false, - "reason": "EuroBert config support and Qwen3-TTS support are unrelated models and modalities." + "reason": "EuroBert config support and Qwen3-TTS support are unrelated model additions with different code paths." }, { "left": "issue:36817", "right": "issue:36879", "accept": false, - "reason": "EuroBert and RF-DETR are different model families; not the same change." + "reason": "EuroBert and RF-DETR are different model families; not the same feature request." }, { "left": "issue:36879", "right": "issue:41211", "accept": false, - "reason": "RF-DETR and DEIMv2 are separate vision-model requests." + "reason": "RF-DETR and DEIMv2 are distinct detection models; no duplicate underlying change." }, { "left": "issue:36879", "right": "issue:43909", "accept": false, - "reason": "RF-DETR is an object-detection model; LFM2.5 Audio is an audio model request." + "reason": "RF-DETR is a vision model request, while LFM2.5 Audio is an audio-model request." }, { "left": "issue:36879", "right": "issue:44121", "accept": false, - "reason": "RF-DETR and OpenAI weight-sparse transformer are unrelated architectures." + "reason": "Different model targets and likely different integration paths; not a single duplicate request." }, { "left": "issue:36879", "right": "issue:43837", "accept": false, - "reason": "RF-DETR and Qwen3-ASR target different model types and tasks." + "reason": "RF-DETR and Qwen3-ASR are different modalities and support requests." }, { "left": "issue:36879", "right": "issue:43671", "accept": false, - "reason": "RF-DETR and Qwen3-TTS are distinct model requests with different modalities." + "reason": "RF-DETR (vision) and Qwen3-TTS (speech generation) are unrelated model additions." }, { "left": "issue:37015", "right": "issue:43837", "accept": false, - "reason": "NeoBERT and Qwen3-ASR are different models; no shared bug/change." + "reason": "NeoBERT and Qwen3-ASR are unrelated models with different architecture and modality." }, { "left": "issue:37015", "right": "issue:44121", "accept": false, - "reason": "NeoBERT support and circuit-sparsity transformer support are unrelated requests." + "reason": "NeoBERT and the weight-sparse transformer request are not the same model or code change." }, { "left": "issue:37015", "right": "issue:43909", "accept": false, - "reason": "NeoBERT is a text encoder model; LFM2.5 Audio is an audio model request." + "reason": "NeoBERT is a BERT-like NLP model; LFM2.5 Audio is an audio model request." }, { "left": "issue:37015", "right": "issue:42503", "accept": false, - "reason": "NeoBERT and ModernVBERT are different model additions, not the same request." + "reason": "NeoBERT and ModernVBERT are different model variants; similar naming only." }, { "left": "issue:36879", "right": "issue:37015", "accept": false, - "reason": "RF-DETR and NeoBERT are unrelated model families." + "reason": "RF-DETR and NeoBERT are distinct model requests across different domains." }, { "left": "issue:41115", "right": "issue:43837", "accept": false, - "reason": "MiniCPM3 architecture support and Qwen3-ASR support are different model/task requests." + "reason": "MiniCPM3 model architecture and Qwen3-ASR support are different requests." }, { "left": "issue:41115", "right": "issue:44121", "accept": false, - "reason": "MiniCPM3 support and weight-sparse transformer support are not the same code-path change." + "reason": "MiniCPM3 and the weight-sparse transformer proposal are not the same underlying model change." }, { "left": "issue:41115", "right": "issue:42503", "accept": false, - "reason": "MiniCPM3 and ModernVBERT are separate model requests." + "reason": "MiniCPM3 and ModernVBERT are separate model additions, not duplicates." }, { "left": "issue:41115", "right": "issue:45027", "accept": false, - "reason": "MiniCPM3 is a general model architecture request; Voxtral TTS support is a different audio-model request." + "reason": "MiniCPM3 and Voxtral TTS support are different model support requests with different modalities." }, { "left": "issue:41211", "right": "issue:43671", "accept": false, - "reason": "DEIMv2 and Qwen3-TTS are unrelated model additions." + "reason": "DEIMv2 and Qwen3-TTS are unrelated model requests; only broad similarity exists." }, { "left": "issue:42503", "right": "issue:43837", "accept": false, - "reason": "ModernVBERT and Qwen3-ASR are different models and different tasks." + "reason": "ModernVBERT and Qwen3-ASR are distinct speech/NLP model requests." }, { "left": "issue:42503", "right": "issue:43671", "accept": false, - "reason": "ModernVBERT and Qwen3-TTS are distinct model requests." + "reason": "ModernVBERT and Qwen3-TTS are different models and different tasks." }, { "left": "issue:42503", "right": "issue:42738", "accept": false, - "reason": "ModernVBERT is a specific model request; 'BERT-like models with RoPE' is a broad category request." + "reason": "ModernVBERT and BERT-like models with RoPE are related only at a naming level, not the same concrete change." }, { "left": "issue:42738", "right": "issue:45027", "accept": false, - "reason": "BERT-like models with RoPE and Voxtral TTS support are not the same underlying change." + "reason": "BERT-like RoPE support and Voxtral TTS support are not the same feature." }, { "left": "issue:43671", "right": "issue:43837", "accept": false, - "reason": "Qwen3-TTS and Qwen3-ASR are related by family name but target different modalities and fixes." + "reason": "Qwen3-TTS and Qwen3-ASR are both audio-related but still different model support requests." }, { "left": "issue:43671", "right": "issue:45027", "accept": false, - "reason": "Qwen3-TTS and Voxtral TTS are different model-specific support requests." + "reason": "Qwen3-TTS and Voxtral-4B-TTS are different TTS model requests, not one duplicate fix." }, { "left": "issue:43671", "right": "issue:45177", "accept": false, - "reason": "Qwen3-TTS and DeepSeek-OCR2 are unrelated model requests." + "reason": "Qwen3-TTS and DeepSeek-OCR2 target different modalities and code paths." }, { "left": "issue:43671", "right": "issue:44121", "accept": false, - "reason": "Qwen3-TTS support and circuit-sparsity transformer support are not the same change." - }, - { - "left": "issue:43671", - "right": "issue:43909", - "accept": false, - "reason": "Qwen3-TTS and LFM2.5 Audio are different audio model requests." + "reason": "TTS support and the weight-sparse transformer request are unrelated model changes." }, { "left": "issue:43837", "right": "issue:45322", "accept": false, - "reason": "Qwen3-ASR and EUPE are different model requests with different purposes." + "reason": "Qwen3-ASR support and EUPE are different model requests in different subsystems." }, { "left": "issue:43837", "right": "issue:43909", "accept": false, - "reason": "Qwen3-ASR and LFM2.5 Audio are distinct audio-model support requests." + "reason": "Qwen3-ASR and LFM2.5 Audio are both audio-related but not the same model or implementation." }, { "left": "issue:43909", "right": "issue:44205", "accept": false, - "reason": "LFM2.5 Audio and SAM3-LiteText are different multimodal model requests." + "reason": "LFM2.5 Audio and SAM3-LiteText are distinct model requests; no shared duplicate change." }, { "left": "issue:43671", "right": "issue:43909", "accept": false, - "reason": "Qwen3-TTS and LFM2.5 Audio are different audio-model support requests." + "reason": "Qwen3-TTS and LFM2.5 Audio are separate audio-model support requests." }, { "left": "issue:44121", "right": "issue:45322", "accept": false, - "reason": "OpenAI weight-sparse transformer and EUPE are unrelated model additions." + "reason": "Weight-sparse transformer support and EUPE are different model proposals with different semantics." }, { "left": "issue:43837", "right": "issue:44121", "accept": false, - "reason": "Qwen3-ASR and circuit-sparsity transformer support do not describe the same code-path change." + "reason": "Qwen3-ASR and the weight-sparse transformer request are unrelated changes." }, { "left": "issue:43837", "right": "issue:45027", "accept": false, - "reason": "Qwen3-ASR and Voxtral TTS are different speech-model support requests." + "reason": "Qwen3-ASR and Voxtral TTS are different speech model support requests." }, { "left": "issue:37015", "right": "issue:45027", "accept": false, - "reason": "NeoBERT and Voxtral TTS are unrelated model requests." + "reason": "NeoBERT and Voxtral TTS are unrelated model additions." }, { "left": "issue:43837", "right": "issue:45177", "accept": false, - "reason": "Qwen3-ASR and DeepSeek-OCR2 are different tasks and model families." + "reason": "Qwen3-ASR and DeepSeek-OCR2 are different modality/model requests." } ] }, @@ -15310,82 +15438,82 @@ "issue:43668|issue:44031", "issue:43668|issue:43813" ], - "prepared_review_unit_hash": "16dfdefcb4a424862aa464d7297d6cb7845cfdca03626deb866d9b1043d9f7e4", + "prepared_review_unit_hash": "2465db6d3a1554098e99f145482d26e3767baebc94884ad91c807f8cff6eda83", "component_budget": { "node_count": 6, "item_count": 6, "soft_pair_count": 8, - "serialized_chars": 6293, - "estimated_input_tokens": 1574, - "estimated_eval_tokens": 3404 + "serialized_chars": 6289, + "estimated_input_tokens": 1573, + "estimated_eval_tokens": 3402 }, "budget": { "node_count": 6, "item_count": 6, "soft_pair_count": 8, - "serialized_chars": 6293, - "estimated_input_tokens": 1574, - "estimated_eval_tokens": 3404 + "serialized_chars": 6289, + "estimated_input_tokens": 1573, + "estimated_eval_tokens": 3402 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These issues are not duplicates of one another; they span unrelated models, configs, typos, and tokenizer warnings. No issue/PR canonical cluster emerges from the provided titles alone.", - "confidence": 0.93, - "canonical_issue_reason": "No clear canonical issue: the set covers distinct bugs/questions in different components (Llama4, Mask2Former, ModernBERT, quantization, tokenizers) with no shared underlying fix.", + "summary": "This cluster is not a duplicate set: the issues span unrelated models, tokenizer warnings, config typing, and a stray typo. No soft edge should be accepted.", + "confidence": 0.98, + "canonical_issue_reason": "No canonical issue: the items are unrelated bugs across different components (Llama4, Mask2Former, ModernBERT, quantization, tokenizer regex warnings).", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44031 is the strongest standalone representative because it describes a concrete, user-visible regression with broader impact and clear reproduction context implied by the version-specific tokenizer warning.", + "best_issue_reason": "No single issue is a good cluster representative; if forced, 43668 is the cleanest standalone bug report, but it does not represent the rest of the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:37934", "right": "issue:43668", "accept": false, - "reason": "Different components and problems: a Llama4 normalization design question versus a ModernBERT config type-hint issue." + "reason": "Different model/component and different bug class: Llama4 norm behavior vs ModernBERT config type hint." }, { "left": "issue:42936", "right": "issue:43240", "accept": false, - "reason": "Mask2Former ignore_value being unused is unrelated to kwargs not being forwarded into a loss function." + "reason": "Mask2former ignore_value being unused is unrelated to kwargs not being forwarded to a loss function." }, { "left": "issue:42936", "right": "issue:44031", "accept": false, - "reason": "These concern different subsystems and failure modes: Mask2Former loss handling versus tokenizer regex warnings." + "reason": "Different subsystems and symptoms: model loss parameter handling vs tokenizer regex warnings." }, { "left": "issue:42936", "right": "issue:43813", "accept": false, - "reason": "A Mask2Former loss-parameter bug is not the same underlying issue as a typo in a quantization attribute name." + "reason": "An unused loss parameter and a typo in a quantization attribute are not the same underlying bug." }, { "left": "issue:42936", "right": "issue:43668", "accept": false, - "reason": "Both mention model config/loss-adjacent topics, but the concrete bugs are unrelated and would not be fixed by one patch." + "reason": "These are separate issues in different models: Mask2former loss wiring vs ModernBERT config typing." }, { "left": "issue:43240", "right": "issue:43668", "accept": false, - "reason": "Passing kwargs to a loss function is a behavior bug, while ModernBERTConfig norm_eps is a type annotation problem." + "reason": "Loss kwargs forwarding bug and a config type-hint error are unrelated." }, { "left": "issue:43668", "right": "issue:44031", "accept": false, - "reason": "A config type-hint mismatch and a tokenizer regex warning regression are unrelated issues." + "reason": "ModernBERTConfig typing issue does not match the tokenizer regex warning regression." }, { "left": "issue:43668", "right": "issue:43813", "accept": false, - "reason": "ModernBERT norm_eps typing and a quantization typo are distinct, non-overlapping fixes." + "reason": "Config type hint incorrectness and a typo in a quantization symbol are different problems." } ] }, @@ -15415,12 +15543,12 @@ "soft_pairs": [ "issue:41084|issue:43037" ], - "prepared_review_unit_hash": "df808b98ceeda2c6ef33bfc35a4e8b5bd0ecbf87a2e897998d183170618a8512", + "prepared_review_unit_hash": "827e6bd2a0944778c7ad390c12ebb457d7fc7e7be8d36660672ebd8ea990a9c2", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1479, + "serialized_chars": 1480, "estimated_input_tokens": 370, "estimated_eval_tokens": 996 }, @@ -15428,7 +15556,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1479, + "serialized_chars": 1480, "estimated_input_tokens": 370, "estimated_eval_tokens": 996 }, @@ -15437,18 +15565,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are about different feature requests and do not look like duplicates: one is for block decoding, the other for DeepSeek v3.2 support.", - "confidence": 0.97, - "canonical_issue_reason": "No canonical issue: the titles and likely scopes are unrelated, so neither issue cleanly subsumes the other.", + "summary": "These are unrelated open issues: one about block decoding behavior and one about adding DeepSeek v3.2 support. No shared targets, filenames, or explicit linkage suggests duplication.", + "confidence": 0.98, + "canonical_issue_reason": "#41084 is the more specific implementation/behavior issue; #43037 is a model-support request and does not appear to be a duplicate of it.", "canonical_pr_reason": null, - "best_issue_reason": "Neither issue is a good cluster representative for a duplicate set because they describe different underlying requests; if forced, #43037 is more specific, but it is not a duplicate of #41084.", + "best_issue_reason": "Neither issue is a strong global canonical for the other because they cover different product areas and different kinds of work.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:41084", "right": "issue:43037", "accept": false, - "reason": "Different topics and no shared code-path evidence; this looks like two separate feature requests, not the same underlying bug or change." + "reason": "Different subjects and likely different code paths: block decoding vs DeepSeek v3.2 model support. Soft similarity alone is not enough to treat them as the same underlying bug or change." } ] }, @@ -15478,7 +15606,7 @@ "soft_pairs": [ "issue:43598|issue:43600" ], - "prepared_review_unit_hash": "fad4c255a15ba146e429de201108d42c070e8fc7b311032e8b2718735a29546f", + "prepared_review_unit_hash": "932a3ee1df1b0e3bccc096c5ca5449d3ecc4bcd6ebff5514aacbb40ab393d1fe", "component_budget": { "node_count": 2, "item_count": 2, @@ -15500,18 +15628,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are related only at a very high level (custom trainers / Trainer behavior) but they describe different problems and different code paths, so they should not be merged as duplicates.", - "confidence": 0.91, - "canonical_issue_reason": "Issue #43598 is the better representative of the cluster because it is a concrete Trainer customization request centered on `remove_unused_column`; it has the clearest, most generalizable framing of the shared theme.", + "summary": "The two issues are both about Trainer extensibility, but they describe different problems: one is about `remove_unused_column` behavior/customization, the other about FLOPs estimation for custom trainers. They should not be clustered as duplicates.", + "confidence": 0.87, + "canonical_issue_reason": "Issue #43598 is the better representative because it states a specific Trainer customization problem and is more directly about a code-path behavior change.", "canonical_pr_reason": null, - "best_issue_reason": "Issue #43598 is the strongest single representative for the cluster, but it should not be treated as the same bug as #43600.", + "best_issue_reason": "Issue #43598 is the strongest issue to keep as the cluster representative; #43600 is a separate Trainer accounting problem rather than the same bug.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43598", "right": "issue:43600", "accept": false, - "reason": "They both mention custom trainers, but #43598 is about making `remove_unused_column` more customizable in `Trainer`, while #43600 is about accounting for custom trainers when estimating FLOPS. Different functionality, different failure modes, and not plausibly one merged fix." + "reason": "They share a broad Trainer/customization theme, but the concrete problems differ: unused-column removal vs FLOPs estimation for custom trainers." } ] }, @@ -15541,12 +15669,12 @@ "soft_pairs": [ "issue:43684|issue:44405" ], - "prepared_review_unit_hash": "6ef2a09e2bc8a74def6e854d4f788412aec0fca355139c61879a19488643b278", + "prepared_review_unit_hash": "f4c92435dc4ee3df0a87e4596eb4d92012908abe7a0885c5b430b3e83c35fb23", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1611, + "serialized_chars": 1612, "estimated_input_tokens": 403, "estimated_eval_tokens": 1062 }, @@ -15554,7 +15682,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1611, + "serialized_chars": 1612, "estimated_input_tokens": 403, "estimated_eval_tokens": 1062 }, @@ -15563,18 +15691,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related only at a broad transformer-model support level, but they target different model families and different AutoModel entry points, so they are not duplicates.", - "confidence": 0.93, - "canonical_issue_reason": "Issue 43684 is the broader registration request and covers a core AutoModel / AutoModelForConditionalGeneration gap for Qwen3-Omni, making it the better canonical issue for this cluster.", + "summary": "These are not duplicates: one request is for Qwen3-Omni registration in AutoModel/AutoModelForConditionalGeneration, while the other is for AutoModelForSequenceClassification support for Qwen3.5. They target different model families and different head/task support.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue fits both; they describe different underlying changes and should remain separate.", "canonical_pr_reason": null, - "best_issue_reason": "43684 is the best representative because it describes a clear library registration deficiency affecting generic model loading, whereas 44405 is a narrower task for a different config and a different task head.", + "best_issue_reason": "Neither issue is a strong global canonical for the cluster because the requests are unrelated in model family and API surface.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43684", "right": "issue:44405", "accept": false, - "reason": "Different underlying requests: one is about registering Qwen3-Omni for AutoModel/AutoModelForConditionalGeneration, the other is about adding AutoModelForSequenceClassification support for Qwen3.5. They involve different configs and API surfaces, so they should not be merged as duplicates." + "reason": "Different underlying requests: Qwen3-Omni model registration vs Qwen3.5 sequence-classification support. They do not fix the same code path or change." } ] }, @@ -15611,12 +15739,12 @@ "issue:43845|issue:43859", "issue:43846|issue:43859" ], - "prepared_review_unit_hash": "bb4e6ec27fbd7e1456e7ce86fc840e8aea730d072af661a711173902499a45eb", + "prepared_review_unit_hash": "279365bda713883de957e004ac3ce915d1284e6cf5084ffe538f356ac363dac6", "component_budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 4267, + "serialized_chars": 4265, "estimated_input_tokens": 1067, "estimated_eval_tokens": 2390 }, @@ -15624,7 +15752,7 @@ "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 4267, + "serialized_chars": 4265, "estimated_input_tokens": 1067, "estimated_eval_tokens": 2390 }, @@ -15633,48 +15761,48 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Three issues share the exact title \"huggingface \" and look like duplicates, but there is no body detail to prove they describe the same underlying bug. \"Fett\" appears unrelated. With only title-level similarity, I would not merge any pair.", - "confidence": 0.33, - "canonical_issue_reason": "issue:43845 is the earliest of the near-identical \"huggingface \" issues, so it is the best representative if this cluster is deduplicated.", + "summary": "The cluster is mostly low-information spam-like issues. The three \"huggingface\" tickets are superficially similar, but there isn\u2019t enough body detail, repro, or target evidence to safely treat them as confirmed duplicates. \"Fett\" looks unrelated.", + "confidence": 0.61, + "canonical_issue_reason": "issue:43845 is the best available anchor only because it is the earliest of the repeated \"huggingface\" reports; however, all of these issues are too sparse to prove a real duplicate bug.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43845 is the cleanest issue to keep as the canonical duplicate target because it is earliest and matches the repeated title exactly.", + "best_issue_reason": "issue:43845 is the most plausible cluster representative for triage since it matches the repeated title pattern and is the earliest candidate, but the report quality is too poor to call it a strong canonical.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43835", "right": "issue:43845", "accept": false, - "reason": "Different titles and no shared body/detail; title overlap is too weak to infer the same bug." + "reason": "Different titles and no shared bug details, targets, or files; similarity alone is not enough." }, { "left": "issue:43835", "right": "issue:43846", "accept": false, - "reason": "Different titles and no evidence beyond generic similarity." + "reason": "Different titles and no evidence they describe the same underlying problem." }, { "left": "issue:43835", "right": "issue:43859", "accept": false, - "reason": "Different titles and no concrete indicator they report the same problem." + "reason": "No substantive content to connect \"Fett\" with the \"huggingface\" reports." }, { "left": "issue:43845", "right": "issue:43846", "accept": false, - "reason": "Same short title, but no description or issue target to confirm they are the same underlying report." + "reason": "Identical titles, but the bodies are too thin to confirm the same bug or one concrete duplicate change." }, { "left": "issue:43845", "right": "issue:43859", "accept": false, - "reason": "Identical title alone is insufficient; the bodies are too sparse to verify the same bug or change." + "reason": "Same broad title only; no repro, target, or code-path evidence tying them together." }, { "left": "issue:43846", "right": "issue:43859", "accept": false, - "reason": "Identical title alone does not establish a duplicate without supporting details." + "reason": "Shared title pattern only; insufficient detail to accept as the same issue." } ] }, @@ -15715,88 +15843,88 @@ "issue:44572|issue:44937", "issue:44573|issue:44937" ], - "prepared_review_unit_hash": "c0b07763fa508d7a153a6361c283bf88e7dba455d9fc5d2a14b469edb0372b87", + "prepared_review_unit_hash": "0b5dbb12d7c3895d4bfaab7863562af247e50e9b60c009ad856995977a29fc64", "component_budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 9, - "serialized_chars": 5791, - "estimated_input_tokens": 1448, - "estimated_eval_tokens": 3152 + "serialized_chars": 5788, + "estimated_input_tokens": 1447, + "estimated_eval_tokens": 3150 }, "budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 9, - "serialized_chars": 5791, - "estimated_input_tokens": 1448, - "estimated_eval_tokens": 3152 + "serialized_chars": 5788, + "estimated_input_tokens": 1447, + "estimated_eval_tokens": 3150 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All five items look like low-signal or spammy closed issues with no substantive body, files, or linked targets. None are strong duplicate candidates; the identical \"\" titles are not enough to treat them as the same underlying issue.", - "confidence": 0.92, - "canonical_issue_reason": "Issue 44371 is the best placeholder canonical because it is the earliest of the generic spam-like issues and has no evidence suggesting a different topic from the other empty items.", + "summary": "This cluster looks like low-information or spam-like issues rather than a real duplicate set. The only somewhat descriptive item is the documentation suggestion issue; the others are placeholder/spam titles.", + "confidence": 0.78, + "canonical_issue_reason": "issue:44418 is the most informative and least spam-like title in the set; the others are obvious placeholders or spam-style posts.", "canonical_pr_reason": null, - "best_issue_reason": "44371 is the safest representative for this cluster: it is generic, earliest, and no item contains enough detail to justify a more specific choice.", + "best_issue_reason": "issue:44418 is the best representative of the cluster because it has the clearest, non-placeholder title, even though the overall cluster is not a strong duplicate group.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44371", "right": "issue:44572", "accept": false, - "reason": "Both are generic spam placeholders, but there is no concrete bug or change described to show they are the same underlying issue." + "reason": "Both are placeholder '' issues, but there is no evidence they describe the same underlying report rather than separate spam posts." }, { "left": "issue:44371", "right": "issue:44573", "accept": false, - "reason": "Same spam-style title, but no substantive body or shared technical context to confirm a true duplicate." + "reason": "Same title is not enough here; these look like independent spam/placeholder issues with no shared substantive content." }, { "left": "issue:44371", "right": "issue:44418", "accept": false, - "reason": "One is a spam placeholder and the other is a vague documentation suggestion; titles are too different and unsupported by details." + "reason": "One is a spam placeholder and the other is a documentation suggestion; they do not appear to be the same issue." }, { "left": "issue:44371", "right": "issue:44937", "accept": false, - "reason": "No overlap in described problem; one is spam-like and the other is a vague phrase with no technical content." + "reason": "A spam placeholder and a specific-sounding title are not enough to conclude a shared underlying bug/change." }, { "left": "issue:44418", "right": "issue:44572", "accept": false, - "reason": "A vague documentation suggestion and a spam placeholder are not the same underlying issue." + "reason": "Documentation enhancement vs placeholder spam; no concrete overlap in problem statement." }, { "left": "issue:44418", "right": "issue:44573", "accept": false, - "reason": "No evidence of shared bug/change; only a superficial similarity score." + "reason": "Documentation enhancement vs placeholder spam; they are not the same underlying issue." }, { "left": "issue:44572", "right": "issue:44573", "accept": false, - "reason": "Identical spam titles do not establish the same bug or change, especially with empty bodies." + "reason": "Identical spam titles alone are insufficient evidence of the same underlying bug/change." }, { "left": "issue:44572", "right": "issue:44937", "accept": false, - "reason": "The titles point to unrelated content, and neither item provides enough detail for duplicate confirmation." + "reason": "Placeholder spam and a specific named-check title do not clearly match the same report." }, { "left": "issue:44573", "right": "issue:44937", "accept": false, - "reason": "No concrete shared problem is described, so this is not safe to merge as a duplicate." + "reason": "No substantive evidence ties these together as the same issue; titles point to different content." } ] }, @@ -15826,7 +15954,7 @@ "soft_pairs": [ "issue:45145|issue:45151" ], - "prepared_review_unit_hash": "7f7cd23ef14eb76e99e0e92fa0c68a5a5e4affb89b0109088164b6329261b157", + "prepared_review_unit_hash": "f213f635b28561fb70ebcb6d958345f272a008a50a1a99f4297dc33fead21fb4", "component_budget": { "node_count": 2, "item_count": 2, @@ -15848,24 +15976,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The pair is likely a duplicate issue cluster, but the evidence is limited to identical titles, similar body lengths, and a soft-similarity signal. #45145 is the best canonical issue because it appeared earlier; however, the available packet does not prove the bodies are identical or that the requests are unquestionably the same.", - "confidence": 0.74, - "canonical_issue_reason": "#45145 is the earliest issue in a pair with the same title and no distinguishing links, targets, or filenames, so it is the best canonical issue to keep.", + "summary": "The two closed issues are very likely duplicates: they share the exact same title, close creation times, and similar body sizes, with soft-similarity support. The evidence is strong enough to treat them as the same issue thread, though not as absolute certainty.", + "confidence": 0.92, + "canonical_issue_reason": "Issue #45145 is the better canonical representative because it was created first and has an inbound reference, making it the slightly more established thread.", "canonical_pr_reason": null, - "best_issue_reason": "#45145 is the best representative for the cluster because it predates #45151 and nothing in the packet distinguishes #45151 as a different request.", + "best_issue_reason": "Issue #45145 is the best representative for the cluster because it is the earliest instance of the identical issue title and already has some reference activity.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45145", "right": "issue:45151", "accept": true, - "reason": "Likely duplicate based on identical titles and similar metadata, but the packet does not expose the full bodies, so this should be treated as a probable rather than proven duplicate." + "reason": "Accept: the issues have the exact same title and similar-sized bodies, and the packet flags soft similarity between them. This is consistent with a duplicate issue rather than a distinct report." } ] }, "evaluator_result": { "accept": false, - "feedback": "The cluster is plausibly duplicated, but the analyst overstates certainty. The packet only shows identical titles, similar body lengths, and a soft-similarity signal; it does not expose the actual issue bodies or prove they are the same request. The soft-edge verdict should be more conservative (e.g. \"likely duplicate\"), not a clear duplicate." + "feedback": "The packet supports a strong similarity/duplicate hypothesis, but the analyst slightly overstates the evidence. In particular, phrases like \"clearly the same underlying proposal\" and the 0.99 confidence go beyond what is directly grounded in the packet, which only shows identical titles, similar body sizes, and soft-similarity evidence. A more conservative wording would be safer." }, "evaluator_used": true, "retried": true, @@ -15946,40 +16074,40 @@ "soft_pairs": [ "issue:45672|issue:45674" ], - "prepared_review_unit_hash": "d9feab0e20b7003fa81e3b23b65224caf0e6b28bef1dcea373c3007cc225796f", + "prepared_review_unit_hash": "5f8153d558f6a927653737f9f3a7de72453ba19fdea0a5211284060859ea898e", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1669, - "estimated_input_tokens": 418, - "estimated_eval_tokens": 1092 + "serialized_chars": 1668, + "estimated_input_tokens": 417, + "estimated_eval_tokens": 1090 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1669, - "estimated_input_tokens": 418, - "estimated_eval_tokens": 1092 + "serialized_chars": 1668, + "estimated_input_tokens": 417, + "estimated_eval_tokens": 1090 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues both mention 4-bit inference, but they describe different failures in different code paths: one is a Gemma4/audio-module torch.finfo() uint8 TypeError, the other is a BitsAndBytesConfig skip_modules assertion regression. They should not be merged as duplicates.", - "confidence": 0.97, - "canonical_issue_reason": "Issue 45674 is the more general and reusable report: it describes a concrete BitsAndBytesConfig regression in 4-bit inference with a clear config-related failure mode.", + "summary": "These are two separate 4-bit inference bugs: one is a Gemma4/audio-module uint8 torch.finfo crash, the other is a BitsAndBytesConfig regression where llm_int8_skip_modules drops the default lm_head exclusion and triggers an AssertionError.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue: the reports describe different failure modes and different root causes/code paths.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45674 is the better cluster representative because it is broader and more directly tied to a specific inference/config bug, while 45672 appears to be a separate model/module-specific failure.", + "best_issue_reason": "If one report must be preferred as the broader/clearer issue, issue 45674 is more general and actionable; 45672 is model-specific and unrelated to the skip_modules regression.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45672", "right": "issue:45674", "accept": false, - "reason": "Different root causes and code paths. 45672 is a Gemma4/audio-module dtype error from torch.finfo() on uint8 weights; 45674 is a BitsAndBytesConfig skip_modules behavior regression causing an AssertionError in 4-bit inference." + "reason": "Reject: different underlying bugs and fixes. 45672 is a Gemma4/audio-module uint8 torch.finfo TypeError during NF4 inference, while 45674 is an AssertionError caused by llm_int8_skip_modules clearing the default lm_head exclusion." } ] }, @@ -16009,7 +16137,7 @@ "soft_pairs": [ "issue:45762|issue:45763" ], - "prepared_review_unit_hash": "6904d6aebe97eb5d480fbe3d64043843dc490e53cfa971aaae31e19d17282996", + "prepared_review_unit_hash": "f92ab211d7336bee30bffa8aa6b52122d78b160533114577dbda37ffd78cc7c8", "component_budget": { "node_count": 2, "item_count": 2, @@ -16031,18 +16159,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two closed issues are just personal-name/date titles with no body or technical context. They do not appear to describe the same bug or request, so they should not be clustered as duplicates.", - "confidence": 0.98, - "canonical_issue_reason": null, + "summary": "The two issues are not duplicates; they are separate closed items with different personal-name/date titles and no body, links, or shared context beyond superficial similarity.", + "confidence": 0.99, + "canonical_issue_reason": "issue:45762 is only a marginally better representative because it was created first, but neither issue is a meaningful canonical duplicate target.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "No strong best issue exists; both are low-information, unrelated-looking issues and should not be merged as duplicates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45762", "right": "issue:45763", "accept": false, - "reason": "The titles reference different people and different dates, with no shared technical subject, body text, or linked targets. This is not evidence of the same underlying issue." + "reason": "Different names and dates with no supporting body or references; superficial resemblance only, not the same underlying issue." } ] }, @@ -16072,40 +16200,40 @@ "soft_pairs": [ "issue:45764|issue:45765" ], - "prepared_review_unit_hash": "2598ed8620b33a14a9b912cb08976c4d6aa26479de9bfaf77a8b578ef5b33662", + "prepared_review_unit_hash": "3ee35c90b902a4631387ad8b0e2ecaedffff191938842c628fa80c70723e9311", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1480, - "estimated_input_tokens": 370, - "estimated_eval_tokens": 996 + "serialized_chars": 1481, + "estimated_input_tokens": 371, + "estimated_eval_tokens": 998 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1480, - "estimated_input_tokens": 370, - "estimated_eval_tokens": 996 + "serialized_chars": 1481, + "estimated_input_tokens": 371, + "estimated_eval_tokens": 998 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two closed issues with only person-name-and-date titles and no body content; they do not appear to describe the same underlying problem.", - "confidence": 0.97, + "summary": "These two closed issues are unrelated: they are just distinct name/date strings with no shared topic, code location, or discussion context. No duplicate relationship is evident.", + "confidence": 0.99, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Neither issue is a good canonical representative: both are sparse, unrelated-looking, and lack any problem description or linkage.", + "best_issue_reason": "Neither issue is a good canonical choice because there is no shared underlying bug or request to consolidate.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45764", "right": "issue:45765", "accept": false, - "reason": "Different names and dates with no body, links, or shared context; no evidence they describe the same issue." + "reason": "The titles refer to different people and different dates; there is no sign of the same underlying issue." } ] }, @@ -16164,154 +16292,154 @@ "pull_request:43823|pull_request:45795", "pull_request:43251|pull_request:43254" ], - "prepared_review_unit_hash": "9ec6aa791bc863f6282a046bb5ae3d4326a7a2a411b44bbbf949a184d658ba03", + "prepared_review_unit_hash": "8e20c9186238d166d1a988d06c7e9922cecb54000799f18c1d40480c07a66b59", "component_budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 19, - "serialized_chars": 35202, - "estimated_input_tokens": 8801, - "estimated_eval_tokens": 17858 + "serialized_chars": 35200, + "estimated_input_tokens": 8800, + "estimated_eval_tokens": 17856 }, "budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 19, - "serialized_chars": 35202, - "estimated_input_tokens": 8801, - "estimated_eval_tokens": 17858 + "serialized_chars": 35200, + "estimated_input_tokens": 8800, + "estimated_eval_tokens": 17856 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is mostly heterogeneous PRs that only share broad Transformers themes. The only high-confidence duplicate pair is the two `fixed_cross_entropy` patches (`43251` and `43254`), which both extend the same function in the same file for the same issue target.", - "confidence": 0.88, + "summary": "This cluster is mostly a set of unrelated PRs that happen to share broad themes like model additions, docs, or refactors. The only strong duplicate-like pair is the two `fixed_cross_entropy` changes, which target the same function and same loss kwargs fix.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "Pick `pull_request:43251` as the canonical PR for the one true duplicate pair: it has the explicit issue link, more review activity, and the same concrete `fixed_cross_entropy` API expansion as `43254`.", + "canonical_pr_reason": "PR 43251 is the better canonical representative: it directly fixes the issue-linked `fixed_cross_entropy` behavior and cleanly forwards the new cross-entropy kwargs. PR 43254 is the same underlying change with a slightly different API wording.", "best_issue_reason": null, - "best_pr_reason": "`pull_request:43251` is the best overall PR anchor because it is the clearest duplicate candidate in the cluster; the remaining PRs are unrelated enough that they should not be merged into the same canonical thread.", + "best_pr_reason": "PR 43251 is the best global representative because it is the clearest, issue-linked implementation of the only true duplicate pair in the cluster.", "soft_edge_verdicts": [ { "left": "pull_request:39895", "right": "pull_request:43424", "accept": false, - "reason": "Both are unrelated feature/test PRs; they only share vague model/runtime context." + "reason": "VideoPrism model addition and an executorch dynamic-shape test are unrelated changes; they do not fix the same bug or implement the same feature." }, { "left": "pull_request:41855", "right": "pull_request:43823", "accept": false, - "reason": "Tokenizer method additions vs a new model implementation; not the same change." + "reason": "Mistral tokenizer methods and adding MobileLLM are unrelated model-level changes." }, { "left": "pull_request:41855", "right": "pull_request:43424", "accept": false, - "reason": "Different subsystems and goals: tokenizer API gaps vs executorch export tests." + "reason": "Tokenizer API additions and executorch export tests are unrelated." }, { "left": "pull_request:42668", "right": "pull_request:43823", "accept": false, - "reason": "Processor robustness changes are unrelated to adding a new model." + "reason": "A processor-loading robustness fix is not the same underlying change as adding a new model implementation." }, { "left": "pull_request:43251", "right": "pull_request:43823", "accept": false, - "reason": "Cross-entropy kwargs fix vs adding MobileLLM; no shared bug/change." + "reason": "A cross-entropy kwargs fix and a MobileLLM model add are unrelated." }, { "left": "pull_request:43251", "right": "pull_request:44827", "accept": false, - "reason": "Same loss utility vs Mistral4 test/model work; too different to be duplicates." + "reason": "Both touch training/model code, but one fixes loss kwargs while the other updates Mistral4 tests and MoE behavior; not the same bug/change." }, { "left": "pull_request:43424", "right": "pull_request:44827", "accept": false, - "reason": "Executorch dynamic-shape test vs Mistral4 test/model changes; unrelated." + "reason": "Executorch export test coverage and Mistral4 test fixes are different test changes for different subsystems." }, { "left": "pull_request:43424", "right": "pull_request:43823", "accept": false, - "reason": "Different feature areas: exportability tests vs new model addition." + "reason": "Dynamic-shape executorch tests and MobileLLM model addition are unrelated." }, { "left": "pull_request:43823", "right": "pull_request:45655", "accept": false, - "reason": "MobileLLM implementation vs config-resolution fix; no concrete overlap." + "reason": "Adding MobileLLM and fixing `cls.config` resolution order are unrelated code paths." }, { "left": "pull_request:43823", "right": "pull_request:43995", "accept": false, - "reason": "Adding a model vs refactoring Falcon output collection; not the same code-path problem." + "reason": "New model addition and Falcon output-interface refactoring are not the same change." }, { "left": "pull_request:43995", "right": "pull_request:44827", "accept": false, - "reason": "Falcon refactor and Mistral4 test fixes are separate model-specific changes." + "reason": "Falcon standardized output refactoring and Mistral4 test fixes do not target the same underlying bug." }, { "left": "pull_request:43995", "right": "pull_request:45213", "accept": false, - "reason": "Falcon output-interface refactor vs model-creation skill scaffolding; unrelated." + "reason": "A model refactor and the model-creation skill/docs are unrelated." }, { "left": "pull_request:44827", "right": "pull_request:45170", "accept": false, - "reason": "Mistral4 test fix vs a layernorm renaming sweep; different bug/change." + "reason": "Mistral4 test fixes and a `layrnorm`\u2192`layernorm` rename pass are unrelated changes." }, { "left": "pull_request:44827", "right": "pull_request:45795", "accept": false, - "reason": "Mistral4 test/model fixes vs docs for audio/video processors; not the same change." + "reason": "Mistral4 tests and audio/video processor docs are unrelated." }, { "left": "pull_request:44827", "right": "pull_request:45655", "accept": false, - "reason": "Model-specific test fix vs class-annotation resolution fix; unrelated." + "reason": "Mistral4 test changes and `cls.config` resolution order fixes are different issues." }, { "left": "pull_request:44827", "right": "pull_request:45213", "accept": false, - "reason": "The shared `modeling_auto.py` touch is incidental; one is a model fix, the other is a docs/skill PR." + "reason": "Test fixes for Mistral4 are unrelated to the model-creation skill draft." }, { "left": "pull_request:43823", "right": "pull_request:45213", "accept": false, - "reason": "New model implementation vs a model-creation skill/docs PR; different artifacts and purpose." + "reason": "Adding MobileLLM and introducing the model-creation skill are unrelated." }, { "left": "pull_request:43823", "right": "pull_request:45795", "accept": false, - "reason": "Model addition vs documentation for processors; only broad topical overlap." + "reason": "MobileLLM model code and docs for audio/video processors are unrelated." }, { "left": "pull_request:43251", "right": "pull_request:43254", "accept": true, - "reason": "Both patch `fixed_cross_entropy` in the same file for the same issue target, adding support for the same kwargs; they are effectively the same bug fix and could be merged together." + "reason": "Both PRs change `src/transformers/loss/loss_utils.py` in the same way: they extend `fixed_cross_entropy` to forward supported `nn.functional.cross_entropy` kwargs for the same underlying loss behavior fix." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet. The only clearly supported duplicate candidate is `pull_request:43251`/`pull_request:43254`: they touch the same file and function, share the same issue target `43240`, and both add the same `fixed_cross_entropy` kwargs support. The other pairwise rejects are conservative and consistent with the diff previews." + "feedback": "Grounded overall. The cluster appears mostly unrelated, and the only clearly mergeable duplicate pair is PR 43251/43254: both touch `src/transformers/loss/loss_utils.py`, share issue target 43240, and implement the same `fixed_cross_entropy` kwargs-forwarding fix with minor API-shape differences. The other soft-edge verdicts stay conservative and do not overstate evidence." }, "evaluator_used": true, "retried": false, @@ -16338,7 +16466,7 @@ "soft_pairs": [ "pull_request:40546|pull_request:42781" ], - "prepared_review_unit_hash": "a14f67ba43916306d6a1e33ffa06b9ce46c1b305157c670bb6d90c43cf72aabd", + "prepared_review_unit_hash": "cb6cc1072a53783e76a994e92ced85a5c190039577b8126d4abc504f3319c677", "component_budget": { "node_count": 2, "item_count": 2, @@ -16360,18 +16488,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related VibeVoice PRs, but they are not duplicates: one introduces the base VibeVoice integration, while the other adds a realtime variant and an acoustic tokenizer, which is a larger follow-on feature.", - "confidence": 0.89, + "summary": "Two open PRs both touch VibeVoice, but they are not the same change: #40546 implements the base VibeVoice integration, while #42781 adds a separate realtime variant plus acoustic-tokenizer work. Shared filenames are mostly the common model scaffold, not evidence of the same concrete fix.", + "confidence": 0.92, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 40546 is the foundational VibeVoice implementation and the cleanest representative of the core model integration.", + "canonical_pr_reason": "#40546 is the better canonical PR because it is the broader, non-draft base VibeVoice implementation and appears to establish the core model package that later work extends.", "best_issue_reason": null, - "best_pr_reason": "PR 40546 is the most central standalone change; PR 42781 appears to be a dependent extension rather than the same change.", + "best_pr_reason": "#40546 is the best representative of the cluster since it covers the primary VibeVoice integration rather than a narrower realtime follow-up.", "soft_edge_verdicts": [ { "left": "pull_request:40546", "right": "pull_request:42781", "accept": false, - "reason": "They overlap on VibeVoice infrastructure and auto registrations, but 42781 adds realtime support plus an acoustic tokenizer, making it a distinct follow-on feature rather than the same concrete change." + "reason": "Related VibeVoice work, but different scope and code path: base model integration versus realtime/acoustic-tokenizer additions. They share scaffold files, not the same underlying change, so they should not be merged as duplicates." } ] }, @@ -16401,7 +16529,7 @@ "soft_pairs": [ "pull_request:40755|pull_request:41224" ], - "prepared_review_unit_hash": "19af67eca9bc99364f5c52e954f4e1e919c16b32d4d6d550be27c5c34ca9adbf", + "prepared_review_unit_hash": "c1c706f1f74c0978886eddd287a0d82ccfc49f26afe881c79cffafcf910ea7d2", "component_budget": { "node_count": 2, "item_count": 2, @@ -16423,18 +16551,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two pull requests are not duplicates: one adds TimesFM covariate forecasting, the other adds DINOv3 image-classification support. They touch different models, code paths, and feature sets, so they should stay separate.", - "confidence": 0.98, + "summary": "These two PRs are unrelated feature additions in different model families: TimesFM covariate forecasting vs DINOv3 image classification support. They do not describe the same bug or change, so the soft duplicate link should be rejected.", + "confidence": 0.99, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "PR #41224 is the more self-contained and concrete change, adding a specific DINOv3 image-classification path; however, it is not a duplicate of the TimesFM covariate PR.", "best_issue_reason": null, - "best_pr_reason": null, + "best_pr_reason": "PR #41224 is the best representative of the cluster because it is narrowly scoped and clearly tied to one model capability, whereas PR #40755 addresses a separate forecasting feature in a different subsystem.", "soft_edge_verdicts": [ { "left": "pull_request:40755", "right": "pull_request:41224", "accept": false, - "reason": "Different underlying changes for different model families (TimesFM forecasting with covariates vs DINOv3 image classification). Shared doc/test/modeling churn is incidental and they could not plausibly be merged into one PR." + "reason": "Different models, tasks, and code paths: TimesFM covariate forecasting vs DINOv3 image classification. They are not the same underlying change and should not be merged as duplicates." } ] }, @@ -16529,46 +16657,46 @@ "pull_request:41797|pull_request:45186", "pull_request:45186|pull_request:45630" ], - "prepared_review_unit_hash": "837fd7abbd1f0bbe53d661f1607a2c57728a6532f0403f9bf32b86bb733daa5b", + "prepared_review_unit_hash": "12af8159060eea929c5fcdf1fc4d8ec4a46cb67384cc3c8bd5b9e113e8496773", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 10429, - "estimated_input_tokens": 2608, - "estimated_eval_tokens": 5472 + "serialized_chars": 10428, + "estimated_input_tokens": 2607, + "estimated_eval_tokens": 5470 }, "budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 10429, - "estimated_input_tokens": 2608, - "estimated_eval_tokens": 5472 + "serialized_chars": 10428, + "estimated_input_tokens": 2607, + "estimated_eval_tokens": 5470 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are three separate model-addition PRs (DeepSeek OCR, Isaac, Kimi2-6). They share mostly boilerplate registration/docs paths, but they implement different models with different codebases and release scopes, so they are not duplicates.", - "confidence": 0.97, + "summary": "These are three unrelated model-addition PRs (DeepSeek OCR, Isaac, Kimi2-6). Their overlaps are mostly boilerplate registration/docs changes, not the same underlying bug or feature, so the soft similarity edges should be rejected.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "No canonical PR: the items introduce different models and do not represent duplicate fixes or the same concrete code change.", "best_issue_reason": null, - "best_pr_reason": null, + "best_pr_reason": "No single best PR for deduplication purposes; each PR targets a different model integration and should stand on its own.", "soft_edge_verdicts": [ { "left": "pull_request:41797", "right": "pull_request:45186", "accept": false, - "reason": "Different model integrations: DeepSeek OCR vs Isaac. Shared files are generic auto-registry/docs scaffolding, not the same concrete code-path change." + "reason": "Different model additions (DeepSeek OCR vs Isaac). Shared auto-registration/docs files are generic scaffolding, not evidence of the same fix." }, { "left": "pull_request:45186", "right": "pull_request:45630", "accept": false, - "reason": "Different model integrations: Isaac vs Kimi2-6. Overlap is limited to common Transformers wiring and documentation, so they should not be merged as one PR." + "reason": "Different model additions (Isaac vs Kimi2-6). The overlap is broad framework wiring, but the concrete changes and model code paths are unrelated." } ] }, @@ -16598,7 +16726,7 @@ "soft_pairs": [ "pull_request:42975|pull_request:42976" ], - "prepared_review_unit_hash": "0187cc80dc93d8fcfaf78a82c89ee7bcc92579834fe141963ea2a4369fdfefea", + "prepared_review_unit_hash": "e1b761349bcc15ce2d757565f7bcd60611c9b468b191b5efb180ea36f0922c64", "component_budget": { "node_count": 2, "item_count": 2, @@ -16620,18 +16748,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two workflow maintenance PRs that both update GitHub Actions versions, but they target different action families and different upgrade goals. They overlap in some workflow files, yet they are not the same underlying change.", - "confidence": 0.92, + "summary": "These are two separate GitHub Actions maintenance PRs: one upgrades core workflow actions for Node 24 compatibility, the other bumps docker/build-push-action for Docker image workflows. They overlap in subsystem but not in the underlying change or fix.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 42975 is the stronger canonical representative: it has a clear stated purpose (Node 24 compatibility) and a broad, coherent set of workflow action upgrades tied to that goal.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "PR 42975 is the best single PR in the cluster because its scope is more clearly defined and it represents the broader dependency-upgrade theme more cleanly than the docker-action bump.", + "best_pr_reason": "PR 42975 is the broader representative of the cluster theme because it updates multiple workflow actions across many CI jobs for Node 24 compatibility; PR 42976 is a narrower Docker action version bump.", "soft_edge_verdicts": [ { "left": "pull_request:42975", "right": "pull_request:42976", "accept": false, - "reason": "Do not merge: 42975 upgrades core GitHub workflow actions like checkout/cache/upload-artifact for Node 24 compatibility, while 42976 updates docker/build-push-action to v6 in Docker image workflows. They are related maintenance, but not the same concrete bug/change and would not naturally be a single PR." + "reason": "Different underlying changes: 42975 updates checkout/cache/upload-artifact and related workflow actions for Node 24 compatibility, while 42976 only bumps docker/build-push-action from v5 to v6 in Docker image workflows. Same broad area, but not the same concrete fix." } ] }, @@ -16674,29 +16802,29 @@ "pull_request:44601|pull_request:45838", "pull_request:45490|pull_request:45838" ], - "prepared_review_unit_hash": "17159348b8513b917aee729703768caab0ac52f4c1d5cd1bd1d2777e6bc56997", + "prepared_review_unit_hash": "8ca98e9bb2b8e0745e9789b25f981163a31083dd71daaf1a8cb04b33782cd19f", "component_budget": { "node_count": 7, "item_count": 7, "soft_pair_count": 9, - "serialized_chars": 23511, - "estimated_input_tokens": 5878, - "estimated_eval_tokens": 12012 + "serialized_chars": 23513, + "estimated_input_tokens": 5879, + "estimated_eval_tokens": 12014 }, "budget": { "node_count": 7, "item_count": 7, "soft_pair_count": 9, - "serialized_chars": 23511, - "estimated_input_tokens": 5878, - "estimated_eval_tokens": 12012 + "serialized_chars": 23513, + "estimated_input_tokens": 5879, + "estimated_eval_tokens": 12014 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All items are separate PRs for different model additions or an unrelated distributed-loading feature; the shared file overlap is mostly docs/auto-registration boilerplate. No soft edge looks like the same underlying change, so every candidate is rejected.", + "summary": "This cluster is not a duplicate set: it mixes unrelated model-addition PRs (ViT NEPA, Molmo, Molmo2, LFM2 audio, CTSM, PP-OCRv6) plus a distributed pipeline-parallel infrastructure PR. The shared files are mostly boilerplate registration/docs, not the same concrete bug or change.", "confidence": 0.98, "canonical_issue_reason": null, "canonical_pr_reason": null, @@ -16707,55 +16835,55 @@ "left": "pull_request:42978", "right": "pull_request:43448", "accept": false, - "reason": "Different model families and code paths: ViT NEPA vs Molmo. Shared docs/auto files are just integration boilerplate, not the same fix." + "reason": "Different model families and different feature scopes: ViT NEPA vs Molmo. Shared docs/auto-registration files are boilerplate, not evidence of the same change." }, { "left": "pull_request:42978", "right": "pull_request:43451", "accept": false, - "reason": "Different additions: ViT NEPA vs Molmo2. These are separate model implementations, not one mergeable change." + "reason": "ViT NEPA and Molmo2 are separate model additions with distinct architectures and code paths; no same underlying bug or mergeable combined PR." }, { "left": "pull_request:42978", "right": "pull_request:45490", "accept": false, - "reason": "ViT NEPA model support and CTSM model support are unrelated changes." + "reason": "ViT NEPA and CTSM are unrelated model additions in different domains; overlap is only in generic transformer model registry/docs." }, { "left": "pull_request:43448", "right": "pull_request:43451", "accept": false, - "reason": "Molmo and Molmo2 are related by name but are distinct model families/versions with separate configs and implementations." + "reason": "Molmo and Molmo2 are related by name and domain, but they are different model families and implementations, not the same concrete change." }, { "left": "pull_request:43448", "right": "pull_request:45490", "accept": false, - "reason": "Molmo support and CTSM support are different model additions, not the same underlying bug or feature." + "reason": "Molmo vs CTSM is unrelated; the shared files are broad auto/model registration updates and do not indicate the same fix." }, { "left": "pull_request:43451", "right": "pull_request:45490", "accept": false, - "reason": "Molmo2 support and CTSM support are unrelated model additions." + "reason": "Molmo2 and CTSM are different model additions with no shared bug or code path beyond generic framework plumbing." }, { "left": "pull_request:43973", "right": "pull_request:45838", "accept": false, - "reason": "Lfm2 Audio adds an audio model; PP-OCRv6 adds text recognition models. Different modalities and code paths." + "reason": "LFM2 audio is an audio ASR model, while PP-OCRv6 text recognition is an OCR vision/text model; these are entirely different changes." }, { "left": "pull_request:44601", "right": "pull_request:45838", "accept": false, - "reason": "Native pipeline-parallel loading support is a distributed infra change, not the same change as adding PP-OCRv6 model support." + "reason": "Pipeline-parallel infrastructure support is orthogonal to PP-OCRv6 model support; they do not fix the same concrete code-path problem." }, { "left": "pull_request:45490", "right": "pull_request:45838", "accept": false, - "reason": "CTSM and PP-OCRv6 are different model families; shared docs/model-registry edits do not indicate a duplicate." + "reason": "CTSM time-series model support and PP-OCRv6 text recognition support are unrelated model additions, not a duplicate or combined PR candidate." } ] }, @@ -16788,7 +16916,7 @@ "pull_request:43291|pull_request:44815", "pull_request:43665|pull_request:44815" ], - "prepared_review_unit_hash": "b0d7537286cd96dff952dedcb94100a40dc921bead251642b27dd9a884fc8844", + "prepared_review_unit_hash": "d954f0ef8b843168b9a9434f7154d50cb201987e19db618df067240dd19f6ad7", "component_budget": { "node_count": 3, "item_count": 3, @@ -16810,30 +16938,30 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The three PRs are not duplicates of one another: one is a Whisper tokenizer/test fix, one is a CLIP/ViT test-only change with no real bug context, and one is an FP8 dequant/model-loading fix. The only thing they share is vague similarity in being fixes.", - "confidence": 0.9, + "summary": "All three pull requests address different changes: a Whisper tokenizer/test fix, unrelated test edits for CLIP/ViT, and a broader dequantization/fine-grained FP8 loading fix. None look like duplicates of the same underlying bug or patch.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 44815 is the most concrete and substantial fix, with a clear code-path change in FP8 dequant/model loading and multiple related files; the others are unrelated test-only or Whisper-specific changes.", + "canonical_pr_reason": "PR 44815 is the strongest canonical PR because it is the most substantive and coherent fix, spanning the dequantization/loading path with a focused change set across implementation and tests.", "best_issue_reason": null, - "best_pr_reason": "PR 44815 is the strongest representative because it addresses a specific runtime bug path rather than only adjusting tests or making a narrow tokenizer tweak.", + "best_pr_reason": "PR 44815 is the best standalone PR: it targets a concrete code-path problem in model loading / FP8 dequantization and has a consistent scope. The other PRs are narrower test-centric changes.", "soft_edge_verdicts": [ { "left": "pull_request:43291", "right": "pull_request:43665", "accept": false, - "reason": "Whisper tokenizer decoding changes are unrelated to the CLIP/ViT test edits; they do not fix the same underlying bug or change." + "reason": "Different areas and intents: Whisper tokenization/test cleanup versus unrelated CLIP/ViT test edits. They do not fix the same bug or change the same code path." }, { "left": "pull_request:43291", "right": "pull_request:44815", "accept": false, - "reason": "Different subsystems and different problems: Whisper decode logic vs FP8 dequant/loading logic." + "reason": "One is a Whisper-specific tokenizer/test fix; the other is a model-loading/FP8 dequantization fix. No shared underlying defect." }, { "left": "pull_request:43665", "right": "pull_request:44815", "accept": false, - "reason": "Test-only CLIP/ViT edits are not the same concrete code-path change as the FP8 dequant fix, so they should not be merged as duplicates." + "reason": "Although both touch tests, 43665 is just arbitrary failing-test assertions, while 44815 fixes dequantization/model-loading behavior. Not the same change and not mergeable as one PR." } ] }, @@ -16863,12 +16991,12 @@ "soft_pairs": [ "pull_request:43488|pull_request:43532" ], - "prepared_review_unit_hash": "9054f002811e467113ecbc1389d35f95f3fb57f3ebe92015fb400a2a7810f6b0", + "prepared_review_unit_hash": "fb99b8e13ee84a1402ed33c35fb5a0f8b44af7b635925c8f5dfd883aea8bf684", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4563, + "serialized_chars": 4562, "estimated_input_tokens": 1141, "estimated_eval_tokens": 2538 }, @@ -16876,7 +17004,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4563, + "serialized_chars": 4562, "estimated_input_tokens": 1141, "estimated_eval_tokens": 2538 }, @@ -16885,18 +17013,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two pull requests are not duplicates: one is a small setup/version/test-bot formatting change, while the other rewires CI/workflow behavior and touches several unrelated files. The only overlap is a shared Vit test file, which is not enough to treat them as the same change.", - "confidence": 0.95, + "summary": "These two PRs are not duplicates: one is a narrow repo-bot/setup/test tweak, the other is a broader CI/workflow/testing utility change. They do not look like the same underlying fix or mergeable into one PR.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 43488 is the better canonical PR because it is the narrower, more self-contained change and does not mix in unrelated workflow edits.", + "canonical_pr_reason": "PR 43532 is the more substantial and representative change in this cluster; PR 43488 is a small, unrelated experimental setup/version tweak.", "best_issue_reason": null, - "best_pr_reason": "PR 43488 is the best representative of this cluster if one must be chosen, since it is the smaller and more focused PR.", + "best_pr_reason": "PR 43532 is the best representative PR because it has the broader, more substantive code and workflow changes, whereas 43488 is a tiny targeted setup/test adjustment.", "soft_edge_verdicts": [ { "left": "pull_request:43488", "right": "pull_request:43532", "accept": false, - "reason": "Reject: these PRs address different code paths and intents. 43488 is about setup/version/test-bot plumbing, while 43532 changes GitHub Actions workflows and notification/testing utilities; the shared Vit test file is incidental and not enough to merge them as one PR." + "reason": "Different change sets and purposes: 43488 adjusts packaging/versioning and a Vit test for repo-bot formatting, while 43532 rewires CI/workflow and testing utilities. They do not fix the same concrete bug or belong in one merged PR." } ] }, @@ -16926,12 +17054,12 @@ "soft_pairs": [ "pull_request:43612|pull_request:43613" ], - "prepared_review_unit_hash": "595500950fc0b4a09ebf8a5d66a7c3bc825eb8b96f109a4233680d6620d61e24", + "prepared_review_unit_hash": "b211b49304801b3f3678ba231aafe9b5aead179ae042955d4d4d4986adf21403", "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 }, @@ -16939,7 +17067,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6082, + "serialized_chars": 6081, "estimated_input_tokens": 1521, "estimated_eval_tokens": 3298 }, @@ -16948,18 +17076,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are closely related pipeline-addition PRs, but they implement different features: concept segmentation vs visual segmentation. The shared scaffolding and doc/test files are not enough to treat them as duplicates.", - "confidence": 0.97, + "summary": "These are two separate pipeline additions for different segmentation tasks: promptable concept segmentation vs promptable visual segmentation. They share some framework/doc scaffolding, but the underlying feature scope and model backends differ, so they are not duplicates.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "No canonical duplicate PR here: PR 43612 and PR 43613 implement different segmentation pipelines with different backends and test suites.", "best_issue_reason": null, - "best_pr_reason": null, + "best_pr_reason": "If one must be used as a representative anchor, PR 43613 is the broader pipeline addition, but it is still not a duplicate of PR 43612.", "soft_edge_verdicts": [ { "left": "pull_request:43612", "right": "pull_request:43613", "accept": false, - "reason": "Different concrete changes and code paths: one adds Promptable Concept Segmentation, the other adds Promptable Visual Segmentation. They share documentation and registration boilerplate, but the targets, pipelines, and model/test coverage differ enough that they would not plausibly merge into one PR." + "reason": "Different feature targets and code paths: concept segmentation with SAM3 vs visual segmentation with SAM/SAM2/EdgeTam/tracker. Shared plumbing alone is not enough to merge them as one change." } ] }, @@ -16989,12 +17117,12 @@ "soft_pairs": [ "pull_request:43651|pull_request:43663" ], - "prepared_review_unit_hash": "933c0ba11bbc4ff084c5107a7d5c66d8d1e9312c88055e185d5f060de1c8970e", + "prepared_review_unit_hash": "d19c5e7f6d947b65fd0e4d27710b75936b33d2f56a91f5ac5445998f16fa16c1", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4292, + "serialized_chars": 4289, "estimated_input_tokens": 1073, "estimated_eval_tokens": 2402 }, @@ -17002,7 +17130,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4292, + "serialized_chars": 4289, "estimated_input_tokens": 1073, "estimated_eval_tokens": 2402 }, @@ -17011,18 +17139,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two PRs both touch `trainer.py` and are trainer extensibility refactors, but they address different code paths: gradient-accumulation loss scaling vs. signature-column filtering. They should not be merged as duplicates.", - "confidence": 0.97, + "summary": "The two PRs both touch `trainer.py`, but they solve different problems: one adds a hook for gradient-accumulation loss scaling, the other refactors signature-column filtering for PEFT/custom trainers. They are not duplicates.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 43651 is the more concrete, self-contained trainer behavior change and is the better representative of this small cluster.", + "canonical_pr_reason": "No canonical PR: the changes address separate trainer code paths and are not mergeable as one fix.", "best_issue_reason": null, - "best_pr_reason": "PR 43651 is the more concrete, self-contained trainer behavior change and is the better representative of this small cluster.", + "best_pr_reason": "Neither PR is a good cluster representative for duplicate triage because they are unrelated enhancements in the same file, not the same underlying change.", "soft_edge_verdicts": [ { "left": "pull_request:43651", "right": "pull_request:43663", "accept": false, - "reason": "Reject: these are different trainer code paths. 43651 changes loss scaling during `training_step` via `_loss_is_scaled_for_ga`; 43663 adds `_get_signature_columns` for column filtering. Shared file/subsystem is not enough to treat them as the same change." + "reason": "Different features and different code paths in `trainer.py` (`training_step` loss scaling vs signature-column filtering); they should not be treated as the same PR." } ] }, @@ -17054,7 +17182,7 @@ "pull_request:43656|pull_request:43842", "pull_request:43836|pull_request:43842" ], - "prepared_review_unit_hash": "56107392d5aefe4a230690ae72c7c2be0ef50da456b018a1610392f8c374c3f0", + "prepared_review_unit_hash": "59435728ce49d97e28a589aa75a7a52bf78e82f6c0457920ba7a99ae060deece", "component_budget": { "node_count": 3, "item_count": 3, @@ -17076,30 +17204,30 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All three PRs target the same TypeAdapter NameError area in `src/transformers/cli/serve.py`. The most focused fix is the later, smaller PR that explicitly handles the pydantic-not-installed case.", - "confidence": 0.7, + "summary": "All three PRs appear to address the same underlying import-time TypeAdapter annotation NameError in `src/transformers/cli/serve.py`, with different levels of scope. The most focused fix is 43842.", + "confidence": 0.81, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #43842 is the most focused and precise fix: it directly addresses the TypeAdapter annotation import problem in the CLI serve path with a small, scoped change and an explicit pydantic-not-installed rationale.", + "canonical_pr_reason": "PR 43842 is the clearest, most minimal fix: it directly prevents runtime evaluation of `TypeAdapter` by moving the import under `TYPE_CHECKING` and quoting the annotation where used.", "best_issue_reason": null, - "best_pr_reason": "PR #43842 is the best representative because it is the narrowest patch, most clearly tied to the runtime NameError, and least entangled with broader annotation refactors.", + "best_pr_reason": "PR 43842 is the best representative of the cluster because it precisely targets the reported NameError with the smallest blast radius and the clearest description.", "soft_edge_verdicts": [ { "left": "pull_request:43656", "right": "pull_request:43842", "accept": true, - "reason": "Both PRs fix the same TypeAdapter NameError in the CLI serve code path, differing mainly in annotation-handling mechanics; they could plausibly be merged into one patch." + "reason": "Both target the same `TypeAdapter` NameError in the CLI serve module; 43656 and 43842 are different implementations of the same import-time annotation problem." }, { "left": "pull_request:43836", "right": "pull_request:43842", "accept": true, - "reason": "Both target the same serve.py TypeAdapter annotation failure and appear to be alternative/overlapping fixes for the same runtime issue, so they look like duplicates." + "reason": "Both are fixing the same concrete `TypeAdapter` annotation/runtime-evaluation bug in `src/transformers/cli/serve.py`, and their changes could plausibly be combined into one PR." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet: all three PRs touch `src/transformers/cli/serve.py` and are about the TypeAdapter annotation NameError class of bugs. The choice of PR #43842 as the narrowest and most explicit fix is supported by its smaller diff and pydantic-specific title. The soft-edge acceptances are conservative enough here because the changes appear compatible and mergeable into a single bugfix patch, though #43656 is somewhat broader than the others." + "feedback": "The summary stays within the packet: 43842 is clearly the most minimal TypeAdapter NameError fix, and the other PRs are plausibly related changes in the same file. The soft-edge judgments are conservative enough, especially since 43836 and 43842 share the same explicit issue target and 43656 is at least title-aligned with the same bug class." }, "evaluator_used": true, "retried": false, @@ -17288,20 +17416,20 @@ "pull_request:44044|pull_request:44129", "pull_request:44044|pull_request:44154" ], - "prepared_review_unit_hash": "cc5177d6d64827bf3fc06c0949062dc630078fa6f8f0c5e7510aa346e59a57aa", + "prepared_review_unit_hash": "d8bd8cec9a2e933dd09b671a715d8bb528789a3b98f071340df4ea6157283ccc", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 78, - "serialized_chars": 84900, - "estimated_input_tokens": 21225, - "estimated_eval_tokens": 42706 + "serialized_chars": 84902, + "estimated_input_tokens": 21226, + "estimated_eval_tokens": 42708 }, "budget": { "node_count": 27, "item_count": 27, "soft_pair_count": 72, - "serialized_chars": 75097, + "serialized_chars": 75099, "estimated_input_tokens": 18775, "estimated_eval_tokens": 37806 }, @@ -17310,252 +17438,210 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of model-specific refactors for standardized output tracing across unrelated architectures. Most pairs only share the same tracking issue or broad refactor theme; only a few same-model/file pairs look like true duplicates.", - "confidence": 0.93, + "summary": "This cluster is mostly a set of model-specific refactors for standardized output tracing. Most pairs share only the umbrella issue target and should not be merged; the true duplicates are the GPT-J/CodeGen family and the GPT-Neo pair.", + "confidence": 0.86, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 44722 is the best representative: it covers the same GPT-J/CodeGen output-tracing migration as the other GPT-J-focused PRs, is later, and appears to subsume the narrower variants.", + "canonical_pr_reason": "PR 44722 is the most complete representative of the duplicated GPT-J/CodeGen refactor family: it explicitly targets the standardized output-tracing change and appears to subsume earlier GPT-J-only revisions.", "best_issue_reason": null, - "best_pr_reason": "PR 44722 is the strongest overall fit because it is the most complete instance of the GPT-J/CodeGen standardized output-tracing change in this set.", + "best_pr_reason": "PR 44722 is the best cluster representative because it is the newest consolidated version of the concrete output-tracing refactor and covers the same code paths as the earlier GPT-J/CodeGen PRs.", "soft_edge_verdicts": [ { "left": "pull_request:43996", "right": "pull_request:44085", "accept": false, - "reason": "Same broad output-tracing theme, but different model families and files; not the same concrete change." + "reason": "Different model families (CVT/FNet vs GPT-J); same umbrella issue target is not enough." }, { "left": "pull_request:43996", "right": "pull_request:44044", "accept": false, - "reason": "Different model families (Cvt/FNet vs DeBERTa); only the refactor theme is shared." + "reason": "CVT/FNet and DeBERTa are unrelated implementations; only the tracking issue is shared." }, { "left": "pull_request:44066", "right": "pull_request:44085", "accept": true, - "reason": "Both touch GPT-J output tracing in the same code path; one PR is just broader (also CodeGen), so they could be merged." + "reason": "Both are GPT-J output-tracing refactors on the same code path; 44085 looks like an overlapping/earlier revision of the same change." }, { "left": "pull_request:44007", "right": "pull_request:44072", "accept": false, - "reason": "Different models and different files; same issue target is too broad to call them duplicates." + "reason": "RegNet and EfficientNet are separate model refactors; shared issue target alone is insufficient." }, { "left": "pull_request:44072", "right": "pull_request:44722", "accept": false, - "reason": "EfficientNet vs GPT-J/CodeGen are unrelated concrete changes." + "reason": "EfficientNet vs GPT-J/CodeGen are different model changes, not the same underlying bug." }, { "left": "pull_request:44066", "right": "pull_request:44072", "accept": false, - "reason": "Different model families; no shared concrete bug or code path." + "reason": "GPT-J/CodeGen refactor and EfficientNet refactor are unrelated changes." }, { "left": "pull_request:44013", "right": "pull_request:44044", "accept": false, - "reason": "MobileNetV2 and DeBERTa output-tracing refactors are unrelated implementations." + "reason": "MobileNetV2 and DeBERTa are distinct model implementations with different refactors." }, { "left": "pull_request:44066", "right": "pull_request:44086", "accept": false, - "reason": "GPT-J/CodeGen versus MGP-STR; same style of refactor, not the same artifact." + "reason": "GPT-J/CodeGen output tracing and MGP-STR output tracing are different model-specific changes." }, { "left": "pull_request:44066", "right": "pull_request:44071", "accept": false, - "reason": "GPT-J/CodeGen and MPT are different code paths." + "reason": "GPT-J/CodeGen and MPT are separate code paths; same broad refactor theme is not enough." }, { "left": "pull_request:44018", "right": "pull_request:44068", "accept": true, - "reason": "Both are GPT-Neo output-tracing refactors in the same model family and file, so they look like the same change." + "reason": "Both PRs refactor GPT-Neo output tracing in the same file and appear to address the same concrete change." }, { "left": "pull_request:44066", "right": "pull_request:44068", "accept": false, - "reason": "GPT-J/CodeGen versus GPT-Neo; shared refactor pattern only." + "reason": "GPT-J/CodeGen and GPT-Neo are different models; they do not fix the same concrete path." }, { "left": "pull_request:44018", "right": "pull_request:44066", "accept": false, - "reason": "GPT-Neo and GPT-J are different model implementations." + "reason": "GPT-Neo and GPT-J/CodeGen are separate model refactors, not mergeable duplicates." }, { "left": "pull_request:44019", "right": "pull_request:44722", "accept": false, - "reason": "ResNet output tracing and GPT-J/CodeGen output tracing are unrelated code paths." + "reason": "ResNet output tracing is unrelated to GPT-J/CodeGen output tracing." }, { "left": "pull_request:44086", "right": "pull_request:44722", "accept": false, - "reason": "MGP-STR and GPT-J/CodeGen are unrelated model families." + "reason": "MGP-STR and GPT-J/CodeGen are different model changes." }, { "left": "pull_request:44019", "right": "pull_request:44085", "accept": false, - "reason": "ResNet and GPT-J are different implementations; only the decorator migration theme overlaps." + "reason": "ResNet and GPT-J refactors are unrelated." }, { "left": "pull_request:44018", "right": "pull_request:44019", "accept": false, - "reason": "GPT-Neo versus ResNet is not the same underlying change." + "reason": "GPT-Neo and ResNet are different models; no shared concrete change." }, { "left": "pull_request:44019", "right": "pull_request:44071", "accept": false, - "reason": "ResNet and MPT are different concrete changes." + "reason": "ResNet and MPT output tracing are separate refactors." }, { "left": "pull_request:44068", "right": "pull_request:44116", "accept": false, - "reason": "GPT-Neo and Flaubert are unrelated model code paths." + "reason": "GPT-Neo and Flaubert are unrelated model changes." }, { "left": "pull_request:44068", "right": "pull_request:44722", "accept": false, - "reason": "GPT-Neo versus GPT-J/CodeGen; same refactor style, different artifact." + "reason": "GPT-Neo refactor is not the same underlying change as GPT-J/CodeGen." }, { "left": "pull_request:44071", "right": "pull_request:44722", "accept": false, - "reason": "MPT and GPT-J/CodeGen are unrelated implementations." + "reason": "MPT and GPT-J/CodeGen are different model-specific output-tracing changes." }, { - "left": "pull_request:44059", - "right": "pull_request:44086", + "left": "pull_request:44018", + "right": "pull_request:44116", "accept": false, - "reason": "GPT-2 and MGP-STR are different model families." + "reason": "GPT-Neo and Flaubert are distinct changes." }, { - "left": "pull_request:44085", + "left": "pull_request:44018", "right": "pull_request:44086", "accept": false, - "reason": "GPT-J and MGP-STR are unrelated concrete changes." - }, - { - "left": "pull_request:44098", - "right": "pull_request:44101", - "accept": false, - "reason": "ViLT and Flaubert/XLM are different code paths." + "reason": "GPT-Neo and MGP-STR are different code paths." }, { "left": "pull_request:44018", - "right": "pull_request:44059", - "accept": false, - "reason": "GPT-Neo versus GPT-2; same general output-tracing refactor, but not the same bug or change." - }, - { - "left": "pull_request:44073", - "right": "pull_request:44161", - "accept": false, - "reason": "VisualBert and LongT5 are unrelated models." - }, - { - "left": "pull_request:44101", - "right": "pull_request:44161", - "accept": false, - "reason": "Flaubert/XLM and LongT5 are different implementations." - }, - { - "left": "pull_request:43997", - "right": "pull_request:44017", - "accept": false, - "reason": "RegNet and Segformer are unrelated model changes." - }, - { - "left": "pull_request:43997", - "right": "pull_request:44056", + "right": "pull_request:44085", "accept": false, - "reason": "RegNet and MPNet do not share the same concrete change." + "reason": "GPT-Neo vs GPT-J are different model refactors." }, { - "left": "pull_request:43997", - "right": "pull_request:44161", + "left": "pull_request:44068", + "right": "pull_request:44086", "accept": false, - "reason": "RegNet and LongT5 are different code paths." + "reason": "GPT-Neo and MGP-STR are unrelated model changes." }, { "left": "pull_request:44071", - "right": "pull_request:44076", + "right": "pull_request:44086", "accept": false, - "reason": "MPT and ImageGPT are unrelated model implementations." + "reason": "MPT and MGP-STR are separate output-tracing refactors." }, { - "left": "pull_request:44074", - "right": "pull_request:44101", + "left": "pull_request:44059", + "right": "pull_request:44086", "accept": false, - "reason": "TextNet and Flaubert/XLM are different artifacts." + "reason": "GPT2 and MGP-STR are different model-specific changes." }, { - "left": "pull_request:43997", - "right": "pull_request:44018", + "left": "pull_request:44085", + "right": "pull_request:44086", "accept": false, - "reason": "RegNet and GPT-Neo are unrelated despite the shared output-tracing theme." + "reason": "GPT-J and MGP-STR are unrelated changes." }, { - "left": "pull_request:44056", - "right": "pull_request:44161", + "left": "pull_request:44059", + "right": "pull_request:44722", "accept": false, - "reason": "MPNet and LongT5 are different model families." + "reason": "GPT2 output tracing is not the same concrete change as GPT-J/CodeGen." }, { - "left": "pull_request:43997", - "right": "pull_request:44071", + "left": "pull_request:44098", + "right": "pull_request:44101", "accept": false, - "reason": "RegNet and MPT are unrelated code paths." + "reason": "ViLT and Flaubert are separate model refactors." }, { - "left": "pull_request:44010", + "left": "pull_request:44018", "right": "pull_request:44071", "accept": false, - "reason": "SqueezeBert and MPT are unrelated model implementations." - }, - { - "left": "pull_request:44010", - "right": "pull_request:44073", - "accept": false, - "reason": "SqueezeBert and VisualBert are different code paths." + "reason": "GPT-Neo and MPT are different models and code paths." }, { "left": "pull_request:44017", "right": "pull_request:44722", "accept": false, - "reason": "Segformer and GPT-J/CodeGen are unrelated concrete changes." + "reason": "Segformer and GPT-J/CodeGen are unrelated output-tracing refactors." }, { "left": "pull_request:44017", "right": "pull_request:44161", "accept": false, - "reason": "Segformer and LongT5 are different model families." + "reason": "Segformer and LongT5 are different model changes." }, { "left": "pull_request:44017", "right": "pull_request:44073", "accept": false, - "reason": "Segformer and VisualBert are unrelated implementations." - }, - { - "left": "pull_request:44068", - "right": "pull_request:44086", - "accept": false, - "reason": "GPT-Neo and MGP-STR are unrelated model code paths." + "reason": "Segformer and VisualBert are unrelated model refactors." }, { "left": "pull_request:44068", @@ -17564,100 +17650,148 @@ "reason": "GPT-Neo and MPT are different models." }, { - "left": "pull_request:44059", - "right": "pull_request:44722", + "left": "pull_request:44071", + "right": "pull_request:44076", "accept": false, - "reason": "GPT-2 and GPT-J/CodeGen are distinct model implementations." + "reason": "MPT and ImageGPT are separate output-tracing changes." }, { - "left": "pull_request:44098", + "left": "pull_request:44074", "right": "pull_request:44101", "accept": false, - "reason": "ViLT versus Flaubert/XLM is not the same underlying change." + "reason": "TextNet and Flaubert are unrelated model refactors." }, { - "left": "pull_request:44010", - "right": "pull_request:44018", + "left": "pull_request:43997", + "right": "pull_request:44017", "accept": false, - "reason": "SqueezeBert and GPT-Neo are different model families." + "reason": "RegNet and Segformer are different models; shared umbrella issue is not enough." }, { - "left": "pull_request:44071", - "right": "pull_request:44129", + "left": "pull_request:43997", + "right": "pull_request:44056", "accept": false, - "reason": "MPT and SpeechT5 are unrelated code paths." + "reason": "RegNet and MPNet are distinct refactors." }, { - "left": "pull_request:43996", - "right": "pull_request:44017", + "left": "pull_request:43997", + "right": "pull_request:44161", "accept": false, - "reason": "FNet/CVT and Segformer are unrelated model implementations." + "reason": "RegNet and LongT5 are unrelated model changes." }, { - "left": "pull_request:43996", + "left": "pull_request:43997", + "right": "pull_request:44074", + "accept": false, + "reason": "RegNet and TextNet are different output-tracing implementations." + }, + { + "left": "pull_request:43997", "right": "pull_request:44129", "accept": false, - "reason": "FNet/CVT and SpeechT5 are different concrete changes." + "reason": "RegNet and SpeechT5 are separate model refactors." }, { "left": "pull_request:44010", - "right": "pull_request:44044", + "right": "pull_request:44071", + "accept": false, + "reason": "SqueezeBert and MPT are different code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44073", + "accept": false, + "reason": "SqueezeBert and VisualBert are unrelated model changes." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44018", "accept": false, - "reason": "SqueezeBert and DeBERTa are different model families." + "reason": "SqueezeBert and GPT-Neo are different model refactors." }, { "left": "pull_request:44044", "right": "pull_request:44074", "accept": false, - "reason": "DeBERTa and TextNet are unrelated changes." + "reason": "DeBERTa and TextNet are separate output-tracing changes." }, { "left": "pull_request:44066", "right": "pull_request:44722", "accept": true, - "reason": "Same GPT-J/CodeGen output-tracing refactor; 44722 looks like a broader follow-up/subsuming PR rather than a different change." + "reason": "Same GPT-J/CodeGen standardized output-tracing refactor family; 44722 looks like the consolidated version of the same concrete change." }, { "left": "pull_request:44085", "right": "pull_request:44722", "accept": true, - "reason": "Both are GPT-J output-tracing refactors on the same file; 44722 just adds the CodeGen companion changes." + "reason": "Both target GPT-J output tracing on the same code path, and 44722 appears to be the consolidated follow-up." }, { "left": "pull_request:44085", "right": "pull_request:44129", "accept": false, - "reason": "GPT-J and SpeechT5 are unrelated model implementations." + "reason": "GPT-J and SpeechT5 are unrelated model refactors." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44007", + "accept": false, + "reason": "CVT/FNet and RegNet are different model changes." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44129", + "accept": false, + "reason": "CVT/FNet and SpeechT5 are unrelated refactors." }, { "left": "pull_request:44044", "right": "pull_request:44129", "accept": false, - "reason": "DeBERTa and SpeechT5 do not share the same concrete bug or change." + "reason": "DeBERTa and SpeechT5 are different model-specific changes." }, { "left": "pull_request:44044", "right": "pull_request:44154", "accept": false, - "reason": "DeBERTa and Vits are unrelated code paths." + "reason": "DeBERTa and VITS are unrelated output-tracing refactors." }, { - "left": "pull_request:44018", - "right": "pull_request:44068", - "accept": true, - "reason": "Both are GPT-Neo output-tracing migrations in the same model family and file, so they look like the same underlying change." + "left": "pull_request:44098", + "right": "pull_request:44017", + "accept": false, + "reason": "ViLT and Segformer are different models." + }, + { + "left": "pull_request:44098", + "right": "pull_request:44073", + "accept": false, + "reason": "ViLT and VisualBert are distinct changes." }, { "left": "pull_request:44101", "right": "pull_request:44116", - "accept": true, - "reason": "Both touch Flaubert output tracing in the same file; one is a broader Flaubert/XLM refactor, the other is a Flaubert-only WIP variant." + "accept": false, + "reason": "Flaubert and VITS are unrelated model refactors." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44161", + "accept": false, + "reason": "MPNet and LongT5 are separate output-tracing changes." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44018", + "accept": false, + "reason": "MPNet and GPT-Neo are different model implementations." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet: it correctly distinguishes broad, model-specific output-tracing refactors from the few same-model/file overlaps. The accepted soft edges are conservative and each pair looks mergeable as a single bugfix PR, especially the GPT-J/CodeGen cluster and the GPT-Neo and Flaubert subset pairs. No pair is overstated as a duplicate beyond what the diffs support." + "feedback": "The summary is grounded in the packet and stays conservative about the soft edges. The accepted pairs all look mergeable as overlapping revisions of the same model-specific output-tracing bugfix, especially the GPT-J/CodeGen family and the GPT-Neo pair. The rejected pairs mostly rely only on the shared umbrella issue and correctly avoid overstating duplication across different model implementations." }, "evaluator_used": true, "retried": false, @@ -17695,14 +17829,14 @@ "pull_request:43997|pull_request:44010", "pull_request:44010|pull_request:44129" ], - "prepared_review_unit_hash": "15ccb1a032239bfd2ef594a76c77f7abb2ed6e901b21f47eb036a0c6d5de0566", + "prepared_review_unit_hash": "039519c8a1d555417675dd12f030731906cab5c5e261950b23c524ec8b6eba9e", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 78, - "serialized_chars": 84900, - "estimated_input_tokens": 21225, - "estimated_eval_tokens": 42706 + "serialized_chars": 84902, + "estimated_input_tokens": 21226, + "estimated_eval_tokens": 42708 }, "budget": { "node_count": 8, @@ -17717,48 +17851,48 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These PRs are thematically similar standardized-output/capture refactors, but they target different model implementations and code paths, so they are not duplicates.", - "confidence": 0.96, + "summary": "This cluster is a set of model-specific refactors to standardize output tracing/capture_outputs across different Transformers architectures. They share a common theme, but they are not the same concrete code change or bug fix.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "No single canonical PR: each item is a separate model-specific refactor (RegNet, SqueezeBert, FocalNet, MPNet, VisualBert, TextNet, ImageGPT, SpeechT5).", + "canonical_pr_reason": "PR 44073 is the clearest representative of the cluster: it explicitly frames the change as output tracing standardization, uses the new decorators, and its scope is easy to understand from the title and diff.", "best_issue_reason": null, - "best_pr_reason": "44129 is the clearest representative because it is explicitly issue-linked and directly about standardized output capture, but it still only applies to SpeechT5.", + "best_pr_reason": "PR 44073 is the best cluster representative because it is a well-scoped, clearly titled example of the broader standardized output-capture refactor and is easier to interpret than the more vague or typo-prone alternatives.", "soft_edge_verdicts": [ { "left": "pull_request:44024", "right": "pull_request:44076", "accept": false, - "reason": "Same broad output-tracing theme, but FocalNet and ImageGPT are different model code paths and not mergeable as one fix." + "reason": "Both are output-tracing refactors, but they touch different model implementations (FocalNet vs ImageGPT) and are separate architecture-specific changes, not one mergeable code-path fix." }, { "left": "pull_request:44073", "right": "pull_request:44074", "accept": false, - "reason": "Both are output-tracing refactors, but they touch different models (VisualBert vs TextNet) with separate implementation details." + "reason": "These are related standardization refactors, but they apply to different models (VisualBert vs TextNet) with different forward paths and test impact; too broad to treat as the same PR." }, { "left": "pull_request:44056", "right": "pull_request:44073", "accept": false, - "reason": "MPNet and VisualBert are unrelated model implementations; this is a shared refactor pattern, not the same underlying change." + "reason": "Both modernize output capture in distinct models, but the affected code paths are unrelated model-specific implementations, so they do not look like one underlying change." }, { "left": "pull_request:44056", "right": "pull_request:44074", "accept": false, - "reason": "Different models and different forward/output plumbing; too broad to be one duplicate PR." + "reason": "MPNet and TextNet are separate model refactors; the similarity is only the shared migration pattern, not the same concrete bug or patch." }, { "left": "pull_request:43997", "right": "pull_request:44010", "accept": false, - "reason": "RegNet and SqueezeBert are distinct modules; similar decorator migration does not make them the same change." + "reason": "RegNet and SqueezeBert are different architectures and the diffs are separate refactors toward standardized output handling, not the same fix." }, { "left": "pull_request:44010", "right": "pull_request:44129", "accept": false, - "reason": "SqueezeBert and SpeechT5 have separate output-tracing implementations and cannot plausibly be merged into one concrete PR." + "reason": "They both relate to output tracing, but SpeechT5 is a different subsystem and includes distinct signature/behavior changes; this is not a single shared change." } ] }, @@ -17838,274 +17972,274 @@ "pull_request:44028|pull_request:44030", "pull_request:44029|pull_request:44030" ], - "prepared_review_unit_hash": "994079c53d2504cfa7b4e32abde074a12e31211b86ed72b3790eae349b417368", + "prepared_review_unit_hash": "450f4f3c6c37f0f2b5b662dd094b6405e54cdd1a39da60f8fb1bbbab3fb0eb1b", "component_budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 40, - "serialized_chars": 40170, - "estimated_input_tokens": 10043, - "estimated_eval_tokens": 20342 + "serialized_chars": 40159, + "estimated_input_tokens": 10040, + "estimated_eval_tokens": 20336 }, "budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 40, - "serialized_chars": 40170, - "estimated_input_tokens": 10043, - "estimated_eval_tokens": 20342 + "serialized_chars": 40159, + "estimated_input_tokens": 10040, + "estimated_eval_tokens": 20336 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All items are model-specific refactors for output tracing/can_return_tuple, but they target different model implementations and code paths, so they are not duplicates of one underlying bug/change.", - "confidence": 0.93, + "summary": "These PRs are all variations of the same refactor theme, but they target different model implementations and code paths, so they are not duplicates of one another. The cluster should be treated as related-but-separate model-specific changes, not one mergeable fix.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "No issue target here; PR 44025 is the best representative because it is a clear, multi-file example of the shared output-tracing refactor pattern.", + "canonical_pr_reason": "PR 44003 is the best representative because it is the broadest coherent change in the set, touching both falcon_mamba and mamba with the new output-capture plumbing. It best reflects the underlying refactor pattern without being just a tiny one-model tweak.", "best_issue_reason": null, - "best_pr_reason": "PR 44025 is the strongest cluster representative: it applies the tracing refactor across two closely related depth-anything files and matches the dominant theme of the set.", + "best_pr_reason": "44003 is the strongest representative PR in this cluster: it spans two closely related model files and introduces the output tracing changes directly, making it more substantive than the single-model refactors.", "soft_edge_verdicts": [ { "left": "pull_request:43998", "right": "pull_request:44002", "accept": false, - "reason": "Different model families; same refactor theme, but not the same concrete change or code path." + "reason": "Same high-level refactor style, but timm_backbone and upernet are different models with different forward paths and outputs; not the same concrete change." }, { "left": "pull_request:43998", "right": "pull_request:44000", "accept": false, - "reason": "Different models and implementations; only the high-level output-tracing pattern overlaps." + "reason": "Different model families and different forward signatures; only the tracing refactor theme is shared." }, { "left": "pull_request:43998", "right": "pull_request:44001", "accept": false, - "reason": "Separate model code paths; not a mergeable duplicate pair." + "reason": "UnivNet is a separate audio model with a distinct code path, so this is not the same underlying change as timm_backbone." }, { "left": "pull_request:43998", "right": "pull_request:44003", "accept": false, - "reason": "Both are tracing refactors, but for different architectures and modules." + "reason": "timm_backbone and mamba/falcon_mamba are unrelated implementations; the shared wording is just a refactor pattern." }, { "left": "pull_request:43998", "right": "pull_request:44029", "accept": false, - "reason": "Unrelated model code paths; shared refactor motif is too broad." + "reason": "Different model and different output-capturing context; not mergeable as one PR." }, { "left": "pull_request:43999", "right": "pull_request:44001", "accept": false, - "reason": "Different model families and forward paths; not the same underlying fix." + "reason": "MobileNetV1 and UnivNet are unrelated models; the overlap is only that both refactor return/output handling." }, { "left": "pull_request:43999", "right": "pull_request:44002", "accept": false, - "reason": "Only the output-tracing refactor theme is shared." + "reason": "MobileNetV1 and UperNet have different forward logic and outputs; this is not the same bug or change." }, { "left": "pull_request:43999", "right": "pull_request:44025", "accept": false, - "reason": "Different model implementations; no shared concrete bug or change." + "reason": "MobileNetV1 and Depth Anything are separate vision models with distinct tracing changes." }, { "left": "pull_request:43999", "right": "pull_request:44026", "accept": false, - "reason": "Separate encoder-decoder vs vision model code paths." + "reason": "VisionEncoderDecoder and MobileNetV1 are unrelated code paths; similar decorator/signature cleanup is not enough to merge." }, { "left": "pull_request:43999", "right": "pull_request:44027", "accept": false, - "reason": "Distinct models; same broad refactor pattern is insufficient." + "reason": "SpeechEncoderDecoder is a different model wrapper and does not fix the same concrete issue as MobileNetV1." }, { "left": "pull_request:44000", "right": "pull_request:44001", "accept": false, - "reason": "Different multimodal/audio-vision model code paths; not duplicates." + "reason": "Vision-text dual encoder and UnivNet are completely different model classes; same refactor motif only." }, { "left": "pull_request:44000", "right": "pull_request:44002", "accept": false, - "reason": "Different subsystems and forward implementations." + "reason": "Different models and different forward contracts; not the same underlying fix." }, { "left": "pull_request:44000", "right": "pull_request:44003", "accept": false, - "reason": "Shared tracing refactor theme only; concrete changes differ." + "reason": "Vision-text dual encoder vs mamba/falcon_mamba are unrelated code paths; shared output-tracing language is superficial." }, { "left": "pull_request:44000", "right": "pull_request:44025", "accept": false, - "reason": "Different model families; cannot be merged as one PR." + "reason": "These are different model families with different outputs and tracing behavior, so they are not duplicates." }, { "left": "pull_request:44001", "right": "pull_request:44002", "accept": false, - "reason": "Same broad goal, but unrelated model code paths." + "reason": "UnivNet and UperNet are unrelated implementations; the PRs are only similar in style." }, { "left": "pull_request:44001", "right": "pull_request:44003", "accept": false, - "reason": "Not the same concrete bug/change." + "reason": "Different model families and different changed files; no evidence of a shared underlying bug." }, { "left": "pull_request:44001", "right": "pull_request:44025", "accept": false, - "reason": "Different model implementations and refactor scope." + "reason": "Depth Anything and UnivNet are separate domains and code paths; not one mergeable fix." }, { "left": "pull_request:44002", "right": "pull_request:44029", "accept": false, - "reason": "Different model families; only theme-level similarity." + "reason": "UperNet and RWKV are unrelated model architectures, so these are not the same change." }, { "left": "pull_request:44002", "right": "pull_request:44003", "accept": false, - "reason": "Separate code paths and architectures." + "reason": "UperNet and mamba/falcon_mamba do not fix the same concrete issue; only the refactor theme matches." }, { "left": "pull_request:44003", "right": "pull_request:44028", "accept": false, - "reason": "Different models; same output-tracing motif is too generic." + "reason": "FalconMamba/Mamba and SuperPoint are unrelated code paths; the shared output-tracing modernization is too broad to merge." }, { "left": "pull_request:44003", "right": "pull_request:44004", "accept": false, - "reason": "Different model architectures and unrelated implementation details." + "reason": "Mamba/falcon_mamba and CodeGen are different model implementations with different tracing behavior." }, { "left": "pull_request:43998", "right": "pull_request:44004", "accept": false, - "reason": "Not the same code path; codegen and timm_backbone are unrelated." + "reason": "TimmBackbone and CodeGen are unrelated modules; similar signature cleanup does not make them duplicates." }, { "left": "pull_request:44001", "right": "pull_request:44004", "accept": false, - "reason": "Separate model families and forward logic." + "reason": "UnivNet and CodeGen are different models and different code paths; not the same concrete fix." }, { "left": "pull_request:44002", "right": "pull_request:44004", "accept": false, - "reason": "Only broad refactor similarity, not a shared bug." + "reason": "UperNet and CodeGen are unrelated implementations; the common theme is only output-tracing refactoring." }, { "left": "pull_request:44000", "right": "pull_request:44004", "accept": false, - "reason": "Different modules and model behavior; not mergeable as one fix." + "reason": "Vision-text dual encoder and CodeGen are unrelated; no shared bug or change beyond generic refactor style." }, { "left": "pull_request:44025", "right": "pull_request:44026", "accept": false, - "reason": "Different model code paths; same refactor pattern only." + "reason": "Depth Anything and VisionEncoderDecoder are different model wrappers and forward paths; similar decorator cleanup is not enough." }, { "left": "pull_request:44025", "right": "pull_request:44027", "accept": false, - "reason": "Distinct encoder-decoder vs depth-estimation code paths." + "reason": "Depth Anything and SpeechEncoderDecoder are separate models with different code paths, so not the same underlying change." }, { "left": "pull_request:44025", "right": "pull_request:44028", "accept": false, - "reason": "Different models and changes; not a duplicate." + "reason": "Depth Anything and SuperPoint are different vision models; these PRs are not duplicates." }, { "left": "pull_request:44025", "right": "pull_request:44029", "accept": false, - "reason": "Unrelated model implementations despite similar tracing refactor." + "reason": "Depth Anything and RWKV address different architectures and output handling." }, { "left": "pull_request:44025", "right": "pull_request:44030", "accept": false, - "reason": "Different concrete model logic and filenames." + "reason": "Depth Anything and DPR are unrelated models; the overlap is only that both refactor output tracing." }, { "left": "pull_request:44026", "right": "pull_request:44029", "accept": false, - "reason": "Different model families; no shared underlying bug." + "reason": "VisionEncoderDecoder and RWKV are unrelated code paths and do not appear to fix the same issue." }, { "left": "pull_request:44026", "right": "pull_request:44027", "accept": false, - "reason": "Separate encoder-decoder integrations; not the same fix." + "reason": "VisionEncoderDecoder and SpeechEncoderDecoder are analogous wrappers, but they are still different model paths and changes." }, { "left": "pull_request:44026", "right": "pull_request:44028", "accept": false, - "reason": "Different code paths and implementation scopes." + "reason": "VisionEncoderDecoder and SuperPoint are separate models; no evidence they are the same concrete change." }, { "left": "pull_request:44026", "right": "pull_request:44030", "accept": false, - "reason": "Different models and refactor details." + "reason": "VisionEncoderDecoder and DPR are different architectures and different forward logic." }, { "left": "pull_request:44027", "right": "pull_request:44028", "accept": false, - "reason": "Different model families and forward implementations." + "reason": "SpeechEncoderDecoder and SuperPoint are unrelated modules; only the refactor theme is shared." }, { "left": "pull_request:44027", "right": "pull_request:44029", "accept": false, - "reason": "No common concrete bug/change beyond a generic tracing refactor." + "reason": "SpeechEncoderDecoder and RWKV are different model families with different output contracts." }, { "left": "pull_request:44027", "right": "pull_request:44030", "accept": false, - "reason": "Different model code paths; not a duplicate pair." + "reason": "SpeechEncoderDecoder and DPR do not share the same underlying bug or code path." }, { "left": "pull_request:44028", "right": "pull_request:44029", "accept": false, - "reason": "Different model architectures and refactor scope." + "reason": "SuperPoint and RWKV are unrelated architectures; not a duplicate fix." }, { "left": "pull_request:44028", "right": "pull_request:44030", "accept": false, - "reason": "Separate models; same theme is too broad." + "reason": "SuperPoint and DPR are different models with different output handling, so they should stay separate." }, { "left": "pull_request:44029", "right": "pull_request:44030", "accept": false, - "reason": "Different model families; not the same code-path fix." + "reason": "RWKV and DPR are unrelated model implementations; similar output-tracing refactors do not make them the same change." } ] }, @@ -18138,52 +18272,52 @@ "pull_request:44662|pull_request:45075", "pull_request:45075|pull_request:45401" ], - "prepared_review_unit_hash": "0ccbb217cac8ab612a3519fe73cd6a1230358c93a616db0a3090677e1f847793", + "prepared_review_unit_hash": "6398e298064659d1e33aa50717ce4f7a9e0ed5ddc16fad98ba731e1af032e828", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 10913, - "estimated_input_tokens": 2729, - "estimated_eval_tokens": 5714 + "serialized_chars": 10912, + "estimated_input_tokens": 2728, + "estimated_eval_tokens": 5712 }, "budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 10913, - "estimated_input_tokens": 2729, - "estimated_eval_tokens": 5714 + "serialized_chars": 10912, + "estimated_input_tokens": 2728, + "estimated_eval_tokens": 5712 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All three pull requests are separate model additions for different architectures (PenguinVL, DeepSeek-OCR-2, Voxtral TTS). They share only common transformer plumbing and docs/registry updates, so they should not be deduplicated or merged together.", + "summary": "These are three separate model-addition PRs (PenguinVL, DeepSeek-OCR-2, Voxtral TTS). They only overlap on common Transformers integration scaffolding and docs, not on the same underlying bug or change, 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": null, + "best_pr_reason": "PR 45075 is the strongest representative of the cluster: it is non-draft, has the highest review activity, and contains a full model integration. The others are for different models, and 45401 is still draft.", "soft_edge_verdicts": [ { "left": "pull_request:44662", "right": "pull_request:45401", "accept": false, - "reason": "Different model families and different implementation surfaces; overlap is limited to shared auto-registry/docs files, which is boilerplate for model additions." + "reason": "Different model implementations (PenguinVL vs Voxtral TTS); shared auto/docs files are generic integration boilerplate, not the same fix/change." }, { "left": "pull_request:44662", "right": "pull_request:45075", "accept": false, - "reason": "These add unrelated models with different configs, processors, and tests. Similarity comes from standard integration files, not the same code-path fix." + "reason": "Different model additions (PenguinVL vs DeepSeek-OCR-2) with distinct configs, processors, and model code; overlap is only in common registration/docs plumbing." }, { "left": "pull_request:45075", "right": "pull_request:45401", "accept": false, - "reason": "Separate model integrations for different tasks and architectures; no evidence they address the same bug or could be merged into one PR." + "reason": "Different concrete changes for different models (DeepSeek-OCR-2 vs Voxtral TTS); they cannot plausibly be merged as one PR for the same code-path problem." } ] }, @@ -18213,7 +18347,7 @@ "soft_pairs": [ "pull_request:44979|pull_request:45363" ], - "prepared_review_unit_hash": "76478190ab9114795c279ff84a84c835e9dae977c85947fda0846ee26af91bea", + "prepared_review_unit_hash": "d49ea25632609450da2ec1e8c905aa7d6a1e7d852ad355c1b1be99fe3fc422ac", "component_budget": { "node_count": 2, "item_count": 2, @@ -18235,18 +18369,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two pull requests are related around module/kernel fusion, but they are not duplicates: PR #44979 introduces a generic Module Fusion API and test scaffolding, while PR #45363 wires n-to-1 kernel fusion through KernelConfig and adds integration/example code. They target different layers of the stack and would not plausibly be merged as one PR.", - "confidence": 0.93, + "summary": "These PRs are related in the broad fusion/kernel-optimization area, but they are not the same change: one introduces a generic Module Fusion API, while the other wires up n-to-1 kernel fusion through KernelConfig and integrations. They should not be treated as duplicates.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45363 is the more concrete end-user feature request: it adds KernelConfig-based fusion integration across modeling utils, hub kernels, and an example. PR #44979 is a lower-level foundation for module fusion, so it is better treated as supporting infrastructure rather than the main duplicate candidate.", + "canonical_pr_reason": "PR #44979 is the more foundational and reusable change: it introduces the generic Module Fusion API and test coverage for the abstraction itself. PR #45363 is a narrower, downstream kernel-fusion configuration feature built on top of that idea.", "best_issue_reason": null, - "best_pr_reason": "PR #45363 is the best representative of the cluster because it has the broader product-facing change and ties together the fusion workflow via KernelConfig. PR #44979 is a separate foundational API addition and not the same change.", + "best_pr_reason": "PR #44979 is the best cluster anchor because it defines the core fusion abstraction rather than one specific application of it, making it the more representative and durable reference point.", "soft_edge_verdicts": [ { "left": "pull_request:44979", "right": "pull_request:45363", "accept": false, - "reason": "Related but distinct: #44979 adds a generic module-fusion API, while #45363 implements KernelConfig-driven n-to-1 kernel fusion integration. Different code paths and feature scope." + "reason": "Same broad fusion theme, but different concrete implementations and code paths: generic module-fusion API vs KernelConfig-driven n-to-1 kernel fusion. Not close enough to be duplicates." } ] }, @@ -18276,7 +18410,7 @@ "soft_pairs": [ "pull_request:45086|pull_request:45317" ], - "prepared_review_unit_hash": "f59fe99ed9ace6d99dc0fe1a10642346f6ff2f16cdad37826e14e432e8736448", + "prepared_review_unit_hash": "ddb9a23444c59c77aa4b77173cc343e32886b6bc0f2103fa316931467bc8f721", "component_budget": { "node_count": 2, "item_count": 2, @@ -18298,18 +18432,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs address the same AttributeError in `_patch_mistral_regex` by switching the pre-tokenizer access from `backend_tokenizer.pre_tokenizer` to `tokenizer.pre_tokenizer`; the second PR is a fuller version with tests.", - "confidence": 0.98, + "summary": "Both PRs target the same Mistral regex patch AttributeError in `src/transformers/tokenization_utils_tokenizers.py` against issue #45081; 45317 is the more complete version because it adds tests in addition to the code fix.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "Prefer pull_request:45317 as the canonical PR because it includes the same code fix plus test coverage, making it the more complete representative.", + "canonical_pr_reason": "PR 45317 is the stronger canonical PR because it fixes the same concrete bug as 45086 and also includes test coverage, making it the more complete representative of the change.", "best_issue_reason": null, - "best_pr_reason": "pull_request:45317 is the best PR candidate: same fix as 45086, same explicit issue target, and it adds tests in addition to the code change.", + "best_pr_reason": "PR 45317 is the best PR candidate: same fix, same target issue, and it adds verification via tests rather than only the code change.", "soft_edge_verdicts": [ { "left": "pull_request:45086", "right": "pull_request:45317", "accept": true, - "reason": "Same underlying bug fix in the same code path and file; the diffs are equivalent except that 45317 adds tests, so they are duplicate PRs for the same change." + "reason": "Same underlying AttributeError fix in `_patch_mistral_regex`, same shared issue target, and the diffs are the same code-path change; 45317 simply adds tests." } ] }, @@ -18339,7 +18473,7 @@ "soft_pairs": [ "pull_request:45115|pull_request:45569" ], - "prepared_review_unit_hash": "9ed13eb7a329ad22d352fc37d67e95d76281be1e6042b115fb1f1214671da029", + "prepared_review_unit_hash": "c6e21db8fe39d78ba480a5b3274c9a008cc414857da0c0e8ab80d9a07a7da12d", "component_budget": { "node_count": 2, "item_count": 2, @@ -18361,18 +18495,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two PRs overlap on NemotronH internals, but they are not the same change: one is a narrow checkpoint conversion-mapping update, the other is a broader model split/refactor with docs/tests. Reject as duplicates.", + "summary": "Both PRs are NemotronH-related, but they address different changes: 45115 is a narrower conversion-mapping/refactor for NemotronH compatibility, while 45569 is the broader model-family split introducing nemotron_h_dense and nemotron_h_sparse plus docs, auto-mappings, and tests. They overlap in shared model files, but not in the same underlying change.", "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45569 is the better cluster anchor because it is the broader end-to-end NemotronH refactor and package split, while #45115 is a much narrower conversion-mapping patch.", + "canonical_pr_reason": "45569 is the better cluster representative because it is the more complete, user-facing NemotronH change set; 45115 is a narrower conversion-mapping/refactor and not the same change.", "best_issue_reason": null, - "best_pr_reason": "PR #45569 is the strongest representative of the cluster: it introduces the concrete model-architecture and auto-mapping changes plus documentation/test updates. #45115 only adjusts weight renaming for conversion.", + "best_pr_reason": "45569 best represents the cluster because it introduces the actual NemotronH dense/sparse model split and updates the surrounding API/docs; 45115 is a smaller compatibility refactor.", "soft_edge_verdicts": [ { "left": "pull_request:45115", "right": "pull_request:45569", "accept": false, - "reason": "They share NemotronH model files, but the concrete changes differ: #45115 adds checkpoint weight-renaming mappings, while #45569 refactors NemotronH into dense/sparse variants with docs and tests. Not the same underlying bug/change." + "reason": "Related theme and shared files, but different concrete changes: conversion-mapping refactor vs architecture/model split with docs and tests. Not the same underlying PR." } ] }, @@ -18402,7 +18536,7 @@ "soft_pairs": [ "pull_request:45332|pull_request:45333" ], - "prepared_review_unit_hash": "bbf35807c8ccb5eff50a8b911e9ec1049861fa7f1d7080771f2336179d68c405", + "prepared_review_unit_hash": "33e974db553f53d158706a9ad7ed5057c48eb64e2274d9acddd01712edd5ba64", "component_budget": { "node_count": 2, "item_count": 2, @@ -18424,18 +18558,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two open PRs form a related stack around heterogeneous-layer support, but they are not duplicates: one adds per-layer config plumbing, the other adds broader model/runtime support on top of it.", - "confidence": 0.88, + "summary": "These are related but distinct PRs: one adds heterogeneous per-layer config infrastructure, the other adds broader heterogeneous modeling/runtime support that appears to build on that config work. They should not be deduplicated.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45332 is the better representative because it is the broader, more complete change: it adds heterogeneous model support and already incorporates the downstream cache/masking/modeling updates that depend on per-layer config.", + "canonical_pr_reason": "PR #45333 is the more foundational, narrowly scoped change: it introduces heterogeneous per-layer configuration support and the shared config helpers/classes. It is a cleaner canonical representative than the broader modeling PR.", "best_issue_reason": null, - "best_pr_reason": "PR #45332 is the strongest standalone candidate for the cluster since it covers the end-to-end feature rather than only the configuration layer.", + "best_pr_reason": "PR #45333 is the best single PR to represent the cluster because it defines the underlying config mechanism that the other PR seems to consume. PR #45332 is broader and depends on the same heterogeneity groundwork, so it is less suitable as the canonical representative.", "soft_edge_verdicts": [ { "left": "pull_request:45332", "right": "pull_request:45333", "accept": false, - "reason": "These are adjacent but distinct changes: #45333 introduces per-layer configuration support, while #45332 builds heterogeneous model/runtime support on top of that. They do not look like the same underlying bug/change and should not be merged as duplicates." + "reason": "Different changes: #45333 adds per-layer heterogeneous config support and helpers, while #45332 adds heterogeneous model/runtime behavior across cache, masking, modeling, and GPT-OSS integration. They are related but not the same bug/change and would not be merged as one PR." } ] }, @@ -18465,12 +18599,12 @@ "soft_pairs": [ "pull_request:45415|pull_request:45425" ], - "prepared_review_unit_hash": "39c4a94d4f1c742fe9cd709cb14dfe187f60448388a6002b8239d4115564f743", + "prepared_review_unit_hash": "dd0c2945a0f477996e6128c1281138bef86a8c9902bc23dc094f012fc87c60d9", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5514, + "serialized_chars": 5513, "estimated_input_tokens": 1379, "estimated_eval_tokens": 3014 }, @@ -18478,7 +18612,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5514, + "serialized_chars": 5513, "estimated_input_tokens": 1379, "estimated_eval_tokens": 3014 }, @@ -18487,18 +18621,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests are typing-related, but they target different changes: PR 45415 is a broad type-checking sweep across many `src/transformers/*` files, while PR 45425 adds typing helpers/protocols and updates a smaller set of core typing-sensitive code paths. They should not be merged as duplicates.", + "summary": "Both PRs are type-checking/typing-oriented, but they address different scopes: #45415 is a broad repo-wide typing cleanup and mypy/config adjustment, while #45425 adds specific typing protocols/aliases for modeling_utils/peft-related code paths. They are related in theme, not the same change.", "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45425 is the better canonical representative for the cluster because it introduces reusable typing infrastructure in `_typing.py` and updates core consumers, making it the more specific and structurally central typing change.", + "canonical_pr_reason": "#45415 is the broader, more representative typing initiative across many files and configuration, so it is the better canonical PR for this cluster.", "best_issue_reason": null, - "best_pr_reason": "PR 45425 is the strongest standalone PR in this cluster: it has a focused code change, clear typing abstractions, and a tighter concrete scope than the broad type-checking cleanup in PR 45415.", + "best_pr_reason": "#45415 is the best overall representative for the cluster because it spans the widest set of typing changes and subsumes the general theme better than the narrower follow-up typing support in #45425.", "soft_edge_verdicts": [ { "left": "pull_request:45415", "right": "pull_request:45425", "accept": false, - "reason": "Both are about typing, but they solve different problems: 45415 is a wide type-checking cleanup across many files, while 45425 adds new typing protocols/utilities and touches different implementation paths. They are not the same underlying change and would not plausibly merge into one PR." + "reason": "Do not merge: both are typing-related, but #45415 is a broad type-checking pass over many modules, while #45425 adds focused protocol/alias support for specific modeling_utils and PEFT integrations. They do not appear to fix the same concrete bug or be mergeable as one PR." } ] }, @@ -18533,12 +18667,12 @@ "pull_request:45817|pull_request:45843", "pull_request:45836|pull_request:45843" ], - "prepared_review_unit_hash": "733589c62f11dea8532215bb5fbd699dfa0ea9a7400985a8ba7e25f37e78f8e7", + "prepared_review_unit_hash": "f22783679188155488146f1254baace45e22287936a170f5ab4f9025f9e1ca71", "component_budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 4, - "serialized_chars": 9782, + "serialized_chars": 9783, "estimated_input_tokens": 2446, "estimated_eval_tokens": 5148 }, @@ -18546,7 +18680,7 @@ "node_count": 4, "item_count": 4, "soft_pair_count": 4, - "serialized_chars": 9782, + "serialized_chars": 9783, "estimated_input_tokens": 2446, "estimated_eval_tokens": 5148 }, @@ -18555,36 +18689,36 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are all related to model-parallel / device-placement work, but they target different models and different concrete failures, so they are not duplicates of one another.", - "confidence": 0.86, + "summary": "These are separate pull requests for different models and different fixes: two model-parallel device-placement bugs, one additional model-parallel bug in Gemma4, and one test skip for MiniCPM-V 4.6. They are related only at a broad theme level, not the same underlying change.", + "confidence": 0.94, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45487 is the best representative because it fixes a concrete device-placement bug in real model code, spans multiple affected implementations, and is more substantial than the test-only or model-specific alternatives.", + "canonical_pr_reason": "PR 45487 is the broadest concrete code fix in the set: it patches device placement in AltCLIP/ChineseCLIP model-parallel paths and includes targeted test updates. The others are different model-specific fixes or a test-only skip, so they are not better canonical representatives.", "best_issue_reason": null, - "best_pr_reason": "PR #45487 is the strongest overall PR candidate for this cluster: it addresses an actual model-parallel code-path defect rather than only adjusting tests, and its fix is narrowly concrete and mergeable.", + "best_pr_reason": "PR 45487 is the best representative PR because it contains an actual bug fix affecting multiple code paths, not just a test adjustment, and it is more substantial than the other single-model patches.", "soft_edge_verdicts": [ { "left": "pull_request:45487", "right": "pull_request:45843", "accept": false, - "reason": "Both are model-parallel fixes, but they affect different architectures (AltCLIP/ChineseCLIP vs DeimV2) and different code paths, so they are not the same change." + "reason": "Both are model-parallel fixes, but for different architectures (AltCLIP/ChineseCLIP vs Deimv2) and different code paths. They are not the same concrete bug and would not plausibly merge into one PR." }, { "left": "pull_request:45487", "right": "pull_request:45836", "accept": false, - "reason": "One is a code fix for model parallel behavior; the other only skips a failing test for MiniCPM-V. Different underlying problem and scope." + "reason": "45487 fixes runtime device-placement logic; 45836 only skips a failing test for MiniCPM-V 4.6. This is not the same underlying change." }, { "left": "pull_request:45817", "right": "pull_request:45843", "accept": false, - "reason": "These fix separate model-specific parallelism issues in Gemma4 and DeimV2, with different device-placement and split-module changes." + "reason": "Gemma4 and Deimv2 have separate model-parallel device-placement fixes in different forward paths. Same broad theme, but distinct bugs." }, { "left": "pull_request:45836", "right": "pull_request:45843", "accept": false, - "reason": "A test skip for an incompatible quantized-cache case is not the same underlying bug as DeimV2's model-parallel code fix." + "reason": "One is a test skip, the other is a code fix for Deimv2 model parallelism. They do not describe the same bug or change." } ] }, @@ -18614,7 +18748,7 @@ "soft_pairs": [ "pull_request:45548|pull_request:45662" ], - "prepared_review_unit_hash": "a194869684d137cf8ea47fe06f5a287bf1e606331a89cb194462f3b42baa57e6", + "prepared_review_unit_hash": "89c24f112bb465c101cbeb673bbc7a94f6aa4cf6de993a72e9894adc74d6cf96", "component_budget": { "node_count": 2, "item_count": 2, @@ -18636,18 +18770,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs are about expert-parallel loading, but they fix different bugs on different code paths: one adds config/has_ep support for accelerate + ZeRO-3, the other prevents expert parameters from being overwritten during FSDP2/rank-0 broadcast loading.", + "summary": "These are related EP/distributed-loading PRs, but they fix different failure modes: one is about preserving/reading distributed_config during load, the other about expert tensors being overwritten during rank-0 broadcast in FSDP2. They should not be treated as duplicates.", "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45662 is the better representative of the cluster\u2019s core EP loading regression because it fixes the actual parameter-sharding/overwrite path; PR 45548 is a separate accelerate + ZeRO-3 configuration compatibility patch.", + "canonical_pr_reason": "No single canonical PR as a duplicate target; 45662 is the broader fix for the rank-0 broadcast overwrite bug, while 45548 addresses a separate accelerate/ZeRO-3 loading path.", "best_issue_reason": null, - "best_pr_reason": "PR 45662 is the strongest single PR here: it addresses a concrete loading bug in the expert-parameter path and has the broader, more direct fix. PR 45548 solves a different EP-related integration issue.", + "best_pr_reason": "45548 is a narrower loader/config serialization fix, but 45662 is the more substantive and multi-file change for the concrete expert-overwrite issue. If forced to pick one representative, 45662 is the better umbrella PR for EP distributed-loading problems.", "soft_edge_verdicts": [ { "left": "pull_request:45548", "right": "pull_request:45662", "accept": false, - "reason": "Related EP-loading theme, but not the same underlying bug/change: 45548 serializes distributed_config and adds has_ep for accelerate/ZeRO-3, while 45662 fixes expert params being overwritten during sharded/FSDP2 loading. Different code paths and mergeable as separate PRs." + "reason": "Different root causes and code paths: 45548 adds distributed_config serialization and a has_ep helper for EP/ZeRO-3 loading via accelerate, while 45662 fixes experts being clobbered by rank-0 broadcast with FSDP2. Shared EP context is too broad to call them the same change." } ] }, @@ -18677,7 +18811,7 @@ "soft_pairs": [ "pull_request:45639|pull_request:45645" ], - "prepared_review_unit_hash": "0cac7dff735e01029d94bda28a6ce94bcb1d8f4eb9cbf9790eb6c2db075837d9", + "prepared_review_unit_hash": "2ef108318803ee0f07caedd4f46ecaaf576d254b8b8f777b1b656b3ec58411d4", "component_budget": { "node_count": 2, "item_count": 2, @@ -18699,18 +18833,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs address the same xdist-related captured_info logging collision in testing utilities; PR #45645 looks like the more complete variant because it also updates CI log collection for the new per-worker files.", - "confidence": 0.86, + "summary": "Both PRs target the same xdist-related captured_info logging problem; PR 45645 is the broader, more complete fix and subsumes the narrower utility change in 45639.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45645 is the broader, more complete fix: it includes the xdist-safe file handling in testing_utils plus the CI-side aggregation needed to preserve captured logs.", + "canonical_pr_reason": "PR 45645 is the better canonical PR because it addresses the captured_info xdist collision in the code and also updates the CI workflow to consume multiple captured_info files, making it the more complete end-to-end fix.", "best_issue_reason": null, - "best_pr_reason": "PR #45645 best represents the underlying change because it covers both the core bug fix and the downstream log-handling adjustment.", + "best_pr_reason": "PR 45645 is the strongest representative: it fixes the xdist-safe output handling and preserves CI debug log collection, whereas 45639 only changes the test utility side.", "soft_edge_verdicts": [ { "left": "pull_request:45639", "right": "pull_request:45645", "accept": true, - "reason": "Same underlying bug/change: both fix captured_info handling under pytest-xdist in testing_utils, and #45645 appears to extend the same fix with CI log aggregation." + "reason": "Both PRs fix the same underlying captured_info/xdist collision path in testing_utils, and 45645 appears to extend rather than diverge from 45639 by adding workflow support for the same artifact change." } ] }, diff --git a/analysis/current/manifest.json b/analysis/current/manifest.json index 79d3b954b56cf1f750c46db54dd508207d0ad7bb..4d50b02137b567b60650d08ecbe6c54042061d61 100644 --- a/analysis/current/manifest.json +++ b/analysis/current/manifest.json @@ -1,8 +1,8 @@ { - "analysis_id": "hybrid-model-20260508t180023z", + "analysis_id": "hybrid-model-20260509t000027z", "archived_artifacts": { - "hybrid": "snapshots/20260508T180023Z/analysis-runs/hybrid-model-20260508t180023z/analysis-report-hybrid.json", - "hybrid_reviews": "snapshots/20260508T180023Z/analysis-runs/hybrid-model-20260508t180023z/analysis-report-hybrid.llm-reviews.json" + "hybrid": "snapshots/20260509T000027Z/analysis-runs/hybrid-model-20260509t000027z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260509T000027Z/analysis-runs/hybrid-model-20260509t000027z/analysis-report-hybrid.llm-reviews.json" }, "artifacts": { "hybrid": "analysis/current/analysis-report-hybrid.json", @@ -10,9 +10,9 @@ }, "channel": "canonical", "model": null, - "published_at": "2026-05-08T18:13:51Z", + "published_at": "2026-05-09T00:11:03Z", "repo": "huggingface/transformers", "schema_version": 1, - "snapshot_id": "20260508T180023Z", + "snapshot_id": "20260509T000027Z", "variant": "hybrid" } diff --git a/snapshots/20260509T000027Z/analysis-runs/hybrid-model-20260509t000027z/analysis-report-hybrid.json b/snapshots/20260509T000027Z/analysis-runs/hybrid-model-20260509t000027z/analysis-report-hybrid.json new file mode 100644 index 0000000000000000000000000000000000000000..215cd30a6313d65840600937b65e6da9a8ac47b3 --- /dev/null +++ b/snapshots/20260509T000027Z/analysis-runs/hybrid-model-20260509t000027z/analysis-report-hybrid.json @@ -0,0 +1,2205 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260509T000027Z", + "generated_at": "2026-05-09T00:11:01Z", + "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-43656-4", + "summary": "Cluster of 1 issues and 3 PRs centered on issue #43824.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 43824, + "canonical_pr_number": 43656, + "issue_numbers": [ + 43824 + ], + "pr_numbers": [ + 43656, + 43836, + 43842 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 43656, + "right_pr_number": 43836, + "code_similarity": 0.36, + "size_similarity": 0.176, + "file_overlap": 1.0, + "area_overlap": 0.051, + "patch_similarity": 0.048, + "shared_filenames": [ + "src/transformers/cli/serve.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/cli/serve.py", + "left_ranges": [ + [ + 11, + 18 + ], + [ + 30, + 36 + ], + [ + 315, + 323 + ], + [ + 665, + 671 + ], + [ + 931, + 937 + ], + [ + 1843, + 1849 + ], + [ + 1868, + 1874 + ] + ], + "right_ranges": [ + [ + 11, + 18 + ], + [ + 359, + 429 + ], + [ + 584, + 590 + ], + [ + 1892, + 1910 + ], + [ + 1917, + 1923 + ] + ] + } + ] + }, + { + "left_pr_number": 43656, + "right_pr_number": 43842, + "code_similarity": 0.405, + "size_similarity": 0.5, + "file_overlap": 1.0, + "area_overlap": 0.0, + "patch_similarity": 0.036, + "shared_filenames": [ + "src/transformers/cli/serve.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/cli/serve.py", + "left_ranges": [ + [ + 11, + 18 + ], + [ + 30, + 36 + ], + [ + 315, + 323 + ], + [ + 665, + 671 + ], + [ + 931, + 937 + ], + [ + 1843, + 1849 + ], + [ + 1868, + 1874 + ] + ], + "right_ranges": [ + [ + 54, + 61 + ], + [ + 587, + 593 + ] + ] + } + ] + }, + { + "left_pr_number": 43836, + "right_pr_number": 43842, + "code_similarity": 0.332, + "size_similarity": 0.088, + "file_overlap": 1.0, + "area_overlap": 0.033, + "patch_similarity": 0.017, + "shared_filenames": [ + "src/transformers/cli/serve.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/cli/serve.py", + "left_ranges": [ + [ + 11, + 18 + ], + [ + 359, + 429 + ], + [ + 584, + 590 + ], + [ + 1892, + 1910 + ], + [ + 1917, + 1923 + ] + ], + "right_ranges": [ + [ + 54, + 61 + ], + [ + 587, + 593 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-43240-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #43240.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 43240, + "canonical_pr_number": 43251, + "issue_numbers": [ + 43240 + ], + "pr_numbers": [ + 43251, + 43254 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 43251, + "right_pr_number": 43254, + "code_similarity": 0.794, + "size_similarity": 0.64, + "file_overlap": 1.0, + "area_overlap": 0.667, + "patch_similarity": 0.882, + "shared_filenames": [ + "src/transformers/loss/loss_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/loss/loss_utils.py", + "left_ranges": [ + [ + 30, + 50 + ] + ], + "right_ranges": [ + [ + 30, + 43 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-41115-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #41115.", + "status": "open", + "confidence": 0.55, + "canonical_issue_number": 41115, + "canonical_pr_number": 45613, + "issue_numbers": [ + 41115 + ], + "pr_numbers": [ + 41116, + 45613 + ], + "evidence_types": [ + "closing_reference" + ], + "pr_comparisons": [ + { + "left_pr_number": 41116, + "right_pr_number": 45613, + "code_similarity": 0.805, + "size_similarity": 0.885, + "file_overlap": 0.75, + "area_overlap": 0.854, + "patch_similarity": 0.693, + "shared_filenames": [ + "docs/source/en/_toctree.yml", + "docs/source/en/model_doc/minicpm3.md", + "src/transformers/models/auto/modeling_auto.py", + "src/transformers/models/minicpm3/__init__.py", + "src/transformers/models/minicpm3/configuration_minicpm3.py", + "src/transformers/models/minicpm3/modeling_minicpm3.py", + "src/transformers/models/minicpm3/modular_minicpm3.py", + "tests/models/minicpm3/__init__.py", + "tests/models/minicpm3/test_modeling_minicpm3.py" + ], + "shared_file_areas": [ + { + "filename": "docs/source/en/_toctree.yml", + "left_ranges": [ + [ + 713, + 720 + ] + ], + "right_ranges": [ + [ + 713, + 720 + ] + ] + }, + { + "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": [ + [ + 699, + 705 + ], + [ + 1301, + 1307 + ] + ] + }, + { + "filename": "src/transformers/models/minicpm3/__init__.py", + "left_ranges": [ + [ + 1, + 29 + ] + ], + "right_ranges": [ + [ + 1, + 29 + ] + ] + }, + { + "filename": "src/transformers/models/minicpm3/configuration_minicpm3.py", + "left_ranges": [ + [ + 1, + 141 + ] + ], + "right_ranges": [ + [ + 1, + 126 + ] + ] + }, + { + "filename": "src/transformers/models/minicpm3/modeling_minicpm3.py", + "left_ranges": [ + [ + 1, + 544 + ] + ], + "right_ranges": [ + [ + 1, + 522 + ] + ] + }, + { + "filename": "src/transformers/models/minicpm3/modular_minicpm3.py", + "left_ranges": [ + [ + 1, + 444 + ] + ], + "right_ranges": [ + [ + 1, + 342 + ] + ] + }, + { + "filename": "tests/models/minicpm3/test_modeling_minicpm3.py", + "left_ranges": [ + [ + 1, + 109 + ] + ], + "right_ranges": [ + [ + 1, + 136 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-45647-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45647.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 45647, + "canonical_pr_number": 45738, + "issue_numbers": [ + 45647 + ], + "pr_numbers": [ + 45737, + 45738 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 45737, + "right_pr_number": 45738, + "code_similarity": 0.526, + "size_similarity": 0.11, + "file_overlap": 0.5, + "area_overlap": 1.0, + "patch_similarity": 0.024, + "shared_filenames": [ + "src/transformers/models/musicgen_melody/modeling_musicgen_melody.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/musicgen_melody/modeling_musicgen_melody.py", + "left_ranges": [ + [ + 491, + 497 + ] + ], + "right_ranges": [ + [ + 491, + 497 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-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-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": 0.532, + "size_similarity": 0.619, + "file_overlap": 0.5, + "area_overlap": 0.543, + "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": [ + [ + 3529, + 3536 + ], + [ + 3753, + 3787 + ] + ], + "right_ranges": [ + [ + 3525, + 3558 + ], + [ + 3782, + 3788 + ] + ] + }, + { + "filename": "tests/utils/test_testing_utils.py", + "left_ranges": [ + [ + 1, + 86 + ] + ], + "right_ranges": [ + [ + 1, + 114 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-43698-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #43698.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 43698, + "canonical_pr_number": 43779, + "issue_numbers": [ + 43698 + ], + "pr_numbers": [ + 43779, + 43816 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 43779, + "right_pr_number": 43816, + "code_similarity": 0.418, + "size_similarity": 0.538, + "file_overlap": 1.0, + "area_overlap": 0.02, + "patch_similarity": 0.02, + "shared_filenames": [ + "src/transformers/integrations/integration_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/integrations/integration_utils.py", + "left_ranges": [ + [ + 2227, + 2241 + ], + [ + 2303, + 2309 + ] + ], + "right_ranges": [ + [ + 2278, + 2291 + ], + [ + 2309, + 2322 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-44018-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 44068, + "issue_numbers": [], + "pr_numbers": [ + 44018, + 44068 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 44018, + "right_pr_number": 44068, + "code_similarity": 0.766, + "size_similarity": 0.939, + "file_overlap": 1.0, + "area_overlap": 0.425, + "patch_similarity": 0.866, + "shared_filenames": [ + "src/transformers/models/gpt_neo/modeling_gpt_neo.py", + "tests/models/gpt_neo/test_modeling_gpt_neo.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/gpt_neo/modeling_gpt_neo.py", + "left_ranges": [ + [ + 26, + 46 + ], + [ + 138, + 143 + ], + [ + 182, + 187 + ], + [ + 283, + 295 + ], + [ + 327, + 341 + ], + [ + 348, + 354 + ], + [ + 360, + 369 + ], + [ + 399, + 406 + ], + [ + 411, + 419 + ], + [ + 428, + 433 + ], + [ + 472, + 492 + ], + [ + 507, + 513 + ], + [ + 519, + 528 + ], + [ + 541, + 559 + ], + [ + 562, + 567 + ], + [ + 595, + 601 + ], + [ + 607, + 614 + ], + [ + 627, + 643 + ], + [ + 685, + 690 + ], + [ + 708, + 714 + ], + [ + 720, + 727 + ], + [ + 740, + 757 + ], + [ + 761, + 766 + ], + [ + 780, + 786 + ], + [ + 791, + 798 + ], + [ + 807, + 822 + ], + [ + 840, + 845 + ] + ], + "right_ranges": [ + [ + 26, + 31 + ], + [ + 34, + 44 + ], + [ + 136, + 141 + ], + [ + 180, + 185 + ], + [ + 281, + 293 + ], + [ + 325, + 339 + ], + [ + 346, + 352 + ], + [ + 358, + 367 + ], + [ + 397, + 404 + ], + [ + 409, + 417 + ], + [ + 426, + 434 + ], + [ + 463, + 483 + ], + [ + 498, + 504 + ], + [ + 510, + 519 + ], + [ + 532, + 550 + ], + [ + 553, + 564 + ], + [ + 586, + 592 + ], + [ + 598, + 605 + ], + [ + 618, + 634 + ], + [ + 676, + 688 + ], + [ + 699, + 705 + ], + [ + 711, + 718 + ], + [ + 731, + 748 + ], + [ + 752, + 762 + ], + [ + 771, + 777 + ], + [ + 782, + 789 + ], + [ + 798, + 813 + ], + [ + 831, + 836 + ] + ] + }, + { + "filename": "tests/models/gpt_neo/test_modeling_gpt_neo.py", + "left_ranges": [ + [ + 458, + 464 + ] + ], + "right_ranges": [ + [ + 458, + 464 + ] + ] + } + ] + } + ] + } + ], + "duplicate_issues": [], + "duplicate_prs": [ + { + "cluster_id": "cluster-41115-3", + "canonical_pr_number": 45613, + "duplicate_pr_numbers": [ + 41116 + ], + "target_issue_number": 41115, + "reason": "PRs in cluster-41115-3 are treated as duplicates because they converge on issue #41115 with closing_reference evidence." + }, + { + "cluster_id": "cluster-41211-3", + "canonical_pr_number": 44339, + "duplicate_pr_numbers": [ + 41356 + ], + "target_issue_number": 41211, + "reason": "PRs in cluster-41211-3 are treated as duplicates because they converge on issue #41211 with closing_reference evidence." + }, + { + "cluster_id": "cluster-43240-3", + "canonical_pr_number": 43251, + "duplicate_pr_numbers": [ + 43254 + ], + "target_issue_number": 43240, + "reason": "PRs in cluster-43240-3 are treated as duplicates because they converge on issue #43240 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-43656-4", + "canonical_pr_number": 43656, + "duplicate_pr_numbers": [ + 43836, + 43842 + ], + "target_issue_number": 43824, + "reason": "PRs in cluster-43656-4 are treated as duplicates because they converge on issue #43824 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-43698-3", + "canonical_pr_number": 43779, + "duplicate_pr_numbers": [ + 43816 + ], + "target_issue_number": 43698, + "reason": "PRs in cluster-43698-3 are treated as duplicates because they converge on issue #43698 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-43979-11", + "canonical_pr_number": 44007, + "duplicate_pr_numbers": [ + 43996, + 44013, + 44044, + 44066, + 44072, + 44085, + 44129, + 44154, + 44722 + ], + "target_issue_number": 43979, + "reason": "PRs in cluster-43979-11 are treated as duplicates because they converge on issue #43979 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-44018-2", + "canonical_pr_number": 44068, + "duplicate_pr_numbers": [ + 44018 + ], + "target_issue_number": null, + "reason": "PRs in cluster-44018-2 are treated as duplicates because they share soft_similarity evidence." + }, + { + "cluster_id": "cluster-45081-3", + "canonical_pr_number": 45317, + "duplicate_pr_numbers": [ + 45086 + ], + "target_issue_number": 45081, + "reason": "PRs in cluster-45081-3 are treated as duplicates because they converge on issue #45081 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-45561-3", + "canonical_pr_number": 45645, + "duplicate_pr_numbers": [ + 45639 + ], + "target_issue_number": 45561, + "reason": "PRs in cluster-45561-3 are treated as duplicates because they converge on issue #45561 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-45647-3", + "canonical_pr_number": 45738, + "duplicate_pr_numbers": [ + 45737 + ], + "target_issue_number": 45647, + "reason": "PRs in cluster-45647-3 are treated as duplicates because they converge on issue #45647 with closing_reference, shared_issue_target evidence." + } + ], + "best_issue": { + "cluster_id": "cluster-45647-3", + "issue_number": 45647, + "reason": "Issue #45647 is the strongest global issue candidate because it is open, belongs to a cluster with 3 artifacts, and carries 12 discussion comments plus 8 inbound references.", + "score": 162.0 + }, + "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/20260509T000027Z/analysis-runs/hybrid-model-20260509t000027z/analysis-report-hybrid.llm-reviews.json b/snapshots/20260509T000027Z/analysis-runs/hybrid-model-20260509t000027z/analysis-report-hybrid.llm-reviews.json new file mode 100644 index 0000000000000000000000000000000000000000..2c72b69356db580c3cd6279a4a610405378a3c64 --- /dev/null +++ b/snapshots/20260509T000027Z/analysis-runs/hybrid-model-20260509t000027z/analysis-report-hybrid.llm-reviews.json @@ -0,0 +1,18988 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260509T000027Z", + "generated_at": "2026-05-09T00:11:01Z", + "model": "gpt-5.4-mini?service_tier=flex", + "reviews": [ + { + "label": "issue", + "component_index": 1, + "component_count": 16, + "review_unit_index": 1, + "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": "4929c4a3069ad42a997018a65be24c829a02d2274479dc5fb99cda6877b28426", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 2477, + "estimated_input_tokens": 620, + "estimated_eval_tokens": 1496 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 2477, + "estimated_input_tokens": 620, + "estimated_eval_tokens": 1496 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The cluster is not a true duplicate set. Issue 16618 is the broad model-loading failure, while 15980 is specifically about a bad auth error for private model downloads and 13244 is about Tapas tokenization differences.", + "confidence": 0.9, + "canonical_issue_reason": "#16618 is the broadest and most central user-facing model-load failure in the set; the others are about different root causes and code paths.", + "canonical_pr_reason": null, + "best_issue_reason": "#16618 best represents the cluster\u2019s shared model-loading theme, even though the other issues are not duplicates of it.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:13244", + "right": "issue:16618", + "accept": false, + "reason": "Tapas tokenization behavior is unrelated to a generic BERT model load failure; different subsystem and bug." + }, + { + "left": "issue:15980", + "right": "issue:16618", + "accept": false, + "reason": "Private-model download auth/error-message handling is a different failure mode from 'can't load bert-base-uncased'; 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": 2, + "component_count": 16, + "review_unit_index": 1, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:24643", + "issue:30277", + "issue:34634", + "issue:34928", + "issue:35545", + "issue:36010", + "issue:36331", + "issue:39290", + "issue:41093", + "issue:41720", + "issue:41762", + "issue:42915", + "issue:43039", + "issue:43404", + "issue:43425", + "issue:43531", + "issue:43541", + "issue:43596", + "issue:43638", + "issue:43716", + "issue:43792", + "issue:43827", + "issue:43828", + "issue:43854", + "issue:43866", + "issue:43901", + "issue:43975", + "issue:44292", + "issue:44322", + "issue:44457", + "issue:44509", + "issue:44512", + "issue:44560", + "issue:44805", + "issue:44841", + "issue:44863", + "issue:44918", + "issue:45070", + "issue:45081", + "issue:45084", + "issue:45161", + "issue:45237", + "issue:45362", + "issue:45464", + "issue:45507", + "issue:45685", + "issue:45800", + "issue:45823" + ], + "soft_pairs": [ + "issue:44322|issue:45464", + "issue:45237|issue:45685", + "issue:43716|issue:45237", + "issue:43828|issue:45237", + "issue:45237|issue:45507", + "issue:44292|issue:45237", + "issue:30277|issue:35545", + "issue:35545|issue:42915", + "issue:35545|issue:41720", + "issue:44509|issue:44512", + "issue:35545|issue:36010", + "issue:44805|issue:45823", + "issue:43638|issue:44805", + "issue:43596|issue:44805", + "issue:44805|issue:45161", + "issue:44805|issue:44918", + "issue:34634|issue:35545", + "issue:30277|issue:43638", + "issue:43596|issue:43638", + "issue:43975|issue:44457", + "issue:43039|issue:45161", + "issue:44292|issue:44457", + "issue:34928|issue:44805", + "issue:24643|issue:30277", + "issue:43638|issue:45161", + "issue:34928|issue:36331", + "issue:43596|issue:45161", + "issue:43716|issue:45507", + "issue:39290|issue:45081", + "issue:43828|issue:45161", + "issue:43425|issue:44292", + "issue:44918|issue:45161", + "issue:34928|issue:41720", + "issue:43716|issue:45685", + "issue:43541|issue:43716", + "issue:43828|issue:44292", + "issue:45507|issue:45685", + "issue:43716|issue:44292", + "issue:44292|issue:44918", + "issue:43039|issue:44918", + "issue:41762|issue:43638", + "issue:43827|issue:44512", + "issue:43716|issue:44560", + "issue:43828|issue:45685", + "issue:44560|issue:45685", + "issue:43039|issue:43828", + "issue:30277|issue:36331", + "issue:34928|issue:36010", + "issue:43039|issue:44805", + "issue:43901|issue:44512", + "issue:41762|issue:43596", + "issue:39290|issue:41720", + "issue:44841|issue:45823", + "issue:43039|issue:44560", + "issue:43866|issue:44863", + "issue:43854|issue:43866", + "issue:39290|issue:43531", + "issue:41093|issue:43638", + "issue:43425|issue:45070", + "issue:43596|issue:45800", + "issue:43792|issue:45070", + "issue:45362|issue:45464", + "issue:43404|issue:43866", + "issue:39290|issue:45823", + "issue:45464|issue:45685", + "issue:43541|issue:45685", + "issue:43541|issue:43828", + "issue:36010|issue:39290", + "issue:44841|issue:45084", + "issue:45800|issue:45823", + "issue:41093|issue:41762" + ], + "prepared_review_unit_hash": "83671578cf33ce3c9529ee4bef7423820903edbe840e6cb4e4d37193ba8486f1", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 71, + "serialized_chars": 48869, + "estimated_input_tokens": 12218, + "estimated_eval_tokens": 24692 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a set of unrelated issue reports with similar-sounding error strings. The only clear duplicate is the DeepSpeed ZeRO-3/Bert index-out-of-bounds pair; the rest differ by model, backend, or feature area.", + "confidence": 0.83, + "canonical_issue_reason": "issue:43596 is the best anchor for the one clear duplicate set: it names the ZeRO-3 init index-out-of-bounds failure concisely and covers the shared BertModel code path.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43596 is the strongest representative issue in this set because it is the clearest, most general description of the shared ZeRO-3/Bert failure.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44322", + "right": "issue:45464", + "accept": false, + "reason": "Different Qwen3.5 failures: one is a config attribute error, the other is a streaming chat/completions runtime issue." + }, + { + "left": "issue:45237", + "right": "issue:45685", + "accept": false, + "reason": "Different models and backends: AMD GPU failure for GPT-OSS vs an MPS histogram op error for moe." + }, + { + "left": "issue:43716", + "right": "issue:45237", + "accept": false, + "reason": "Different model families and symptoms; this is a dtype mismatch in Mistral-3 image preprocessing, not the GPT-OSS AMD GPU crash." + }, + { + "left": "issue:43828", + "right": "issue:45237", + "accept": false, + "reason": "Both are runtime/backend errors, but the concrete failures and affected models are unrelated." + }, + { + "left": "issue:45237", + "right": "issue:45507", + "accept": false, + "reason": "Different code paths: GPT-OSS AMD GPU failure vs GraniteMoEHybrid calling an invalid method." + }, + { + "left": "issue:44292", + "right": "issue:45237", + "accept": false, + "reason": "Different model/quantization issue; NVFP4 runtime failure is not the same bug as the AMD GPU crash." + }, + { + "left": "issue:30277", + "right": "issue:35545", + "accept": false, + "reason": "Different subsystems: DeepSpeed collective mismatch vs ONNX export error." + }, + { + "left": "issue:35545", + "right": "issue:42915", + "accept": false, + "reason": "Different model/export problems; ModernBERT ONNX export is unrelated to Qwen3Moe FP8 failure." + }, + { + "left": "issue:35545", + "right": "issue:41720", + "accept": false, + "reason": "Unrelated failures: ONNX export for ModernBERT vs CUDA assert with Qwen3 auto device mapping." + }, + { + "left": "issue:44509", + "right": "issue:44512", + "accept": false, + "reason": "Both are v5 docs regressions, but they reference different removed items and do not look like one concrete bug." + }, + { + "left": "issue:35545", + "right": "issue:36010", + "accept": false, + "reason": "Different issues: ONNX export vs GenerationMixin import path breakage." + }, + { + "left": "issue:44805", + "right": "issue:45823", + "accept": false, + "reason": "Both mention shape/mismatch-type failures, but the affected models and failure modes are different." + }, + { + "left": "issue:43638", + "right": "issue:44805", + "accept": false, + "reason": "Different error families: ZeRO-3 index-out-of-bounds on Bert vs mask/index shape mismatch." + }, + { + "left": "issue:43596", + "right": "issue:44805", + "accept": false, + "reason": "Same general category of indexing errors, but not the same model path or concrete bug." + }, + { + "left": "issue:44805", + "right": "issue:45161", + "accept": false, + "reason": "Different models and execution modes; mask-shape mismatch is unrelated to GPT-OSS tensor-parallel behavior." + }, + { + "left": "issue:44805", + "right": "issue:44918", + "accept": false, + "reason": "Different failure sites: mask/index mismatch vs Qwen3.5 embedding unpacking in TRL SFT trainer." + }, + { + "left": "issue:34634", + "right": "issue:35545", + "accept": false, + "reason": "BarkProcessor voice_preset bug is unrelated to ModernBERT ONNX export." + }, + { + "left": "issue:30277", + "right": "issue:43638", + "accept": false, + "reason": "Both involve DeepSpeed, but one is a collective mismatch and the other is a ZeRO-3 index-out-of-bounds init bug." + }, + { + "left": "issue:43596", + "right": "issue:43638", + "accept": true, + "reason": "Same ZeRO-3 index-out-of-bounds failure on BertModel; the second report is the non-pretrained Bert variant of the same underlying bug." + }, + { + "left": "issue:43975", + "right": "issue:44457", + "accept": false, + "reason": "Different functionality: detokenization regression vs LoRA merge/save/reload output mismatch." + }, + { + "left": "issue:43039", + "right": "issue:45161", + "accept": false, + "reason": "Liger Kernel cross-entropy routing and GPT-OSS tensor-parallel failure are unrelated." + }, + { + "left": "issue:44292", + "right": "issue:44457", + "accept": false, + "reason": "Different areas entirely: Qwen NVFP4 runtime failure vs LoRA merge/save consistency." + }, + { + "left": "issue:34928", + "right": "issue:44805", + "accept": false, + "reason": "Different distributed-training bugs: activation checkpointing/FSDP size mismatch vs mask/index mismatch." + }, + { + "left": "issue:24643", + "right": "issue:30277", + "accept": false, + "reason": "Both are DeepSpeed-related, but the concrete failures are different and not the same bug." + }, + { + "left": "issue:43638", + "right": "issue:45161", + "accept": false, + "reason": "Bert ZeRO-3 init failure is unrelated to GPT-OSS tensor-parallel behavior." + }, + { + "left": "issue:34928", + "right": "issue:36331", + "accept": false, + "reason": "Activation-checkpointing size mismatch and CustomTrainer signature mismatch are unrelated." + }, + { + "left": "issue:43596", + "right": "issue:45161", + "accept": false, + "reason": "Different models and code paths; ZeRO-3 Bert init bug is not the GPT-OSS TP issue." + }, + { + "left": "issue:43716", + "right": "issue:45507", + "accept": false, + "reason": "Different bugs: image preprocessor/model dtype mismatch vs invalid method call in GraniteMoEHybrid." + }, + { + "left": "issue:39290", + "right": "issue:45081", + "accept": false, + "reason": "Different model/config issues: Gemma3 sliding-window config vs Mistral tokenizer regex patch crash." + }, + { + "left": "issue:43828", + "right": "issue:45161", + "accept": false, + "reason": "Phi-tiny-MoE autocast dtype mismatch is unrelated to GPT-OSS tensor-parallel failures." + }, + { + "left": "issue:44918", + "right": "issue:45161", + "accept": false, + "reason": "Qwen3.5 embedding unpacking in TRL is not the same bug as GPT-OSS tensor-parallel support." + }, + { + "left": "issue:43039", + "right": "issue:44918", + "accept": false, + "reason": "Different areas: Liger Kernel cross-entropy vs TRL SFT input embedding unpacking." + }, + { + "left": "issue:41762", + "right": "issue:43638", + "accept": false, + "reason": "Both are ZeRO-3 index errors, but they involve different model families and are not clearly the same concrete bug." + }, + { + "left": "issue:43827", + "right": "issue:44512", + "accept": false, + "reason": "Both are docs regressions, but they refer to different removed v5 documentation surfaces." + }, + { + "left": "issue:43716", + "right": "issue:44560", + "accept": false, + "reason": "Different multimodal models and different runtime errors; dtype mismatch is not the same as StopIteration in video embedding." + }, + { + "left": "issue:43828", + "right": "issue:45685", + "accept": false, + "reason": "Autocast dtype mismatch and MPS histogram op missing are unrelated backend bugs." + }, + { + "left": "issue:44560", + "right": "issue:45685", + "accept": false, + "reason": "Different model/backend issues: Qwen3-vl video StopIteration vs MPS unsupported histogram op." + }, + { + "left": "issue:43039", + "right": "issue:43828", + "accept": false, + "reason": "Different failure modes in different models; cross_entropy routing is not the same as autocast dtype mismatch." + }, + { + "left": "issue:30277", + "right": "issue:36331", + "accept": false, + "reason": "DeepSpeed collective mismatch and trainer method signature mismatch are unrelated." + }, + { + "left": "issue:34928", + "right": "issue:36010", + "accept": false, + "reason": "Different problems: activation checkpointing/FSDP shape mismatch vs missing GenerationMixin import." + }, + { + "left": "issue:43039", + "right": "issue:44805", + "accept": false, + "reason": "Liger Kernel cross-entropy call issue is unrelated to mask/index shape mismatch." + }, + { + "left": "issue:43901", + "right": "issue:44512", + "accept": false, + "reason": "Both are docs issues, but one is about return_all_scores and the other about the removed transformers run command." + }, + { + "left": "issue:41762", + "right": "issue:43596", + "accept": false, + "reason": "Same broad ZeRO-3 symptom, but the affected models differ enough that this is not clearly one duplicate bug." + }, + { + "left": "issue:39290", + "right": "issue:41720", + "accept": false, + "reason": "Gemma3 sliding_window_pattern config and Qwen3 auto device mapping CUDA assert are unrelated." + }, + { + "left": "issue:44841", + "right": "issue:45823", + "accept": false, + "reason": "Processor failure for Voxtral is unrelated to Gemma4 PLE device mismatch in forward." + }, + { + "left": "issue:43039", + "right": "issue:44560", + "accept": false, + "reason": "Different subsystems entirely: Liger Kernel cross-entropy vs Qwen3-vl video StopIteration." + }, + { + "left": "issue:43866", + "right": "issue:44863", + "accept": false, + "reason": "Checkpoint loading problems, but for different models and likely different root causes." + }, + { + "left": "issue:43854", + "right": "issue:43866", + "accept": false, + "reason": "Different model-loading failures for different models; no clear shared bug." + }, + { + "left": "issue:39290", + "right": "issue:43531", + "accept": false, + "reason": "Related Qwen/Gemma sliding-window themes, but different model families and concrete failures." + }, + { + "left": "issue:41093", + "right": "issue:43638", + "accept": false, + "reason": "Both are index-related errors, but the mask/tensor size mismatch is different from the ZeRO-3 empty-dimension init bug." + }, + { + "left": "issue:43425", + "right": "issue:45070", + "accept": false, + "reason": "Torch compatibility and pydantic model regression are unrelated." + }, + { + "left": "issue:43596", + "right": "issue:45800", + "accept": false, + "reason": "Different DeepSpeed/Bert init bug vs torch/transformers version incompatibility." + }, + { + "left": "issue:43792", + "right": "issue:45070", + "accept": false, + "reason": "Whisper loading failure and PretrainedConfig pydantic breakage are unrelated." + }, + { + "left": "issue:45362", + "right": "issue:45464", + "accept": false, + "reason": "Both involve Qwen3.5 and chat, but one is a transformers chat crash and the other is a streaming chat/completions API failure; not clearly the same bug." + }, + { + "left": "issue:43404", + "right": "issue:43866", + "accept": false, + "reason": "Different issue classes: tied lm_head weight bug vs corrupted checkpoint/load failure." + }, + { + "left": "issue:39290", + "right": "issue:45823", + "accept": false, + "reason": "Gemma3 sliding-window config and Gemma4 device-mapping mismatch are different model issues." + }, + { + "left": "issue:45464", + "right": "issue:45685", + "accept": false, + "reason": "Chat/completions streaming failure on Qwen3.5 is unrelated to the MPS histogram op bug." + }, + { + "left": "issue:43541", + "right": "issue:45685", + "accept": false, + "reason": "Mixtral grouped_mm tracing error and MPS histogram op missing are unrelated." + }, + { + "left": "issue:43541", + "right": "issue:43828", + "accept": false, + "reason": "Grouped_mm tracing failure in Mixtral is unrelated to Phi-tiny-MoE autocast dtype mismatch." + }, + { + "left": "issue:36010", + "right": "issue:39290", + "accept": false, + "reason": "GenerationMixin import failure and Gemma3 sliding-window config issue are unrelated." + }, + { + "left": "issue:44841", + "right": "issue:45084", + "accept": false, + "reason": "Processor failure and template compilation TypeError are different bugs." + }, + { + "left": "issue:45800", + "right": "issue:45823", + "accept": false, + "reason": "Torch/transformers version incompatibility and Gemma4 device mismatch are unrelated." + }, + { + "left": "issue:41093", + "right": "issue:41762", + "accept": false, + "reason": "Both are index errors, but the mask/index mismatch is not the same as the ZeRO-3 empty-dimension load failure." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is well grounded in the packet and stays conservative. The only accepted duplicate pair, issue:43596/issue:43638, is supported by the shared DeepSpeed ZeRO-3 + Bert index-out-of-bounds titles. The other soft-edge rejections are cautious and consistent with the visible model/backend/feature differences in the packet." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 2, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:39186", + "issue:39692", + "issue:40990", + "issue:41093", + "issue:42915", + "issue:43054", + "issue:43296", + "issue:43366", + "issue:43404", + "issue:43531", + "issue:43541", + "issue:43572", + "issue:43596", + "issue:43644", + "issue:43645", + "issue:43716", + "issue:43828", + "issue:43866", + "issue:43873", + "issue:43950", + "issue:43975", + "issue:43994", + "issue:44291", + "issue:44292", + "issue:44360", + "issue:44387", + "issue:44403", + "issue:44492", + "issue:44512", + "issue:44530", + "issue:44534", + "issue:44589", + "issue:44863", + "issue:44918", + "issue:44928", + "issue:44960", + "issue:45005", + "issue:45020", + "issue:45070", + "issue:45161", + "issue:45399", + "issue:45464", + "issue:45507", + "issue:45542", + "issue:45632", + "issue:45685", + "issue:45698", + "issue:45800" + ], + "soft_pairs": [ + "issue:43296|issue:43366", + "issue:43572|issue:45070", + "issue:44387|issue:45005", + "issue:43541|issue:45161", + "issue:43828|issue:45464", + "issue:44291|issue:45070", + "issue:43366|issue:45464", + "issue:45020|issue:45800", + "issue:43531|issue:45070", + "issue:43541|issue:44589", + "issue:41093|issue:43596", + "issue:43054|issue:43994", + "issue:45542|issue:45800", + "issue:43716|issue:44918", + "issue:44918|issue:45685", + "issue:45632|issue:45698", + "issue:43404|issue:45005", + "issue:43645|issue:45399", + "issue:40990|issue:44960", + "issue:44530|issue:45005", + "issue:42915|issue:45399", + "issue:44360|issue:44512", + "issue:44863|issue:45005", + "issue:43873|issue:45005", + "issue:39186|issue:44918", + "issue:44492|issue:44512", + "issue:44918|issue:45507", + "issue:43975|issue:44292", + "issue:43950|issue:44534", + "issue:43644|issue:43950", + "issue:44292|issue:45507", + "issue:39692|issue:43054", + "issue:43828|issue:44928", + "issue:43866|issue:44403" + ], + "prepared_review_unit_hash": "6e7ca954d63323d5d398617986dcb10cb3c111d7f6cc0eebe0f8990c8386375a", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37468, + "estimated_input_tokens": 9367, + "estimated_eval_tokens": 18990 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most items are unrelated. The only clear duplicate cluster is the Transformers v5 non-persistent-buffer corruption regression, best anchored by issue 43950.", + "confidence": 0.92, + "canonical_issue_reason": "Issue 43950 is the most explicit statement of the underlying regression and matches the small duplicate cluster around non-persistent buffers being corrupted on load.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43950 is the best overall representative because it describes the core failure mode clearly and centrally, rather than a narrower symptom.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43296", + "right": "issue:43366", + "accept": false, + "reason": "Different problems: PaddleOCR-VL loading under transformers/vLLM vs GGUF support for gpt-oss architecture." + }, + { + "left": "issue:43572", + "right": "issue:45070", + "accept": false, + "reason": "Both are v5 regressions, but one is a missing StableLm config field and the other is a pydantic PretrainedConfig field breakage; not the same bug." + }, + { + "left": "issue:44387", + "right": "issue:45005", + "accept": false, + "reason": "Different failure modes: int4 quantization OOM vs tied-weights handling in translation models." + }, + { + "left": "issue:43541", + "right": "issue:45161", + "accept": false, + "reason": "Both involve MoE models, but one is a grouped_mm tracing failure and the other is a tensor-parallel-only issue for GPT-OSS." + }, + { + "left": "issue:43828", + "right": "issue:45464", + "accept": false, + "reason": "Both are dtype/runtime issues, but the model path and failure context differ: Phi-tiny-MoE autocast mismatch vs Qwen3.5 streaming inference failure." + }, + { + "left": "issue:44291", + "right": "issue:45070", + "accept": false, + "reason": "Different code paths: init_empty_weights/_is_hf_initialized argument handling vs pydantic PretrainedConfig field regression." + }, + { + "left": "issue:43366", + "right": "issue:45464", + "accept": false, + "reason": "Unrelated: GGUF/gpt-oss architecture support vs chat/completions streaming failure on Qwen3.5." + }, + { + "left": "issue:45020", + "right": "issue:45800", + "accept": false, + "reason": "Both are recent-version breakages, but one is remote_code model loading and the other is a torch/transformers compatibility issue." + }, + { + "left": "issue:43531", + "right": "issue:45070", + "accept": false, + "reason": "Different bugs: Qwen3-MoE sliding_window behavior vs PretrainedConfig field breakage in v5.4.0." + }, + { + "left": "issue:43541", + "right": "issue:44589", + "accept": false, + "reason": "Both hit model execution errors, but grouped_mm tracing and missing Float8 storage are separate underlying failures." + }, + { + "left": "issue:41093", + "right": "issue:43596", + "accept": false, + "reason": "Both are IndexErrors, but the mask-shape mismatch and the zero-sized tensor access are different root causes." + }, + { + "left": "issue:43054", + "right": "issue:43994", + "accept": false, + "reason": "Both mention siglip2, but one is degraded embedding quality and the other is nonsensical output with AutoModel/pipeline; not enough evidence of the same bug." + }, + { + "left": "issue:45542", + "right": "issue:45800", + "accept": false, + "reason": "Different regressions: missing TensorFlow backend handling vs torch 2.4.1 / transformers 5.8.0 incompatibility." + }, + { + "left": "issue:43716", + "right": "issue:44918", + "accept": false, + "reason": "Different model families and failure paths: Mistral-3 image preprocessor dtype mismatch vs Qwen3.5 input-embedding unpacking in TRL SFT." + }, + { + "left": "issue:44918", + "right": "issue:45685", + "accept": false, + "reason": "Both involve MoE-ish models, but one is TRL unpacking and the other is an MPS histogram op limitation; unrelated." + }, + { + "left": "issue:45632", + "right": "issue:45698", + "accept": false, + "reason": "Both involve custom-module loading, but one is a trust_remote_code cache-path collision and the other is loading the wrong custom module after save_pretrained; different concrete bugs." + }, + { + "left": "issue:43404", + "right": "issue:45005", + "accept": false, + "reason": "Related to tied weights in broad terms, but one is Mistral3 lm_head tying and the other is translation-model tied-weight issues in v5." + }, + { + "left": "issue:43645", + "right": "issue:45399", + "accept": false, + "reason": "Notebook custom-model initialization vs flash-attn2 fallback blocking are unrelated." + }, + { + "left": "issue:40990", + "right": "issue:44960", + "accept": false, + "reason": "Different symptoms and model families: high perplexity on gpt-oss-20b vs GLM5 issues." + }, + { + "left": "issue:44530", + "right": "issue:45005", + "accept": false, + "reason": "PagedAttentionCache / Qwen3.5.0 linear_attention crash is unrelated to translation tied-weight regressions." + }, + { + "left": "issue:42915", + "right": "issue:45399", + "accept": false, + "reason": "FineGrainedFP8Config failure for Qwen3-MoE is unrelated to flash-attn2 fallback blocking." + }, + { + "left": "issue:44360", + "right": "issue:44512", + "accept": false, + "reason": "No shared underlying bug: a DSA indexer algorithm note vs stale docs about the removed run command." + }, + { + "left": "issue:44863", + "right": "issue:45005", + "accept": false, + "reason": "NemotronH checkpoint-loading failure is unrelated to tied weights in translation models." + }, + { + "left": "issue:43873", + "right": "issue:45005", + "accept": false, + "reason": "Quantization/offloading behavior and translation tied-weight handling are different problems." + }, + { + "left": "issue:39186", + "right": "issue:44918", + "accept": false, + "reason": "FSDP weight dimensionality error is unrelated to Qwen3.5 embedding unpacking in TRL." + }, + { + "left": "issue:44492", + "right": "issue:44512", + "accept": false, + "reason": "Different doc issues: cache-strategy typo vs removed transformers run command." + }, + { + "left": "issue:44918", + "right": "issue:45507", + "accept": false, + "reason": "Different failure paths: TRL input-embedding unpacking vs GraniteMoEHybrid calling an invalid method." + }, + { + "left": "issue:43975", + "right": "issue:44292", + "accept": false, + "reason": "Different bugs: detokenization regression vs init_empty_weights constructor argument mismatch." + }, + { + "left": "issue:43950", + "right": "issue:44534", + "accept": true, + "reason": "Same underlying regression: Transformers v5 corrupts/fills non-persistent buffers during load." + }, + { + "left": "issue:43644", + "right": "issue:43950", + "accept": true, + "reason": "Same bug cluster: non-persistent buffers being filled with junk/corrupted on from_pretrained in v5." + }, + { + "left": "issue:44292", + "right": "issue:45507", + "accept": false, + "reason": "init_empty_weights argument handling and GraniteMoEHybrid invalid-method calls are unrelated." + }, + { + "left": "issue:39692", + "right": "issue:43054", + "accept": false, + "reason": "SigLIP2 docs example errors are unrelated to the later report that text embeddings are worse than SigLip." + }, + { + "left": "issue:43828", + "right": "issue:44928", + "accept": false, + "reason": "Both are Qwen/MoE-adjacent runtime issues, but the autocast dtype mismatch and RLHF NaN explosion have different causes." + }, + { + "left": "issue:43866", + "right": "issue:44403", + "accept": false, + "reason": "Checkpoint corruption is unrelated to the generic loading-noise discussion." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded overall and conservative in the soft-edge calls. The only clearly supported duplicate cluster in the packet is the Transformers v5 non-persistent-buffer corruption regression around issues 43950/44534/43644, and the representative choice is reasonable." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 3, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:36331", + "issue:38175", + "issue:41553", + "issue:41628", + "issue:41762", + "issue:42491", + "issue:42915", + "issue:42947", + "issue:43054", + "issue:43257", + "issue:43404", + "issue:43493", + "issue:43638", + "issue:43643", + "issue:43742", + "issue:43782", + "issue:43792", + "issue:43824", + "issue:43827", + "issue:43828", + "issue:43856", + "issue:43866", + "issue:43872", + "issue:43881", + "issue:43883", + "issue:43950", + "issue:43975", + "issue:44220", + "issue:44368", + "issue:44387", + "issue:44403", + "issue:44451", + "issue:44488", + "issue:44509", + "issue:44589", + "issue:44661", + "issue:44960", + "issue:44991", + "issue:45020", + "issue:45081", + "issue:45237", + "issue:45356", + "issue:45399", + "issue:45520", + "issue:45685", + "issue:45701", + "issue:45800", + "issue:45823" + ], + "soft_pairs": [ + "issue:36331|issue:41762", + "issue:41628|issue:45800", + "issue:43872|issue:44589", + "issue:41553|issue:43883", + "issue:38175|issue:43054", + "issue:43054|issue:45701", + "issue:41553|issue:43881", + "issue:42947|issue:44387", + "issue:44403|issue:44451", + "issue:43975|issue:45520", + "issue:43643|issue:45020", + "issue:43054|issue:43493", + "issue:43975|issue:45237", + "issue:44403|issue:44488", + "issue:42947|issue:43856", + "issue:43881|issue:43883", + "issue:41762|issue:42915", + "issue:42491|issue:44368", + "issue:45685|issue:45823", + "issue:43975|issue:44451", + "issue:43824|issue:43872", + "issue:45020|issue:45823", + "issue:44991|issue:45399", + "issue:44661|issue:45020", + "issue:43828|issue:45823", + "issue:45020|issue:45356", + "issue:43782|issue:44960", + "issue:45081|issue:45399", + "issue:43827|issue:44509", + "issue:43404|issue:43950", + "issue:43792|issue:44220", + "issue:43742|issue:45399", + "issue:42915|issue:45020", + "issue:43638|issue:45823", + "issue:43257|issue:43866" + ], + "prepared_review_unit_hash": "14f1adbe699b6788f5de374c28a2482f974afbde359bd892b81e2f1a897ea4f8", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 38124, + "estimated_input_tokens": 9531, + "estimated_eval_tokens": 19318 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The packet supports two concrete duplicate pockets only: the SigLIP2 regression cluster (issue:38175, issue:43054, issue:43493) and the docs/pipeline-removal cleanup pair (issue:43827, issue:44509). The remaining soft pairs are thematic similarities, not the same underlying bug or change.", + "confidence": 0.84, + "canonical_issue_reason": "issue:43054 is the clearest global canonical issue because it sits at the center of the supported SigLIP2 pocket, linking the user-visible symptom in issue:38175 with the implementation discrepancy in issue:43493.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43054 is the strongest representative issue overall: it anchors the only non-doc duplicate pocket with explicit support, while the docs pocket is a smaller cleanup pair.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:36331", + "right": "issue:41762", + "accept": false, + "reason": "Different failures: a custom trainer signature mismatch vs a ZeRO-3 loading IndexError." + }, + { + "left": "issue:41628", + "right": "issue:45800", + "accept": false, + "reason": "Import/export API issue vs a torch/transformers version incompatibility; not the same bug." + }, + { + "left": "issue:43872", + "right": "issue:44589", + "accept": false, + "reason": "Both involve storage/quantization behavior, but the concrete errors and code paths differ." + }, + { + "left": "issue:41553", + "right": "issue:43883", + "accept": false, + "reason": "Different model-loading problems with no clear shared failure mechanism." + }, + { + "left": "issue:38175", + "right": "issue:43054", + "accept": true, + "reason": "Same SigLIP2 regression: one reports the bad output behavior, the other is the broader symptom cluster." + }, + { + "left": "issue:43054", + "right": "issue:45701", + "accept": false, + "reason": "SigLIP2 regression vs a generic tokenization-change report; too broad to be the same bug." + }, + { + "left": "issue:41553", + "right": "issue:43881", + "accept": false, + "reason": "Bad tokenizer error message vs a model load failure for a different model family." + }, + { + "left": "issue:42947", + "right": "issue:44387", + "accept": false, + "reason": "Gradient checkpointing/LoRA issue vs int4 reserved-memory/OOM regression; different concrete problems." + }, + { + "left": "issue:44403", + "right": "issue:44451", + "accept": false, + "reason": "Generic loading noise vs inability to load a specific model; not the same underlying defect." + }, + { + "left": "issue:43975", + "right": "issue:45520", + "accept": false, + "reason": "Tokenizer detokenization regression vs a flash_attn import KeyError; unrelated failures." + }, + { + "left": "issue:43643", + "right": "issue:45020", + "accept": false, + "reason": "Related in theme, but not clearly the same concrete remote_code bug." + }, + { + "left": "issue:43054", + "right": "issue:43493", + "accept": true, + "reason": "Same SigLIP2 bug; issue:43493 describes the HF-vs-JAX discrepancy that explains issue:43054." + }, + { + "left": "issue:43975", + "right": "issue:45237", + "accept": false, + "reason": "Different model/use-case failures with no shared code-path evidence." + }, + { + "left": "issue:44403", + "right": "issue:44488", + "accept": false, + "reason": "Two separate model loading failures with no concrete match." + }, + { + "left": "issue:42947", + "right": "issue:43856", + "accept": false, + "reason": "LoRA/checkpointing behavior vs Qwen3 MoE memory use; same area, different bug." + }, + { + "left": "issue:43881", + "right": "issue:43883", + "accept": false, + "reason": "Different load-time errors for different models; not the same defect." + }, + { + "left": "issue:41762", + "right": "issue:42915", + "accept": false, + "reason": "Both are loading/training issues, but one is a ZeRO-3 IndexError and the other is a Qwen3 MoE FP8 config failure." + }, + { + "left": "issue:42491", + "right": "issue:44368", + "accept": false, + "reason": "LoRA/Qwen3 MoE compatibility issue vs a tie_word_embeddings warning; too indirect to merge." + }, + { + "left": "issue:45685", + "right": "issue:45823", + "accept": false, + "reason": "MPS histogram limitation vs Gemma4 device mismatch; different device/runtime bugs." + }, + { + "left": "issue:43975", + "right": "issue:44451", + "accept": false, + "reason": "Different tokenizer/model-specific regressions." + }, + { + "left": "issue:43824", + "right": "issue:43872", + "accept": false, + "reason": "ImportError for Qwen2.5 VL vs bitsandbytes Int8Params kwarg incompatibility; unrelated." + }, + { + "left": "issue:45020", + "right": "issue:45823", + "accept": false, + "reason": "Broad remote_code breakage vs a Gemma4 PLE device-mapping issue; not the same bug." + }, + { + "left": "issue:44991", + "right": "issue:45399", + "accept": false, + "reason": "Tokenizer loading regression vs flash-attn fallback blocking; no shared concrete failure." + }, + { + "left": "issue:44661", + "right": "issue:45020", + "accept": false, + "reason": "Tokenizer-mapping registration bug vs broader remote_code regressions; not the same underlying change." + }, + { + "left": "issue:43828", + "right": "issue:45823", + "accept": false, + "reason": "Autocast dtype mismatch vs device mismatch; different execution-path errors." + }, + { + "left": "issue:45020", + "right": "issue:45356", + "accept": false, + "reason": "Broad remote_code issue vs a specific tokenizer codec/warning regression." + }, + { + "left": "issue:43782", + "right": "issue:44960", + "accept": false, + "reason": "Different model-loading failures with no evidence of a shared root cause." + }, + { + "left": "issue:45081", + "right": "issue:45399", + "accept": false, + "reason": "Mistral tokenizer regex crash vs flash-attn fallback logic; unrelated issues." + }, + { + "left": "issue:43827", + "right": "issue:44509", + "accept": true, + "reason": "Same docs cleanup after v5 pipeline removals; both report stale pipeline references in documentation." + }, + { + "left": "issue:43404", + "right": "issue:43950", + "accept": false, + "reason": "Weight-tying/model head issue vs non-persistent buffer corruption; different code paths." + }, + { + "left": "issue:43792", + "right": "issue:44220", + "accept": false, + "reason": "Whisper failing to run and an fbank feature-extraction issue are not clearly the same concrete bug." + }, + { + "left": "issue:43742", + "right": "issue:45399", + "accept": false, + "reason": "Different loading/runtime failures for unrelated models." + }, + { + "left": "issue:42915", + "right": "issue:45020", + "accept": false, + "reason": "Qwen3 MoE FP8 config failure vs broad remote_code regressions; not a duplicate." + }, + { + "left": "issue:43638", + "right": "issue:45823", + "accept": false, + "reason": "ZeRO-3 IndexError vs Gemma4 device-mapping mismatch; different failures." + }, + { + "left": "issue:43257", + "right": "issue:43866", + "accept": false, + "reason": "Qwen3 MoE weight conversion problem vs a corrupted checkpoint report; similar family, different failure." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The SigLIP2 and docs/pipeline-removal pockets are grounded, but the summary overstates the evidence by claiming a separate DeepSpeed ZeRO-3 cluster that is not actually supported by the packet's soft pairs or accepted verdicts. Keep the clustering summary limited to the pockets with explicit evidence in the packet." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 4, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:24643", + "issue:30064", + "issue:36010", + "issue:36331", + "issue:38175", + "issue:41093", + "issue:41553", + "issue:42617", + "issue:42915", + "issue:43257", + "issue:43366", + "issue:43531", + "issue:43550", + "issue:43632", + "issue:43645", + "issue:43646", + "issue:43824", + "issue:43828", + "issue:43854", + "issue:43872", + "issue:43940", + "issue:43950", + "issue:43957", + "issue:43994", + "issue:44155", + "issue:44291", + "issue:44479", + "issue:44488", + "issue:44530", + "issue:44560", + "issue:44589", + "issue:44661", + "issue:44805", + "issue:44863", + "issue:44910", + "issue:44918", + "issue:44928", + "issue:44991", + "issue:45161", + "issue:45305", + "issue:45313", + "issue:45325", + "issue:45538", + "issue:45584", + "issue:45588", + "issue:45663", + "issue:45823", + "issue:45834" + ], + "soft_pairs": [ + "issue:43940|issue:45313", + "issue:43872|issue:44291", + "issue:45161|issue:45823", + "issue:43854|issue:44863", + "issue:41553|issue:44991", + "issue:42915|issue:43872", + "issue:44928|issue:45588", + "issue:43257|issue:45305", + "issue:36331|issue:41093", + "issue:41553|issue:44488", + "issue:43872|issue:45538", + "issue:44530|issue:44863", + "issue:41093|issue:44805", + "issue:44479|issue:44560", + "issue:24643|issue:30064", + "issue:43366|issue:43828", + "issue:45663|issue:45823", + "issue:43366|issue:43531", + "issue:43950|issue:45834", + "issue:43550|issue:44910", + "issue:43645|issue:43646", + "issue:43366|issue:44589", + "issue:43950|issue:45325", + "issue:44560|issue:44918", + "issue:42617|issue:43366", + "issue:43824|issue:43957", + "issue:42915|issue:43957", + "issue:44661|issue:45823", + "issue:36010|issue:42915", + "issue:38175|issue:43994", + "issue:43632|issue:45834", + "issue:44155|issue:45584" + ], + "prepared_review_unit_hash": "e0e825cc8aa59d66b2a8fc5afd5352843599b5a487604b70542677d771151cec", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37425, + "estimated_input_tokens": 9357, + "estimated_eval_tokens": 18970 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly heterogeneous cluster with a few tight duplicate subclusters around Transformers v5 init/loading regressions and a couple of near-exact symptom repeats. The SigLIP2 pair should stay separate on this evidence packet: related symptoms, but no confirmed shared root cause or duplicate-level proof.", + "confidence": 0.77, + "canonical_issue_reason": "issue:43632 is the best anchor because it is the broadest central regression here: a Transformers v5 init/loading break that overlaps with several nearby loading failures.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43632 is the strongest single representative of the cluster\u2019s main theme: loading/initialization regressions tied to the Transformers v5 `_is_hf_initialized` change.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43940", + "right": "issue:45313", + "accept": false, + "reason": "Both are DeepSpeed weight-loading problems, but they affect different Qwen variants and failure modes." + }, + { + "left": "issue:43872", + "right": "issue:44291", + "accept": true, + "reason": "Same underlying `_is_hf_initialized` regression: a new Transformers v5 kwarg breaks constructor/loading paths in different surfaces." + }, + { + "left": "issue:45161", + "right": "issue:45823", + "accept": false, + "reason": "Same broad model area, but these are different bugs: TP/GPT-OSS MoE vs Gemma4 device-mismatch in forward." + }, + { + "left": "issue:43854", + "right": "issue:44863", + "accept": false, + "reason": "Different model-loading failures for different architectures; no clear shared code path." + }, + { + "left": "issue:41553", + "right": "issue:44991", + "accept": false, + "reason": "Both concern tokenizers, but one is a bad error message and the other is a loading regression for a different model." + }, + { + "left": "issue:42915", + "right": "issue:43872", + "accept": false, + "reason": "Quantization/FP8 failure vs `_is_hf_initialized` constructor incompatibility; different root causes." + }, + { + "left": "issue:44928", + "right": "issue:45588", + "accept": false, + "reason": "Both mention attention/video-ish behavior, but the concrete failures and code paths are unrelated." + }, + { + "left": "issue:43257", + "right": "issue:45305", + "accept": false, + "reason": "Both involve DeepSpeed, but one is missing MOE weight conversion and the other is GAS gradient averaging." + }, + { + "left": "issue:36331", + "right": "issue:41093", + "accept": false, + "reason": "Trainer API regression vs tensor-mask shape mismatch; unrelated bugs." + }, + { + "left": "issue:41553", + "right": "issue:44488", + "accept": false, + "reason": "Both are loading-related, but they are different tokenizer/model issues with different failure mechanisms." + }, + { + "left": "issue:43872", + "right": "issue:45538", + "accept": false, + "reason": "`_is_hf_initialized` constructor issue vs CLIPTokenizer max-length constant bug; unrelated." + }, + { + "left": "issue:44530", + "right": "issue:44863", + "accept": false, + "reason": "PagedAttention cache crash vs checkpoint-loading failure for NemotronH; different bugs." + }, + { + "left": "issue:41093", + "right": "issue:44805", + "accept": true, + "reason": "Same concrete mask/tensor shape-mismatch error; the later report is a repeat of the same bug class." + }, + { + "left": "issue:44479", + "right": "issue:44560", + "accept": false, + "reason": "Both are Qwen video regressions, but the titles describe different concrete failures and model scopes." + }, + { + "left": "issue:24643", + "right": "issue:30064", + "accept": false, + "reason": "Unrelated: DeepSpeed training error vs image processor segmentation-map bug." + }, + { + "left": "issue:43366", + "right": "issue:43828", + "accept": false, + "reason": "GGUF support request vs autocast dtype mismatch; no shared bug." + }, + { + "left": "issue:45663", + "right": "issue:45823", + "accept": false, + "reason": "Same model family, but FSDP2 shared_kv_states rebuild vs device_map auto mismatch are different issues." + }, + { + "left": "issue:43366", + "right": "issue:43531", + "accept": false, + "reason": "GGUF gpt-oss support and Qwen3-MoE sliding-window handling are unrelated." + }, + { + "left": "issue:43950", + "right": "issue:45834", + "accept": false, + "reason": "Non-persistent buffer corruption vs long OCR index error; unrelated." + }, + { + "left": "issue:43550", + "right": "issue:44910", + "accept": false, + "reason": "Different attention/compile failures on different models and code paths." + }, + { + "left": "issue:43645", + "right": "issue:43646", + "accept": true, + "reason": "Near-identical Transformers 5.0 custom-model initialization regression; same underlying breakage with duplicate reporting." + }, + { + "left": "issue:43366", + "right": "issue:44589", + "accept": false, + "reason": "GGUF support request vs missing Float8 storage type; unrelated." + }, + { + "left": "issue:43950", + "right": "issue:45325", + "accept": false, + "reason": "Different regressions: buffer corruption in from_pretrained vs Qwen2.5-VL rope-index scaling." + }, + { + "left": "issue:44560", + "right": "issue:44918", + "accept": false, + "reason": "Both are Qwen-related, but one is a video StopIteration bug and the other is an SFT embedding unpacking failure." + }, + { + "left": "issue:42617", + "right": "issue:43366", + "accept": false, + "reason": "3D parallel script failure vs GGUF support request; unrelated." + }, + { + "left": "issue:43824", + "right": "issue:43957", + "accept": false, + "reason": "Import error for a specific class vs meta-device loading regression; different issues." + }, + { + "left": "issue:42915", + "right": "issue:43957", + "accept": false, + "reason": "FP8 config failure vs meta-device loading regression; different code paths." + }, + { + "left": "issue:44661", + "right": "issue:45823", + "accept": false, + "reason": "Tokenizer-mapping helper bug vs Gemma4 device mismatch; unrelated." + }, + { + "left": "issue:36010", + "right": "issue:42915", + "accept": false, + "reason": "Import error for GenerationMixin vs Qwen3Moe FP8 failure; unrelated." + }, + { + "left": "issue:38175", + "right": "issue:43994", + "accept": false, + "reason": "Both are SigLIP2 output-quality regressions, but the packet does not establish that they are duplicates or share a confirmed root cause." + }, + { + "left": "issue:43632", + "right": "issue:45834", + "accept": false, + "reason": "Core Transformers v5 init-flag regression vs Kosmos2.5 long-input OCR index error; unrelated." + }, + { + "left": "issue:44155", + "right": "issue:45584", + "accept": false, + "reason": "AudioFlamingo3 batch leakage bug vs Whisper empty-transcription failure; unrelated." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "Mostly grounded, but one soft-edge verdict is too strong for the evidence packet: accepting issue:38175 \u2194 issue:43994 as the same underlying bug goes beyond what the titles support. They are both SigLIP2 output-quality regressions, but the packet does not show they are duplicates or share a confirmed root cause. The rest of the cluster framing is conservative and reasonable, but this over-acceptance means I can\u2019t approve the evaluation as written." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 5, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:36010", + "issue:38175", + "issue:39404", + "issue:39692", + "issue:41720", + "issue:42491", + "issue:42915", + "issue:43257", + "issue:43317", + "issue:43404", + "issue:43454", + "issue:43596", + "issue:43638", + "issue:43644", + "issue:43645", + "issue:43708", + "issue:43720", + "issue:43854", + "issue:43856", + "issue:43873", + "issue:43931", + "issue:43994", + "issue:44075", + "issue:44387", + "issue:44451", + "issue:44488", + "issue:44534", + "issue:44560", + "issue:44661", + "issue:44756", + "issue:44805", + "issue:44877", + "issue:44910", + "issue:44960", + "issue:44977", + "issue:44991", + "issue:45030", + "issue:45276", + "issue:45292", + "issue:45313", + "issue:45335", + "issue:45357", + "issue:45362", + "issue:45406", + "issue:45488", + "issue:45588", + "issue:45812" + ], + "soft_pairs": [ + "issue:44877|issue:45030", + "issue:44560|issue:44805", + "issue:43638|issue:44661", + "issue:43317|issue:44756", + "issue:43596|issue:44661", + "issue:44661|issue:44805", + "issue:38175|issue:39692", + "issue:43644|issue:44534", + "issue:45313|issue:45357", + "issue:45488|issue:45812", + "issue:44960|issue:44977", + "issue:44451|issue:44488", + "issue:42491|issue:43720", + "issue:43854|issue:43931", + "issue:41720|issue:42915", + "issue:43708|issue:44075", + "issue:42491|issue:45276", + "issue:43257|issue:43854", + "issue:43856|issue:43873", + "issue:42915|issue:44756", + "issue:39692|issue:43994", + "issue:42491|issue:43931", + "issue:43856|issue:44387", + "issue:43644|issue:43645", + "issue:44910|issue:45588", + "issue:45362|issue:45406", + "issue:42491|issue:45292", + "issue:44488|issue:44991", + "issue:36010|issue:39404", + "issue:45276|issue:45335", + "issue:43404|issue:43454" + ], + "prepared_review_unit_hash": "f2a83a50b2e6438511ed909e6a28227a033cd2a5b2f37ffc668f394923db9fab", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36412, + "estimated_input_tokens": 9103, + "estimated_eval_tokens": 18462 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous, but there are two clear duplicate families: the v5 non-persistent buffer junk regression and the resize_token_embeddings/output_embeddings regression. Most other pairs are only loosely related by subsystem or model family and should stay separate.", + "confidence": 0.74, + "canonical_issue_reason": "issue:45292 is the best umbrella issue for the resize_token_embeddings regression: it is the most generic title and cleanly covers the narrower Gemma4/T5Gemma variants.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45292 is the strongest cluster anchor among the issues shown because it represents a broad, recurring code-path bug rather than a model-specific symptom.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44877", + "right": "issue:45030", + "accept": false, + "reason": "Both are config-validation load failures, but they affect different model families and likely different validation rules." + }, + { + "left": "issue:44560", + "right": "issue:44805", + "accept": false, + "reason": "Different symptoms and code paths: video StopIteration in Qwen3-VL embedding versus a mask/tensor shape IndexError." + }, + { + "left": "issue:43638", + "right": "issue:44661", + "accept": false, + "reason": "One is a ZeRO-3/BertModel indexing failure; the other is an add-new-model-like tokenizer mapping failure." + }, + { + "left": "issue:43317", + "right": "issue:44756", + "accept": false, + "reason": "GPU+CPU offload load failure is unrelated to disabling mmap to avoid OOM." + }, + { + "left": "issue:43596", + "right": "issue:44661", + "accept": false, + "reason": "Different failure modes: ZeRO-3 with BertModel versus tokenizer mapping logic." + }, + { + "left": "issue:44661", + "right": "issue:44805", + "accept": false, + "reason": "Tokenizer mapping bug and tensor mask shape mismatch are unrelated." + }, + { + "left": "issue:38175", + "right": "issue:39692", + "accept": false, + "reason": "Same model family, but one is zero probabilities and the other is a docs/example mismatch plus quantization failure; not the same concrete bug." + }, + { + "left": "issue:43644", + "right": "issue:44534", + "accept": true, + "reason": "These describe the same Transformers v5 regression where non-persistent buffers get filled with junk." + }, + { + "left": "issue:45313", + "right": "issue:45357", + "accept": false, + "reason": "Different Qwen3.5 issues: ZeRO-3 loading of language_model weights versus incorrect visual encoder keys on save_pretrained." + }, + { + "left": "issue:45488", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different model families and different tokenization failures." + }, + { + "left": "issue:44960", + "right": "issue:44977", + "accept": false, + "reason": "Unrelated models and failures: GLM5 versus Qwen3.5 flash-attention generation." + }, + { + "left": "issue:44451", + "right": "issue:44488", + "accept": false, + "reason": "Both are model loading failures, but they involve different models and no shared concrete failure is evident." + }, + { + "left": "issue:42491", + "right": "issue:43720", + "accept": false, + "reason": "Different problems: hf4.x to hf5.x LoRA/Qwen3 MoE incompatibility versus BitNet packed-weight unpacking during accelerate load." + }, + { + "left": "issue:43854", + "right": "issue:43931", + "accept": false, + "reason": "Different models and different loading errors; no same underlying bug is evident." + }, + { + "left": "issue:41720", + "right": "issue:42915", + "accept": false, + "reason": "Both involve Qwen3, but one is device-mapping cudaErrorAssert and the other is FineGrainedFP8Config failure." + }, + { + "left": "issue:43708", + "right": "issue:44075", + "accept": false, + "reason": "Trainer resume/max_steps math is unrelated to SGD arguments being ignored." + }, + { + "left": "issue:42491", + "right": "issue:45276", + "accept": false, + "reason": "Different model families and failure modes; one is Qwen3 MoE compatibility, the other is resize_token_embeddings propagation." + }, + { + "left": "issue:43257", + "right": "issue:43854", + "accept": false, + "reason": "Both concern loading, but one is Qwen3 MoE weight conversion and the other is a GLM-4.7-Flash unit-test load issue." + }, + { + "left": "issue:43856", + "right": "issue:43873", + "accept": false, + "reason": "Training memory usage and quantization offloading are only broadly related, not the same bug." + }, + { + "left": "issue:42915", + "right": "issue:44756", + "accept": false, + "reason": "Qwen3 MoE FP8 failure is unrelated to Strix Halo mmap OOM mitigation." + }, + { + "left": "issue:39692", + "right": "issue:43994", + "accept": false, + "reason": "Both mention SigLIP2, but one is a documentation/example error and the other is a model/pipeline output regression." + }, + { + "left": "issue:42491", + "right": "issue:43931", + "accept": false, + "reason": "Different Qwen3-related loading bugs with no shared concrete code path shown." + }, + { + "left": "issue:43856", + "right": "issue:44387", + "accept": false, + "reason": "Different memory issues: MoE training memory usage versus int4 quantization reserved-memory OOM." + }, + { + "left": "issue:43644", + "right": "issue:43645", + "accept": false, + "reason": "Both are Transformers v5 regressions, but one is about non-persistent buffers and the other about custom model initialization in notebooks." + }, + { + "left": "issue:44910", + "right": "issue:45588", + "accept": false, + "reason": "Both touch flash-attention, but one is a packed-sequence position_ids crash and the other is an s_aux=None AttributeError." + }, + { + "left": "issue:45362", + "right": "issue:45406", + "accept": false, + "reason": "Different serving/chat crashes and different model classes." + }, + { + "left": "issue:44488", + "right": "issue:44991", + "accept": false, + "reason": "Tokenizer load failures, but for different models and likely different root causes." + }, + { + "left": "issue:45276", + "right": "issue:45335", + "accept": true, + "reason": "Both report resize_token_embeddings failing to update output/decoder embeddings; 45335 is the narrower T5Gemma variant of the same bug." + }, + { + "left": "issue:43404", + "right": "issue:43454", + "accept": false, + "reason": "Both are weight-tying problems, but they affect different model classes and are not clearly the same concrete bug." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative overall. The two accepted pairs are strongly supported by the titles: issue:43644/44534 are the same non-persistent buffer junk regression, and issue:45276/45335 are the same resize_token_embeddings/output_embeddings bug family. The rest of the soft-edge verdicts are appropriately cautious and do not overclaim evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 6, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:39404", + "issue:39692", + "issue:41720", + "issue:42947", + "issue:43257", + "issue:43335", + "issue:43381", + "issue:43425", + "issue:43611", + "issue:43645", + "issue:43646", + "issue:43708", + "issue:43828", + "issue:43854", + "issue:43873", + "issue:43906", + "issue:43931", + "issue:43957", + "issue:44155", + "issue:44164", + "issue:44291", + "issue:44292", + "issue:44387", + "issue:44402", + "issue:44451", + "issue:44479", + "issue:44488", + "issue:44530", + "issue:44589", + "issue:44683", + "issue:44756", + "issue:44863", + "issue:44877", + "issue:44912", + "issue:44928", + "issue:44960", + "issue:45216", + "issue:45276", + "issue:45292", + "issue:45335", + "issue:45588", + "issue:45663", + "issue:45684", + "issue:45698", + "issue:45701", + "issue:45704", + "issue:45759", + "issue:45823" + ], + "soft_pairs": [ + "issue:44164|issue:44291", + "issue:44402|issue:44488", + "issue:43425|issue:43957", + "issue:39692|issue:44960", + "issue:42947|issue:45704", + "issue:43335|issue:45663", + "issue:43873|issue:44387", + "issue:45663|issue:45759", + "issue:44155|issue:45823", + "issue:43854|issue:45216", + "issue:44387|issue:44756", + "issue:44291|issue:44756", + "issue:44292|issue:44912", + "issue:42947|issue:43381", + "issue:45276|issue:45292", + "issue:45684|issue:45698", + "issue:43381|issue:44928", + "issue:44164|issue:44756", + "issue:41720|issue:44155", + "issue:45588|issue:45663", + "issue:44479|issue:44877", + "issue:43646|issue:43708", + "issue:43854|issue:44960", + "issue:43957|issue:44756", + "issue:45292|issue:45335", + "issue:44402|issue:44451", + "issue:43828|issue:43957", + "issue:43645|issue:44164", + "issue:43257|issue:44863", + "issue:43425|issue:43828", + "issue:44589|issue:44756", + "issue:44683|issue:45823", + "issue:44479|issue:45663", + "issue:43906|issue:44530", + "issue:39404|issue:45701", + "issue:43611|issue:43645", + "issue:43425|issue:45823", + "issue:43931|issue:44863" + ], + "prepared_review_unit_hash": "c33d2c8c9ee12860070b42abffa5b69b45015f4a329947c72d3b08d35119f4ed", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 38, + "serialized_chars": 39075, + "estimated_input_tokens": 9769, + "estimated_eval_tokens": 19794 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous. The only convincing duplicate-style pair is the `resize_token_embeddings` regression around output embeddings; the rest are distinct bugs affecting different models/code paths.", + "confidence": 0.83, + "canonical_issue_reason": "issue:45292 is the broad, generic report for the `resize_token_embeddings` regression; issue:45276 is a model-specific Gemma4 repro of the same output-embedding problem.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45292 is the clearest representative issue because it states the underlying bug in the most general form and can subsume the more specific Gemma4 variant.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44164", + "right": "issue:44291", + "accept": false, + "reason": "Different bugs: save/from_pretrained extra_state handling vs `init_empty_weights` passing an unexpected `_is_hf_initialized` arg." + }, + { + "left": "issue:44402", + "right": "issue:44488", + "accept": false, + "reason": "Both are model-loading failures, but for different models and different root causes." + }, + { + "left": "issue:43425", + "right": "issue:43957", + "accept": false, + "reason": "Torch version incompatibility is not the same as `torch.device(\"meta\")` loading breakage." + }, + { + "left": "issue:39692", + "right": "issue:44960", + "accept": false, + "reason": "Documentation/example errors vs a GLM5 model issue; no shared underlying defect." + }, + { + "left": "issue:42947", + "right": "issue:45704", + "accept": false, + "reason": "Gradient checkpointing/LoRA behavior is unrelated to the apex RMSNorm memory leak." + }, + { + "left": "issue:43335", + "right": "issue:45663", + "accept": false, + "reason": "Different model families and failures: SwitchTransformers sparse-layer config vs Gemma4 FSDP2 `KeyError`." + }, + { + "left": "issue:43873", + "right": "issue:44387", + "accept": false, + "reason": "Quantization/offloading behavior and int4 CUDA reserved-memory OOM are related only broadly, not the same bug." + }, + { + "left": "issue:45663", + "right": "issue:45759", + "accept": false, + "reason": "Gemma4 FSDP2 attention-state rebuild is unrelated to Qwen multimodal `text_config` unwrapping." + }, + { + "left": "issue:44155", + "right": "issue:45823", + "accept": false, + "reason": "AudioFlamingo3 batch leakage and Gemma4 device-map mismatch are different code paths." + }, + { + "left": "issue:43854", + "right": "issue:45216", + "accept": false, + "reason": "Different regressions: unit-test loading for GLM-4.7-Flash vs Qwen3.5 save_pretrained correctness." + }, + { + "left": "issue:44387", + "right": "issue:44756", + "accept": false, + "reason": "Int4 OOM from reserved memory growth is not the same as disabling mmap on Strix Halo." + }, + { + "left": "issue:44291", + "right": "issue:44756", + "accept": false, + "reason": "Unexpected init arg during meta init vs mmap/OOM on Strix Halo are unrelated." + }, + { + "left": "issue:44292", + "right": "issue:44912", + "accept": false, + "reason": "Qwen3-VL shape mismatch and MXFP4 fallback to bf16 are different model-loading failures." + }, + { + "left": "issue:42947", + "right": "issue:43381", + "accept": false, + "reason": "Both mention gradient checkpointing, but one is PEFT LoRA effectiveness and the other is eval-mode prohibition." + }, + { + "left": "issue:45276", + "right": "issue:45292", + "accept": true, + "reason": "Same underlying `resize_token_embeddings` bug: the generic output-embedding regression in 45292 matches the Gemma4-specific repro in 45276." + }, + { + "left": "issue:45684", + "right": "issue:45698", + "accept": false, + "reason": "Read-only permission propagation in save_pretrained is a different bug from loading the wrong custom module after save_pretrained." + }, + { + "left": "issue:43381", + "right": "issue:44928", + "accept": false, + "reason": "Eval-mode checkpointing restriction is unrelated to the RLHF NaN/SDPA fallback issue." + }, + { + "left": "issue:44164", + "right": "issue:44756", + "accept": false, + "reason": "extra_state save/load handling is unrelated to Strix Halo mmap memory pressure." + }, + { + "left": "issue:41720", + "right": "issue:44155", + "accept": false, + "reason": "Qwen3 device-map cuda assert and AudioFlamingo3 batch token leakage are different failures." + }, + { + "left": "issue:45588", + "right": "issue:45663", + "accept": false, + "reason": "Flash-attention crash on `s_aux=None` is not the same as Gemma4 FSDP2 shared-state rebuilding." + }, + { + "left": "issue:44479", + "right": "issue:44877", + "accept": false, + "reason": "Qwen video-input regression and strict-config loading for granite_speech are different issues." + }, + { + "left": "issue:43646", + "right": "issue:43708", + "accept": false, + "reason": "Custom model initialization in notebooks is not the same as incorrect `max_steps` on checkpoint resume." + }, + { + "left": "issue:43854", + "right": "issue:44960", + "accept": false, + "reason": "GLM-4.7-Flash loading failure and GLM5 are different model-specific reports." + }, + { + "left": "issue:43957", + "right": "issue:44756", + "accept": false, + "reason": "Meta-device loading breakage is unrelated to mmap being disabled on Strix Halo." + }, + { + "left": "issue:45292", + "right": "issue:45335", + "accept": false, + "reason": "Both are `resize_token_embeddings` regressions, but they hit different components (`output_embeddings` vs `decoder.embed_tokens`) and don't clearly share one exact code-path bug." + }, + { + "left": "issue:44402", + "right": "issue:44451", + "accept": false, + "reason": "Different model-loading failures for different checkpoints; too model-specific to merge." + }, + { + "left": "issue:43828", + "right": "issue:43957", + "accept": false, + "reason": "Dtype mismatch under autocast is unrelated to meta-device loading breakage." + }, + { + "left": "issue:43645", + "right": "issue:44164", + "accept": false, + "reason": "Custom-model notebook initialization and extra_state save/load are different problems." + }, + { + "left": "issue:43257", + "right": "issue:44863", + "accept": false, + "reason": "Qwen3 MoE deepspeed conversion and NemotronH checkpoint loading are unrelated." + }, + { + "left": "issue:43425", + "right": "issue:43828", + "accept": false, + "reason": "Torch compatibility complaints and an autocast dtype mismatch are separate failures." + }, + { + "left": "issue:44589", + "right": "issue:44756", + "accept": false, + "reason": "Float8 storage lookup error and Strix Halo mmap OOM are unrelated." + }, + { + "left": "issue:44683", + "right": "issue:45823", + "accept": false, + "reason": "Compiled flex_attention on torch>=2.9 and Gemma4 PLE device mismatch are different bugs." + }, + { + "left": "issue:44479", + "right": "issue:45663", + "accept": false, + "reason": "Video-input regression in Qwen models is unrelated to Gemma4 FSDP2 attention-state rebuilding." + }, + { + "left": "issue:43906", + "right": "issue:44530", + "accept": false, + "reason": "An isolated reproduction of a different issue is not the same as the Qwen3.5 PagedAttentionCache crash." + }, + { + "left": "issue:39404", + "right": "issue:45701", + "accept": false, + "reason": "Whisper `return_language` pipeline behavior and tokenizer-version drift are unrelated." + }, + { + "left": "issue:43611", + "right": "issue:43645", + "accept": false, + "reason": "Both concern Transformers 5.0 custom-model breakage, but the failure modes are different (`base_model_prefix` loading vs notebook redefinition/init issues)." + }, + { + "left": "issue:43425", + "right": "issue:45823", + "accept": false, + "reason": "Torch 2.10 incompatibility is not the same as the Gemma4 PLE device mismatch." + }, + { + "left": "issue:43931", + "right": "issue:44863", + "accept": false, + "reason": "Qwen3-VL weight-shape mismatch and NemotronH checkpoint loading are different model-loading bugs." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The cluster summary is grounded and appropriately conservative. The only accepted duplicate-style pair, issue:45276 with issue:45292, is well supported by the titles and high similarity around the same `resize_token_embeddings`/embedding regression. The other verdicts stay cautious by rejecting pairs that only share broad themes (e.g. model loading, custom-model breakage, or checkpointing) without clear evidence of the same root cause. " + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 7, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:32090", + "issue:34567", + "issue:36032", + "issue:39404", + "issue:40444", + "issue:40990", + "issue:42947", + "issue:43299", + "issue:43421", + "issue:43425", + "issue:43452", + "issue:43582", + "issue:43596", + "issue:43611", + "issue:43632", + "issue:43643", + "issue:43644", + "issue:43716", + "issue:43856", + "issue:43883", + "issue:43901", + "issue:43931", + "issue:43994", + "issue:44079", + "issue:44164", + "issue:44206", + "issue:44220", + "issue:44265", + "issue:44479", + "issue:44488", + "issue:44509", + "issue:44530", + "issue:44589", + "issue:44661", + "issue:44671", + "issue:44683", + "issue:44756", + "issue:44805", + "issue:44811", + "issue:44898", + "issue:44912", + "issue:44918", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45584", + "issue:45759" + ], + "soft_pairs": [ + "issue:43425|issue:43716", + "issue:44530|issue:45084", + "issue:43901|issue:44509", + "issue:39404|issue:44912", + "issue:44265|issue:44898", + "issue:43299|issue:43931", + "issue:44479|issue:44805", + "issue:43425|issue:44918", + "issue:44589|issue:45084", + "issue:43856|issue:44756", + "issue:40990|issue:42947", + "issue:39404|issue:44671", + "issue:43994|issue:44079", + "issue:44164|issue:45759", + "issue:36032|issue:43452", + "issue:43596|issue:44683", + "issue:44671|issue:44991", + "issue:44991|issue:45081", + "issue:32090|issue:39404", + "issue:44164|issue:44479", + "issue:44488|issue:45081", + "issue:44661|issue:44683", + "issue:40990|issue:43421", + "issue:43582|issue:43632", + "issue:43611|issue:43644", + "issue:44206|issue:44220", + "issue:43716|issue:45084", + "issue:43883|issue:44488", + "issue:44488|issue:44987", + "issue:43643|issue:44530", + "issue:34567|issue:40444", + "issue:44811|issue:45584" + ], + "prepared_review_unit_hash": "1163698b639cdca620e89879b4c252e7b39c452a96920a2ac760163011e477e7", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37334, + "estimated_input_tokens": 9334, + "estimated_eval_tokens": 18924 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: the soft pairs span unrelated regressions, model-loading issues, audio/Whisper bugs, quantization, docs, and training behavior. I would reject all suggested duplicate edges.", + "confidence": 0.95, + "canonical_issue_reason": "No single canonical issue stands out; the items are distinct bugs/regressions across different subsystems, so collapsing them would over-merge unrelated reports.", + "canonical_pr_reason": null, + "best_issue_reason": "No good global representative emerges because the cluster is not a duplicate set but a mix of separate issues.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43425", + "right": "issue:43716", + "accept": false, + "reason": "Unrelated problems: torch version incompatibility vs image preprocessor/model dtype mismatch." + }, + { + "left": "issue:44530", + "right": "issue:45084", + "accept": false, + "reason": "Different failures: Qwen3.5 PagedAttentionCache linear_attention crash vs template compilation error." + }, + { + "left": "issue:43901", + "right": "issue:44509", + "accept": false, + "reason": "Both are docs issues, but they concern different pipeline/task documentation changes." + }, + { + "left": "issue:39404", + "right": "issue:44912", + "accept": false, + "reason": "Whisper pipeline return_language regression is unrelated to MXFP4 quantization fallback." + }, + { + "left": "issue:44265", + "right": "issue:44898", + "accept": false, + "reason": "Different code paths: torch.export with torch_compilable_check vs Perceiver non-default resolution interpolation." + }, + { + "left": "issue:43299", + "right": "issue:43931", + "accept": false, + "reason": "Both involve Qwen3-VL loading, but one is a v5 regression for MoE models and the other is a weight-shape mismatch on a specific checkpoint." + }, + { + "left": "issue:44479", + "right": "issue:44805", + "accept": false, + "reason": "Video-input regression for Qwen VL models is not the same as a generic mask/index shape mismatch." + }, + { + "left": "issue:43425", + "right": "issue:44918", + "accept": false, + "reason": "Torch incompatibility is unrelated to TRL SFT embedding unpacking for Qwen3.5." + }, + { + "left": "issue:44589", + "right": "issue:45084", + "accept": false, + "reason": "Float8 storage lookup failure is unrelated to a template-node compilation error." + }, + { + "left": "issue:43856", + "right": "issue:44756", + "accept": false, + "reason": "Both mention memory, but one is Qwen3 MoE training efficiency and the other is Strix Halo mmap OOM." + }, + { + "left": "issue:40990", + "right": "issue:42947", + "accept": false, + "reason": "High perplexity on a model/dataset is unrelated to LoRA gradient checkpointing ineffectiveness." + }, + { + "left": "issue:39404", + "right": "issue:44671", + "accept": false, + "reason": "Whisper pipeline behavior and CamemBERT masked LM predictions are separate model-family bugs." + }, + { + "left": "issue:43994", + "right": "issue:44079", + "accept": false, + "reason": "Incorrect SigLIP outputs are not the same as ModelOutput key assignment when values were previously None." + }, + { + "left": "issue:44164", + "right": "issue:45759", + "accept": false, + "reason": "save/from_pretrained extra_state handling is unrelated to AutoModelForCausalLM.from_config text_config unwrapping." + }, + { + "left": "issue:36032", + "right": "issue:43452", + "accept": false, + "reason": "Tokenizer special-token method conflict is unrelated to gguf_file loading failures." + }, + { + "left": "issue:43596", + "right": "issue:44683", + "accept": false, + "reason": "DeepSpeed ZeRO-3/BertModel initialization error is unrelated to compiled flex_attention on newer torch." + }, + { + "left": "issue:44671", + "right": "issue:44991", + "accept": false, + "reason": "CamemBERT MLM prediction issues are unrelated to tokenizer loading for EMBEDDIA/est-roberta." + }, + { + "left": "issue:44991", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer load failure and Mistral regex patch crash are different model-specific bugs." + }, + { + "left": "issue:32090", + "right": "issue:39404", + "accept": false, + "reason": "Trainer GPU broadcast NoneType error is unrelated to Whisper return_language behavior." + }, + { + "left": "issue:44164", + "right": "issue:44479", + "accept": false, + "reason": "Checkpoint save/load extra_state handling is unrelated to Qwen video-input regressions." + }, + { + "left": "issue:44488", + "right": "issue:45081", + "accept": false, + "reason": "Loading cjvt/sleng-bert and patching Mistral regex are unrelated tokenizer issues." + }, + { + "left": "issue:44661", + "right": "issue:44683", + "accept": false, + "reason": "Model registration in TOKENIZER_MAPPING_NAMES is unrelated to compiled flex_attention on torch >= 2.9." + }, + { + "left": "issue:40990", + "right": "issue:43421", + "accept": false, + "reason": "Perplexity regression on gpt-oss-20b is unrelated to tokenizers backend post-processor updates." + }, + { + "left": "issue:43582", + "right": "issue:43632", + "accept": false, + "reason": "Apple Silicon caching_allocator_warmup TypeError is unrelated to the _is_hf_initialized flag regression." + }, + { + "left": "issue:43611", + "right": "issue:43644", + "accept": false, + "reason": "Base_model_prefix loading bug and non-persistent buffer junk are different Transformers v5 regressions." + }, + { + "left": "issue:44206", + "right": "issue:44220", + "accept": false, + "reason": "Unsupported center argument in LasrFeatureExtractor is not the same as a generic _torch_extract_fbank_features issue." + }, + { + "left": "issue:43716", + "right": "issue:45084", + "accept": false, + "reason": "Mistral-3 dtype mismatch and non-template-node compilation failure are unrelated." + }, + { + "left": "issue:43883", + "right": "issue:44488", + "accept": false, + "reason": "Missing all_tied_weights_keys on Molmo is unrelated to sleng-bert loading failure." + }, + { + "left": "issue:44488", + "right": "issue:44987", + "accept": false, + "reason": "sleng-bert tokenizer loading and physical-intelligence/fast loading failures are different model-specific issues." + }, + { + "left": "issue:43643", + "right": "issue:44530", + "accept": false, + "reason": "trust_remote_code AutoConfig field loss is unrelated to Qwen3.5 PagedAttentionCache crashes." + }, + { + "left": "issue:34567", + "right": "issue:40444", + "accept": false, + "reason": "TrainerState num_input_tokens_seen not updating is unrelated to iterable-dataset multi-image fine-tuning failures." + }, + { + "left": "issue:44811", + "right": "issue:45584", + "accept": false, + "reason": "Ignoring skip_special_tokens in batch_decode is a different Whisper bug from empty-transcription generation after align_special_tokens." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 8, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:36032", + "issue:42915", + "issue:43262", + "issue:43296", + "issue:43317", + "issue:43322", + "issue:43377", + "issue:43404", + "issue:43452", + "issue:43482", + "issue:43531", + "issue:43540", + "issue:43541", + "issue:43611", + "issue:43632", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43742", + "issue:43819", + "issue:43856", + "issue:43883", + "issue:43994", + "issue:44062", + "issue:44220", + "issue:44265", + "issue:44291", + "issue:44403", + "issue:44451", + "issue:44466", + "issue:44530", + "issue:44534", + "issue:44589", + "issue:44610", + "issue:44683", + "issue:44756", + "issue:44849", + "issue:44877", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45127", + "issue:45216", + "issue:45325", + "issue:45357", + "issue:45507" + ], + "soft_pairs": [ + "issue:43644|issue:43646", + "issue:44451|issue:44987", + "issue:36032|issue:42915", + "issue:43541|issue:45084", + "issue:43452|issue:43482", + "issue:42915|issue:44530", + "issue:43317|issue:43482", + "issue:44849|issue:45325", + "issue:43883|issue:44991", + "issue:43404|issue:44466", + "issue:36032|issue:43645", + "issue:45084|issue:45507", + "issue:43262|issue:43377", + "issue:43646|issue:44534", + "issue:44987|issue:44991", + "issue:43994|issue:44610", + "issue:43322|issue:43540", + "issue:44403|issue:45081", + "issue:43377|issue:43819", + "issue:42915|issue:45084", + "issue:43645|issue:44534", + "issue:44451|issue:45081", + "issue:45216|issue:45357", + "issue:44265|issue:44610", + "issue:44987|issue:45081", + "issue:44220|issue:44683", + "issue:43404|issue:45127", + "issue:43632|issue:44756", + "issue:43296|issue:43531", + "issue:42915|issue:44291", + "issue:36032|issue:43742", + "issue:44062|issue:44987", + "issue:43531|issue:44877", + "issue:42915|issue:44589", + "issue:43611|issue:44534", + "issue:43317|issue:43856" + ], + "prepared_review_unit_hash": "2fbb473898070b9e4ad3cfd7aff1c52f1202cd81bb17c6611d38ae3c5c2f4830", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 36, + "serialized_chars": 37916, + "estimated_input_tokens": 9479, + "estimated_eval_tokens": 19214 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a loose thematic cluster of mostly unrelated Transformers v5 regressions around loading, initialization, tokenizers, and model-specific edge cases. The similarities are largely topical rather than the same underlying bug, so none of the soft pairs should be merged as duplicates.", + "confidence": 0.91, + "canonical_issue_reason": "issue:43646 is the broadest and most representative of the recurring v5 loading/initialization breakages; it best serves as an umbrella among the set, even though the cluster is heterogeneous.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43646 is the best single issue to anchor the cluster because it is general, central, and closest in scope to many of the other v5 initialization/loading reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43644", + "right": "issue:43646", + "accept": false, + "reason": "Both are v5 regressions, but one is about non-persistent buffers being initialized with junk while the other is about custom model initialization; different bugs and code paths." + }, + { + "left": "issue:44451", + "right": "issue:44987", + "accept": false, + "reason": "Both are model-loading failures, but they affect different models and likely different regressions; too broad to be the same underlying bug." + }, + { + "left": "issue:36032", + "right": "issue:42915", + "accept": false, + "reason": "T5 tokenizer import/load error vs Qwen3MoE FineGrainedFP8Config failure; unrelated subsystems and failure modes." + }, + { + "left": "issue:43541", + "right": "issue:45084", + "accept": false, + "reason": "Mixtral torch dynamo tracing error and template compilation error are distinct runtime/compiler issues, not the same bug." + }, + { + "left": "issue:43452", + "right": "issue:43482", + "accept": false, + "reason": "Both mention GGUF/loading, but one is about gguf_file API handling and the other is a Qwen2.5-GGUF v5 load failure; not the same concrete defect." + }, + { + "left": "issue:42915", + "right": "issue:44530", + "accept": false, + "reason": "Qwen3MoE FineGrainedFP8Config failure and PagedAttentionCache linear_attention crash are separate Qwen-related issues with different code paths." + }, + { + "left": "issue:43317", + "right": "issue:43482", + "accept": false, + "reason": "device_map=auto offload/dequantized-model loading is unrelated to Qwen2.5-GGUF loading failure." + }, + { + "left": "issue:44849", + "right": "issue:45325", + "accept": false, + "reason": "Both are Qwen3.5/Qwen2.5-VL regressions, but one is about output_hidden_states and the other about rope/position_ids scaling; different behavior." + }, + { + "left": "issue:43883", + "right": "issue:44991", + "accept": false, + "reason": "Molmo tied-weights attribute error is unrelated to tokenizer loading for EMBEDDIA/est-roberta." + }, + { + "left": "issue:43404", + "right": "issue:44466", + "accept": false, + "reason": "Both involve tied lm_head weights, but one is a missing tie in Mistral3 and the other is device-dependent serialization inconsistency; related theme, different bugs." + }, + { + "left": "issue:36032", + "right": "issue:43645", + "accept": false, + "reason": "Tokenizer special-token conflict vs notebook custom-model initialization breakage; no same failure path." + }, + { + "left": "issue:45084", + "right": "issue:45507", + "accept": false, + "reason": "Template compilation error and GraniteMoEHybrid invalid method call are unrelated code-path failures." + }, + { + "left": "issue:43262", + "right": "issue:43377", + "accept": false, + "reason": "Audio processor chat-template sampling-rate default bug vs MIMI encoder padding-mask bug; both audio-adjacent but distinct." + }, + { + "left": "issue:43646", + "right": "issue:44534", + "accept": false, + "reason": "General custom-model initialization breakage vs non-persistent buffer junk initialization; similar release theme, not the same defect." + }, + { + "left": "issue:44987", + "right": "issue:44991", + "accept": false, + "reason": "Different load failures affecting different artifacts/models; no evidence they share the same root cause." + }, + { + "left": "issue:43994", + "right": "issue:44610", + "accept": false, + "reason": "SigLIP2 nonsensical outputs and OmDet-Turbo processor size mismatch are different model/processor bugs." + }, + { + "left": "issue:43322", + "right": "issue:43540", + "accept": false, + "reason": "Llava Next segmentation fault and Qwen3OmniMoe video ValueError are separate multimodal model failures." + }, + { + "left": "issue:44403", + "right": "issue:45081", + "accept": false, + "reason": "Generic transformer-loading noise and Mistral regex patch crash are not the same underlying issue." + }, + { + "left": "issue:43377", + "right": "issue:43819", + "accept": false, + "reason": "Missing padding mask in MIMI encoder and DAC latent mismatch/STE issue are unrelated audio-model bugs." + }, + { + "left": "issue:42915", + "right": "issue:45084", + "accept": false, + "reason": "FineGrainedFP8Config failure and template compilation failure do not share a concrete code path." + }, + { + "left": "issue:43645", + "right": "issue:44534", + "accept": false, + "reason": "Notebook custom model initialization vs non-persistent buffer junk are different regressions." + }, + { + "left": "issue:44451", + "right": "issue:45081", + "accept": false, + "reason": "Loading a specific model checkpoint and Mistral tokenizer regex patch crash are unrelated." + }, + { + "left": "issue:45216", + "right": "issue:45357", + "accept": false, + "reason": "Both are Qwen3.5 save_pretrained regressions, but one is incorrect checkpoint content generally and the other is visual encoder key serialization specifically; not the same bug." + }, + { + "left": "issue:44265", + "right": "issue:44610", + "accept": false, + "reason": "torch.export/exportability issue with torch_compilable_check is unrelated to an image processor size mismatch." + }, + { + "left": "issue:44987", + "right": "issue:45081", + "accept": false, + "reason": "Different model-loading regressions with different mechanisms and affected components." + }, + { + "left": "issue:44220", + "right": "issue:44683", + "accept": false, + "reason": "fbank feature extraction and compiled flex_attention failures are unrelated audio vs attention/compiler bugs." + }, + { + "left": "issue:43404", + "right": "issue:45127", + "accept": false, + "reason": "Tied lm_head weight bug in Mistral3 and LoRA merge collapse with extended vocab are related only at a high level; different concrete failure modes." + }, + { + "left": "issue:43632", + "right": "issue:44756", + "accept": false, + "reason": "_is_hf_initialized flag regression and mmap/OOM on Strix Halo are unrelated issues." + }, + { + "left": "issue:43296", + "right": "issue:43531", + "accept": false, + "reason": "PaddleOCR-VL load failure in vLLM and Qwen3-MoE sliding_window issue are separate model-specific problems." + }, + { + "left": "issue:42915", + "right": "issue:44291", + "accept": false, + "reason": "FineGrainedFP8Config failure and unexpected _is_hf_initialized argument in init_empty_weights are different v5 regressions." + }, + { + "left": "issue:36032", + "right": "issue:43742", + "accept": false, + "reason": "T5 tokenizer special-token conflict and MobileLLM loading KeyError are unrelated model-loading bugs." + }, + { + "left": "issue:44062", + "right": "issue:44987", + "accept": false, + "reason": "AddedToken special-argument duplication and a model load failure for physical-intelligence/fast are distinct tokenizer vs model issues." + }, + { + "left": "issue:43531", + "right": "issue:44877", + "accept": false, + "reason": "Qwen3-MoE sliding_window problem and granite_speech strict config loading issue are unrelated." + }, + { + "left": "issue:42915", + "right": "issue:44589", + "accept": false, + "reason": "FineGrainedFP8Config failure and missing Float8 storage type are different quantization/runtime issues." + }, + { + "left": "issue:43611", + "right": "issue:44534", + "accept": false, + "reason": "base_model_prefix loading breakage and non-persistent buffer junk are separate v5 regressions with different symptoms." + }, + { + "left": "issue:43317", + "right": "issue:43856", + "accept": false, + "reason": "device_map=auto dequantized offload failure is unrelated to Qwen3 MoE training memory usage." + } + ] + }, + "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": 16, + "review_unit_index": 9, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:33666", + "issue:36032", + "issue:40444", + "issue:40990", + "issue:41553", + "issue:41720", + "issue:42175", + "issue:43296", + "issue:43377", + "issue:43421", + "issue:43425", + "issue:43454", + "issue:43482", + "issue:43525", + "issue:43531", + "issue:43582", + "issue:43611", + "issue:43632", + "issue:43646", + "issue:43792", + "issue:43818", + "issue:43994", + "issue:44079", + "issue:44265", + "issue:44291", + "issue:44403", + "issue:44462", + "issue:44466", + "issue:44479", + "issue:44534", + "issue:44568", + "issue:44589", + "issue:44683", + "issue:44811", + "issue:44829", + "issue:44898", + "issue:44991", + "issue:45020", + "issue:45072", + "issue:45081", + "issue:45325", + "issue:45357", + "issue:45479", + "issue:45491", + "issue:45538", + "issue:45542", + "issue:45663", + "issue:45812" + ], + "soft_pairs": [ + "issue:41553|issue:43582", + "issue:44479|issue:45325", + "issue:40990|issue:44811", + "issue:43425|issue:44403", + "issue:42175|issue:45542", + "issue:44462|issue:45812", + "issue:43818|issue:45357", + "issue:44079|issue:45663", + "issue:36032|issue:41720", + "issue:43421|issue:44568", + "issue:43792|issue:44589", + "issue:43632|issue:44534", + "issue:44403|issue:44991", + "issue:44683|issue:45020", + "issue:44829|issue:45479", + "issue:43482|issue:45538", + "issue:33666|issue:40444", + "issue:43482|issue:43525", + "issue:44291|issue:44589", + "issue:40444|issue:45325", + "issue:43377|issue:45491", + "issue:41720|issue:45081", + "issue:42175|issue:43296", + "issue:43531|issue:44291", + "issue:43611|issue:43646", + "issue:44265|issue:45072", + "issue:43632|issue:45072", + "issue:43454|issue:44466", + "issue:44079|issue:44534", + "issue:43994|issue:44898" + ], + "prepared_review_unit_hash": "794874f4f9a11081157d6c6881a5a5bcda21c0feb8de5341cb393de4db9c94ba", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 37013, + "estimated_input_tokens": 9254, + "estimated_eval_tokens": 18764 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose set of unrelated transformers bugs, many around v5 regressions, tokenizer loading, vision-language models, and backend/integration issues. I found no true duplicate PRs, and all soft edges should be rejected.", + "confidence": 0.42, + "canonical_issue_reason": "Issue 45020 is the best umbrella candidate: it is broad, recent, and explicitly frames a class of regressions in recent transformers versions affecting remote_code models.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45020 is the strongest global representative because it is the most general open issue in the set and could plausibly anchor related v5 regression reports, even though the cluster is not a tight duplicate group.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41553", + "right": "issue:43582", + "accept": false, + "reason": "Different subsystems and failures: tokenizer loading error message vs Apple Silicon caching allocator TypeError." + }, + { + "left": "issue:44479", + "right": "issue:45325", + "accept": false, + "reason": "Both are Qwen VL/video-related, but one is a broad video-input regression and the other is a still-image temporal position_ids scaling bug." + }, + { + "left": "issue:40990", + "right": "issue:44811", + "accept": false, + "reason": "Perplexity on GPT-OSS/WikiText and Whisper batch_decode skipping special tokens are unrelated bugs." + }, + { + "left": "issue:43425", + "right": "issue:44403", + "accept": false, + "reason": "Torch 2.10 compatibility and noisy transformer loading messages are different issues." + }, + { + "left": "issue:42175", + "right": "issue:45542", + "accept": false, + "reason": "Both mention TensorFlow installation, but one is missing TF from the torch extra and the other is a backend detection/runtime error with tensorboard-only installs." + }, + { + "left": "issue:44462", + "right": "issue:45812", + "accept": false, + "reason": "Tokenizer.json being ignored and Granite token IDs being wrong are different tokenizer-path failures." + }, + { + "left": "issue:43818", + "right": "issue:45357", + "accept": false, + "reason": "Video-LLaVA architecture/weight-sharing bug vs Qwen3.5 save_pretrained visual-key regression." + }, + { + "left": "issue:44079", + "right": "issue:45663", + "accept": false, + "reason": "ModelOutput key assignment and Gemma-4 shared_kv_states KeyError are unrelated code paths." + }, + { + "left": "issue:36032", + "right": "issue:41720", + "accept": false, + "reason": "T5 tokenizer method conflict and Qwen3 auto device mapping cuda assert are unrelated." + }, + { + "left": "issue:43421", + "right": "issue:44568", + "accept": false, + "reason": "Runtime post-processor updates for special tokens and mdeberta add_special_tokens behavior are different tokenization bugs." + }, + { + "left": "issue:43792", + "right": "issue:44589", + "accept": false, + "reason": "Whisper failing to run and missing Float8 storage support are unrelated." + }, + { + "left": "issue:43632", + "right": "issue:44534", + "accept": false, + "reason": "_is_hf_initialized handling and non-persistent buffer junk are separate v5 regressions." + }, + { + "left": "issue:44403", + "right": "issue:44991", + "accept": false, + "reason": "Generic loading noise and est-roberta tokenizer loading failure are not the same bug." + }, + { + "left": "issue:44683", + "right": "issue:45020", + "accept": false, + "reason": "Compiled flex_attention on torch>=2.9 and remote_code model breakage are unrelated." + }, + { + "left": "issue:44829", + "right": "issue:45479", + "accept": false, + "reason": "Both affect training, but flash_attention_3 degeneration and single-label-classification zero-loss are distinct root causes." + }, + { + "left": "issue:43482", + "right": "issue:45538", + "accept": false, + "reason": "Qwen2.5-GGUF loading failure and CLIPTokenizer model_max_length are unrelated tokenizer/model-loading issues." + }, + { + "left": "issue:33666", + "right": "issue:40444", + "accept": false, + "reason": "Both are Qwen2 VL training reports, but multi-GPU training and iterable-dataset multi-image failure are different problems." + }, + { + "left": "issue:43482", + "right": "issue:43525", + "accept": false, + "reason": "Qwen2.5-GGUF loading and Llama4Config pad_token_id missing are unrelated." + }, + { + "left": "issue:44291", + "right": "issue:44589", + "accept": false, + "reason": "init_empty_weights argument handling and Float8 storage lookup are different failures." + }, + { + "left": "issue:40444", + "right": "issue:45325", + "accept": false, + "reason": "IterableDataset multi-image prompting and still-image rope-index scaling are separate Qwen2.5-VL bugs." + }, + { + "left": "issue:43377", + "right": "issue:45491", + "accept": false, + "reason": "MIMI batch-vs-single padding-mask mismatch and Gemma3 NaN embeddings with sliding windows are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:45081", + "accept": false, + "reason": "Qwen3 auto device map cuda assert and Mistral tokenizer regex patch crash are different code paths." + }, + { + "left": "issue:42175", + "right": "issue:43296", + "accept": false, + "reason": "Packaging/backend install issue vs PaddleOCR-VL load failure in vLLM are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:44291", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and init_empty_weights argument errors are different bugs." + }, + { + "left": "issue:43611", + "right": "issue:43646", + "accept": false, + "reason": "Both are v5 init regressions, but base_model_prefix loading and custom model initialization are not the same underlying defect." + }, + { + "left": "issue:44265", + "right": "issue:45072", + "accept": false, + "reason": "torch.export/torch_compilable_check failure and bfloat16 dtype mismatch are unrelated." + }, + { + "left": "issue:43632", + "right": "issue:45072", + "accept": false, + "reason": "_is_hf_initialized breakage and bfloat16 dtype mismatches are different issues." + }, + { + "left": "issue:43454", + "right": "issue:44466", + "accept": false, + "reason": "Tied lm_head weights generation bug and device-dependent serialization of lm_head.weight are related in theme but not clearly the same concrete bug." + }, + { + "left": "issue:44079", + "right": "issue:44534", + "accept": false, + "reason": "ModelOutput key assignment and non-persistent buffer initialization are unrelated." + }, + { + "left": "issue:43994", + "right": "issue:44898", + "accept": false, + "reason": "Siglip2 nonsensical outputs and Perceiver resize/interpolation failure are separate model-specific bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 10, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:35707", + "issue:39692", + "issue:40444", + "issue:40990", + "issue:41720", + "issue:42915", + "issue:43262", + "issue:43296", + "issue:43317", + "issue:43377", + "issue:43454", + "issue:43482", + "issue:43526", + "issue:43550", + "issue:43572", + "issue:43606", + "issue:43644", + "issue:43701", + "issue:43746", + "issue:43792", + "issue:43819", + "issue:43873", + "issue:43881", + "issue:43883", + "issue:43957", + "issue:44079", + "issue:44186", + "issue:44265", + "issue:44291", + "issue:44387", + "issue:44466", + "issue:44479", + "issue:44488", + "issue:44610", + "issue:44617", + "issue:44683", + "issue:44877", + "issue:44898", + "issue:44991", + "issue:45072", + "issue:45325", + "issue:45357", + "issue:45381", + "issue:45584", + "issue:45759", + "issue:45800", + "issue:45823", + "issue:45834" + ], + "soft_pairs": [ + "issue:45325|issue:45834", + "issue:43792|issue:43957", + "issue:42915|issue:43482", + "issue:44079|issue:45072", + "issue:40444|issue:45381", + "issue:45325|issue:45381", + "issue:43644|issue:44079", + "issue:44387|issue:44617", + "issue:43881|issue:44488", + "issue:43296|issue:45800", + "issue:44479|issue:45823", + "issue:35707|issue:40444", + "issue:44291|issue:44877", + "issue:43317|issue:43873", + "issue:43606|issue:44265", + "issue:43701|issue:45325", + "issue:44186|issue:44265", + "issue:43606|issue:43819", + "issue:43262|issue:43746", + "issue:41720|issue:45823", + "issue:43454|issue:43550", + "issue:43526|issue:43883", + "issue:43377|issue:44610", + "issue:43454|issue:45072", + "issue:40990|issue:45584", + "issue:43377|issue:44898", + "issue:43572|issue:45759", + "issue:39692|issue:40444", + "issue:43881|issue:44991", + "issue:44466|issue:45357", + "issue:43296|issue:44991", + "issue:43792|issue:44683" + ], + "prepared_review_unit_hash": "ba687a57a9e9a9e67ff20604f5bbc0e4c2e53f9629f401056dea2839d0373b0a", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37301, + "estimated_input_tokens": 9326, + "estimated_eval_tokens": 18908 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The items are mostly a mix of unrelated bug reports across different models and subsystems; there is no strong duplicate cluster, so I rejected all soft matches.", + "confidence": 0.96, + "canonical_issue_reason": "No single issue cleanly serves as a canonical duplicate: the set spans unrelated loading, quantization, multimodal, tokenizer, and serialization problems.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue is a clear global representative of a duplicate group; the closest overlaps are still only broad subsystem similarity, not the same concrete bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45325", + "right": "issue:45834", + "accept": false, + "reason": "Both involve multimodal position/index handling, but they target different models and different failure modes (still-image temporal scaling vs long OCR index error)." + }, + { + "left": "issue:43792", + "right": "issue:43957", + "accept": false, + "reason": "Both are loading/runtime failures, but the models and root causes differ; this is too broad to be the same bug." + }, + { + "left": "issue:42915", + "right": "issue:43482", + "accept": false, + "reason": "Different model families and different failure modes: FP8 config failure vs GGUF/v5 loading failure." + }, + { + "left": "issue:44079", + "right": "issue:45072", + "accept": false, + "reason": "Unrelated bugs: ModelOutput key assignment vs bfloat16 dtype mismatches." + }, + { + "left": "issue:40444", + "right": "issue:45381", + "accept": false, + "reason": "Both are Qwen2.5-VL-related, but one is training with multiple images and the other is video vision position IDs; not the same bug." + }, + { + "left": "issue:45325", + "right": "issue:45381", + "accept": false, + "reason": "Related to vision position ids, but one is still-image temporal scaling and the other is video input position IDs; not concrete enough to merge." + }, + { + "left": "issue:43644", + "right": "issue:44079", + "accept": false, + "reason": "Non-persistent buffer initialization and ModelOutput key assignment are unrelated issues." + }, + { + "left": "issue:44387", + "right": "issue:44617", + "accept": false, + "reason": "Both mention OOM, but one is int4 quantization memory regression and the other is a model-specific video OOM." + }, + { + "left": "issue:43881", + "right": "issue:44488", + "accept": false, + "reason": "Different models and likely unrelated loading failures." + }, + { + "left": "issue:43296", + "right": "issue:45800", + "accept": false, + "reason": "A vLLM/model-loading issue and a torch/transformers version incompatibility are not the same bug." + }, + { + "left": "issue:44479", + "right": "issue:45823", + "accept": false, + "reason": "Both are multimodal regressions, but video-input regression and PLE device mismatch are different code-path problems." + }, + { + "left": "issue:35707", + "right": "issue:40444", + "accept": false, + "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen2.5-VL finetuning with multiple images." + }, + { + "left": "issue:44291", + "right": "issue:44877", + "accept": false, + "reason": "init_empty_weights/_is_hf_initialized is a loading API break; granite_speech strict config is a separate config issue." + }, + { + "left": "issue:43317", + "right": "issue:43873", + "accept": false, + "reason": "Both involve offloading/quantization, but one is a gpu+cpu load failure and the other is a broader offloading complaint; too imprecise to treat as one bug." + }, + { + "left": "issue:43606", + "right": "issue:44265", + "accept": false, + "reason": "CPU offload device mismatch and torch.export/export compatibility are unrelated." + }, + { + "left": "issue:43701", + "right": "issue:45325", + "accept": false, + "reason": "Checkpoint key mismatch and Qwen2.5-VL rope/index scaling are unrelated." + }, + { + "left": "issue:44186", + "right": "issue:44265", + "accept": false, + "reason": "Tokenizer padding/truncation crashes and torch.export with torch_compilable_check are different subsystems." + }, + { + "left": "issue:43606", + "right": "issue:43819", + "accept": false, + "reason": "Both mention model computation correctness, but one is CPU offload device mismatch and the other is DAC.from_latents/STE mismatch." + }, + { + "left": "issue:43262", + "right": "issue:43746", + "accept": false, + "reason": "Audio chat-template sampling rate defaults and GraniteSpeech PEFT checkpoint loading are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:45823", + "accept": false, + "reason": "Both are device_map=auto failures, but different models and different failure types (cuda assert vs device mismatch)." + }, + { + "left": "issue:43454", + "right": "issue:43550", + "accept": false, + "reason": "Weight tying in AyaVision and torch.compile/SDPA in Bamba are unrelated bugs." + }, + { + "left": "issue:43526", + "right": "issue:43883", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and Molmo tied-weights attribute errors are unrelated." + }, + { + "left": "issue:43377", + "right": "issue:44610", + "accept": false, + "reason": "Padding-mask batch consistency and processor image-size mismatch are different issues." + }, + { + "left": "issue:43454", + "right": "issue:45072", + "accept": false, + "reason": "AyaVision weight tying and dtype mismatches in SwitchTransformers/TimmWrapperModel are unrelated." + }, + { + "left": "issue:40990", + "right": "issue:45584", + "accept": false, + "reason": "High perplexity on gpt-oss-20b and Whisper empty-transcription generation failure are unrelated." + }, + { + "left": "issue:43572", + "right": "issue:45759", + "accept": false, + "reason": "Missing StableLmConfig pad_token_idx and composite Qwen text_config unwrapping are different config-loading problems." + }, + { + "left": "issue:39692", + "right": "issue:40444", + "accept": false, + "reason": "SigLIP2 docs errors/quantization failure and Qwen2.5-VL finetuning failure are unrelated." + }, + { + "left": "issue:43881", + "right": "issue:44991", + "accept": false, + "reason": "Different model/tokenizer loading failures with no shared concrete code path." + }, + { + "left": "issue:44466", + "right": "issue:45357", + "accept": false, + "reason": "Both involve serialization, but one is lm_head weight persistence and the other is visual encoder key naming; not the same bug." + }, + { + "left": "issue:43296", + "right": "issue:44991", + "accept": false, + "reason": "PaddleOCR-VL loading in vLLM and tokenizer loading for EMBEDDIA/est-roberta are unrelated." + }, + { + "left": "issue:43792", + "right": "issue:44683", + "accept": false, + "reason": "Whisper runtime failure and compiled flex_attention on torch>=2.9 are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 11, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:34634", + "issue:35707", + "issue:36331", + "issue:40990", + "issue:41720", + "issue:42617", + "issue:42915", + "issue:43295", + "issue:43316", + "issue:43317", + "issue:43322", + "issue:43404", + "issue:43450", + "issue:43454", + "issue:43526", + "issue:43550", + "issue:43577", + "issue:43606", + "issue:43708", + "issue:43720", + "issue:43749", + "issue:43873", + "issue:43881", + "issue:44008", + "issue:44186", + "issue:44387", + "issue:44423", + "issue:44479", + "issue:44534", + "issue:44610", + "issue:44756", + "issue:44877", + "issue:44898", + "issue:44913", + "issue:44918", + "issue:44977", + "issue:44991", + "issue:45072", + "issue:45081", + "issue:45161", + "issue:45357", + "issue:45406", + "issue:45632", + "issue:45684", + "issue:45701", + "issue:45704", + "issue:45812", + "issue:45823" + ], + "soft_pairs": [ + "issue:43454|issue:43720", + "issue:43873|issue:45701", + "issue:41720|issue:44977", + "issue:44008|issue:45823", + "issue:45632|issue:45684", + "issue:43404|issue:45357", + "issue:34634|issue:35707", + "issue:43577|issue:43708", + "issue:44186|issue:45072", + "issue:44423|issue:45406", + "issue:35707|issue:41720", + "issue:43873|issue:44756", + "issue:44877|issue:44991", + "issue:43749|issue:44534", + "issue:42617|issue:42915", + "issue:43450|issue:45701", + "issue:43550|issue:43720", + "issue:36331|issue:42915", + "issue:44610|issue:44898", + "issue:44186|issue:44610", + "issue:44387|issue:45704", + "issue:36331|issue:42617", + "issue:44479|issue:44918", + "issue:43526|issue:43881", + "issue:43316|issue:44913", + "issue:42915|issue:43317", + "issue:43550|issue:43606", + "issue:45081|issue:45812", + "issue:41720|issue:43322", + "issue:40990|issue:45701", + "issue:43295|issue:45701", + "issue:41720|issue:45161" + ], + "prepared_review_unit_hash": "6882e637f516a7a21e8789a1c0e940258931d5fb1773d34e055cb80a0ff0597c", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37387, + "estimated_input_tokens": 9347, + "estimated_eval_tokens": 18950 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous rather than a true duplicate set: it mixes unrelated bugs in generation, tokenization, loading/offload, training, and multimodal processors. None of the soft pairs appear to be the same concrete underlying bug or change.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue stands out; the items span unrelated code paths and model families, so there is no clear duplicate anchor.", + "canonical_pr_reason": null, + "best_issue_reason": "No global best issue for this cluster: the issues are thematically similar only at a very broad level, not as one shared defect.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43454", + "right": "issue:43720", + "accept": false, + "reason": "Both concern model-loading problems, but one is lm_head tying and the other is packed-weight unpacking in accelerate; different bugs and code paths." + }, + { + "left": "issue:43873", + "right": "issue:45701", + "accept": false, + "reason": "Quantization/offload behavior versus tokenizer/version-dependent tokenization changes; too broad and unrelated." + }, + { + "left": "issue:41720", + "right": "issue:44977", + "accept": false, + "reason": "Different Qwen variants and different failures: device-map cuda assert vs flash-attention generation regression." + }, + { + "left": "issue:44008", + "right": "issue:45823", + "accept": false, + "reason": "Gemma3n audio tensor name collision is unrelated to Gemma4 PLE device mismatch during forward." + }, + { + "left": "issue:45632", + "right": "issue:45684", + "accept": false, + "reason": "Cache-path collision for trust_remote_code is unrelated to save_pretrained permission propagation." + }, + { + "left": "issue:43404", + "right": "issue:45357", + "accept": false, + "reason": "Weight tying in Mistral3 conditional generation is a different issue from Qwen3.5 visual encoder key saving." + }, + { + "left": "issue:34634", + "right": "issue:35707", + "accept": false, + "reason": "BarkProcessor voice_preset failure is unrelated to progressive generation with inputs_embeds and past_key_values." + }, + { + "left": "issue:43577", + "right": "issue:43708", + "accept": false, + "reason": "Blip2 dtype loading bug versus Trainer resume/max_steps calculation bug; different subsystems." + }, + { + "left": "issue:44186", + "right": "issue:45072", + "accept": false, + "reason": "Tokenizer crash on NER/padding is unrelated to dtype mismatches in SwitchTransformers/TimmWrapperModel inference." + }, + { + "left": "issue:44423", + "right": "issue:45406", + "accept": false, + "reason": "Both are serve/multimodal crashes, but one is continuous-batching string handling and the other is a missing _tokenizer attribute in Gemma4Processor." + }, + { + "left": "issue:35707", + "right": "issue:41720", + "accept": false, + "reason": "Both involve generation, but the concrete failures are different: cache/input-embed handling vs device-map cuda assert." + }, + { + "left": "issue:43873", + "right": "issue:44756", + "accept": false, + "reason": "Quantization offload issues and disabling mmap for Strix Halo OOM are not the same underlying problem." + }, + { + "left": "issue:44877", + "right": "issue:44991", + "accept": false, + "reason": "Strict config rejection for granite_speech is unrelated to tokenizer loading failure for EMBEDDIA/est-roberta." + }, + { + "left": "issue:43749", + "right": "issue:44534", + "accept": false, + "reason": "FSDP CPU RAM efficient loading is a loading-path issue; non-persistent buffer junk is a different v5 buffer initialization bug." + }, + { + "left": "issue:42617", + "right": "issue:42915", + "accept": false, + "reason": "3d_parallel.py runtime failure is unrelated to Qwen3Moe failing with FineGrainedFP8Config." + }, + { + "left": "issue:43450", + "right": "issue:45701", + "accept": false, + "reason": "Video processor batch-shape bug is unrelated to version-dependent tokenization changes." + }, + { + "left": "issue:43550", + "right": "issue:43720", + "accept": false, + "reason": "Bamba torch.compile/SDPA failure and BitNet packed-weight loading failure are different model/backend bugs." + }, + { + "left": "issue:36331", + "right": "issue:42915", + "accept": false, + "reason": "Trainer compute_loss signature change is unrelated to Qwen3Moe FP8 configuration loading." + }, + { + "left": "issue:44610", + "right": "issue:44898", + "accept": false, + "reason": "Both are image-resolution related, but they affect different models and different preprocessing/model expectations." + }, + { + "left": "issue:44186", + "right": "issue:44610", + "accept": false, + "reason": "LayoutLMv2 tokenizer crashes and OmDet-Turbo image-size mismatch are different processor bugs." + }, + { + "left": "issue:44387", + "right": "issue:45704", + "accept": false, + "reason": "Int4 quantization memory regression in Transformers 5.x is unrelated to apex FusedRMSNorm memory leak in T5." + }, + { + "left": "issue:36331", + "right": "issue:42617", + "accept": false, + "reason": "CustomTrainer compute_loss API break and 3d_parallel.py failure share no concrete code path." + }, + { + "left": "issue:44479", + "right": "issue:44918", + "accept": false, + "reason": "Video-input regression in Qwen2.5/Qwen3.5 VL is different from TRL SFT issues unpacking Qwen3.5 input embeddings." + }, + { + "left": "issue:43526", + "right": "issue:43881", + "accept": false, + "reason": "BeitImageProcessorFast reduce_labels bug is unrelated to glm-4v-9b loading failure." + }, + { + "left": "issue:43316", + "right": "issue:44913", + "accept": false, + "reason": "Both are config reload discrepancies, but for different configs and different fields; not the same bug." + }, + { + "left": "issue:42915", + "right": "issue:43317", + "accept": false, + "reason": "Qwen3Moe FineGrainedFP8Config failure is unrelated to dequantized model loading with gpu+cpu offload." + }, + { + "left": "issue:43550", + "right": "issue:43606", + "accept": false, + "reason": "Different failures: torch.compile/SDPA in Bamba versus CPU offload device mismatch in Bark-small." + }, + { + "left": "issue:45081", + "right": "issue:45812", + "accept": false, + "reason": "Mistral regex patch crash on tokenizer internals is unrelated to Granite token ID regression." + }, + { + "left": "issue:41720", + "right": "issue:43322", + "accept": false, + "reason": "Qwen3 auto device-map cuda assert and Llava Next segfault are different model-loading failures." + }, + { + "left": "issue:40990", + "right": "issue:45701", + "accept": false, + "reason": "High perplexity on gpt-oss-20b is not the same as a tokenization change regression." + }, + { + "left": "issue:43295", + "right": "issue:45701", + "accept": false, + "reason": "Processor.tokenizer regression with images is a different issue from generic tokenization changes across versions." + }, + { + "left": "issue:41720", + "right": "issue:45161", + "accept": false, + "reason": "Auto device-map cuda assert on Qwen3 is unrelated to TP-only failure on GPT-OSS MoE." + } + ] + }, + "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": 16, + "review_unit_index": 12, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:30064", + "issue:34567", + "issue:34634", + "issue:35707", + "issue:36010", + "issue:41720", + "issue:42617", + "issue:42915", + "issue:43322", + "issue:43388", + "issue:43421", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43572", + "issue:43606", + "issue:43653", + "issue:43720", + "issue:43746", + "issue:43782", + "issue:43873", + "issue:43927", + "issue:44062", + "issue:44186", + "issue:44361", + "issue:44368", + "issue:44451", + "issue:44492", + "issue:44509", + "issue:44514", + "issue:44545", + "issue:44589", + "issue:44671", + "issue:44743", + "issue:44779", + "issue:44898", + "issue:44987", + "issue:44991", + "issue:45042", + "issue:45059", + "issue:45072", + "issue:45357", + "issue:45406", + "issue:45464", + "issue:45701", + "issue:45704", + "issue:45759", + "issue:45812" + ], + "soft_pairs": [ + "issue:43531|issue:43572", + "issue:44671|issue:44743", + "issue:43421|issue:45812", + "issue:41720|issue:42617", + "issue:36010|issue:42617", + "issue:43540|issue:43782", + "issue:44186|issue:44361", + "issue:45406|issue:45464", + "issue:43421|issue:45701", + "issue:30064|issue:34634", + "issue:44492|issue:44509", + "issue:44779|issue:45812", + "issue:42915|issue:43322", + "issue:43606|issue:43720", + "issue:43388|issue:45059", + "issue:43550|issue:43653", + "issue:44514|issue:44545", + "issue:43531|issue:45759", + "issue:43606|issue:43653", + "issue:43421|issue:43927", + "issue:34567|issue:35707", + "issue:44368|issue:45357", + "issue:43873|issue:45704", + "issue:44451|issue:44991", + "issue:44898|issue:45072", + "issue:41720|issue:43746", + "issue:44062|issue:44451", + "issue:44987|issue:45042", + "issue:43531|issue:44589" + ], + "prepared_review_unit_hash": "6fc1ad4f48ef898c0c5371e76ab68c8e5fe6cbb58f62401b988e720fdd2258d3", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36380, + "estimated_input_tokens": 9095, + "estimated_eval_tokens": 18446 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Only one clear duplicate pair stands out: the two Qwen2_5_VLProcessor.apply_chat_template crash reports. All other soft links are broad similarity matches but describe different bugs, models, or code paths and should remain separate.", + "confidence": 0.89, + "canonical_issue_reason": "Issue 44545 is the better representative of the exact duplicate pair with 44514: same crash, same title/behavior, and slightly more inbound references/discussion.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44545 is the strongest issue candidate overall because it is the clearest exact duplicate target in this set and has the most signal among the pair.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43531", + "right": "issue:43572", + "accept": false, + "reason": "Both mention Qwen3-MoE / config-related problems, but one is a sliding_window bug and the other is a missing pad_token_idx regression; different failures." + }, + { + "left": "issue:44671", + "right": "issue:44743", + "accept": false, + "reason": "CamemBERT masked-LM prediction errors and Qwen3 modular cache state resets are unrelated bugs in different models and code paths." + }, + { + "left": "issue:43421", + "right": "issue:45812", + "accept": false, + "reason": "One is about runtime post-processor updates for special tokens; the other is a silent tokenizer ID regression across Granite models. Related area, different concrete bug." + }, + { + "left": "issue:41720", + "right": "issue:42617", + "accept": false, + "reason": "Qwen3 auto device mapping on A800 causing cudaErrorAssert is unrelated to failing to run 3d_parallel.py." + }, + { + "left": "issue:36010", + "right": "issue:42617", + "accept": false, + "reason": "An import error for GenerationMixin is a different issue from a script execution failure." + }, + { + "left": "issue:43540", + "right": "issue:43782", + "accept": false, + "reason": "Qwen3OmniMoe video-input processing and Qwen3VL weight_only loading are distinct model/load-path bugs." + }, + { + "left": "issue:44186", + "right": "issue:44361", + "accept": false, + "reason": "Both are tokenizer crashes, but they involve different tokenizers, tasks, and failure modes; not the same underlying bug." + }, + { + "left": "issue:45406", + "right": "issue:45464", + "accept": false, + "reason": "Gemma4Processor serve crash and Qwen3.5 streaming chat/completions failure are different processor/runtime problems." + }, + { + "left": "issue:43421", + "right": "issue:45701", + "accept": false, + "reason": "Runtime special-token/post-processor updates and a general tokenization-version change are not the same concrete defect." + }, + { + "left": "issue:30064", + "right": "issue:34634", + "accept": false, + "reason": "Void segmentation map processing and BarkProcessor voice_preset behavior are unrelated processor bugs." + }, + { + "left": "issue:44492", + "right": "issue:44509", + "accept": false, + "reason": "A cache-strategy typo and stale docs for removed pipeline tasks are documentation issues of different scope." + }, + { + "left": "issue:44779", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenization regressions, but for different model families and symptoms; no evidence of one shared bug." + }, + { + "left": "issue:42915", + "right": "issue:43322", + "accept": false, + "reason": "Qwen3Moe FP8 failure and Llava Next segmentation fault are unrelated model-specific crashes." + }, + { + "left": "issue:43606", + "right": "issue:43720", + "accept": false, + "reason": "CPU offload device mismatch for Bark-small and packed-weights unpacking for BitNet are different accelerate/loading bugs." + }, + { + "left": "issue:43388", + "right": "issue:45059", + "accept": false, + "reason": "Mask2Former label dropping in gather_for_metrics is unrelated to SAM3 text+box prompt weirdness." + }, + { + "left": "issue:43550", + "right": "issue:43653", + "accept": false, + "reason": "torch.compile/SDPA failure in Bamba-9B-v2 and tokenizer special-token registration in BigBirdTokenizer are unrelated." + }, + { + "left": "issue:44514", + "right": "issue:44545", + "accept": true, + "reason": "These appear to be the same exact bug report: identical title and same apply_chat_template crash on batched input with padding=False." + }, + { + "left": "issue:43531", + "right": "issue:45759", + "accept": false, + "reason": "Sliding_window behavior in Qwen3-MoE and AutoModelForCausalLM.from_config text_config unwrapping are different issues." + }, + { + "left": "issue:43606", + "right": "issue:43653", + "accept": false, + "reason": "Device mismatch under CPU offload and a tokenizer mask-token decode issue are not the same underlying problem." + }, + { + "left": "issue:43421", + "right": "issue:43927", + "accept": false, + "reason": "Special-token post-processor refresh at runtime and DiaConfig losing custom token IDs on save/load are different tokenizer/config bugs." + }, + { + "left": "issue:34567", + "right": "issue:35707", + "accept": false, + "reason": "TrainerState token counter not updating and progressive generation with inputs_embeds/past_key_values are unrelated training vs generation issues." + }, + { + "left": "issue:44368", + "right": "issue:45357", + "accept": false, + "reason": "A warning about tie_word_embeddings and incorrect save_pretrained visual encoder keys are different Qwen3.5 fine-tuning/save issues." + }, + { + "left": "issue:43873", + "right": "issue:45704", + "accept": false, + "reason": "Quantization/offloading not working and a T5 apex memory leak are different mechanisms and symptoms." + }, + { + "left": "issue:44451", + "right": "issue:44991", + "accept": false, + "reason": "Cannot load one specific model and tokenizer loading failures for another model are separate compatibility issues." + }, + { + "left": "issue:44898", + "right": "issue:45072", + "accept": false, + "reason": "Perceiver non-default resolution handling and bfloat16 dtype mismatches in SwitchTransformers/TimmWrapperModel are different bugs." + }, + { + "left": "issue:41720", + "right": "issue:43746", + "accept": false, + "reason": "Qwen3 auto device mapping CUDA assert and GraniteSpeech PEFT local checkpoint loading are unrelated failures." + }, + { + "left": "issue:44062", + "right": "issue:44451", + "accept": false, + "reason": "AddedToken special-argument collision and inability to load a model are not the same code-path." + }, + { + "left": "issue:44987", + "right": "issue:45042", + "accept": false, + "reason": "A model loading failure for physical-intelligence/fast and a PIL backend torchvision dependency regression are different issues." + }, + { + "left": "issue:43531", + "right": "issue:44589", + "accept": false, + "reason": "Qwen3-MoE sliding_window and a missing Float8 storage type are unrelated bugs." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded overall. The exact-duplicate call on 44514/44545 is supported by the packet: identical title, same crash description, and 44545 has slightly more inbound references/discussion. The remaining soft-edge rejections are conservative and stay within what the titles/metadata support. No PR-pair mergeability concerns apply here." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 13, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:30064", + "issue:33666", + "issue:34634", + "issue:35707", + "issue:36010", + "issue:39692", + "issue:41720", + "issue:42886", + "issue:42915", + "issue:43335", + "issue:43381", + "issue:43421", + "issue:43525", + "issue:43531", + "issue:43540", + "issue:43575", + "issue:43596", + "issue:43618", + "issue:43653", + "issue:43697", + "issue:43716", + "issue:43720", + "issue:43756", + "issue:43782", + "issue:43819", + "issue:43844", + "issue:43873", + "issue:43994", + "issue:44008", + "issue:44077", + "issue:44186", + "issue:44220", + "issue:44610", + "issue:44625", + "issue:44661", + "issue:44671", + "issue:44756", + "issue:44843", + "issue:44918", + "issue:44936", + "issue:45081", + "issue:45183", + "issue:45356", + "issue:45381", + "issue:45479", + "issue:45588", + "issue:45704", + "issue:45759" + ], + "soft_pairs": [ + "issue:45588|issue:45759", + "issue:33666|issue:35707", + "issue:43421|issue:43653", + "issue:43381|issue:43844", + "issue:42886|issue:44843", + "issue:39692|issue:44610", + "issue:44918|issue:44936", + "issue:43421|issue:44186", + "issue:43525|issue:43531", + "issue:44625|issue:45479", + "issue:43782|issue:45381", + "issue:45081|issue:45356", + "issue:34634|issue:41720", + "issue:43994|issue:44671", + "issue:43381|issue:45704", + "issue:39692|issue:43873", + "issue:42915|issue:45759", + "issue:43596|issue:44220", + "issue:43540|issue:44008", + "issue:43618|issue:43697", + "issue:44220|issue:44661", + "issue:45183|issue:45479", + "issue:44077|issue:44625", + "issue:43575|issue:45704", + "issue:43720|issue:43819", + "issue:30064|issue:36010", + "issue:43335|issue:43756", + "issue:44756|issue:45759", + "issue:43653|issue:43720", + "issue:43716|issue:44008" + ], + "prepared_review_unit_hash": "405c6bc498835c3765518eddf60049deee7554a9a015e55fb853e6d04f903d4b", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36813, + "estimated_input_tokens": 9204, + "estimated_eval_tokens": 18664 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is a heterogeneous mix of unrelated bug reports across different models, tokenizers, processors, and training/runtime paths. None of the proposed soft pairs look like the same underlying issue, so no duplicate merge is warranted.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45588", + "right": "issue:45759", + "accept": false, + "reason": "Different subsystems and failures: flash-attention sink handling vs Qwen multimodal config unwrapping." + }, + { + "left": "issue:33666", + "right": "issue:35707", + "accept": false, + "reason": "Qwen2-VL multi-GPU training is unrelated to progressive generation with inputs_embeds/past_key_values." + }, + { + "left": "issue:43421", + "right": "issue:43653", + "accept": false, + "reason": "Both involve tokenizers, but one is runtime post-processor updates and the other is BigBird mask-token registration/decode behavior." + }, + { + "left": "issue:43381", + "right": "issue:43844", + "accept": false, + "reason": "Gradient checkpointing eval-mode behavior is unrelated to ZeRO-3 gradient growth during training." + }, + { + "left": "issue:42886", + "right": "issue:44843", + "accept": false, + "reason": "Both mention offline/tokenizer loading, but one is generic cache loading and the other is a specific _patch_mistral_regex model_info() call." + }, + { + "left": "issue:39692", + "right": "issue:44610", + "accept": false, + "reason": "Different model-specific processor mismatches: SigLIP2 docs vs OmDet-Turbo image size expectation." + }, + { + "left": "issue:44918", + "right": "issue:44936", + "accept": false, + "reason": "Unpacking Qwen3.5 input embeddings with TRL SFT is not the same as trainer.evaluate() failing after train()." + }, + { + "left": "issue:43421", + "right": "issue:44186", + "accept": false, + "reason": "Different tokenizer bugs: dynamic special-token post-processing vs LayoutLMv2 batching/NER crash." + }, + { + "left": "issue:43525", + "right": "issue:43531", + "accept": false, + "reason": "Missing pad_token_id in Llama4Config is unrelated to Qwen3-MoE sliding_window behavior." + }, + { + "left": "issue:44625", + "right": "issue:45479", + "accept": false, + "reason": "Config propagation of num_labels is a different bug from single-label classification degenerating to zero loss." + }, + { + "left": "issue:43782", + "right": "issue:45381", + "accept": false, + "reason": "Different Qwen VL issues: weight_only loading failure vs video vision_position_ids being wrong." + }, + { + "left": "issue:45081", + "right": "issue:45356", + "accept": false, + "reason": "Both touch fix_mistral_regex, but one crashes on backend_tokenizer and the other is a codec-handling regression." + }, + { + "left": "issue:34634", + "right": "issue:41720", + "accept": false, + "reason": "Bark voice_preset parsing is unrelated to Qwen3 auto device mapping cudaErrorAssert." + }, + { + "left": "issue:43994", + "right": "issue:44671", + "accept": false, + "reason": "Different models and outputs: SigLIP2 pipeline behavior vs CamemBERT masked LM predictions." + }, + { + "left": "issue:43381", + "right": "issue:45704", + "accept": false, + "reason": "Gradient checkpointing eval-mode support is unrelated to a T5 apex.RMSNorm memory leak." + }, + { + "left": "issue:39692", + "right": "issue:43873", + "accept": false, + "reason": "SigLIP2 example issues are unrelated to quantization offloading behavior." + }, + { + "left": "issue:42915", + "right": "issue:45759", + "accept": false, + "reason": "Qwen3Moe FP8 loading failure is unrelated to Qwen multimodal config unwrapping." + }, + { + "left": "issue:43596", + "right": "issue:44220", + "accept": false, + "reason": "Deepspeed ZeRO-3/BertModel index error is unrelated to _torch_extract_fbank_features()." + }, + { + "left": "issue:43540", + "right": "issue:44008", + "accept": false, + "reason": "Video input validation in Qwen3OmniMoe is unrelated to Gemma3n's audio_mel_mask name collision." + }, + { + "left": "issue:43618", + "right": "issue:43697", + "accept": false, + "reason": "CLIP attentions assignment regression is unrelated to RTDetrV2 differing outputs in v5." + }, + { + "left": "issue:44220", + "right": "issue:44661", + "accept": false, + "reason": "Audio feature extraction bug is unrelated to tokenizer mapping-name handling in add-new-model-like." + }, + { + "left": "issue:45183", + "right": "issue:45479", + "accept": false, + "reason": "Incorrect input validation messaging is unrelated to the num_labels=1 zero-loss classification bug." + }, + { + "left": "issue:44077", + "right": "issue:44625", + "accept": false, + "reason": "Optional post_init policy for patchtsmixer is unrelated to config propagation of num_labels." + }, + { + "left": "issue:43575", + "right": "issue:45759", + "accept": false, + "reason": "Tensor-parallel OOM on a large Qwen2 model is unrelated to multimodal text_config unwrapping." + }, + { + "left": "issue:43720", + "right": "issue:43819", + "accept": false, + "reason": "Packed-weight loading in BitNet is unrelated to DAC.from_latents missing STE behavior." + }, + { + "left": "issue:30064", + "right": "issue:36010", + "accept": false, + "reason": "Image processor void-segmentation handling is unrelated to the GenerationMixin import error." + }, + { + "left": "issue:43335", + "right": "issue:43756", + "accept": false, + "reason": "SwitchTransformers sparse-layer construction is unrelated to Smollm3 RoPE layer dropping." + }, + { + "left": "issue:44756", + "right": "issue:45759", + "accept": false, + "reason": "Strix Halo mmap OOM mitigation is unrelated to Qwen multimodal config unwrapping." + }, + { + "left": "issue:43653", + "right": "issue:43720", + "accept": false, + "reason": "BigBirdTokenizer special-token decode failure is unrelated to BitNet packed-weight loading." + }, + { + "left": "issue:43716", + "right": "issue:44008", + "accept": false, + "reason": "Mistral-3 image dtype mismatch is unrelated to Gemma3n audio tensor attribute collision." + } + ] + }, + "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": 16, + "review_unit_index": 14, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:30064", + "issue:36331", + "issue:39692", + "issue:42915", + "issue:43295", + "issue:43322", + "issue:43421", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43577", + "issue:43582", + "issue:43606", + "issue:43643", + "issue:43653", + "issue:43746", + "issue:43782", + "issue:43818", + "issue:43819", + "issue:43825", + "issue:43827", + "issue:43873", + "issue:43994", + "issue:44062", + "issue:44186", + "issue:44442", + "issue:44448", + "issue:44479", + "issue:44485", + "issue:44488", + "issue:44509", + "issue:44561", + "issue:44671", + "issue:44811", + "issue:44821", + "issue:44843", + "issue:44877", + "issue:44898", + "issue:44977", + "issue:45020", + "issue:45072", + "issue:45081", + "issue:45084", + "issue:45381", + "issue:45584", + "issue:45685", + "issue:45834" + ], + "soft_pairs": [ + "issue:43295|issue:45834", + "issue:42915|issue:43746", + "issue:43746|issue:45381", + "issue:43550|issue:45072", + "issue:43746|issue:43873", + "issue:43550|issue:43819", + "issue:44186|issue:44898", + "issue:44843|issue:45081", + "issue:43531|issue:43643", + "issue:43643|issue:44877", + "issue:43421|issue:44811", + "issue:43577|issue:43818", + "issue:44062|issue:44488", + "issue:43540|issue:45381", + "issue:43653|issue:43819", + "issue:44448|issue:44671", + "issue:43782|issue:44977", + "issue:45584|issue:45834", + "issue:43825|issue:43994", + "issue:43322|issue:45685", + "issue:44561|issue:45020", + "issue:44821|issue:45084", + "issue:39692|issue:45834", + "issue:30064|issue:36331", + "issue:44485|issue:44509", + "issue:44479|issue:45834", + "issue:43825|issue:43827", + "issue:44186|issue:44442", + "issue:43577|issue:43582", + "issue:43606|issue:45072" + ], + "prepared_review_unit_hash": "b2e34c56af20a190fb916a286cc7ac8a0f78e9c785ded4d241f90ccb884a4fd2", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36226, + "estimated_input_tokens": 9057, + "estimated_eval_tokens": 18370 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose semantic bucket of mostly unrelated Transformers issues. The soft pairs share broad keywords (Qwen, tokenizer, pipeline, video, config, etc.) but not the same concrete bug or change, so none should be merged as duplicates.", + "confidence": 0.9, + "canonical_issue_reason": "No canonical issue: the similarities are superficial and span distinct models, code paths, and failure modes rather than one underlying defect.", + "canonical_pr_reason": null, + "best_issue_reason": "No single issue is a good global representative because the set is heterogeneous and does not form a true duplicate cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43295", + "right": "issue:45834", + "accept": false, + "reason": "Different regressions: custom processor.tokenizer access vs OCR long-input indexing." + }, + { + "left": "issue:42915", + "right": "issue:43746", + "accept": false, + "reason": "Unrelated bugs: FineGrainedFP8Config loading vs GraniteSpeech adapter/local-checkpoint loading." + }, + { + "left": "issue:43746", + "right": "issue:45381", + "accept": false, + "reason": "Different code paths: PEFT/local checkpoint loading vs Qwen2.5-VL video position ids." + }, + { + "left": "issue:43550", + "right": "issue:45072", + "accept": false, + "reason": "Different failures: torch.compile SDPA on Bamba vs dtype mismatch in SwitchTransformers/TimmWrapperModel." + }, + { + "left": "issue:43746", + "right": "issue:43873", + "accept": false, + "reason": "Same broad area of loading/offloading, but distinct bugs and models; not the same defect." + }, + { + "left": "issue:43550", + "right": "issue:43819", + "accept": false, + "reason": "Unrelated: SDPA compile bug vs DAC latents/STE mismatch." + }, + { + "left": "issue:44186", + "right": "issue:44898", + "accept": false, + "reason": "Different subsystems: LayoutLMv2 tokenizer padding/NER crash vs Perceiver image resolution handling." + }, + { + "left": "issue:44843", + "right": "issue:45081", + "accept": false, + "reason": "Both touch _patch_mistral_regex, but one is offline model_info access and the other is a backend_tokenizer AttributeError; separate bugs." + }, + { + "left": "issue:43531", + "right": "issue:43643", + "accept": false, + "reason": "Different issues: Qwen3-MoE sliding_window vs AutoConfig trust_remote_code field loss." + }, + { + "left": "issue:43643", + "right": "issue:44877", + "accept": false, + "reason": "Both involve config loading, but one loses fields and the other is strict config rejection; not the same bug." + }, + { + "left": "issue:43421", + "right": "issue:44811", + "accept": false, + "reason": "TokenizersBackend post-processor updates vs Whisper batch_decode skip_special_tokens behavior." + }, + { + "left": "issue:43577", + "right": "issue:43818", + "accept": false, + "reason": "Different model bugs: BLIP2 dtype persistence vs Video-LLaVA temporal attention/weight sharing." + }, + { + "left": "issue:44062", + "right": "issue:44488", + "accept": false, + "reason": "AddedToken special-arg duplication vs tokenizer load failure for cjvt/sleng-bert." + }, + { + "left": "issue:43540", + "right": "issue:45381", + "accept": false, + "reason": "Different video regressions in different models and failure modes." + }, + { + "left": "issue:43653", + "right": "issue:43819", + "accept": false, + "reason": "Tokenizer special-token registration bug vs DAC forward/from_latents mismatch." + }, + { + "left": "issue:44448", + "right": "issue:44671", + "accept": false, + "reason": "Both are output quality regressions, but on different models/tasks and likely different causes." + }, + { + "left": "issue:43782", + "right": "issue:44977", + "accept": false, + "reason": "Qwen3VL weight_only load error vs Qwen3.5 flash-attention generation issue." + }, + { + "left": "issue:45584", + "right": "issue:45834", + "accept": false, + "reason": "Whisper empty-transcription generation vs Kosmos2.5 long-OCR index error." + }, + { + "left": "issue:43825", + "right": "issue:43994", + "accept": false, + "reason": "Error-message wording about pipeline tasks vs SigLIP2 model/pipeline output correctness." + }, + { + "left": "issue:43322", + "right": "issue:45685", + "accept": false, + "reason": "Segfault loading Llava Next vs MPS histogram_int dtype error in MoE; unrelated." + }, + { + "left": "issue:44561", + "right": "issue:45020", + "accept": false, + "reason": "General remote_code breakage is too broad; these are not the same concrete regression." + }, + { + "left": "issue:44821", + "right": "issue:45084", + "accept": false, + "reason": "AutoImageProcessor URL loading vs template-node compilation TypeError." + }, + { + "left": "issue:39692", + "right": "issue:45834", + "accept": false, + "reason": "SigLIP2 docs/example mistakes vs Kosmos2.5 OCR input indexing." + }, + { + "left": "issue:30064", + "right": "issue:36331", + "accept": false, + "reason": "Image segmentation-map handling vs CustomTrainer loss signature regression." + }, + { + "left": "issue:44485", + "right": "issue:44509", + "accept": false, + "reason": "GLM-5 RoPE discussion vs docs mentioning removed pipeline tasks." + }, + { + "left": "issue:44479", + "right": "issue:45834", + "accept": false, + "reason": "Qwen video-input regression vs Kosmos2.5 OCR indexing; different models and symptoms." + }, + { + "left": "issue:43825", + "right": "issue:43827", + "accept": false, + "reason": "Both are about v5 pipeline removals, but one is an error-message issue and the other is stale docs; not the same underlying bug." + }, + { + "left": "issue:44186", + "right": "issue:44442", + "accept": false, + "reason": "LayoutLMv2 tokenizer crash vs AutoTokenizer loading FastSpeech2ConformerTokenizer." + }, + { + "left": "issue:43577", + "right": "issue:43582", + "accept": false, + "reason": "BLIP2 dtype retention vs Apple Silicon caching_allocator_warmup TypeError." + }, + { + "left": "issue:43606", + "right": "issue:45072", + "accept": false, + "reason": "CPU offload device mismatch vs bfloat16 dtype mismatch in a different model." + } + ] + }, + "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": 16, + "review_unit_index": 15, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:34634", + "issue:36010", + "issue:38175", + "issue:39692", + "issue:42175", + "issue:43262", + "issue:43329", + "issue:43404", + "issue:43452", + "issue:43526", + "issue:43582", + "issue:43638", + "issue:43644", + "issue:43650", + "issue:43653", + "issue:43697", + "issue:43720", + "issue:43742", + "issue:43746", + "issue:43749", + "issue:43782", + "issue:43819", + "issue:43854", + "issue:43906", + "issue:44060", + "issue:44368", + "issue:44554", + "issue:44589", + "issue:44610", + "issue:44671", + "issue:44805", + "issue:44811", + "issue:44857", + "issue:44912", + "issue:44936", + "issue:44977", + "issue:44991", + "issue:45020", + "issue:45042", + "issue:45072", + "issue:45081", + "issue:45084", + "issue:45406", + "issue:45459", + "issue:45538", + "issue:45600", + "issue:45701", + "issue:45803" + ], + "soft_pairs": [ + "issue:34634|issue:36010", + "issue:44805|issue:44936", + "issue:43452|issue:44991", + "issue:44610|issue:45072", + "issue:42175|issue:45081", + "issue:43720|issue:45072", + "issue:44060|issue:44368", + "issue:44671|issue:45701", + "issue:45081|issue:45406", + "issue:43404|issue:43782", + "issue:43650|issue:43906", + "issue:43582|issue:44857", + "issue:45459|issue:45803", + "issue:43262|issue:43819", + "issue:43526|issue:43697", + "issue:45084|issue:45538", + "issue:43644|issue:43749", + "issue:43329|issue:45600", + "issue:43653|issue:44912", + "issue:43742|issue:45020", + "issue:43638|issue:44936", + "issue:43746|issue:44811", + "issue:44589|issue:45042", + "issue:43854|issue:44977", + "issue:38175|issue:44610", + "issue:39692|issue:44912", + "issue:43526|issue:44554" + ], + "prepared_review_unit_hash": "4222a258e0ca48c4fce29a00cf860f2e12a19fce9c539c7d0a9e10efd42298d0", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 36042, + "estimated_input_tokens": 9011, + "estimated_eval_tokens": 18278 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly unrelated issue set; the soft-similarity links look like false positives across different models, processors, loaders, and runtime failures rather than true duplicates.", + "confidence": 0.81, + "canonical_issue_reason": "No single canonical issue stands out: the items cover many distinct bugs in different subsystems and model families.", + "canonical_pr_reason": null, + "best_issue_reason": "If forced to pick a representative, issue:45020 is the broadest/open report, but it is still too generic to serve as a real duplicate anchor.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:34634", + "right": "issue:36010", + "accept": false, + "reason": "Different failures: Bark voice preset behavior vs a GenerationMixin import error." + }, + { + "left": "issue:44805", + "right": "issue:44936", + "accept": false, + "reason": "Different bugs: a mask/index shape mismatch vs trainer state breaking evaluate() after train()." + }, + { + "left": "issue:43452", + "right": "issue:44991", + "accept": false, + "reason": "Both are loading issues, but they affect different tokenizer/model paths and do not appear to be the same defect." + }, + { + "left": "issue:44610", + "right": "issue:45072", + "accept": false, + "reason": "Different models and symptoms: OmDet-Turbo processor size mismatch vs dtype mismatch in SwitchTransformers/TimmWrapperModel." + }, + { + "left": "issue:42175", + "right": "issue:45081", + "accept": false, + "reason": "Unrelated: backend packaging problem vs a Mistral tokenizer regex crash." + }, + { + "left": "issue:43720", + "right": "issue:45072", + "accept": false, + "reason": "Different problems: BitNet packed-weight loading vs bfloat16 dtype mismatches." + }, + { + "left": "issue:44060", + "right": "issue:44368", + "accept": false, + "reason": "Related theme of tied-weight warnings, but the model families and underlying warning bug differ." + }, + { + "left": "issue:44671", + "right": "issue:45701", + "accept": false, + "reason": "CamemBERT masked-LM predictions vs a broad tokenization-regression report; not the same concrete bug." + }, + { + "left": "issue:45081", + "right": "issue:45406", + "accept": false, + "reason": "Different processor/tokenizer crashes in different code paths and models." + }, + { + "left": "issue:43404", + "right": "issue:43782", + "accept": false, + "reason": "Different generation/loading issues: untied lm_head weights vs weight_only load error." + }, + { + "left": "issue:43650", + "right": "issue:43906", + "accept": false, + "reason": "No clear overlap; one is a vague data request, the other is an isolated reproduction of a separate upstream issue." + }, + { + "left": "issue:43582", + "right": "issue:44857", + "accept": false, + "reason": "Different runtime environments and failures: Apple Silicon allocator warmup TypeError vs CUDA float16 AMP crash." + }, + { + "left": "issue:45459", + "right": "issue:45803", + "accept": false, + "reason": "Both involve exception handling, but they are different functions and different error-hiding bugs." + }, + { + "left": "issue:43262", + "right": "issue:43819", + "accept": false, + "reason": "Audio sampling-rate defaulting bug vs DAC STE mismatch; unrelated code paths." + }, + { + "left": "issue:43526", + "right": "issue:43697", + "accept": false, + "reason": "Different model issues: BEiT Fast label reduction vs RTDetrV2 output divergence." + }, + { + "left": "issue:45084", + "right": "issue:45538", + "accept": false, + "reason": "Template compilation error vs tokenizer max_length constant; unrelated defects." + }, + { + "left": "issue:43644", + "right": "issue:43749", + "accept": false, + "reason": "Different serialization/loading bugs: non-persistent buffer corruption vs FSDP CPU RAM efficient loading failure." + }, + { + "left": "issue:43329", + "right": "issue:45600", + "accept": false, + "reason": "Different scopes: multimodal token counting video-path bug vs stale auto_mappings CI breakage." + }, + { + "left": "issue:43653", + "right": "issue:44912", + "accept": false, + "reason": "Tokenizer special-token registration bug vs quantization fallback to bf16; not the same issue." + }, + { + "left": "issue:43742", + "right": "issue:45020", + "accept": false, + "reason": "MobileLLM loading KeyError is model-specific; the other is a broad remote_code regression report." + }, + { + "left": "issue:43638", + "right": "issue:44936", + "accept": false, + "reason": "Different causes and workflows: Deepspeed ZeRO3 index error vs trainer evaluate() state regression." + }, + { + "left": "issue:43746", + "right": "issue:44811", + "accept": false, + "reason": "Different processor loading bug vs Whisper batch_decode skip_special_tokens handling." + }, + { + "left": "issue:44589", + "right": "issue:45042", + "accept": false, + "reason": "Unrelated: missing Float8 storage object vs PIL image processor unexpectedly requiring torchvision." + }, + { + "left": "issue:43854", + "right": "issue:44977", + "accept": false, + "reason": "Different model-loading/test failure vs flash-attention generation bug." + }, + { + "left": "issue:38175", + "right": "issue:44610", + "accept": false, + "reason": "Different symptom and subsystem: zero probabilities in SigLIP2 vs processor/model input-size mismatch." + }, + { + "left": "issue:39692", + "right": "issue:44912", + "accept": false, + "reason": "SigLIP2 docs/example errors are not the same as MXFP4 quantization fallback behavior." + }, + { + "left": "issue:43526", + "right": "issue:44554", + "accept": false, + "reason": "Different models and failure modes; BEiT Fast label reduction is unrelated to the MPS attention correctness bug." + } + ] + }, + "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": 16, + "review_unit_index": 16, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:33357", + "issue:34567", + "issue:34634", + "issue:36010", + "issue:36331", + "issue:40990", + "issue:41720", + "issue:42915", + "issue:43257", + "issue:43262", + "issue:43317", + "issue:43329", + "issue:43450", + "issue:43452", + "issue:43454", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43577", + "issue:43582", + "issue:43643", + "issue:43645", + "issue:43697", + "issue:43720", + "issue:43742", + "issue:43749", + "issue:43756", + "issue:43819", + "issue:43927", + "issue:43931", + "issue:44291", + "issue:44479", + "issue:44560", + "issue:44568", + "issue:44610", + "issue:44811", + "issue:44821", + "issue:44849", + "issue:44912", + "issue:44987", + "issue:45042", + "issue:45356", + "issue:45357", + "issue:45381", + "issue:45488", + "issue:45538", + "issue:45701", + "issue:45759" + ], + "soft_pairs": [ + "issue:36010|issue:43645", + "issue:43454|issue:43577", + "issue:34634|issue:36331", + "issue:43819|issue:43927", + "issue:43257|issue:43329", + "issue:43550|issue:44912", + "issue:43720|issue:44610", + "issue:43749|issue:44291", + "issue:44568|issue:45488", + "issue:43531|issue:45042", + "issue:43262|issue:44811", + "issue:43643|issue:45759", + "issue:44568|issue:45356", + "issue:36010|issue:36331", + "issue:44849|issue:45701", + "issue:43329|issue:43450", + "issue:34567|issue:40990", + "issue:43329|issue:44479", + "issue:43257|issue:43931", + "issue:43697|issue:45357", + "issue:44291|issue:45042", + "issue:43452|issue:43582", + "issue:41720|issue:43540", + "issue:42915|issue:43742", + "issue:43317|issue:45538", + "issue:33357|issue:34567", + "issue:43927|issue:44912", + "issue:44987|issue:45538", + "issue:42915|issue:44821", + "issue:44479|issue:45381", + "issue:42915|issue:45538", + "issue:43257|issue:43756", + "issue:43540|issue:44560", + "issue:43540|issue:44479" + ], + "prepared_review_unit_hash": "47732b72f8fa57bead48e94c59be81e7cbb474e04a5258a0135f932cb4d88b43", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 38058, + "estimated_input_tokens": 9515, + "estimated_eval_tokens": 19286 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a heterogeneous cluster of mostly unrelated Transformers v5 regressions/load/load-time bugs. The soft links are mainly superficial topic overlap (tokenization, video, loading, quantization) rather than the same underlying bug, so none should be merged as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": "No true canonical duplicate stands out; if forced to anchor the cluster, issue 45701 is the broadest umbrella for the recurring version-dependent tokenization regression theme, but it does not cover the full set.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45701 is the best available anchor only because it is broad and open with several inbound references; it is still too generic to represent the whole cluster cleanly.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:36010", + "right": "issue:43645", + "accept": false, + "reason": "Different breakages: GenerationMixin import path vs notebook custom-model initialization under v5." + }, + { + "left": "issue:43454", + "right": "issue:43577", + "accept": false, + "reason": "Unrelated model-specific bugs: AyaVision weight tying vs Blip2 dtype staying float32." + }, + { + "left": "issue:34634", + "right": "issue:36331", + "accept": false, + "reason": "Different APIs and symptoms: BarkProcessor voice_preset vs Trainer.compute_loss kwargs." + }, + { + "left": "issue:43819", + "right": "issue:43927", + "accept": false, + "reason": "Different code paths: DAC latent conversion mismatch vs DiaConfig losing custom token IDs on save/load." + }, + { + "left": "issue:43257", + "right": "issue:43329", + "accept": false, + "reason": "Qwen3 MoE weight conversion issue is distinct from multimodal video token counting undefined-variable bug." + }, + { + "left": "issue:43550", + "right": "issue:44912", + "accept": false, + "reason": "Different failures: torch.compile/SDPA on Bamba vs MXFP4 quantization fallback on gpt-oss-20b." + }, + { + "left": "issue:43720", + "right": "issue:44610", + "accept": false, + "reason": "Packed-weight loading bug in BitNet is unrelated to OmDet-Turbo processor image-size mismatch." + }, + { + "left": "issue:43749", + "right": "issue:44291", + "accept": false, + "reason": "Both are loading-related, but one is FSDP CPU RAM efficient loading and the other is init_empty_weights/_is_hf_initialized." + }, + { + "left": "issue:44568", + "right": "issue:45488", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different tokenizers and different implementation paths." + }, + { + "left": "issue:43531", + "right": "issue:45042", + "accept": false, + "reason": "Sliding-window handling in Qwen3-MoE is unrelated to PIL backend processor torchvision dependency changes." + }, + { + "left": "issue:43262", + "right": "issue:44811", + "accept": false, + "reason": "Audio chat-template sampling-rate default bug is unrelated to Whisper batch_decode skip_special_tokens handling." + }, + { + "left": "issue:43643", + "right": "issue:45759", + "accept": false, + "reason": "Different config bugs: trust_remote_code field loss vs from_config not unwrapping text_config." + }, + { + "left": "issue:44568", + "right": "issue:45356", + "accept": false, + "reason": "Different tokenizer regressions: add_special_tokens behavior vs Kimi-K2.5 codec handling/warning issue." + }, + { + "left": "issue:36010", + "right": "issue:36331", + "accept": false, + "reason": "Unrelated incompatibilities: GenerationMixin import breakage vs custom trainer loss signature change." + }, + { + "left": "issue:44849", + "right": "issue:45701", + "accept": false, + "reason": "Qwen3.5 hidden-states output bug is a model forward-pass issue, not the broad tokenization regression described by 45701." + }, + { + "left": "issue:43329", + "right": "issue:43450", + "accept": false, + "reason": "Both are video-related, but one is undefined vars in token counting and the other is batched-shape handling in video processors." + }, + { + "left": "issue:34567", + "right": "issue:40990", + "accept": false, + "reason": "TrainerState token-count tracking and gpt-oss perplexity are unrelated symptoms and subsystems." + }, + { + "left": "issue:43329", + "right": "issue:44479", + "accept": false, + "reason": "Both mention video regressions, but one is a specific undefined-variable bug and the other is a broader v5.3.0 regression across multiple models." + }, + { + "left": "issue:43257", + "right": "issue:43931", + "accept": false, + "reason": "Different load-time failures: Qwen3 MoE weight conversion vs Qwen3-VL shape mismatch." + }, + { + "left": "issue:43697", + "right": "issue:45357", + "accept": false, + "reason": "RTDetrV2 output drift and Qwen3.5 save_pretrained key corruption are unrelated model-specific issues." + }, + { + "left": "issue:44291", + "right": "issue:45042", + "accept": false, + "reason": "init_empty_weights constructor-arg bug is unrelated to PIL backend requiring torchvision." + }, + { + "left": "issue:43452", + "right": "issue:43582", + "accept": false, + "reason": "gguf_file loading breakage and Apple Silicon caching_allocator_warmup TypeError are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:43540", + "accept": false, + "reason": "Different Qwen model families and different failures: cudaErrorAssert on auto device map vs video-processing ValueError." + }, + { + "left": "issue:42915", + "right": "issue:43742", + "accept": false, + "reason": "FineGrainedFP8Config failure on Qwen3Moe is unrelated to MobileLLM-125M key error on load." + }, + { + "left": "issue:43317", + "right": "issue:45538", + "accept": false, + "reason": "Dequantized model offload failure is unrelated to CLIPTokenizer model_max_length behavior." + }, + { + "left": "issue:33357", + "right": "issue:34567", + "accept": false, + "reason": "MacOS CLIP model bus error and TrainerState token counter bug are unrelated." + }, + { + "left": "issue:43927", + "right": "issue:44912", + "accept": false, + "reason": "DiaConfig token-ID persistence and gpt-oss quantization fallback are different bugs." + }, + { + "left": "issue:44987", + "right": "issue:45538", + "accept": false, + "reason": "Loading physical-intelligence/fast under >=5.1.0 is unrelated to CLIPTokenizer max_length." + }, + { + "left": "issue:42915", + "right": "issue:44821", + "accept": false, + "reason": "Qwen3Moe FP8 failure is unrelated to AutoImageProcessor URL loading." + }, + { + "left": "issue:44479", + "right": "issue:45381", + "accept": false, + "reason": "Both are Qwen video regressions, but one is a broad multi-model regression and the other is a specific qwen2.5-vl vision_position_ids issue." + }, + { + "left": "issue:44568", + "right": "issue:45488", + "accept": false, + "reason": "Both are tokenizer-related, but they concern different tokenizers and distinct regressions." + }, + { + "left": "issue:42915", + "right": "issue:45538", + "accept": false, + "reason": "FineGrainedFP8Config on Qwen3Moe is unrelated to CLIPTokenizer max_length." + }, + { + "left": "issue:43257", + "right": "issue:43756", + "accept": false, + "reason": "Qwen3 MoE weight conversion and Smollm3 RoPE-layer dropping are unrelated model/config bugs." + }, + { + "left": "issue:43540", + "right": "issue:44560", + "accept": false, + "reason": "Qwen3OmniMoe video-processing ValueError is unrelated to gpt-oss-20b MXFP4 loading fallback." + }, + { + "left": "issue:43540", + "right": "issue:44479", + "accept": false, + "reason": "Both are video-related, but they affect different model families and different failure modes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 17, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:28282", + "issue:34567", + "issue:34689", + "issue:36010", + "issue:36331", + "issue:38175", + "issue:42175", + "issue:43257", + "issue:43452", + "issue:43493", + "issue:43526", + "issue:43540", + "issue:43577", + "issue:43643", + "issue:43646", + "issue:43696", + "issue:43697", + "issue:43742", + "issue:43749", + "issue:43756", + "issue:43824", + "issue:43906", + "issue:43994", + "issue:44062", + "issue:44291", + "issue:44488", + "issue:44589", + "issue:44737", + "issue:44756", + "issue:44792", + "issue:44821", + "issue:44898", + "issue:44964", + "issue:44991", + "issue:45042", + "issue:45072", + "issue:45081", + "issue:45137", + "issue:45216", + "issue:45325", + "issue:45375", + "issue:45381", + "issue:45405", + "issue:45479", + "issue:45588", + "issue:45600", + "issue:45823" + ], + "soft_pairs": [ + "issue:43257|issue:43696", + "issue:43577|issue:45072", + "issue:45137|issue:45405", + "issue:44737|issue:45588", + "issue:38175|issue:44898", + "issue:44062|issue:44991", + "issue:43742|issue:44291", + "issue:43540|issue:45823", + "issue:43906|issue:45381", + "issue:43646|issue:43749", + "issue:44062|issue:45081", + "issue:43749|issue:44756", + "issue:43643|issue:45375", + "issue:34567|issue:36331", + "issue:43493|issue:43994", + "issue:36010|issue:42175", + "issue:42175|issue:43824", + "issue:44291|issue:44821", + "issue:43452|issue:44488", + "issue:44792|issue:44964", + "issue:43526|issue:45479", + "issue:28282|issue:34689", + "issue:45042|issue:45081", + "issue:43756|issue:45325", + "issue:43742|issue:44589", + "issue:45405|issue:45600", + "issue:43697|issue:45216" + ], + "prepared_review_unit_hash": "42b4363cdf6d8fbaac6f451ebce7cdf6426a41c9adefc7a485c0162880744724", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35477, + "estimated_input_tokens": 8870, + "estimated_eval_tokens": 17996 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly unrelated bug reports with one plausible duplicate pair around SigLIP2 model correctness. The rest differ by model, API surface, or failure mode and should stay separate.", + "confidence": 0.81, + "canonical_issue_reason": "issue:43493 is the best canonical issue: it describes the underlying SigLIP2 implementation discrepancy at a higher level, while issue:43994 is a user-facing symptom on the same model/checkpoint.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43493 is the strongest representative for the SigLIP2 bug because it frames the root behavior difference, not just one downstream reproduction.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43257", + "right": "issue:43696", + "accept": false, + "reason": "Different problems: Qwen3 MoE weight conversion under accelerate/deepspeed vs GPT-oss CUDA OOM." + }, + { + "left": "issue:43577", + "right": "issue:45072", + "accept": false, + "reason": "Both mention dtype, but they involve different models and different code paths; not the same concrete bug." + }, + { + "left": "issue:45137", + "right": "issue:45405", + "accept": false, + "reason": "DeepSpeed ZeRO3 deque error is unrelated to a PEFT version bump." + }, + { + "left": "issue:44737", + "right": "issue:45588", + "accept": false, + "reason": "XLNet CPU allocation bug and flash-attention crash on sink-less models are unrelated." + }, + { + "left": "issue:38175", + "right": "issue:44898", + "accept": false, + "reason": "Different models and symptoms; one is SigLIP2 zero probabilities, the other is Perceiver resolution handling." + }, + { + "left": "issue:44062", + "right": "issue:44991", + "accept": false, + "reason": "Tokenizer AddedToken special-arg duplication is a different failure from est-roberta tokenizer loading." + }, + { + "left": "issue:43742", + "right": "issue:44291", + "accept": false, + "reason": "MobileLLM key error and init_empty_weights/_is_hf_initialized type error are different load-time issues." + }, + { + "left": "issue:43540", + "right": "issue:45823", + "accept": false, + "reason": "Qwen3Omni video-input error and Gemma4 PLE device mismatch are distinct model-forward bugs." + }, + { + "left": "issue:43906", + "right": "issue:45381", + "accept": false, + "reason": "The reproduction of issue 38071 is not the same bug as Qwen2.5-VL vision_position_ids scaling." + }, + { + "left": "issue:43646", + "right": "issue:43749", + "accept": false, + "reason": "AutoConfig missing fields with trust_remote_code is unrelated to FSDP CPU RAM efficient loading." + }, + { + "left": "issue:44062", + "right": "issue:45081", + "accept": false, + "reason": "AddedToken special-arg duplication and Mistral regex patching crash are different tokenizer bugs." + }, + { + "left": "issue:43749", + "right": "issue:44756", + "accept": false, + "reason": "Both are memory/loading related, but the causes and fixes are different." + }, + { + "left": "issue:43643", + "right": "issue:45375", + "accept": false, + "reason": "Both involve missing config fields, but one is generic trust_remote_code behavior and the other is a specific Qwen3_5MoeVisionConfig field being dropped by @strict." + }, + { + "left": "issue:34567", + "right": "issue:36331", + "accept": false, + "reason": "TrainerState token counting not updating is unrelated to CustomTrainer.compute_loss signature breakage." + }, + { + "left": "issue:43493", + "right": "issue:43994", + "accept": true, + "reason": "Same SigLIP2 underlying model bug: both report incorrect outputs for google/siglip2-base-patch16-224, with one describing the implementation mismatch and the other the user-visible bad results." + }, + { + "left": "issue:36010", + "right": "issue:42175", + "accept": false, + "reason": "GenerationMixin import error is unrelated to backend dependency packaging." + }, + { + "left": "issue:42175", + "right": "issue:43824", + "accept": false, + "reason": "TensorFlow missing from a torch install extra and Qwen2_5_VL import failure are different packaging/import issues." + }, + { + "left": "issue:44291", + "right": "issue:44821", + "accept": false, + "reason": "Unexpected _is_hf_initialized argument and AutoImageProcessor URL loading are unrelated." + }, + { + "left": "issue:43452", + "right": "issue:44488", + "accept": false, + "reason": "gguf_file handling for AutoTokenizer/AutoModel and a model loading failure for cjvt/sleng-bert are different problems." + }, + { + "left": "issue:44792", + "right": "issue:44964", + "accept": false, + "reason": "Janus image-generation test failure and Phi-4-multimodal loading failure are distinct multimodal issues." + }, + { + "left": "issue:43526", + "right": "issue:45479", + "accept": false, + "reason": "BEiT image processor label reduction bug is unrelated to sequence-classification loss degeneracy." + }, + { + "left": "issue:28282", + "right": "issue:34689", + "accept": false, + "reason": "PyTorch-missing ImportError and Llama 3.2 Vision model-loading regression are not the same bug." + }, + { + "left": "issue:45042", + "right": "issue:45081", + "accept": false, + "reason": "Torchvision dependency regression in PIL image processors is unrelated to the Mistral regex tokenizer crash." + }, + { + "left": "issue:43756", + "right": "issue:45325", + "accept": false, + "reason": "Both touch RoPE/positioning, but they affect different models and different logic." + }, + { + "left": "issue:43742", + "right": "issue:44589", + "accept": false, + "reason": "MobileLLM key error and missing Float8 storage object are different loading failures." + }, + { + "left": "issue:45405", + "right": "issue:45600", + "accept": false, + "reason": "Unreleased MIN_PEFT_VERSION bump and stale auto_mappings references are unrelated CI/package issues." + }, + { + "left": "issue:43697", + "right": "issue:45216", + "accept": false, + "reason": "RTDetrV2 output drift in v5 and Qwen3.5 save_pretrained corruption are different regressions." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is grounded in the packet and stays conservative. The SigLIP2 pair (43493/43994) is the only clearly plausible duplicate based on the shared model/checkpoint and overlapping symptom, while the other soft-edge pairs are reasonably rejected as different bugs, models, or failure modes. No overstatement beyond the evidence is apparent." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 18, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:24643", + "issue:28282", + "issue:29127", + "issue:30064", + "issue:33666", + "issue:38175", + "issue:41720", + "issue:41762", + "issue:42175", + "issue:43295", + "issue:43452", + "issue:43531", + "issue:43577", + "issue:43582", + "issue:43643", + "issue:43653", + "issue:43688", + "issue:43697", + "issue:43720", + "issue:43742", + "issue:43819", + "issue:43906", + "issue:44361", + "issue:44442", + "issue:44464", + "issue:44466", + "issue:44484", + "issue:44514", + "issue:44561", + "issue:44568", + "issue:44625", + "issue:44704", + "issue:44743", + "issue:44811", + "issue:44821", + "issue:44843", + "issue:44877", + "issue:44898", + "issue:44936", + "issue:44938", + "issue:44977", + "issue:44991", + "issue:45005", + "issue:45290", + "issue:45356", + "issue:45381", + "issue:45440", + "issue:45823" + ], + "soft_pairs": [ + "issue:44484|issue:44704", + "issue:43531|issue:44821", + "issue:44514|issue:45381", + "issue:28282|issue:29127", + "issue:43582|issue:44977", + "issue:44625|issue:44877", + "issue:43643|issue:44561", + "issue:43452|issue:43577", + "issue:43720|issue:44898", + "issue:42175|issue:44938", + "issue:44466|issue:45005", + "issue:44514|issue:45290", + "issue:41762|issue:44936", + "issue:43452|issue:44843", + "issue:43295|issue:45356", + "issue:43819|issue:44811", + "issue:43653|issue:44568", + "issue:44936|issue:45823", + "issue:24643|issue:29127", + "issue:30064|issue:33666", + "issue:44361|issue:44442", + "issue:38175|issue:43295", + "issue:43653|issue:44361", + "issue:44464|issue:44743", + "issue:43577|issue:43697", + "issue:44843|issue:44991", + "issue:41720|issue:43906", + "issue:44977|issue:45356", + "issue:43688|issue:45440", + "issue:43531|issue:43742" + ], + "prepared_review_unit_hash": "f004e77b9379f33936cc73d52812cc263920772b256022cada7903b672c5a63a", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36783, + "estimated_input_tokens": 9196, + "estimated_eval_tokens": 18648 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is not a single duplicate cluster; it contains many unrelated bug reports across tokenizers, configs, processors, generation, and v5 regressions. I rejected all soft pairs because they do not appear to describe the same concrete underlying bug or change.", + "confidence": 0.93, + "canonical_issue_reason": "No true canonical issue fits the whole set; the cluster is heterogeneous. If forced to pick an anchor, issue 44466 is the most concrete recent v5 regression and the closest to another soft pair about tied weights.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44466 is the least-bad representative anchor, but only loosely; it is about tied-weight serialization in v5 and does not duplicate the rest of the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44484", + "right": "issue:44704", + "accept": false, + "reason": "Different problems: shard-size/default save_pretrained behavior vs kwargs not reaching cached_file in AutoProcessor." + }, + { + "left": "issue:43531", + "right": "issue:44821", + "accept": false, + "reason": "Unrelated: Qwen3-MoE sliding_window behavior vs loading AutoImageProcessor from URL." + }, + { + "left": "issue:44514", + "right": "issue:45381", + "accept": false, + "reason": "Both are Qwen2.5-VL, but one is chat-template batching and the other is video vision_position_ids; distinct bugs." + }, + { + "left": "issue:28282", + "right": "issue:29127", + "accept": false, + "reason": "Importing PyTorch vs LayoutLMv3 box-content validation are unrelated." + }, + { + "left": "issue:43582", + "right": "issue:44977", + "accept": false, + "reason": "Apple Silicon allocator TypeError vs Qwen3.5 flash-attention generation issue; different code paths." + }, + { + "left": "issue:44625", + "right": "issue:44877", + "accept": false, + "reason": "Qwen3.5 num_labels propagation vs granite_speech config strictness are separate config-loading bugs." + }, + { + "left": "issue:43643", + "right": "issue:44561", + "accept": false, + "reason": "Both mention trust_remote_code, but one is missing returned config fields and the other is a removed helper breaking models." + }, + { + "left": "issue:43452", + "right": "issue:43577", + "accept": false, + "reason": "gguf loading/tokenizer breakage vs Blip2 dtype not propagating; no shared underlying defect." + }, + { + "left": "issue:43720", + "right": "issue:44898", + "accept": false, + "reason": "BitNet packed-weight loading vs Perceiver interpolation failure are different model-specific bugs." + }, + { + "left": "issue:42175", + "right": "issue:44938", + "accept": false, + "reason": "Backend dependency packaging vs Python 3.14 import/load failure are unrelated." + }, + { + "left": "issue:44466", + "right": "issue:45005", + "accept": false, + "reason": "Both involve tied weights, but one is a device-dependent serialization regression and the other is a broader translation-model report; not the same concrete bug." + }, + { + "left": "issue:44514", + "right": "issue:45290", + "accept": false, + "reason": "Different chat-template crashes: batched padding behavior vs assistant tool-call messages with no content." + }, + { + "left": "issue:41762", + "right": "issue:44936", + "accept": false, + "reason": "Gemma3 DeepSpeed ZeRO-3 loading failure vs trainer.evaluate after train; separate training/runtime issues." + }, + { + "left": "issue:43452", + "right": "issue:44843", + "accept": false, + "reason": "gguf file support vs offline model_info call in Mistral regex patch are unrelated tokenizer/loading problems." + }, + { + "left": "issue:43295", + "right": "issue:45356", + "accept": false, + "reason": "Processor/tokenizer regression vs Kimi-K2.5 codec/regex warning regression; different symptoms and code paths." + }, + { + "left": "issue:43819", + "right": "issue:44811", + "accept": false, + "reason": "DAC latent conversion mismatch vs Whisper batch_decode skip_special_tokens handling are unrelated." + }, + { + "left": "issue:43653", + "right": "issue:44568", + "accept": false, + "reason": "Both are tokenizer special-token issues, but BigBird mask-token registration and mdeberta BOS/EOS omission are different bugs." + }, + { + "left": "issue:44936", + "right": "issue:45823", + "accept": false, + "reason": "Trainer lifecycle failure vs Gemma4 PLE device mismatch during forward are unrelated." + }, + { + "left": "issue:24643", + "right": "issue:29127", + "accept": false, + "reason": "DeepSpeed training weight-dimension error vs LayoutLMv3 box-info validation are not the same bug." + }, + { + "left": "issue:30064", + "right": "issue:33666", + "accept": false, + "reason": "Void segmentation-map processing and Qwen2-VL multi-GPU training are different image/model workflows." + }, + { + "left": "issue:44361", + "right": "issue:44442", + "accept": false, + "reason": "MLukeTokenizer task AttributeError and AutoTokenizer failing on FastSpeech2ConformerTokenizer are separate tokenizer-load issues." + }, + { + "left": "issue:38175", + "right": "issue:43295", + "accept": false, + "reason": "SigLIP2 zero probabilities vs processor.tokenizer regression are unrelated." + }, + { + "left": "issue:43653", + "right": "issue:44361", + "accept": false, + "reason": "BigBird special-token decoding bug vs MLukeTokenizer task AttributeError are distinct tokenizer issues." + }, + { + "left": "issue:44464", + "right": "issue:44743", + "accept": false, + "reason": "Chunked generation with compiled forward and Qwen3.5 recurrent-state reset are different generation/cache bugs." + }, + { + "left": "issue:43577", + "right": "issue:43697", + "accept": false, + "reason": "Blip2 dtype propagation vs RTDetrV2 output divergence are unrelated model behaviors." + }, + { + "left": "issue:44843", + "right": "issue:44991", + "accept": false, + "reason": "Offline model_info regression vs tokenizer loading failure for a specific model are not the same defect." + }, + { + "left": "issue:41720", + "right": "issue:43906", + "accept": false, + "reason": "Qwen3 auto device mapping cudaErrorAssert vs an isolated reproduction of another issue are not duplicates." + }, + { + "left": "issue:44977", + "right": "issue:45356", + "accept": false, + "reason": "Qwen3.5 flash-attention generation bug vs Kimi tokenizer codec regression are unrelated." + }, + { + "left": "issue:43688", + "right": "issue:45440", + "accept": false, + "reason": "Aux-loss normalization in MoE models vs DeepSeekV3 implementation divergence are separate MoE issues." + }, + { + "left": "issue:43531", + "right": "issue:43742", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and MobileLLM key error do not share a concrete bug path." + } + ] + }, + "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": 16, + "review_unit_index": 19, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:28282", + "issue:32090", + "issue:33357", + "issue:33666", + "issue:34567", + "issue:35141", + "issue:36010", + "issue:36331", + "issue:39692", + "issue:41720", + "issue:43122", + "issue:43262", + "issue:43295", + "issue:43329", + "issue:43334", + "issue:43452", + "issue:43493", + "issue:43504", + "issue:43525", + "issue:43550", + "issue:43572", + "issue:43577", + "issue:43606", + "issue:43653", + "issue:43688", + "issue:43720", + "issue:43742", + "issue:43749", + "issue:43756", + "issue:43825", + "issue:43906", + "issue:43927", + "issue:44361", + "issue:44451", + "issue:44464", + "issue:44554", + "issue:44568", + "issue:44610", + "issue:44821", + "issue:44849", + "issue:44913", + "issue:44977", + "issue:45325", + "issue:45357", + "issue:45588", + "issue:45704", + "issue:45812" + ], + "soft_pairs": [ + "issue:43653|issue:43927", + "issue:44464|issue:44977", + "issue:43577|issue:45588", + "issue:43550|issue:43927", + "issue:43550|issue:44361", + "issue:32090|issue:34567", + "issue:43504|issue:43550", + "issue:43504|issue:43720", + "issue:43122|issue:43906", + "issue:32090|issue:33666", + "issue:28282|issue:35141", + "issue:43504|issue:44610", + "issue:43452|issue:44451", + "issue:44849|issue:45357", + "issue:44568|issue:45812", + "issue:43742|issue:44821", + "issue:41720|issue:44464", + "issue:33666|issue:36010", + "issue:43756|issue:44913", + "issue:43329|issue:45325", + "issue:43688|issue:43749", + "issue:43493|issue:43825", + "issue:43525|issue:43572", + "issue:43262|issue:43295", + "issue:43606|issue:43927", + "issue:43720|issue:43927", + "issue:28282|issue:32090", + "issue:28282|issue:33357", + "issue:43577|issue:44554", + "issue:33666|issue:36331", + "issue:39692|issue:43493", + "issue:43720|issue:44361", + "issue:43688|issue:45704", + "issue:43334|issue:43572" + ], + "prepared_review_unit_hash": "e7bbd8d92b83c3a7f0361853b46666fa40d0deb54687ea18cc764990025a23c4", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 34, + "serialized_chars": 37612, + "estimated_input_tokens": 9403, + "estimated_eval_tokens": 19062 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is broad and mostly heterogeneous, with many separate Transformers regressions across tokenization, loading, trainer, and multimodal code paths. I found no PRs, and I do not see any soft-pair that looks like the same underlying bug with enough certainty to merge.", + "confidence": 0.42, + "canonical_issue_reason": "issue:45812 is the strongest representative of the cluster\u2019s broad v4\u2192v5 tokenizer regression theme: it is open, concrete, and impacts multiple Granite models rather than a single narrow repro.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45812 is the best global issue candidate because it describes a high-impact, cross-model tokenizer regression and is phrased broadly enough to anchor related reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43653", + "right": "issue:43927", + "accept": false, + "reason": "Different failure classes: special-token decode behavior in BigBirdTokenizer vs config token ID persistence and generation IndexError in DiaConfig." + }, + { + "left": "issue:44464", + "right": "issue:44977", + "accept": false, + "reason": "Different code paths: compiled chunked generation inconsistency vs Qwen3.5 flash-attention generation failure." + }, + { + "left": "issue:43577", + "right": "issue:45588", + "accept": false, + "reason": "Unrelated bugs: Blip2 dtype loading stays float32 vs flash_attention.py crashing on sink-less models with s_aux=None." + }, + { + "left": "issue:43550", + "right": "issue:43927", + "accept": false, + "reason": "Different areas: torch.compile/SDPA runtime failure vs custom token ID loss and generation IndexError after save/load." + }, + { + "left": "issue:43550", + "right": "issue:44361", + "accept": false, + "reason": "Model execution bug vs tokenizer AttributeError; they do not share the same underlying path." + }, + { + "left": "issue:32090", + "right": "issue:34567", + "accept": false, + "reason": "Trainer broadcast NoneType error and TrainerState token-count update bug are separate trainer issues." + }, + { + "left": "issue:43504", + "right": "issue:43550", + "accept": false, + "reason": "Legacy preset loading failure for BeitForSemanticSegmentation is not the same as Bamba SDPA/compile failure." + }, + { + "left": "issue:43504", + "right": "issue:43720", + "accept": false, + "reason": "Different load-time problems: legacy field handling vs BitNet packed-weight unpacking during accelerate loading." + }, + { + "left": "issue:43122", + "right": "issue:43906", + "accept": false, + "reason": "Both mention tokenization, but 43906 is only an isolated reproduction of another issue and does not clearly match the same tokenizer regression." + }, + { + "left": "issue:32090", + "right": "issue:33666", + "accept": false, + "reason": "Trainer GPU broadcast error is unrelated to Qwen2-VL multi-GPU training behavior." + }, + { + "left": "issue:28282", + "right": "issue:35141", + "accept": false, + "reason": "ImportError for missing PyTorch is unrelated to embedding reinitialization after resize." + }, + { + "left": "issue:43504", + "right": "issue:44610", + "accept": false, + "reason": "Model preset loading and processor/model image-size mismatch are different bugs." + }, + { + "left": "issue:43452", + "right": "issue:44451", + "accept": false, + "reason": "GGUF loading breakage and inability to load a specific model are not the same code-path issue." + }, + { + "left": "issue:44849", + "right": "issue:45357", + "accept": false, + "reason": "Hidden-states generation bug and incorrect visual encoder keys on save are separate Qwen3.5 problems." + }, + { + "left": "issue:44568", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions, but one is mdeberta special-token handling and the other is Granite token IDs; no clear shared underlying bug." + }, + { + "left": "issue:43742", + "right": "issue:44821", + "accept": false, + "reason": "KeyError loading MobileLLM is unrelated to AutoImageProcessor loading from URL." + }, + { + "left": "issue:41720", + "right": "issue:44464", + "accept": false, + "reason": "cudaErrorAssert under auto device mapping is a different failure than compiled-forward chunked generation inconsistency." + }, + { + "left": "issue:33666", + "right": "issue:36010", + "accept": false, + "reason": "Multi-GPU training for Qwen2-VL and GenerationMixin import failure are unrelated." + }, + { + "left": "issue:43756", + "right": "issue:44913", + "accept": false, + "reason": "Smollm3 RoPE-layer mismatch and GPTNeoX rotary_pct reload defaulting are different model/config bugs." + }, + { + "left": "issue:43329", + "right": "issue:45325", + "accept": false, + "reason": "Video multimodal token counting bug is not the same as still-image temporal position scaling in Qwen2.5-VL." + }, + { + "left": "issue:43688", + "right": "issue:43749", + "accept": false, + "reason": "Aux-loss normalization and FSDP CPU RAM efficient loading are separate problems." + }, + { + "left": "issue:43493", + "right": "issue:43825", + "accept": false, + "reason": "SigLIP2 implementation discrepancy and pipeline error-message regression do not share the same bug." + }, + { + "left": "issue:43525", + "right": "issue:43572", + "accept": false, + "reason": "Similar theme, but different configs and missing fields: pad_token_id in Llama4Config vs pad_token_idx in StableLmConfig." + }, + { + "left": "issue:43262", + "right": "issue:43295", + "accept": false, + "reason": "Audio chat-template sample-rate defaulting is unrelated to processor.tokenizer access/images-to-tokenizer regression." + }, + { + "left": "issue:43606", + "right": "issue:43927", + "accept": false, + "reason": "CPU offload device mismatch is a different bug from custom token ID loss and generation failure." + }, + { + "left": "issue:43720", + "right": "issue:43927", + "accept": false, + "reason": "BitNet packed-weight loading failure and DiaConfig token persistence/generation bug are unrelated." + }, + { + "left": "issue:28282", + "right": "issue:32090", + "accept": false, + "reason": "Missing PyTorch ImportError is not the same as Trainer GPU broadcast of NoneType." + }, + { + "left": "issue:28282", + "right": "issue:33357", + "accept": false, + "reason": "PyTorch missing ImportError and MacOS bus error with CLIP are unrelated failures." + }, + { + "left": "issue:43577", + "right": "issue:44554", + "accept": false, + "reason": "Blip2 dtype loading and MPS attention correctness with differing head dims are distinct issues." + }, + { + "left": "issue:33666", + "right": "issue:36331", + "accept": false, + "reason": "Multi-GPU training issue and unexpected compute_loss keyword argument are separate trainer-related bugs." + }, + { + "left": "issue:39692", + "right": "issue:43493", + "accept": false, + "reason": "Documentation example errors are not the same as a SigLIP2 implementation discrepancy." + }, + { + "left": "issue:43720", + "right": "issue:44361", + "accept": false, + "reason": "BitNet accelerate-loading bug and MLukeTokenizer AttributeError are different subsystems." + }, + { + "left": "issue:43688", + "right": "issue:45704", + "accept": false, + "reason": "Aux-loss normalization and apex FusedRMSNorm memory leak are unrelated." + }, + { + "left": "issue:43334", + "right": "issue:43572", + "accept": false, + "reason": "Qwen3-VL missing pad_token_id and StableLmConfig missing pad_token_idx are different model/config regressions." + } + ] + }, + "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": 16, + "review_unit_index": 20, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:33357", + "issue:33666", + "issue:39692", + "issue:43257", + "issue:43278", + "issue:43295", + "issue:43299", + "issue:43334", + "issue:43493", + "issue:43504", + "issue:43540", + "issue:43572", + "issue:43575", + "issue:43606", + "issue:43646", + "issue:43653", + "issue:43688", + "issue:43756", + "issue:43824", + "issue:43844", + "issue:43927", + "issue:43994", + "issue:44112", + "issue:44291", + "issue:44361", + "issue:44423", + "issue:44466", + "issue:44479", + "issue:44488", + "issue:44493", + "issue:44521", + "issue:44589", + "issue:44625", + "issue:44734", + "issue:44792", + "issue:44843", + "issue:44964", + "issue:44977", + "issue:45083", + "issue:45200", + "issue:45229", + "issue:45290", + "issue:45405", + "issue:45520", + "issue:45701", + "issue:45759", + "issue:45823" + ], + "soft_pairs": [ + "issue:44964|issue:45405", + "issue:43540|issue:44977", + "issue:44479|issue:44625", + "issue:43493|issue:43844", + "issue:43334|issue:45759", + "issue:45200|issue:45229", + "issue:43756|issue:45083", + "issue:44488|issue:44843", + "issue:43257|issue:43299", + "issue:43278|issue:43844", + "issue:43646|issue:44466", + "issue:43824|issue:44589", + "issue:45229|issue:45823", + "issue:44589|issue:45520", + "issue:43688|issue:45701", + "issue:39692|issue:43295", + "issue:33357|issue:33666", + "issue:43572|issue:44291", + "issue:43504|issue:43606", + "issue:44423|issue:44734", + "issue:43927|issue:44361", + "issue:43504|issue:43653", + "issue:44625|issue:45759", + "issue:43575|issue:45405", + "issue:43606|issue:44112", + "issue:44792|issue:45405", + "issue:43994|issue:44493", + "issue:44521|issue:45290" + ], + "prepared_review_unit_hash": "952113fe6001b6fa7a1620e5b3cbdfe2c4b8fede404dd9e471e5233024ea3480", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35705, + "estimated_input_tokens": 8927, + "estimated_eval_tokens": 18110 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose collection of unrelated Transformers bug reports, mostly spanning Qwen/Gemma/SigLIP/multimodal loading, config propagation, serving, and tokenizer regressions. There is no clear duplicate center and no PRs in the set.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue stands out; the issues cover many different models and failure modes rather than one shared bug.", + "canonical_pr_reason": null, + "best_issue_reason": "Best treated as a non-duplicate cluster: the most prominent theme is broad regressions in model/config loading, but the reports are still too diverse to collapse into one issue.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44964", + "right": "issue:45405", + "accept": false, + "reason": "Different problems: multimodal model loading failure vs an unreleased PEFT version bump." + }, + { + "left": "issue:43540", + "right": "issue:44977", + "accept": false, + "reason": "Both are Qwen-related, but one is a video processing ValueError and the other is a flash-attention generation issue." + }, + { + "left": "issue:44479", + "right": "issue:44625", + "accept": false, + "reason": "Unrelated Qwen bugs: video input regression vs num_labels propagation in config." + }, + { + "left": "issue:43493", + "right": "issue:43844", + "accept": false, + "reason": "Different subsystems and symptoms: SigLIP2 implementation mismatch vs DeepSpeed gradient scaling anomaly." + }, + { + "left": "issue:43334", + "right": "issue:45759", + "accept": false, + "reason": "Both involve Qwen multimodal configs, but one is a missing pad_token_id attribute and the other is from_config text_config unwrapping." + }, + { + "left": "issue:45200", + "right": "issue:45229", + "accept": false, + "reason": "Text-only Gemma4 token-type defaulting is unrelated to multi-GPU OOM." + }, + { + "left": "issue:43756", + "right": "issue:45083", + "accept": false, + "reason": "Different model families and bugs: Smollm3 RoPE layer loss vs qwen3_omni_moe helper behavior." + }, + { + "left": "issue:44488", + "right": "issue:44843", + "accept": false, + "reason": "Unrelated: model load failure for cjvt/sleng-bert vs offline tokenization patch calling model_info()." + }, + { + "left": "issue:43257", + "right": "issue:43299", + "accept": false, + "reason": "Both are Qwen MoE loading issues, but one is accelerate/deepspeed weight conversion and the other is a dev-load regression for Qwen3VL MoE." + }, + { + "left": "issue:43278", + "right": "issue:43844", + "accept": false, + "reason": "Different issues: embedding dtype drift in evaluation vs abnormal gradient growth under ZeRO-3." + }, + { + "left": "issue:43646", + "right": "issue:44466", + "accept": false, + "reason": "Custom model init breakage is not the same as lm_head.weight serialization varying by device." + }, + { + "left": "issue:43824", + "right": "issue:44589", + "accept": false, + "reason": "Different failures: missing Qwen2_5_VL import vs missing Float8 storage type." + }, + { + "left": "issue:45229", + "right": "issue:45823", + "accept": false, + "reason": "Same model family, but OOM during multi-GPU inference is not the same bug as device mismatch in forward." + }, + { + "left": "issue:44589", + "right": "issue:45520", + "accept": false, + "reason": "Unrelated runtime errors: Float8 storage lookup vs flash_attn KeyError on Python 3.13." + }, + { + "left": "issue:43688", + "right": "issue:45701", + "accept": false, + "reason": "Different topics entirely: auxiliary loss normalization vs tokenization changes across versions." + }, + { + "left": "issue:39692", + "right": "issue:43295", + "accept": false, + "reason": "Both are multimodal/processor issues, but one is a doc example mismatch and the other is a regression in processor.tokenizer usage." + }, + { + "left": "issue:33357", + "right": "issue:33666", + "accept": false, + "reason": "Completely unrelated: MacOS bus error with CLIP vs Qwen2-VL multi-GPU training." + }, + { + "left": "issue:43572", + "right": "issue:44291", + "accept": false, + "reason": "Both are initialization/config regressions, but they affect different code paths and error signatures." + }, + { + "left": "issue:43504", + "right": "issue:43606", + "accept": false, + "reason": "Different models and failures: legacy field in Beit loading vs CPU offload device mismatch in bark-small." + }, + { + "left": "issue:44423", + "right": "issue:44734", + "accept": false, + "reason": "Both concern serve, but one crashes on multimodal input conversion and the other on KV-cache continuation indexing." + }, + { + "left": "issue:43927", + "right": "issue:44361", + "accept": false, + "reason": "Different tokenizer/config bugs with different failure modes and models." + }, + { + "left": "issue:43504", + "right": "issue:43653", + "accept": false, + "reason": "Beit legacy-field loading failure is unrelated to BigBirdTokenizer special-token registration." + }, + { + "left": "issue:44625", + "right": "issue:45759", + "accept": false, + "reason": "Related Qwen config area, but num_labels propagation and text_config unwrapping are distinct bugs." + }, + { + "left": "issue:43575", + "right": "issue:45405", + "accept": false, + "reason": "TP OOM on Qwen2-57B is unrelated to a version constraint bump." + }, + { + "left": "issue:43606", + "right": "issue:44112", + "accept": false, + "reason": "A real model bug vs a stale CI test; not the same underlying issue." + }, + { + "left": "issue:44792", + "right": "issue:45405", + "accept": false, + "reason": "Janus image-generation test failure is unrelated to PEFT version availability." + }, + { + "left": "issue:43994", + "right": "issue:44493", + "accept": false, + "reason": "SigLIP2 pipeline/nonsensical output issue is unrelated to unexpected position-id key warnings." + }, + { + "left": "issue:44521", + "right": "issue:45290", + "accept": false, + "reason": "Both use apply_chat_template, but one is all-zero assistant masks and the other is a crash on tool-call messages with no content." + } + ] + }, + "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": 16, + "review_unit_index": 21, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:15354", + "issue:28218", + "issue:29942", + "issue:30333", + "issue:33357", + "issue:34567", + "issue:37428", + "issue:38175", + "issue:43232", + "issue:43278", + "issue:43295", + "issue:43334", + "issue:43450", + "issue:43493", + "issue:43525", + "issue:43540", + "issue:43572", + "issue:43688", + "issue:43701", + "issue:43824", + "issue:43844", + "issue:43873", + "issue:43957", + "issue:43975", + "issue:44315", + "issue:44464", + "issue:44466", + "issue:44554", + "issue:44568", + "issue:44610", + "issue:44849", + "issue:44898", + "issue:44936", + "issue:44964", + "issue:44987", + "issue:45072", + "issue:45084", + "issue:45127", + "issue:45198", + "issue:45200", + "issue:45325", + "issue:45356", + "issue:45412", + "issue:45446", + "issue:45520", + "issue:45584", + "issue:45701", + "issue:45823" + ], + "soft_pairs": [ + "issue:44464|issue:45584", + "issue:29942|issue:30333", + "issue:43844|issue:44964", + "issue:28218|issue:34567", + "issue:45084|issue:45520", + "issue:44554|issue:45446", + "issue:44568|issue:45701", + "issue:43688|issue:43873", + "issue:43540|issue:45200", + "issue:15354|issue:38175", + "issue:43232|issue:44849", + "issue:45200|issue:45823", + "issue:45072|issue:45198", + "issue:44315|issue:44849", + "issue:43278|issue:43701", + "issue:28218|issue:33357", + "issue:44554|issue:45356", + "issue:45200|issue:45325", + "issue:43295|issue:44610", + "issue:44466|issue:45127", + "issue:44898|issue:45412", + "issue:43493|issue:43701", + "issue:43572|issue:44987", + "issue:37428|issue:43957", + "issue:43334|issue:43525", + "issue:43572|issue:43957", + "issue:43873|issue:44464", + "issue:43295|issue:43450", + "issue:43844|issue:44936", + "issue:43824|issue:43975", + "issue:43295|issue:43873", + "issue:38175|issue:45412", + "issue:37428|issue:43824" + ], + "prepared_review_unit_hash": "232ad769121337cd3469cb8d9f5da0af93f68ab86d9f4133d2830354539193cc", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37692, + "estimated_input_tokens": 9423, + "estimated_eval_tokens": 19102 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is highly heterogeneous: it spans tokenization, generation, multimodal processors, training state, import errors, and device/dtype issues. I found no soft pairs that look like the same underlying bug/change, so no duplicate merges are justified.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 45701 is the broadest, most umbrella-like report (tokenization regressions across versions) and is the closest thing to a cluster anchor, though the cluster is not truly one duplicate set.", + "canonical_pr_reason": null, + "best_issue_reason": "45701 is the least-bad global representative because it is broad and user-facing, but it still does not meaningfully cover the rest of the unrelated issues.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44464", + "right": "issue:45584", + "accept": false, + "reason": "Different code paths: compiled forward/chunked generation vs Whisper empty-transcription generation after special-token alignment." + }, + { + "left": "issue:29942", + "right": "issue:30333", + "accept": false, + "reason": "Unrelated subsystems: FlashAttention2 test failures vs MLflow reporting after failed training." + }, + { + "left": "issue:43844", + "right": "issue:44964", + "accept": false, + "reason": "Different bugs: ZeRO-3 gradient growth vs model loading failure for Phi-4 multimodal." + }, + { + "left": "issue:28218", + "right": "issue:34567", + "accept": false, + "reason": "Tokenizer spacing bug vs TrainerState token counter not updating; no shared underlying change." + }, + { + "left": "issue:45084", + "right": "issue:45520", + "accept": false, + "reason": "Compilation/parsing error vs import_utils flash-attn KeyError on Python 3.13." + }, + { + "left": "issue:44554", + "right": "issue:45446", + "accept": false, + "reason": "MPS attention correctness issue vs a PyTorch version check/import bug in flex_attention." + }, + { + "left": "issue:44568", + "right": "issue:45701", + "accept": false, + "reason": "Specific special-token insertion regression vs a broader tokenization-version regression; not the same concrete bug." + }, + { + "left": "issue:43688", + "right": "issue:43873", + "accept": false, + "reason": "Aux-loss normalization bug vs offloading/quantization behavior; different functionality." + }, + { + "left": "issue:43540", + "right": "issue:45200", + "accept": false, + "reason": "Video input processing error in Qwen3OmniMoe vs Gemma4 multimodal token-type defaults." + }, + { + "left": "issue:15354", + "right": "issue:38175", + "accept": false, + "reason": "TorchScript GeneratorExp export failure vs SigLIP2 zero-probability inference issue." + }, + { + "left": "issue:43232", + "right": "issue:44849", + "accept": false, + "reason": "Generation kwargs update after sync_gpus vs Qwen3.5 hidden-states bug; unrelated generation paths." + }, + { + "left": "issue:45200", + "right": "issue:45823", + "accept": false, + "reason": "Different Gemma4 issues: missing mm_token_type_ids defaults vs device mismatch with device_map='auto'." + }, + { + "left": "issue:45072", + "right": "issue:45198", + "accept": false, + "reason": "BF16 dtype mismatch issues vs Wav2Vec2 save/tokenization failure; no same code path." + }, + { + "left": "issue:44315", + "right": "issue:44849", + "accept": false, + "reason": "Liger Kernel application in model_init vs Qwen3.5 hidden-states output bug." + }, + { + "left": "issue:43278", + "right": "issue:43701", + "accept": false, + "reason": "Training/eval dtype change vs resume_from_checkpoint key mismatch; unrelated." + }, + { + "left": "issue:28218", + "right": "issue:33357", + "accept": false, + "reason": "Tokenizer added-space bug vs MacOS CLIP bus error; unrelated." + }, + { + "left": "issue:44554", + "right": "issue:45356", + "accept": false, + "reason": "Attention correctness on MPS vs Kimi-K2.5 tokenizer regression; different model paths." + }, + { + "left": "issue:45200", + "right": "issue:45325", + "accept": false, + "reason": "Gemma4 token-type defaulting vs Qwen2.5-VL RoPE temporal-position scaling; unrelated multimodal bugs." + }, + { + "left": "issue:43295", + "right": "issue:44610", + "accept": false, + "reason": "Processor API regression vs image-size mismatch in OmDet-Turbo; not the same defect." + }, + { + "left": "issue:44466", + "right": "issue:45127", + "accept": false, + "reason": "Serialization/tied-weights inconsistency vs LoRA merge collapse with extended vocab; related theme but different bug." + }, + { + "left": "issue:44898", + "right": "issue:45412", + "accept": false, + "reason": "Perceiver resolution handling vs RT-DETR memory release; unrelated." + }, + { + "left": "issue:43493", + "right": "issue:43701", + "accept": false, + "reason": "SigLIP2 implementation discrepancy vs checkpoint resume key mismatch; no shared bug." + }, + { + "left": "issue:43572", + "right": "issue:44987", + "accept": false, + "reason": "StableLm config attribute regression vs a model-loading failure in a different package/version path." + }, + { + "left": "issue:37428", + "right": "issue:43957", + "accept": false, + "reason": "Flash-attn import symbol error vs meta-device model loading breakage; different failures." + }, + { + "left": "issue:43334", + "right": "issue:43525", + "accept": false, + "reason": "Same symptom class (missing pad_token_id) but different models/configs; insufficient evidence of one shared bug." + }, + { + "left": "issue:43572", + "right": "issue:43957", + "accept": false, + "reason": "Config attribute missing vs meta-device loading regression; not the same underlying issue." + }, + { + "left": "issue:43873", + "right": "issue:44464", + "accept": false, + "reason": "Quantization/offloading behavior vs compiled-forward generation inconsistency; unrelated." + }, + { + "left": "issue:43295", + "right": "issue:43450", + "accept": false, + "reason": "Processor/tokenizer regression vs batched video output shape bug; different concrete problems." + }, + { + "left": "issue:43844", + "right": "issue:44936", + "accept": false, + "reason": "DeepSpeed gradient anomaly vs evaluate-after-train failure; no common root cause evident." + }, + { + "left": "issue:43824", + "right": "issue:43975", + "accept": false, + "reason": "Import error for a specific Qwen2.5-VL class vs DeepSeek tokenizer detokenization regression; unrelated." + }, + { + "left": "issue:43295", + "right": "issue:43873", + "accept": false, + "reason": "Processor API change vs quantization offloading; different subsystems." + }, + { + "left": "issue:38175", + "right": "issue:45412", + "accept": false, + "reason": "SigLIP2 inference probabilities vs RT-DETR memory cleanup; unrelated." + }, + { + "left": "issue:37428", + "right": "issue:43824", + "accept": false, + "reason": "Flash-attn import failure vs missing Qwen2_5_VL export; different import issues 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": 2, + "component_count": 16, + "review_unit_index": 22, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:37428", + "issue:38175", + "issue:39692", + "issue:42915", + "issue:43262", + "issue:43278", + "issue:43335", + "issue:43425", + "issue:43450", + "issue:43493", + "issue:43526", + "issue:43550", + "issue:43575", + "issue:43606", + "issue:43611", + "issue:43653", + "issue:43819", + "issue:43824", + "issue:43825", + "issue:43827", + "issue:43856", + "issue:43901", + "issue:43931", + "issue:44112", + "issue:44206", + "issue:44351", + "issue:44360", + "issue:44442", + "issue:44448", + "issue:44462", + "issue:44514", + "issue:44521", + "issue:44561", + "issue:44610", + "issue:44743", + "issue:44792", + "issue:44857", + "issue:45083", + "issue:45127", + "issue:45198", + "issue:45200", + "issue:45245", + "issue:45290", + "issue:45381", + "issue:45563", + "issue:45584", + "issue:45588", + "issue:45593" + ], + "soft_pairs": [ + "issue:42915|issue:43824", + "issue:43278|issue:43856", + "issue:44514|issue:44521", + "issue:44442|issue:45198", + "issue:38175|issue:43493", + "issue:43827|issue:43901", + "issue:43653|issue:44112", + "issue:43450|issue:43493", + "issue:43450|issue:43575", + "issue:44792|issue:45563", + "issue:44360|issue:44561", + "issue:45127|issue:45245", + "issue:43819|issue:45083", + "issue:44206|issue:45584", + "issue:43819|issue:44112", + "issue:43262|issue:43450", + "issue:44610|issue:44857", + "issue:37428|issue:43425", + "issue:44857|issue:45593", + "issue:43278|issue:43825", + "issue:45290|issue:45381", + "issue:43335|issue:43550", + "issue:43606|issue:45083", + "issue:45200|issue:45588", + "issue:44448|issue:44743", + "issue:37428|issue:44351", + "issue:43611|issue:43931", + "issue:39692|issue:43526", + "issue:44462|issue:44561" + ], + "prepared_review_unit_hash": "ef307fc5de44a06f9e98b036f2836d749c0ddd4e97e1ab241a63cb8b35a4b22d", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36392, + "estimated_input_tokens": 9098, + "estimated_eval_tokens": 18452 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: the soft pairs are mostly different bugs in different models/subsystems, so none should be merged as duplicates. All soft edges are rejected.", + "confidence": 0.97, + "canonical_issue_reason": "No clear canonical issue: the set covers unrelated bugs across imports, models, processors, and docs rather than one underlying defect.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44521 is the strongest standalone representative: it is open, discussion-heavy, and describes a concrete multimodal apply_chat_template failure.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42915", + "right": "issue:43824", + "accept": false, + "reason": "Different problems: Qwen3Moe FP8 config failure vs a Qwen2.5-VL import error." + }, + { + "left": "issue:43278", + "right": "issue:43856", + "accept": false, + "reason": "Unrelated symptoms: evaluation dtype drift vs Qwen3 MoE training memory usage." + }, + { + "left": "issue:44514", + "right": "issue:44521", + "accept": false, + "reason": "Both involve apply_chat_template, but one is a batched padding crash and the other is all-zero assistant masks; different bugs." + }, + { + "left": "issue:44442", + "right": "issue:45198", + "accept": false, + "reason": "Both are tokenizer-related, but they affect different model families with different failure modes." + }, + { + "left": "issue:38175", + "right": "issue:43493", + "accept": false, + "reason": "Same model family, but zero probabilities and HF-vs-JAX discrepancy are not clearly the same concrete bug." + }, + { + "left": "issue:43827", + "right": "issue:43901", + "accept": false, + "reason": "Both are docs issues around pipelines, but they reference different APIs and different pages." + }, + { + "left": "issue:43653", + "right": "issue:44112", + "accept": false, + "reason": "A tokenizer special-token bug and a GraniteSpeech CI device-override failure are unrelated." + }, + { + "left": "issue:43450", + "right": "issue:43493", + "accept": false, + "reason": "Video batch shape bug vs SigLIP2 implementation discrepancy; different code paths." + }, + { + "left": "issue:43450", + "right": "issue:43575", + "accept": false, + "reason": "Incorrect video processor output shape is unrelated to Qwen2-57B TP OOM." + }, + { + "left": "issue:44792", + "right": "issue:45563", + "accept": false, + "reason": "A Janus image-generation test failure and a paged generate warning are different issues." + }, + { + "left": "issue:44360", + "right": "issue:44561", + "accept": false, + "reason": "DSA indexer discussion vs trust_remote_code breakage from removed helper; no shared defect." + }, + { + "left": "issue:45127", + "right": "issue:45245", + "accept": false, + "reason": "LoRA/tied embeddings collapse is unrelated to the category-count runtime error." + }, + { + "left": "issue:43819", + "right": "issue:45083", + "accept": false, + "reason": "DAC forward/latents mismatch and qwen3_omni_moe length helper behavior are separate bugs." + }, + { + "left": "issue:44206", + "right": "issue:45584", + "accept": false, + "reason": "Whisper empty-transcription failure and LasrFeatureExtractor center-arg crash are different audio bugs." + }, + { + "left": "issue:43819", + "right": "issue:44112", + "accept": false, + "reason": "DAC reconstruction mismatch is unrelated to GraniteSpeech CI device override behavior." + }, + { + "left": "issue:43262", + "right": "issue:43450", + "accept": false, + "reason": "Audio sampling-rate defaulting and video batch-shape issues are unrelated." + }, + { + "left": "issue:44610", + "right": "issue:44857", + "accept": false, + "reason": "Processor size mismatch and AMP/CUDA loss crash are different vision/detection failures." + }, + { + "left": "issue:37428", + "right": "issue:43425", + "accept": false, + "reason": "Missing internal import and Torch 2.10 incompatibility are too broad and not the same bug." + }, + { + "left": "issue:44857", + "right": "issue:45593", + "accept": false, + "reason": "A mixed-precision loss crash and missing auxiliary losses are unrelated D-FINE issues." + }, + { + "left": "issue:43278", + "right": "issue:43825", + "accept": false, + "reason": "Embedding dtype drift and a stale pipeline error message are different regression types." + }, + { + "left": "issue:45290", + "right": "issue:45381", + "accept": false, + "reason": "Chat-template tool-call crash and Qwen2.5-VL vision_position_ids error are distinct multimodal bugs." + }, + { + "left": "issue:43335", + "right": "issue:43550", + "accept": false, + "reason": "SwitchTransformers config construction and Bamba torch.compile/SDPA failure are unrelated." + }, + { + "left": "issue:43606", + "right": "issue:45083", + "accept": false, + "reason": "CPU offload device mismatch and qwen3_omni_moe feature-length helper behavior are unrelated." + }, + { + "left": "issue:45200", + "right": "issue:45588", + "accept": false, + "reason": "Gemma mm_token_type_ids handling and flash_attention s_aux=None are different model-path failures." + }, + { + "left": "issue:44448", + "right": "issue:44743", + "accept": false, + "reason": "Pegasus output differences across versions and qwen3_5 recurrent-state reset are unrelated." + }, + { + "left": "issue:37428", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they concern different missing symbols and likely different regressions." + }, + { + "left": "issue:43611", + "right": "issue:43931", + "accept": false, + "reason": "Base_model_prefix loading breakage and Qwen3-VL weight-shape mismatch are different loading problems." + }, + { + "left": "issue:39692", + "right": "issue:43526", + "accept": false, + "reason": "SigLIP2 doc example issues and BeitImageProcessorFast reduce_labels behavior are unrelated." + }, + { + "left": "issue:44462", + "right": "issue:44561", + "accept": false, + "reason": "AutoTokenizer ignoring tokenizer.json and removed is_torch_fx_available are different loading/runtime regressions." + } + ] + }, + "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": 16, + "review_unit_index": 23, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:28218", + "issue:29127", + "issue:29942", + "issue:32090", + "issue:34689", + "issue:35141", + "issue:38175", + "issue:39692", + "issue:40444", + "issue:43335", + "issue:43388", + "issue:43504", + "issue:43526", + "issue:43550", + "issue:43606", + "issue:43653", + "issue:43701", + "issue:43720", + "issue:43819", + "issue:43927", + "issue:44112", + "issue:44206", + "issue:44292", + "issue:44448", + "issue:44493", + "issue:44561", + "issue:44568", + "issue:44779", + "issue:44855", + "issue:44857", + "issue:44977", + "issue:45072", + "issue:45083", + "issue:45198", + "issue:45229", + "issue:45310", + "issue:45372", + "issue:45459", + "issue:45478", + "issue:45491", + "issue:45520", + "issue:45563", + "issue:45588", + "issue:45593", + "issue:45698", + "issue:45712", + "issue:45803", + "issue:45823" + ], + "soft_pairs": [ + "issue:44206|issue:45823", + "issue:44561|issue:44779", + "issue:43388|issue:43526", + "issue:43335|issue:43653", + "issue:43819|issue:44493", + "issue:45310|issue:45478", + "issue:43550|issue:44112", + "issue:44855|issue:45083", + "issue:29942|issue:32090", + "issue:43335|issue:43504", + "issue:43335|issue:45083", + "issue:39692|issue:44493", + "issue:44292|issue:45520", + "issue:38175|issue:44493", + "issue:44448|issue:44779", + "issue:28218|issue:35141", + "issue:43335|issue:45072", + "issue:43335|issue:45593", + "issue:43720|issue:44112", + "issue:44977|issue:45229", + "issue:44448|issue:44977", + "issue:44568|issue:45491", + "issue:43606|issue:44857", + "issue:45459|issue:45712", + "issue:44857|issue:45072", + "issue:29127|issue:32090", + "issue:45198|issue:45803", + "issue:44977|issue:45563", + "issue:43927|issue:45698", + "issue:34689|issue:35141", + "issue:45372|issue:45588", + "issue:40444|issue:43526", + "issue:43701|issue:45083", + "issue:43550|issue:45083" + ], + "prepared_review_unit_hash": "4e47f8eb728e23f5fd36f1cf01050aa4dc0d7d9044de3b02bcdd1533f48fe9fe", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 38209, + "estimated_input_tokens": 9553, + "estimated_eval_tokens": 19362 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most items are unrelated bugs across different models/subsystems. Only 45310 and 45478 look like the same underlying regression; the rest should stay separate.", + "confidence": 0.94, + "canonical_issue_reason": "If a canonical issue is needed, issue 45310 is the best anchor because 45478 is effectively the same Qwen3.5 MoE from_pretrained error with the same title and context, and 45310 is the earlier, fuller report.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45310 is the strongest issue representative in this set, but only for the exact duplicate pair with 45478; the broader collection is too heterogeneous to treat as one issue cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44206", + "right": "issue:45823", + "accept": false, + "reason": "Different bugs: unsupported feature arg in LasrFeatureExtractor vs Gemma4 PLE device mismatch." + }, + { + "left": "issue:44561", + "right": "issue:44779", + "accept": false, + "reason": "Different code paths: v5 fx helper removal breaking trust_remote_code vs DeepSeek tokenizer output regression." + }, + { + "left": "issue:43388", + "right": "issue:43526", + "accept": false, + "reason": "Both involve labels, but one is gather_for_metrics batching logic and the other is BeitImageProcessorFast label reduction." + }, + { + "left": "issue:43335", + "right": "issue:43653", + "accept": false, + "reason": "Unrelated: SwitchTransformers sparse-layer config bug vs BigBirdTokenizer special-token registration issue." + }, + { + "left": "issue:43819", + "right": "issue:44493", + "accept": false, + "reason": "Different failures: DAC latent/STE mismatch vs unexpected position-id keys across models." + }, + { + "left": "issue:45310", + "right": "issue:45478", + "accept": true, + "reason": "Same issue title and same Qwen3.5 MoE from_pretrained regression; these are duplicates." + }, + { + "left": "issue:43550", + "right": "issue:44112", + "accept": false, + "reason": "Different model/subsystem: Bamba SDPA torch.compile bug vs GraniteSpeech CI device override test failure." + }, + { + "left": "issue:44855", + "right": "issue:45083", + "accept": false, + "reason": "Different regressions: Python 3.13 import/IndentationError vs qwen3_omni_moe feature-length helper behavior." + }, + { + "left": "issue:29942", + "right": "issue:32090", + "accept": false, + "reason": "Flash Attention 2 test failures are unrelated to Trainer _gpu_broadcast_one NoneType broadcast errors." + }, + { + "left": "issue:43335", + "right": "issue:43504", + "accept": false, + "reason": "Different models and causes: SwitchTransformers config vs BEiT semantic-segmentation legacy-field loading." + }, + { + "left": "issue:43335", + "right": "issue:45083", + "accept": false, + "reason": "Different bugs in different components; same broad theme of model/config issues is not enough." + }, + { + "left": "issue:39692", + "right": "issue:44493", + "accept": false, + "reason": "SigLIP2 doc/example errors are unrelated to the position-id key regression." + }, + { + "left": "issue:44292", + "right": "issue:45520", + "accept": false, + "reason": "Different issues: Qwen-3 NVFP4 runtime vs import_utils flash_attn KeyError on Python 3.13." + }, + { + "left": "issue:38175", + "right": "issue:44493", + "accept": false, + "reason": "siglip2 zero-probability output bug is unrelated to position-id key warnings/errors." + }, + { + "left": "issue:44448", + "right": "issue:44779", + "accept": false, + "reason": "Different models and failures: Pegasus v4/v5 output drift vs DeepSeek tokenizer regression." + }, + { + "left": "issue:28218", + "right": "issue:35141", + "accept": false, + "reason": "Tokenizer spacing bug vs embedding reinitialization after resize; distinct underlying problems." + }, + { + "left": "issue:43335", + "right": "issue:45072", + "accept": false, + "reason": "Different failure modes: sparse-layer config bug vs bfloat16 dtype mismatch in inference." + }, + { + "left": "issue:43335", + "right": "issue:45593", + "accept": false, + "reason": "SwitchTransformers config bug and D-FINE auxiliary-loss behavior are unrelated." + }, + { + "left": "issue:43720", + "right": "issue:44112", + "accept": false, + "reason": "Packed-weight loading bug is unrelated to GraniteSpeech stale device-override CI failure." + }, + { + "left": "issue:44977", + "right": "issue:45229", + "accept": false, + "reason": "Different problems: flash-attention generation issue vs Gemma4 multi-GPU OOM." + }, + { + "left": "issue:44448", + "right": "issue:44977", + "accept": false, + "reason": "Different models and symptoms; not the same generation bug." + }, + { + "left": "issue:44568", + "right": "issue:45491", + "accept": false, + "reason": "Tokenizer BOS/EOS omission is unrelated to Gemma3 NaN embeddings from sliding-window attention." + }, + { + "left": "issue:43606", + "right": "issue:44857", + "accept": false, + "reason": "CPU offload device mismatch and float16 AMP CUDA crash are distinct bugs." + }, + { + "left": "issue:45459", + "right": "issue:45712", + "accept": false, + "reason": "Tokenizer error masking is unrelated to leftover dummy class cleanup/check_repo issues." + }, + { + "left": "issue:44857", + "right": "issue:45072", + "accept": false, + "reason": "Different components and symptoms: AMP crash in LwDetrImageLoss vs dtype mismatch in SwitchTransformers/TimmWrapperModel." + }, + { + "left": "issue:29127", + "right": "issue:32090", + "accept": false, + "reason": "LayoutLMv3 error-message clarity issue is unrelated to Trainer broadcast TypeError." + }, + { + "left": "issue:45198", + "right": "issue:45803", + "accept": false, + "reason": "Wav2Vec2 save/tokenization failure is unrelated to Fuyu bare-except exception handling." + }, + { + "left": "issue:43927", + "right": "issue:45698", + "accept": false, + "reason": "Both concern save/load, but one loses custom token IDs while the other loads the wrong custom module; different root causes." + }, + { + "left": "issue:34689", + "right": "issue:35141", + "accept": false, + "reason": "Llama 3.2 Vision model-loading regression is unrelated to embedding reinitialization after resize." + }, + { + "left": "issue:45372", + "right": "issue:45588", + "accept": false, + "reason": "Gemma processor import dependency break and flash_attention.py s_aux=None crash are different problems." + }, + { + "left": "issue:40444", + "right": "issue:43526", + "accept": false, + "reason": "Qwen2.5-VL iterable multi-image finetuning failure is unrelated to BEiT label-reduction behavior." + }, + { + "left": "issue:43701", + "right": "issue:45083", + "accept": false, + "reason": "resume_from_checkpoint key mismatch is unrelated to qwen3_omni_moe feature-length helper behavior." + }, + { + "left": "issue:43550", + "right": "issue:45083", + "accept": false, + "reason": "Bamba torch.compile SDPA issue is unrelated to qwen3_omni_moe output-length helper behavior." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and conservative overall. The only accepted soft pair, issue:45310 vs issue:45478, is supported by the identical title and matching Qwen3.5 MoE from_pretrained regression. The remaining pair judgments stay appropriately cautious and do not overclaim shared root causes." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 24, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:25251", + "issue:29942", + "issue:30064", + "issue:32090", + "issue:33357", + "issue:35141", + "issue:36010", + "issue:37428", + "issue:39692", + "issue:42915", + "issue:42994", + "issue:43065", + "issue:43066", + "issue:43278", + "issue:43381", + "issue:43412", + "issue:43421", + "issue:43653", + "issue:43873", + "issue:43976", + "issue:43994", + "issue:44186", + "issue:44206", + "issue:44265", + "issue:44361", + "issue:44448", + "issue:44451", + "issue:44462", + "issue:44466", + "issue:44479", + "issue:44493", + "issue:44734", + "issue:44743", + "issue:44857", + "issue:44871", + "issue:44977", + "issue:45198", + "issue:45245", + "issue:45372", + "issue:45406", + "issue:45412", + "issue:45440", + "issue:45459", + "issue:45491", + "issue:45520", + "issue:45563", + "issue:45588", + "issue:45823" + ], + "soft_pairs": [ + "issue:43412|issue:45588", + "issue:45245|issue:45563", + "issue:43873|issue:44493", + "issue:29942|issue:35141", + "issue:36010|issue:37428", + "issue:45372|issue:45459", + "issue:43066|issue:44462", + "issue:44857|issue:45588", + "issue:45588|issue:45823", + "issue:44448|issue:44466", + "issue:43278|issue:43381", + "issue:43065|issue:45412", + "issue:25251|issue:32090", + "issue:43653|issue:44743", + "issue:44451|issue:45520", + "issue:43976|issue:45440", + "issue:44186|issue:45198", + "issue:42915|issue:42994", + "issue:43994|issue:45412", + "issue:43421|issue:45491", + "issue:44361|issue:45198", + "issue:44871|issue:44977", + "issue:44265|issue:45198", + "issue:44206|issue:44479", + "issue:29942|issue:30064", + "issue:43065|issue:43994", + "issue:44734|issue:45406", + "issue:33357|issue:39692" + ], + "prepared_review_unit_hash": "a87ac3cf7f88d35d752e6b6b4cd1042ee21803b528f3a621811311a5e36e601d", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36149, + "estimated_input_tokens": 9038, + "estimated_eval_tokens": 18332 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: the soft pairs span unrelated tokenizer, export, flash-attention, quantization, generation, and model-loading issues. None look like the same underlying bug or a mergeable duplicate PR pair, so all soft edges are rejected.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue stands out because the cluster is not a coherent duplicate set; the pairings cover many unrelated failure modes across different subsystems.", + "canonical_pr_reason": null, + "best_issue_reason": "If a representative had to be chosen, issue 45588 is the most central among the flash-attention-adjacent subset, but it is still a poor global representative for the mixed cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43412", + "right": "issue:45588", + "accept": false, + "reason": "Gemma3n export failures and a sink-less flash_attention AttributeError are different code paths and symptoms." + }, + { + "left": "issue:45245", + "right": "issue:45563", + "accept": false, + "reason": "Category-count overflow in segmentation is unrelated to a stale num_return_sequences warning in paged generate()." + }, + { + "left": "issue:43873", + "right": "issue:44493", + "accept": false, + "reason": "Quantization/offloading behavior and an unexpected position_id key are distinct bugs with different triggers." + }, + { + "left": "issue:29942", + "right": "issue:35141", + "accept": false, + "reason": "Flash Attention 2 test failures do not match the token-embedding resize/post_init reinitialization bug." + }, + { + "left": "issue:36010", + "right": "issue:37428", + "accept": false, + "reason": "These are separate import errors for different symbols from different modules." + }, + { + "left": "issue:45372", + "right": "issue:45459", + "accept": false, + "reason": "Gemma 4 processor import breakage from mistral_common is unrelated to protobuf masking tokenizer errors." + }, + { + "left": "issue:43066", + "right": "issue:44462", + "accept": false, + "reason": "Both involve tokenizers, but one is a decoder-type regression and the other is AutoTokenizer ignoring tokenizer.json; not the same fix." + }, + { + "left": "issue:44857", + "right": "issue:45588", + "accept": false, + "reason": "An AMP/CUDA loss crash and a flash_attention s_aux=None AttributeError are different failures." + }, + { + "left": "issue:45588", + "right": "issue:45823", + "accept": false, + "reason": "flash_attention sink-less crash and Gemma4 PLE device-mismatch are unrelated model/runtime issues." + }, + { + "left": "issue:44448", + "right": "issue:44466", + "accept": false, + "reason": "Pegasus output drift and lm_head.weight serialization inconsistency are both v5 regressions, but not the same concrete code-path bug." + }, + { + "left": "issue:43278", + "right": "issue:43381", + "accept": false, + "reason": "Embedding dtype drift in eval and gradient-checkpointing in eval mode are distinct behaviors." + }, + { + "left": "issue:43065", + "right": "issue:45412", + "accept": false, + "reason": "A dummy Conv2d in Sam3PixelDecoder is unrelated to RT-DETR memory not releasing on deletion." + }, + { + "left": "issue:25251", + "right": "issue:32090", + "accept": false, + "reason": "Pipeline top_k nesting and Trainer _gpu_broadcast_one NoneType errors are unrelated." + }, + { + "left": "issue:43653", + "right": "issue:44743", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and Qwen modular recurrent-state reset are different tokenizer/model bugs." + }, + { + "left": "issue:44451", + "right": "issue:45520", + "accept": false, + "reason": "A model-loading failure for ScandiBERT and a Python 3.13 flash_attn import KeyError are unrelated." + }, + { + "left": "issue:43976", + "right": "issue:45440", + "accept": false, + "reason": "Python version compatibility and DeepseekV3MoE behavioral divergence are different issues." + }, + { + "left": "issue:44186", + "right": "issue:45198", + "accept": false, + "reason": "LayoutLMv2Tokenizer NER/padding crashes and Wav2Vec2 save_pretrained/tokenization failures are separate tokenizer bugs." + }, + { + "left": "issue:42915", + "right": "issue:42994", + "accept": false, + "reason": "Qwen3Moe FineGrainedFP8Config failure and quantized model saving failure are different quantization problems." + }, + { + "left": "issue:43994", + "right": "issue:45412", + "accept": false, + "reason": "SigLIP2 AutoModel/pipeline nonsensical results and RT-DETR memory leaks do not share a bug path." + }, + { + "left": "issue:43421", + "right": "issue:45491", + "accept": false, + "reason": "Runtime special-token post-processing updates and Gemma3 NaN embeddings with mixed-length batches are unrelated." + }, + { + "left": "issue:44361", + "right": "issue:45198", + "accept": false, + "reason": "MLukeTokenizer AttributeError on tasks and Wav2Vec2 save/tokenization failure are different model-specific issues." + }, + { + "left": "issue:44871", + "right": "issue:44977", + "accept": false, + "reason": "Gemma-3 EOS config inconsistency and Qwen3.5 flash-attention generation failure are unrelated." + }, + { + "left": "issue:44265", + "right": "issue:45198", + "accept": false, + "reason": "torch.export failure with torch_compilable_check and Wav2Vec2 save/tokenization failure are not the same defect." + }, + { + "left": "issue:44206", + "right": "issue:44479", + "accept": false, + "reason": "An unsupported feature-extractor arg and a video-input regression in Qwen models are separate issues." + }, + { + "left": "issue:29942", + "right": "issue:30064", + "accept": false, + "reason": "Flash Attention 2 test failures are unrelated to void segmentation map processing." + }, + { + "left": "issue:43065", + "right": "issue:43994", + "accept": false, + "reason": "Sam3PixelDecoder's dummy Conv2d and SigLIP2 model/processor mismatch are different model-family bugs." + }, + { + "left": "issue:44734", + "right": "issue:45406", + "accept": false, + "reason": "KV-cache continuation indexing in serve and Gemma4Processor missing _tokenizer are distinct failures." + }, + { + "left": "issue:33357", + "right": "issue:39692", + "accept": false, + "reason": "A MacOS bus error on a community CLIP model and documentation example errors for SigLIP2 are not the same bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 25, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:29127", + "issue:29942", + "issue:32090", + "issue:33357", + "issue:34567", + "issue:36010", + "issue:41720", + "issue:42915", + "issue:43262", + "issue:43278", + "issue:43329", + "issue:43334", + "issue:43381", + "issue:43441", + "issue:43526", + "issue:43531", + "issue:43575", + "issue:43653", + "issue:43701", + "issue:43723", + "issue:43756", + "issue:43824", + "issue:43873", + "issue:43931", + "issue:44186", + "issue:44206", + "issue:44279", + "issue:44442", + "issue:44464", + "issue:44554", + "issue:44610", + "issue:44743", + "issue:44811", + "issue:44857", + "issue:44977", + "issue:45072", + "issue:45103", + "issue:45229", + "issue:45245", + "issue:45372", + "issue:45406", + "issue:45412", + "issue:45440", + "issue:45491", + "issue:45588", + "issue:45593", + "issue:45834" + ], + "soft_pairs": [ + "issue:29127|issue:34567", + "issue:44610|issue:45412", + "issue:45103|issue:45372", + "issue:43756|issue:45440", + "issue:43873|issue:45229", + "issue:43723|issue:44554", + "issue:45372|issue:45406", + "issue:43278|issue:43575", + "issue:43381|issue:43701", + "issue:43526|issue:45834", + "issue:43653|issue:44442", + "issue:41720|issue:45229", + "issue:43278|issue:43329", + "issue:45440|issue:45593", + "issue:29942|issue:36010", + "issue:43334|issue:43531", + "issue:32090|issue:33357", + "issue:43526|issue:44186", + "issue:44977|issue:45245", + "issue:43824|issue:44279", + "issue:43526|issue:44857", + "issue:42915|issue:43931", + "issue:43262|issue:44206", + "issue:44464|issue:45491", + "issue:43441|issue:45588", + "issue:44442|issue:45072", + "issue:44743|issue:44811", + "issue:43575|issue:43701" + ], + "prepared_review_unit_hash": "172ebcef6fa69da1d861ff450aeb74e386fa38594486b83101c931a37de3ca32", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35524, + "estimated_input_tokens": 8881, + "estimated_eval_tokens": 18018 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: the soft-similarity pairs share broad vocabulary at most, but they describe distinct bugs, models, and code paths. I would not merge any of the suggested pairs as duplicates.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 45229 is the best anchor only by process criteria: it is a concrete, well-described open bug with high discussion/inbound activity, while the rest of the set appears unrelated.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45229 is the strongest representative issue: detailed, reproducible, and actively discussed. It is still not a duplicate of the other items, just the most substantive single issue in the mixed cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:29127", + "right": "issue:34567", + "accept": false, + "reason": "LayoutLMv3 box-content error message vs TrainerState token counter update; different components and symptoms." + }, + { + "left": "issue:44610", + "right": "issue:45412", + "accept": false, + "reason": "Processor image-size mismatch vs RT-DETR memory not released on deletion; unrelated bugs." + }, + { + "left": "issue:45103", + "right": "issue:45372", + "accept": false, + "reason": "auto_docstring annotations crash vs Gemma4 processor dependency import failure; different failure modes." + }, + { + "left": "issue:43756", + "right": "issue:45440", + "accept": false, + "reason": "Smollm3 RoPE-layer bug vs DeepSeekV3MoE implementation divergence; different model logic." + }, + { + "left": "issue:43873", + "right": "issue:45229", + "accept": false, + "reason": "Quantization offloading issue vs Gemma4 multi-GPU CUDA OOM; not the same code-path problem." + }, + { + "left": "issue:43723", + "right": "issue:44554", + "accept": false, + "reason": "Tokenizer loading regression vs MPS attention correctness bug; unrelated areas." + }, + { + "left": "issue:45372", + "right": "issue:45406", + "accept": false, + "reason": "Gemma4 loading import error vs Gemma4Processor missing _tokenizer in serve; same family, different bugs." + }, + { + "left": "issue:43278", + "right": "issue:43575", + "accept": false, + "reason": "Embedding dtype drift in eval vs Qwen2-57B tensor-parallel OOM; distinct issues." + }, + { + "left": "issue:43381", + "right": "issue:43701", + "accept": false, + "reason": "Gradient checkpointing eval-mode error vs checkpoint resume key mismatch; different training paths." + }, + { + "left": "issue:43526", + "right": "issue:45834", + "accept": false, + "reason": "BeitImageProcessorFast reduce_labels bug vs Kosmos2.5 long-ocr index error; unrelated processors." + }, + { + "left": "issue:43653", + "right": "issue:44442", + "accept": false, + "reason": "BigBirdTokenizer special-token decode bug vs AutoTokenizer loading failure for a speech tokenizer; different defects." + }, + { + "left": "issue:41720", + "right": "issue:45229", + "accept": false, + "reason": "Qwen3 auto device mapping cuda assert vs Gemma4 multi-GPU OOM; different models and failures." + }, + { + "left": "issue:43278", + "right": "issue:43329", + "accept": false, + "reason": "BF16-to-FP32 eval dtype issue vs multimodal token counting video-branch undefined variables; unrelated code paths." + }, + { + "left": "issue:45440", + "right": "issue:45593", + "accept": false, + "reason": "DeepSeekV3MoE divergence vs D-FINE auxiliary-loss handling; different model behavior." + }, + { + "left": "issue:29942", + "right": "issue:36010", + "accept": false, + "reason": "Flash Attention 2 test failures vs GenerationMixin import error; unrelated regression types." + }, + { + "left": "issue:43334", + "right": "issue:43531", + "accept": false, + "reason": "Qwen3-VL pad_token_id loading error vs Qwen3-MoE sliding_window issue; different configuration bugs." + }, + { + "left": "issue:32090", + "right": "issue:33357", + "accept": false, + "reason": "Trainer _gpu_broadcast_one NoneType vs MacOS CLIP bus error; unrelated runtime failures." + }, + { + "left": "issue:43526", + "right": "issue:44186", + "accept": false, + "reason": "BeitImageProcessorFast label reduction bug vs LayoutLMv2Tokenizer padding/truncation crash; different components." + }, + { + "left": "issue:44977", + "right": "issue:45245", + "accept": false, + "reason": "Qwen3.5 flash-attention generation bug vs category-count runtime limit; no shared underlying bug." + }, + { + "left": "issue:43824", + "right": "issue:44279", + "accept": false, + "reason": "Specific Qwen2.5-VL import error vs vague dependency issue; not the same concrete problem." + }, + { + "left": "issue:43526", + "right": "issue:44857", + "accept": false, + "reason": "BeitImageProcessorFast label reduction bug vs LwDetrImageLoss AMP/CUDA crash; different models and code paths." + }, + { + "left": "issue:42915", + "right": "issue:43931", + "accept": false, + "reason": "Qwen3Moe FineGrainedFP8Config failure vs Qwen3-VL weight-shape mismatch; same family only broadly, not same bug." + }, + { + "left": "issue:43262", + "right": "issue:44206", + "accept": false, + "reason": "Audio chat-template sampling-rate default vs unsupported center arg in LasrFeatureExtractor; unrelated audio bugs." + }, + { + "left": "issue:44464", + "right": "issue:45491", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency vs Gemma3 NaN embeddings from sliding-window padding; different symptoms and models." + }, + { + "left": "issue:43441", + "right": "issue:45588", + "accept": false, + "reason": "Ministral-3 flash-attention failure vs sink-less flash_attention.py s_aux=None crash; both mention flash attention but not the same code-path issue." + }, + { + "left": "issue:44442", + "right": "issue:45072", + "accept": false, + "reason": "FastSpeech2ConformerTokenizer loading failure vs dtype mismatch in bfloat16 inference; unrelated issues." + }, + { + "left": "issue:44743", + "right": "issue:44811", + "accept": false, + "reason": "Qwen3_5 recurrent-state reset bug vs Whisper batch_decode skip_special_tokens regression; different components." + }, + { + "left": "issue:43575", + "right": "issue:43701", + "accept": false, + "reason": "Qwen2-57B tensor-parallel OOM vs resume_from_checkpoint key mismatch; 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": 2, + "component_count": 16, + "review_unit_index": 26, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:28218", + "issue:29127", + "issue:32090", + "issue:33290", + "issue:33357", + "issue:34567", + "issue:34689", + "issue:35141", + "issue:38175", + "issue:39692", + "issue:41628", + "issue:42907", + "issue:43316", + "issue:43381", + "issue:43452", + "issue:43493", + "issue:43525", + "issue:43550", + "issue:43575", + "issue:43611", + "issue:43646", + "issue:43761", + "issue:43827", + "issue:43873", + "issue:43931", + "issue:44077", + "issue:44079", + "issue:44188", + "issue:44387", + "issue:44442", + "issue:44479", + "issue:44568", + "issue:44743", + "issue:44898", + "issue:44964", + "issue:44977", + "issue:45216", + "issue:45290", + "issue:45372", + "issue:45440", + "issue:45479", + "issue:45491", + "issue:45584", + "issue:45588", + "issue:45698", + "issue:45759", + "issue:45803" + ], + "soft_pairs": [ + "issue:43646|issue:44077", + "issue:43452|issue:45588", + "issue:44442|issue:45803", + "issue:29127|issue:35141", + "issue:43575|issue:44387", + "issue:44977|issue:45584", + "issue:33357|issue:38175", + "issue:43316|issue:43525", + "issue:44964|issue:45440", + "issue:43761|issue:44079", + "issue:44479|issue:45290", + "issue:28218|issue:29127", + "issue:42907|issue:45698", + "issue:43931|issue:45216", + "issue:28218|issue:32090", + "issue:33290|issue:34567", + "issue:39692|issue:45491", + "issue:44442|issue:44898", + "issue:34689|issue:43611", + "issue:43761|issue:44188", + "issue:43381|issue:43873", + "issue:43493|issue:43827", + "issue:43550|issue:44743", + "issue:44568|issue:45479", + "issue:43316|issue:45759", + "issue:41628|issue:45372" + ], + "prepared_review_unit_hash": "2ea87565962f40d5eaec5961bf7c676cfe94c331df6f12769902b75b44f9a920", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 26, + "serialized_chars": 35204, + "estimated_input_tokens": 8801, + "estimated_eval_tokens": 17858 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "All suggested duplicate edges look like soft lexical matches rather than the same underlying bug or change. The items span unrelated tokenizer, model-loading, quantization, config, and generation regressions, so there is no single coherent canonical issue for the cluster.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue fits this set: the items cover many unrelated bugs, and the proposed duplicate pairs are not the same concrete code-path problem.", + "canonical_pr_reason": null, + "best_issue_reason": "No global best issue is suitable as a cluster representative because the cluster is heterogeneous and does not collapse to one bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43646", + "right": "issue:44077", + "accept": false, + "reason": "Both mention initialization/post_init, but one is about custom model init in Transformers 5.0 and the other is a patchtsmixer-specific policy change; not the same bug." + }, + { + "left": "issue:43452", + "right": "issue:45588", + "accept": false, + "reason": "One is about gguf_file loading for tokenizer/model classes; the other is a flash-attention crash on s_aux=None. Different code paths and failures." + }, + { + "left": "issue:44442", + "right": "issue:45803", + "accept": false, + "reason": "Tokenizer loading failure for FastSpeech2ConformerTokenizer is unrelated to FuyuBatchFeature swallowing exceptions during tensor conversion." + }, + { + "left": "issue:29127", + "right": "issue:35141", + "accept": false, + "reason": "LayoutLMv3 error-message clarity and embedding reinitialization in post_init are unrelated issues." + }, + { + "left": "issue:43575", + "right": "issue:44387", + "accept": false, + "reason": "Both are OOM reports, but one is tensor-parallel loading for Qwen2-57B and the other is increased reserved memory under int4 quantization; different causes and fixes." + }, + { + "left": "issue:44977", + "right": "issue:45584", + "accept": false, + "reason": "Qwen3.5 flash-attention generation failure and Whisper empty-transcription failure after align_special_tokens are different generation paths." + }, + { + "left": "issue:33357", + "right": "issue:38175", + "accept": false, + "reason": "A MacOS bus error on a community CLIP model is not the same as zero probabilities on siglip2-base-patch16-224." + }, + { + "left": "issue:43316", + "right": "issue:43525", + "accept": false, + "reason": "Different config API bugs: Gemma3TextConfig parity vs missing pad_token_id on Llama4Config." + }, + { + "left": "issue:44964", + "right": "issue:45440", + "accept": false, + "reason": "Phi-4 multimodal loading regression and DeepseekV3MoE divergence from remote implementation are distinct model-specific issues." + }, + { + "left": "issue:43761", + "right": "issue:44079", + "accept": false, + "reason": "CLIPVisionModel returning hidden_states=None is not the same bug as ModelOutput key assignment when a prior value was None." + }, + { + "left": "issue:44479", + "right": "issue:45290", + "accept": false, + "reason": "Video-input regression for Qwen VL models and chat-template crashes on tool-call assistant messages are separate failures." + }, + { + "left": "issue:28218", + "right": "issue:29127", + "accept": false, + "reason": "Tokenizer adds an extra space after added tokens; the other is a LayoutLMv3 error-message issue. No shared code-path." + }, + { + "left": "issue:42907", + "right": "issue:45698", + "accept": false, + "reason": "Saving dequantized models and loading the wrong custom module after save_pretrained are related to persistence generally, but not the same defect." + }, + { + "left": "issue:43931", + "right": "issue:45216", + "accept": false, + "reason": "Weight-shape mismatch on Qwen3-VL loading is different from incorrect saved checkpoints in Qwen3.5 save_pretrained." + }, + { + "left": "issue:28218", + "right": "issue:32090", + "accept": false, + "reason": "Tokenizer spacing behavior and a Trainer NoneType GPU broadcast TypeError are unrelated." + }, + { + "left": "issue:33290", + "right": "issue:34567", + "accept": false, + "reason": "Adafactor/deepspeed OOM and TrainerState.num_input_tokens_seen not updating are different trainer/runtime issues." + }, + { + "left": "issue:39692", + "right": "issue:45491", + "accept": false, + "reason": "SigLIP2 documentation example problems and Gemma3 NaN embeddings under mixed-length batching are unrelated." + }, + { + "left": "issue:44442", + "right": "issue:44898", + "accept": false, + "reason": "AutoTokenizer loading failure for FastSpeech2ConformerTokenizer is not the same as Perceiver classification failing at non-default resolution." + }, + { + "left": "issue:34689", + "right": "issue:43611", + "accept": false, + "reason": "Both are loading-related, but one is a specific Llama 3.2 Vision regression and the other is a broad base_model_prefix compatibility break." + }, + { + "left": "issue:43761", + "right": "issue:44188", + "accept": false, + "reason": "Hidden_states output handling and attention-kernel divergence under torch.compile are different layers and symptoms." + }, + { + "left": "issue:43381", + "right": "issue:43873", + "accept": false, + "reason": "Gradient checkpointing in eval mode is unrelated to offloading failures under quantization." + }, + { + "left": "issue:43493", + "right": "issue:43827", + "accept": false, + "reason": "SigLIP2 implementation mismatch and docs still using pipeline() after v5 removals are not the same underlying bug." + }, + { + "left": "issue:43550", + "right": "issue:44743", + "accept": false, + "reason": "torch.compile/SDPA failure in Bamba-9B-v2 is distinct from recurrent state reset behavior in modular_qwen3_5." + }, + { + "left": "issue:44568", + "right": "issue:45479", + "accept": false, + "reason": "Missing BOS/EOS insertion for a tokenizer is unrelated to zero-loss behavior in sequence classification." + }, + { + "left": "issue:43316", + "right": "issue:45759", + "accept": false, + "reason": "Gemma3TextConfig API inconsistency and Qwen multimodal config unwrapping in from_config are separate configuration-loading issues." + }, + { + "left": "issue:41628", + "right": "issue:45372", + "accept": false, + "reason": "Missing AutoImageProcessor import and Gemma 4 processor loading failing on ReasoningEffort import are different import errors with different dependencies." + } + ] + }, + "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": 16, + "review_unit_index": 27, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:25251", + "issue:28282", + "issue:29127", + "issue:32090", + "issue:33290", + "issue:33357", + "issue:34689", + "issue:36010", + "issue:41950", + "issue:42175", + "issue:43065", + "issue:43232", + "issue:43295", + "issue:43316", + "issue:43381", + "issue:43388", + "issue:43404", + "issue:43441", + "issue:43475", + "issue:43540", + "issue:43550", + "issue:43606", + "issue:43643", + "issue:43701", + "issue:43720", + "issue:43742", + "issue:43761", + "issue:43825", + "issue:43856", + "issue:43873", + "issue:44077", + "issue:44162", + "issue:44190", + "issue:44442", + "issue:44479", + "issue:44496", + "issue:44514", + "issue:44617", + "issue:44792", + "issue:44855", + "issue:44871", + "issue:44877", + "issue:44913", + "issue:45127", + "issue:45137", + "issue:45200", + "issue:45584", + "issue:45656" + ], + "soft_pairs": [ + "issue:43643|issue:44913", + "issue:43720|issue:44442", + "issue:32090|issue:33290", + "issue:44190|issue:44496", + "issue:43873|issue:44871", + "issue:34689|issue:36010", + "issue:43381|issue:43856", + "issue:44077|issue:44479", + "issue:44855|issue:45137", + "issue:33290|issue:45656", + "issue:29127|issue:33357", + "issue:43316|issue:44877", + "issue:43540|issue:44514", + "issue:43381|issue:43742", + "issue:42175|issue:43316", + "issue:33290|issue:34689", + "issue:43404|issue:43441", + "issue:44162|issue:44792", + "issue:25251|issue:28282", + "issue:43232|issue:43388", + "issue:41950|issue:43475", + "issue:43295|issue:43825", + "issue:44871|issue:45584", + "issue:45127|issue:45200", + "issue:43550|issue:43701", + "issue:43606|issue:43701", + "issue:43065|issue:44617", + "issue:43295|issue:44871", + "issue:43761|issue:44479" + ], + "prepared_review_unit_hash": "47078b07b8b5ad6ca9d8e46fab6d1bce89a45b91cdbe6b208dd869a8fd16cd38", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36423, + "estimated_input_tokens": 9106, + "estimated_eval_tokens": 18468 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No duplicate cluster here: the proposed links are all false positives across unrelated bugs/regressions in configs, loading, generation, training, and pipelines.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue: the items are heterogeneous and do not form one underlying bug cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "No strong global best issue; none of these issues is a duplicate hub, so a canonical representative would be arbitrary.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43643", + "right": "issue:44913", + "accept": false, + "reason": "Both are config/loading related, but one is trust_remote_code field loss and the other is GPTNeoX rotary_pct reload; different bugs." + }, + { + "left": "issue:43720", + "right": "issue:44442", + "accept": false, + "reason": "Packed-weight accelerate loading vs AutoTokenizer failing on a specific tokenizer class; unrelated failure modes." + }, + { + "left": "issue:32090", + "right": "issue:33290", + "accept": false, + "reason": "Trainer NoneType broadcast error vs DeepSpeed+Adafactor OOM; different subsystems and symptoms." + }, + { + "left": "issue:44190", + "right": "issue:44496", + "accept": false, + "reason": "Local dataset script issue vs unrecognized model/config loading error; unrelated." + }, + { + "left": "issue:43873", + "right": "issue:44871", + "accept": false, + "reason": "Quantization/offload behavior vs Gemma eos_token_id mismatch; different code paths." + }, + { + "left": "issue:34689", + "right": "issue:36010", + "accept": false, + "reason": "Llama 3.2 Vision loading regression vs GenerationMixin import error; not the same bug." + }, + { + "left": "issue:43381", + "right": "issue:43856", + "accept": false, + "reason": "Gradient checkpointing eval-mode error vs Qwen3 MoE memory usage; different problems." + }, + { + "left": "issue:44077", + "right": "issue:44479", + "accept": false, + "reason": "patchtsmixer post_init policy change vs video input regression in Qwen models; unrelated." + }, + { + "left": "issue:44855", + "right": "issue:45137", + "accept": false, + "reason": "Python 3.13 decorator parsing issue vs DeepSpeed ZeRO3 deque underflow; unrelated." + }, + { + "left": "issue:33290", + "right": "issue:45656", + "accept": false, + "reason": "OOM during Adafactor+DeepSpeed vs optimizer.step being called twice; different bugs." + }, + { + "left": "issue:29127", + "right": "issue:33357", + "accept": false, + "reason": "LayoutLMv3 box info validation vs MacOS bus error for CLIP; unrelated." + }, + { + "left": "issue:43316", + "right": "issue:44877", + "accept": false, + "reason": "Gemma3TextConfig API discrepancy vs strict config blocking granite_speech loading; different config issues." + }, + { + "left": "issue:43540", + "right": "issue:44514", + "accept": false, + "reason": "Both are Qwen multimodal video/chat issues, but they hit different inputs and code paths; not the same underlying bug." + }, + { + "left": "issue:43381", + "right": "issue:43742", + "accept": false, + "reason": "Gradient checkpointing eval-mode issue vs MobileLLM key error on load; unrelated." + }, + { + "left": "issue:42175", + "right": "issue:43316", + "accept": false, + "reason": "Packaging backend omission vs Gemma3TextConfig API inconsistency; unrelated." + }, + { + "left": "issue:33290", + "right": "issue:34689", + "accept": false, + "reason": "DeepSpeed Adafactor OOM vs Llama 3.2 Vision model-loading regression; different bugs." + }, + { + "left": "issue:43404", + "right": "issue:43441", + "accept": false, + "reason": "lm_head weight tying bug vs FlashAttention failure in Ministral-3; not the same change." + }, + { + "left": "issue:44162", + "right": "issue:44792", + "accept": false, + "reason": "ESM2 breakage vs Janus image-generation test failure; unrelated." + }, + { + "left": "issue:25251", + "right": "issue:28282", + "accept": false, + "reason": "Pipeline top_k nesting bug vs missing PyTorch ImportError; completely different." + }, + { + "left": "issue:43232", + "right": "issue:43388", + "accept": false, + "reason": "Generation kwargs update bug vs gather_for_metrics label truncation; different code paths." + }, + { + "left": "issue:41950", + "right": "issue:43475", + "accept": false, + "reason": "Video-classification processor lookup bug vs Sam3Video missing attribute; different model/pipeline issues." + }, + { + "left": "issue:43295", + "right": "issue:43825", + "accept": false, + "reason": "Processor/tokenizer regression vs incorrect pipeline translation error message; unrelated." + }, + { + "left": "issue:44871", + "right": "issue:45584", + "accept": false, + "reason": "Gemma-3 eos_token_id mismatch vs Whisper empty-transcription handling after token alignment; different bugs." + }, + { + "left": "issue:45127", + "right": "issue:45200", + "accept": false, + "reason": "LoRA merge with extended vocab collapse vs Gemma 4 mm_token_type_ids defaulting; unrelated." + }, + { + "left": "issue:43550", + "right": "issue:43701", + "accept": false, + "reason": "Bamba torch.compile/SDPA failure vs resume_from_checkpoint key mismatch; unrelated." + }, + { + "left": "issue:43606", + "right": "issue:43701", + "accept": false, + "reason": "CPU offload device mismatch vs resume_from_checkpoint key mismatch; different issues." + }, + { + "left": "issue:43065", + "right": "issue:44617", + "accept": false, + "reason": "Sam3PixelDecoder dummy layer concern vs Sam3Video CUDA OOM; not the same bug." + }, + { + "left": "issue:43295", + "right": "issue:44871", + "accept": false, + "reason": "Processor/tokenizer regression vs Gemma-3 eos_token_id inconsistency; unrelated." + }, + { + "left": "issue:43761", + "right": "issue:44479", + "accept": false, + "reason": "CLIPVisionModel hidden_states regression vs multi-model video input regression; different models 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": 2, + "component_count": 16, + "review_unit_index": 28, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:33290", + "issue:35141", + "issue:36331", + "issue:41720", + "issue:42617", + "issue:43065", + "issue:43452", + "issue:43575", + "issue:43646", + "issue:43673", + "issue:43742", + "issue:43761", + "issue:43827", + "issue:43873", + "issue:43906", + "issue:43931", + "issue:43994", + "issue:44060", + "issue:44077", + "issue:44190", + "issue:44242", + "issue:44464", + "issue:44479", + "issue:44496", + "issue:44521", + "issue:44534", + "issue:44610", + "issue:44617", + "issue:44734", + "issue:44737", + "issue:44743", + "issue:44792", + "issue:44855", + "issue:44871", + "issue:44913", + "issue:44936", + "issue:44964", + "issue:44977", + "issue:45127", + "issue:45245", + "issue:45290", + "issue:45357", + "issue:45381", + "issue:45584", + "issue:45588", + "issue:45600", + "issue:45698", + "issue:45803" + ], + "soft_pairs": [ + "issue:33290|issue:35141", + "issue:33290|issue:36331", + "issue:43673|issue:44743", + "issue:43452|issue:44496", + "issue:43575|issue:43742", + "issue:44936|issue:45698", + "issue:43646|issue:44913", + "issue:42617|issue:43575", + "issue:43761|issue:44242", + "issue:43575|issue:43873", + "issue:44077|issue:44534", + "issue:43931|issue:44060", + "issue:44464|issue:44871", + "issue:44479|issue:44734", + "issue:44617|issue:45584", + "issue:43827|issue:43994", + "issue:44792|issue:44977", + "issue:43906|issue:45290", + "issue:43065|issue:44610", + "issue:44534|issue:44913", + "issue:44190|issue:44964", + "issue:41720|issue:45245", + "issue:44792|issue:45600", + "issue:44737|issue:45803", + "issue:44855|issue:45588", + "issue:44521|issue:45381", + "issue:45127|issue:45357" + ], + "prepared_review_unit_hash": "f439e8512982ffa495ebc8ac3a126c3e45a333f2d0041656995161125d455f1c", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35865, + "estimated_input_tokens": 8967, + "estimated_eval_tokens": 18190 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These are mostly unrelated issues that only look similar on the surface. I would reject every soft edge as a duplicate match; there is no single underlying bug/change spanning the set.", + "confidence": 0.95, + "canonical_issue_reason": "No true canonical issue emerges here: the items cover unrelated bugs, regressions, and model-specific failures across different subsystems.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43873 is the best generic triage anchor only because it is open and has the most discussion/inbound references, but it does not subsume the other items.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:33290", + "right": "issue:35141", + "accept": false, + "reason": "OOM in Adafactor/DeepSpeed is unrelated to embedding reinitialization after resize." + }, + { + "left": "issue:33290", + "right": "issue:36331", + "accept": false, + "reason": "Different failures: optimizer OOM versus Trainer API keyword mismatch." + }, + { + "left": "issue:43673", + "right": "issue:44743", + "accept": false, + "reason": "Both mention cache, but one is a missing GenerationMixin cache in v5.0.0 and the other is recurrent state reset in Qwen3.5." + }, + { + "left": "issue:43452", + "right": "issue:44496", + "accept": false, + "reason": "Model/tokenizer loading with gguf_file is a different bug from missing model_type config resolution." + }, + { + "left": "issue:43575", + "right": "issue:43742", + "accept": false, + "reason": "OOM loading a large Qwen model is not the same as a KeyError on a small MobileLLM load." + }, + { + "left": "issue:44936", + "right": "issue:45698", + "accept": false, + "reason": "Trainer evaluate-after-train behavior is unrelated to loading the wrong custom module after save_pretrained." + }, + { + "left": "issue:43646", + "right": "issue:44913", + "accept": false, + "reason": "Custom model initialization breakage is not the same as GPTNeoX rotary_pct not persisting on reload." + }, + { + "left": "issue:42617", + "right": "issue:43575", + "accept": false, + "reason": "A 3d_parallel.py runtime issue is not the same as Qwen2 TP memory exhaustion." + }, + { + "left": "issue:43761", + "right": "issue:44242", + "accept": false, + "reason": "CLIPVision hidden_states regression and MoE load-balancing loss omission are different code paths." + }, + { + "left": "issue:43575", + "right": "issue:43873", + "accept": false, + "reason": "Qwen2 TP OOM and quantization offloading behavior are related only broadly to memory, not the same bug." + }, + { + "left": "issue:44077", + "right": "issue:44534", + "accept": false, + "reason": "Optional post_init policy and junk-filled non-persistent buffers are distinct v5 issues." + }, + { + "left": "issue:43931", + "right": "issue:44060", + "accept": false, + "reason": "Shape mismatch when loading Qwen3-VL is not the same as an incorrect tied-weights warning in Qwen3-Next." + }, + { + "left": "issue:44464", + "right": "issue:44871", + "accept": false, + "reason": "Compiled-forward generation inconsistency is unrelated to eos_token_id config mismatch." + }, + { + "left": "issue:44479", + "right": "issue:44734", + "accept": false, + "reason": "Video-input regression in Qwen VL models is different from KV-cache continuation tensor indexing in serve." + }, + { + "left": "issue:44617", + "right": "issue:45584", + "accept": false, + "reason": "Sam3Video OOM and Whisper empty-transcription generation failure are unrelated." + }, + { + "left": "issue:43827", + "right": "issue:43994", + "accept": false, + "reason": "Docs still referencing pipeline() is not the same as SIGLIP2 producing bad outputs with pipeline." + }, + { + "left": "issue:44792", + "right": "issue:44977", + "accept": false, + "reason": "Janus image-generation test failure and Qwen3.5 flash-attention generation issues are different model-specific bugs." + }, + { + "left": "issue:43906", + "right": "issue:45290", + "accept": false, + "reason": "An isolated reproduction of an existing issue is not the same as apply_chat_template crashing on tool-call messages." + }, + { + "left": "issue:43065", + "right": "issue:44610", + "accept": false, + "reason": "A dummy Conv2d placeholder in Sam3PixelDecoder is unrelated to OmDet-Turbo processor size mismatch." + }, + { + "left": "issue:44534", + "right": "issue:44913", + "accept": false, + "reason": "Non-persistent buffer initialization junk is not the same as rotary_pct reverting on config reload." + }, + { + "left": "issue:44190", + "right": "issue:44964", + "accept": false, + "reason": "Local dataset loading failure in a training script is unrelated to loading a multimodal model." + }, + { + "left": "issue:41720", + "right": "issue:45245", + "accept": false, + "reason": "CUDA assert on A800 with auto device mapping is a different failure from category-count overflow." + }, + { + "left": "issue:44792", + "right": "issue:45600", + "accept": false, + "reason": "A janus generation test failure is unrelated to stale auto_mappings references breaking CI." + }, + { + "left": "issue:44737", + "right": "issue:45803", + "accept": false, + "reason": "CPU tensor creation in XLNet and bare except swallowing errors in Fuyu are unrelated bugs." + }, + { + "left": "issue:44855", + "right": "issue:45588", + "accept": false, + "reason": "Python 3.13 torch.jit parsing error is unrelated to flash_attention s_aux handling." + }, + { + "left": "issue:44521", + "right": "issue:45381", + "accept": false, + "reason": "All-zero assistant masks and wrong video vision_position_ids are different multimodal template bugs." + }, + { + "left": "issue:45127", + "right": "issue:45357", + "accept": false, + "reason": "LoRA merge collapse with extended vocab is unrelated to saving incorrect visual encoder keys." + } + ] + }, + "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": 16, + "review_unit_index": 29, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:36246", + "issue:38175", + "issue:38617", + "issue:41628", + "issue:43065", + "issue:43116", + "issue:43295", + "issue:43335", + "issue:43404", + "issue:43479", + "issue:43540", + "issue:43550", + "issue:43643", + "issue:43644", + "issue:43650", + "issue:43761", + "issue:43976", + "issue:44077", + "issue:44112", + "issue:44186", + "issue:44190", + "issue:44220", + "issue:44246", + "issue:44464", + "issue:44466", + "issue:44496", + "issue:44561", + "issue:44683", + "issue:44734", + "issue:44855", + "issue:44877", + "issue:44913", + "issue:44933", + "issue:44938", + "issue:45003", + "issue:45216", + "issue:45245", + "issue:45276", + "issue:45362", + "issue:45375", + "issue:45446", + "issue:45542", + "issue:45563", + "issue:45584", + "issue:45600", + "issue:45632", + "issue:45803", + "issue:45834" + ], + "soft_pairs": [ + "issue:43335|issue:44855", + "issue:44938|issue:45542", + "issue:44077|issue:44877", + "issue:43550|issue:44190", + "issue:44496|issue:45803", + "issue:43761|issue:45834", + "issue:43644|issue:44913", + "issue:43976|issue:44855", + "issue:43540|issue:44734", + "issue:43650|issue:44220", + "issue:45276|issue:45584", + "issue:44112|issue:45600", + "issue:43335|issue:44190", + "issue:43116|issue:44190", + "issue:43295|issue:45276", + "issue:45245|issue:45584", + "issue:43479|issue:45375", + "issue:36246|issue:41628", + "issue:44933|issue:45362", + "issue:44464|issue:45563", + "issue:43404|issue:45216", + "issue:44186|issue:44734", + "issue:44734|issue:45632", + "issue:44683|issue:45446", + "issue:38175|issue:43065", + "issue:44466|issue:44913", + "issue:44561|issue:45632", + "issue:44246|issue:45003", + "issue:36246|issue:38617", + "issue:43643|issue:44496" + ], + "prepared_review_unit_hash": "81a4b020da0db0c47d6d4390cd2f9deb030510fed43cc1a7b1ac558580c344df", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36697, + "estimated_input_tokens": 9175, + "estimated_eval_tokens": 18606 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is mostly heterogeneous: several unrelated import/config/regression issues plus a few loosely similar model-loading bugs. No soft pair looks like the same underlying bug/change, so none are accepted as duplicates.", + "confidence": 0.39, + "canonical_issue_reason": "Issue 43643 is the broadest representative of the cluster\u2019s config/loading regression theme, but it is still not a duplicate of the other items.", + "canonical_pr_reason": null, + "best_issue_reason": "43643 is the best single issue anchor because it is concrete, well-scoped, and closest to the recurring config/serialization-loading failures in this set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43335", + "right": "issue:44855", + "accept": false, + "reason": "Different models and failure modes: SwitchTransformers sparse-layer construction vs a Python 3.13 import/parsing error in DebertaV2." + }, + { + "left": "issue:44938", + "right": "issue:45542", + "accept": false, + "reason": "Both are runtime/load problems, but one is a Python 3.14 import failure and the other is a backend-missing TensorFlow/tensorboard error; not the same bug." + }, + { + "left": "issue:44077", + "right": "issue:44877", + "accept": false, + "reason": "Both involve config handling, but one is PatchTSMixer post_init validation and the other is strict loading of granite_speech config; different concrete issues." + }, + { + "left": "issue:43550", + "right": "issue:44190", + "accept": false, + "reason": "Unrelated: Bamba torch.compile/SDPA failure vs example script failing to load a local dataset." + }, + { + "left": "issue:44496", + "right": "issue:45803", + "accept": false, + "reason": "Different code paths: missing model_type in config loading vs Fuyu tensor conversion swallowing exceptions." + }, + { + "left": "issue:43761", + "right": "issue:45834", + "accept": false, + "reason": "Different multimodal models and bugs: CLIPVisionModel hidden_states regression vs Kosmos2.5 OCR index error." + }, + { + "left": "issue:43644", + "right": "issue:44913", + "accept": false, + "reason": "Both concern saved state/config behavior, but one is non-persistent buffers junk in v5 and the other is GPTNeoX rotary_pct reload behavior; not the same bug." + }, + { + "left": "issue:43976", + "right": "issue:44855", + "accept": false, + "reason": "Same Python-version area, but the root causes differ: v5.1.0 load failure vs an import-time IndentationError in DebertaV2." + }, + { + "left": "issue:43540", + "right": "issue:44734", + "accept": false, + "reason": "Different subsystems: Qwen3OmniMoe video-processing validation vs serve KV-cache continuation indexing." + }, + { + "left": "issue:43650", + "right": "issue:44220", + "accept": false, + "reason": "No meaningful overlap; one issue is essentially a placeholder/spam title and the other is a feature-specific audio feature extraction bug." + }, + { + "left": "issue:45276", + "right": "issue:45584", + "accept": false, + "reason": "Both are generation/model issues, but one is Gemma4 embedding resizing and the other is Whisper empty transcription after align_special_tokens; not the same bug." + }, + { + "left": "issue:44112", + "right": "issue:45600", + "accept": false, + "reason": "Both mention CI, but the failing behavior is different: GraniteSpeech stale device override test vs auto_mappings referencing removed Sam3LiteText configs." + }, + { + "left": "issue:43335", + "right": "issue:44190", + "accept": false, + "reason": "Different domains: SwitchTransformers config bug vs image-classification example script failing to load a local dataset." + }, + { + "left": "issue:43116", + "right": "issue:44190", + "accept": false, + "reason": "Both are example-script issues, but one is multi-label classification output logic and the other is local dataset loading; different bugs." + }, + { + "left": "issue:43295", + "right": "issue:45276", + "accept": false, + "reason": "Separate regressions: processor/tokenizer image-passing breakage vs Gemma4 token embedding resizing not propagating." + }, + { + "left": "issue:45245", + "right": "issue:45584", + "accept": false, + "reason": "Completely different failures: category cardinality overflow vs Whisper empty-transcription generation." + }, + { + "left": "issue:43479", + "right": "issue:45375", + "accept": false, + "reason": "Both are config-field problems, but one is default vision/audio config initialization from None and the other is a missing deepstack_visual_indexes field dropped by strict parsing." + }, + { + "left": "issue:36246", + "right": "issue:41628", + "accept": false, + "reason": "Both are import/export problems, but they are for different symbols and paths: Qwen2_5_VLImageProcessor submodule import vs top-level AutoImageProcessor import." + }, + { + "left": "issue:44933", + "right": "issue:45362", + "accept": false, + "reason": "Different issues: a missing import from image_utils vs Qwen3.5 chat runtime crash." + }, + { + "left": "issue:44464", + "right": "issue:45563", + "accept": false, + "reason": "Different generation bugs: inconsistent compiled-forward chunked generation vs a stale warning for num_return_sequences." + }, + { + "left": "issue:43404", + "right": "issue:45216", + "accept": false, + "reason": "Different model/regression targets: tied lm_head weights in Mistral3ForConditionalGeneration vs incorrect Qwen3.5 save_pretrained checkpoint output." + }, + { + "left": "issue:44186", + "right": "issue:44734", + "accept": false, + "reason": "Different code paths: LayoutLMv2Tokenizer NER/padding crash vs transformers serve KV-cache continuation indexing crash." + }, + { + "left": "issue:44734", + "right": "issue:45632", + "accept": false, + "reason": "Different problems entirely: serve tensor indexing vs trust_remote_code cache-path collisions." + }, + { + "left": "issue:44683", + "right": "issue:45446", + "accept": false, + "reason": "Related to flex_attention, but one is a compiled runtime failure on torch>=2.9 and the other is a version-check/import gating bug; not the same bug." + }, + { + "left": "issue:38175", + "right": "issue:43065", + "accept": false, + "reason": "Unrelated model bugs: siglip2 zero probabilities vs Sam3PixelDecoder dummy Conv2d behavior." + } + ] + }, + "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": 16, + "review_unit_index": 30, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:41950", + "issue:42175", + "issue:42907", + "issue:43295", + "issue:43299", + "issue:43381", + "issue:43388", + "issue:43441", + "issue:43479", + "issue:43525", + "issue:43526", + "issue:43531", + "issue:43577", + "issue:43673", + "issue:43688", + "issue:43824", + "issue:43827", + "issue:43854", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44062", + "issue:44075", + "issue:44112", + "issue:44220", + "issue:44242", + "issue:44279", + "issue:44336", + "issue:44351", + "issue:44368", + "issue:44496", + "issue:44933", + "issue:44964", + "issue:45020", + "issue:45042", + "issue:45200", + "issue:45216", + "issue:45278", + "issue:45335", + "issue:45357", + "issue:45405", + "issue:45406", + "issue:45563", + "issue:45584", + "issue:45593", + "issue:45600", + "issue:45632", + "issue:45759" + ], + "soft_pairs": [ + "issue:45406|issue:45632", + "issue:45563|issue:45584", + "issue:41950|issue:44016", + "issue:44964|issue:45600", + "issue:43388|issue:43688", + "issue:42907|issue:43441", + "issue:44112|issue:44964", + "issue:43295|issue:45335", + "issue:43854|issue:44964", + "issue:44016|issue:44220", + "issue:43673|issue:44336", + "issue:44496|issue:45042", + "issue:43299|issue:45216", + "issue:43531|issue:45600", + "issue:43526|issue:43901", + "issue:42175|issue:44279", + "issue:45335|issue:45584", + "issue:44075|issue:44368", + "issue:43381|issue:45593", + "issue:43299|issue:45357", + "issue:43901|issue:44496", + "issue:44496|issue:45759", + "issue:44351|issue:45278", + "issue:44242|issue:45200", + "issue:44016|issue:44062", + "issue:43388|issue:43827", + "issue:41950|issue:45020", + "issue:43976|issue:45405", + "issue:43577|issue:45216", + "issue:43479|issue:43525", + "issue:43824|issue:44933" + ], + "prepared_review_unit_hash": "72b52a4670ebfd826be10249859445674f1bccc6d04598e5e66344d336c0e074", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36992, + "estimated_input_tokens": 9248, + "estimated_eval_tokens": 18752 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These issues are mostly unrelated Transformers regressions spanning model loading, processor/config handling, docs, and training/runtime behavior. No soft pair looks like the same underlying bug or change.", + "confidence": 0.92, + "canonical_issue_reason": "issue:45278 is the broadest open regression report in the set and best serves as an umbrella representative for the cluster\u2019s general post-upgrade import/load breakage theme.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the strongest cluster anchor: open, generic, and closer to the repeated import/error reports than the more model-specific bugs.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45406", + "right": "issue:45632", + "accept": false, + "reason": "Different failures: Gemma4Processor missing _tokenizer vs trust_remote_code cache-path collision." + }, + { + "left": "issue:45563", + "right": "issue:45584", + "accept": false, + "reason": "A stale warning about num_return_sequences is unrelated to Whisper failing on empty transcription after align_special_tokens." + }, + { + "left": "issue:41950", + "right": "issue:44016", + "accept": false, + "reason": "Video-classification processor lookup is unrelated to a notebook syntax error." + }, + { + "left": "issue:44964", + "right": "issue:45600", + "accept": false, + "reason": "Phi-4 multimodal loading is a different bug than auto_mappings referencing removed Sam3LiteText configs." + }, + { + "left": "issue:43388", + "right": "issue:43688", + "accept": false, + "reason": "Batch label dropping in gather_for_metrics is unrelated to auxiliary-loss normalization in OLMoE/GPT Oss." + }, + { + "left": "issue:42907", + "right": "issue:43441", + "accept": false, + "reason": "Dequantized save failure for Ministrals/Devstrals is not the same as FlashAttention loading failure for Ministral-3." + }, + { + "left": "issue:44112", + "right": "issue:44964", + "accept": false, + "reason": "A stale GraniteSpeech CI test is unrelated to Phi-4 multimodal model loading." + }, + { + "left": "issue:43295", + "right": "issue:45335", + "accept": false, + "reason": "processor.tokenizer / image passing regression is not the same as t5gemma embedding resize not propagating to decoder.embed_tokens." + }, + { + "left": "issue:43854", + "right": "issue:44964", + "accept": false, + "reason": "GLM-4.7-Flash unit-test loading failure is a different model/config issue than Phi-4 multimodal loading." + }, + { + "left": "issue:44016", + "right": "issue:44220", + "accept": false, + "reason": "Notebook syntax error and _torch_extract_fbank_features() bug are unrelated." + }, + { + "left": "issue:43673", + "right": "issue:44336", + "accept": false, + "reason": "Generation cache missing in chunked_prefill is unrelated to ANSI codes in loading_report." + }, + { + "left": "issue:44496", + "right": "issue:45042", + "accept": false, + "reason": "Unrecognized model/config handling is unrelated to PIL image processors unnecessarily requiring torchvision." + }, + { + "left": "issue:43299", + "right": "issue:45216", + "accept": false, + "reason": "Qwen3VL model loading failure and Qwen3.5 save_pretrained regression are different code paths." + }, + { + "left": "issue:43531", + "right": "issue:45600", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior is unrelated to stale Sam3LiteText auto_mappings references." + }, + { + "left": "issue:43526", + "right": "issue:43901", + "accept": false, + "reason": "BeitImageProcessorFast reduce_labels bug is unrelated to a docs/behavior mismatch in TextClassificationPipeline." + }, + { + "left": "issue:42175", + "right": "issue:44279", + "accept": false, + "reason": "Backend packaging dependency omission is too broad and different from a generic transformers dependency issue." + }, + { + "left": "issue:45335", + "right": "issue:45584", + "accept": false, + "reason": "Embedding resize propagation in t5gemma is unrelated to Whisper empty-transcription generation failure." + }, + { + "left": "issue:44075", + "right": "issue:44368", + "accept": false, + "reason": "SGD args not being used is unrelated to a tie_word_embeddings warning in Qwen3.5." + }, + { + "left": "issue:43381", + "right": "issue:45593", + "accept": false, + "reason": "Gradient checkpointing in eval mode and D-FINE auxiliary-loss handling are different bugs." + }, + { + "left": "issue:43299", + "right": "issue:45357", + "accept": false, + "reason": "Both touch Qwen3.5, but one is a loading bug and the other is a save_pretrained key regression; not the same fix." + }, + { + "left": "issue:43901", + "right": "issue:44496", + "accept": false, + "reason": "A docs mismatch about return_all_scores is unrelated to unrecognized model config loading." + }, + { + "left": "issue:44496", + "right": "issue:45759", + "accept": false, + "reason": "Unrecognized model/config errors are not the same as AutoModelForCausalLM.from_config failing to unwrap text_config for composite multimodal configs." + }, + { + "left": "issue:44351", + "right": "issue:45278", + "accept": false, + "reason": "Missing HybridCache export is a specific import bug, while 45278 is a broad report of many import errors; not the same root cause." + }, + { + "left": "issue:44242", + "right": "issue:45200", + "accept": false, + "reason": "Router-logit auxiliary-loss behavior is unrelated to mm_token_type_ids defaults in Gemma 4." + }, + { + "left": "issue:44016", + "right": "issue:44062", + "accept": false, + "reason": "Notebook syntax error and AddedToken special-argument TypeError are unrelated." + }, + { + "left": "issue:43388", + "right": "issue:43827", + "accept": false, + "reason": "Metric label truncation is unrelated to docs still mentioning pipeline()." + }, + { + "left": "issue:41950", + "right": "issue:45020", + "accept": false, + "reason": "Image-processor lookup in video classification is not the same as the broad remote_code breakage report." + }, + { + "left": "issue:43976", + "right": "issue:45405", + "accept": false, + "reason": "Python version compatibility and unreleased MIN_PEFT_VERSION are unrelated packaging/versioning issues." + }, + { + "left": "issue:43577", + "right": "issue:45216", + "accept": false, + "reason": "Blip2 dtype propagation and Qwen3.5 checkpoint save regression are different model-specific bugs." + }, + { + "left": "issue:43479", + "right": "issue:43525", + "accept": false, + "reason": "Default multimodal config initialization is unrelated to Llama4Config missing pad_token_id." + }, + { + "left": "issue:43824", + "right": "issue:44933", + "accept": false, + "reason": "Missing Qwen2_5_VL import and a nonexistent image_utils import are different import/export 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": 16, + "review_unit_index": 31, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:32090", + "issue:35141", + "issue:38175", + "issue:39692", + "issue:41950", + "issue:42994", + "issue:43065", + "issue:43295", + "issue:43299", + "issue:43317", + "issue:43441", + "issue:43450", + "issue:43650", + "issue:43653", + "issue:43723", + "issue:43784", + "issue:43854", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44162", + "issue:44261", + "issue:44263", + "issue:44279", + "issue:44297", + "issue:44462", + "issue:44466", + "issue:44521", + "issue:44617", + "issue:44623", + "issue:44671", + "issue:44683", + "issue:44749", + "issue:44779", + "issue:44898", + "issue:44987", + "issue:45081", + "issue:45276", + "issue:45290", + "issue:45325", + "issue:45397", + "issue:45405", + "issue:45542", + "issue:45593", + "issue:45600", + "issue:45712", + "issue:45823", + "issue:45834" + ], + "soft_pairs": [ + "issue:41950|issue:45823", + "issue:44683|issue:45542", + "issue:43299|issue:43854", + "issue:44279|issue:45405", + "issue:38175|issue:44617", + "issue:39692|issue:43901", + "issue:32090|issue:35141", + "issue:41950|issue:43650", + "issue:43295|issue:44779", + "issue:44462|issue:45081", + "issue:44297|issue:44462", + "issue:43976|issue:44279", + "issue:45600|issue:45712", + "issue:43976|issue:44987", + "issue:44521|issue:45325", + "issue:43441|issue:45834", + "issue:44462|issue:44779", + "issue:43450|issue:44279", + "issue:42994|issue:43317", + "issue:44261|issue:44263", + "issue:44466|issue:45276", + "issue:43065|issue:44898", + "issue:43441|issue:44683", + "issue:43650|issue:45290", + "issue:43723|issue:44162", + "issue:44623|issue:44749", + "issue:43653|issue:45593", + "issue:44623|issue:45397", + "issue:44671|issue:44779", + "issue:43784|issue:45542", + "issue:44016|issue:45823" + ], + "prepared_review_unit_hash": "dea1b6d1ffcf15dce0bf29c24243a125d7c563f957024d87ba1fc4162f977c2e", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36943, + "estimated_input_tokens": 9236, + "estimated_eval_tokens": 18728 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a heterogeneous cluster: mostly tokenizer/processor/model-loading regressions in Transformers v5, plus several unrelated vision, quantization, video, and runtime issues. None of the listed soft pairs look like true duplicates or the same code change, so all soft edges should be rejected. No PRs are present.", + "confidence": 0.55, + "canonical_issue_reason": "issue:44462 is the clearest representative of the tokenizer-loading regression subset: it describes a concrete AutoTokenizer v5 failure with repository-provided tokenizer.json and broad user impact.", + "canonical_pr_reason": null, + "best_issue_reason": "Best overall issue candidate because it is specific, reproducible, and broad enough to anchor the tokenizer-loading subgroup, even though the overall cluster is not a tight duplicate set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41950", + "right": "issue:45823", + "accept": false, + "reason": "Different bugs: video-classification processor lookup vs Gemma4 PLE device-mismatch during forward." + }, + { + "left": "issue:44683", + "right": "issue:45542", + "accept": false, + "reason": "Unrelated failures: compiled flex_attention on torch>=2.9 vs tensorboard/TensorFlow backend installation issue." + }, + { + "left": "issue:43299", + "right": "issue:43854", + "accept": false, + "reason": "Both are model-loading problems, but for different models and different failure modes." + }, + { + "left": "issue:44279", + "right": "issue:45405", + "accept": false, + "reason": "Dependency/version complaints, but one is a generic transformers dependency issue and the other is a PEFT minimum-version bump." + }, + { + "left": "issue:38175", + "right": "issue:44617", + "accept": false, + "reason": "Different model/runtime symptoms: SIGLIP2 zero probabilities vs Sam3Video CUDA OOM." + }, + { + "left": "issue:39692", + "right": "issue:43901", + "accept": false, + "reason": "Documentation/example errors versus a pipeline docs behavior mismatch; not the same bug." + }, + { + "left": "issue:32090", + "right": "issue:35141", + "accept": false, + "reason": "Training GPU broadcast TypeError is unrelated to embedding reinitialization after resize_token_embeddings." + }, + { + "left": "issue:41950", + "right": "issue:43650", + "accept": false, + "reason": "Video-classification processor bug is unrelated to a placeholder issue title with no technical overlap." + }, + { + "left": "issue:43295", + "right": "issue:44779", + "accept": false, + "reason": "Different tokenizer regressions: processor.tokenizer/image handling vs Deepseek tokenizer output correctness." + }, + { + "left": "issue:44462", + "right": "issue:45081", + "accept": false, + "reason": "Both involve tokenizer loading, but one is AutoTokenizer ignoring tokenizer.json and the other is a Mistral regex patch crash." + }, + { + "left": "issue:44297", + "right": "issue:44462", + "accept": false, + "reason": "Serialization mismatch in tokenizer_config.json is a different path from AutoTokenizer ignoring tokenizer.json." + }, + { + "left": "issue:43976", + "right": "issue:44279", + "accept": false, + "reason": "Python version support issue vs generic dependency issue; not the same underlying bug." + }, + { + "left": "issue:45600", + "right": "issue:45712", + "accept": false, + "reason": "Both are repo maintenance/CI-adjacent, but one is auto_mappings references and the other is dummy_pt_objects cleanup." + }, + { + "left": "issue:43976", + "right": "issue:44987", + "accept": false, + "reason": "Different version breakages affecting different model loads; no shared code-path evidence." + }, + { + "left": "issue:44521", + "right": "issue:45325", + "accept": false, + "reason": "apply_chat_template assistant mask bug is unrelated to Qwen2.5-VL rope index scaling." + }, + { + "left": "issue:43441", + "right": "issue:45834", + "accept": false, + "reason": "FlashAttention failure for Ministral-3 is unrelated to Kosmos2.5 OCR index error." + }, + { + "left": "issue:44462", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer-related, but one is repository tokenizer loading and the other is incorrect Deepseek tokenization results." + }, + { + "left": "issue:43450", + "right": "issue:44279", + "accept": false, + "reason": "Video processor batched-shape bug is unrelated to a generic dependency issue." + }, + { + "left": "issue:42994", + "right": "issue:43317", + "accept": false, + "reason": "Quantized model saving failure is a different problem from dequantized model loading with device_map=auto." + }, + { + "left": "issue:44261", + "right": "issue:44263", + "accept": false, + "reason": "Precision/error discussion for MLA q_a_layernorm is unrelated to torch.split return values in GlmMoeDsaIndexer." + }, + { + "left": "issue:44466", + "right": "issue:45276", + "accept": false, + "reason": "Both concern resizing/tied weights behavior, but one is serialization of lm_head.weight and the other is Gemma4 embed/output modules not updating." + }, + { + "left": "issue:43065", + "right": "issue:44898", + "accept": false, + "reason": "Dummy Conv2d in Sam3PixelDecoder and Perceiver non-default resolution failure are different model-specific bugs." + }, + { + "left": "issue:43441", + "right": "issue:44683", + "accept": false, + "reason": "Different attention-path failures: Ministral-3 FlashAttention vs compiled flex_attention on newer torch." + }, + { + "left": "issue:43650", + "right": "issue:45290", + "accept": false, + "reason": "Placeholder issue title is unrelated to apply_chat_template crashing on tool-call assistant messages." + }, + { + "left": "issue:43723", + "right": "issue:44162", + "accept": false, + "reason": "Tokenizer loading regression in v5 is unrelated to the ESM2 model breakage." + }, + { + "left": "issue:44623", + "right": "issue:44749", + "accept": false, + "reason": "processor.save_pretrained missing files is unrelated to a Chinese report about slower filtering after upgrade." + }, + { + "left": "issue:43653", + "right": "issue:45593", + "accept": false, + "reason": "BigBirdTokenizer special-token registration bug is unrelated to D-FINE auxiliary-loss behavior." + }, + { + "left": "issue:44623", + "right": "issue:45397", + "accept": false, + "reason": "Both mention saving/loading-ish flows, but one is processor saving and the other is Gemma-4 zero3 from_pretrained." + }, + { + "left": "issue:44671", + "right": "issue:44779", + "accept": false, + "reason": "CamemBERT masked LM prediction regression is not the same as Deepseek tokenizer output regression." + }, + { + "left": "issue:43784", + "right": "issue:45542", + "accept": false, + "reason": "Missing nn import in sentence-transformers is unrelated to the TensorFlow backend/tensorboard installation issue." + }, + { + "left": "issue:44016", + "right": "issue:45823", + "accept": false, + "reason": "Notebook syntax error and Gemma4 device mismatch are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 32, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:16998", + "issue:22355", + "issue:36246", + "issue:41628", + "issue:41950", + "issue:42491", + "issue:42757", + "issue:43122", + "issue:43262", + "issue:43299", + "issue:43352", + "issue:43450", + "issue:43525", + "issue:43643", + "issue:43650", + "issue:43673", + "issue:43761", + "issue:43824", + "issue:43881", + "issue:43901", + "issue:43976", + "issue:43994", + "issue:44016", + "issue:44188", + "issue:44220", + "issue:44242", + "issue:44246", + "issue:44291", + "issue:44295", + "issue:44297", + "issue:44336", + "issue:44351", + "issue:44361", + "issue:44373", + "issue:44485", + "issue:44496", + "issue:44521", + "issue:44561", + "issue:44589", + "issue:44617", + "issue:44871", + "issue:44964", + "issue:44987", + "issue:45250", + "issue:45362", + "issue:45397", + "issue:45542", + "issue:45593" + ], + "soft_pairs": [ + "issue:44361|issue:45593", + "issue:44987|issue:45542", + "issue:43901|issue:43994", + "issue:43122|issue:43650", + "issue:44016|issue:44291", + "issue:43262|issue:44521", + "issue:43673|issue:43761", + "issue:44188|issue:44336", + "issue:43525|issue:43881", + "issue:43525|issue:43643", + "issue:44297|issue:45397", + "issue:43450|issue:44617", + "issue:42491|issue:43299", + "issue:44589|issue:45362", + "issue:41628|issue:43976", + "issue:43901|issue:44246", + "issue:44336|issue:44373", + "issue:44242|issue:45593", + "issue:16998|issue:22355", + "issue:43352|issue:45250", + "issue:44295|issue:45397", + "issue:44246|issue:44496", + "issue:41950|issue:44220", + "issue:36246|issue:42757", + "issue:36246|issue:44351", + "issue:44336|issue:44485", + "issue:43673|issue:44561", + "issue:43824|issue:43976", + "issue:43976|issue:44964", + "issue:44246|issue:44871" + ], + "prepared_review_unit_hash": "b7d8353c3fc15026bda0ec89449862d647de1f74630d7e3d9fc27b30743de44b", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36378, + "estimated_input_tokens": 9095, + "estimated_eval_tokens": 18446 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is heterogeneous: mostly unrelated import/model-loading regressions plus a few docs/perf/UI issues. None of the soft edges is strong enough to merge, so there is no reliable duplicate cluster here.", + "confidence": 0.81, + "canonical_issue_reason": "No single issue cleanly subsumes the set; if one representative is needed, 44521 is the strongest standalone bug report because it is concrete, reproducible, and has the most discussion/inbound references among the active issues.", + "canonical_pr_reason": null, + "best_issue_reason": "44521 is the best overall issue candidate: it has a clear failure mode, active discussion, and enough traction to serve as a representative bug report, even though it does not duplicate the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44361", + "right": "issue:45593", + "accept": false, + "reason": "Different bugs: MLukeTokenizer task-time AttributeError vs D-FINE missing auxiliary losses when denoising is off." + }, + { + "left": "issue:44987", + "right": "issue:45542", + "accept": false, + "reason": "Unrelated failures: a transformers>=5.1.0 model-loading regression vs a TensorFlow backend/import issue." + }, + { + "left": "issue:43901", + "right": "issue:43994", + "accept": false, + "reason": "Docs mismatch for return_all_scores is unrelated to SigLIP2 nonsensical outputs." + }, + { + "left": "issue:43122", + "right": "issue:43650", + "accept": false, + "reason": "Tokenization regression and a vague unrelated report; no shared code-path bug." + }, + { + "left": "issue:44016", + "right": "issue:44291", + "accept": false, + "reason": "Notebook syntax error and init_empty_weights argument mismatch are unrelated." + }, + { + "left": "issue:43262", + "right": "issue:44521", + "accept": false, + "reason": "Both mention apply_chat_template, but one is audio sampling-rate defaulting and the other is multimodal assistant_mask generation; different bugs." + }, + { + "left": "issue:43673", + "right": "issue:43761", + "accept": false, + "reason": "Generation cache regression and CLIP hidden_states regression are different model/code paths." + }, + { + "left": "issue:44188", + "right": "issue:44336", + "accept": false, + "reason": "Torch.compile attention branching bug vs ANSI codes in loading_report; unrelated." + }, + { + "left": "issue:43525", + "right": "issue:43881", + "accept": false, + "reason": "Both are model-loading errors, but they affect different models and have different failure causes." + }, + { + "left": "issue:43525", + "right": "issue:43643", + "accept": false, + "reason": "Missing pad_token_id on Llama4Config is not the same as trust_remote_code returning missing fields." + }, + { + "left": "issue:44297", + "right": "issue:45397", + "accept": false, + "reason": "Tokenizer save metadata mismatch is unrelated to gemma-4 zero3 from_pretrained loading failure." + }, + { + "left": "issue:43450", + "right": "issue:44617", + "accept": false, + "reason": "Batched video processor shape bug and CUDA OOM are not the same underlying problem." + }, + { + "left": "issue:42491", + "right": "issue:43299", + "accept": false, + "reason": "Both involve Qwen3/MoE and v5-era breakage, but one is LoRA compatibility and the other is base Qwen3VL model loading; not clearly the same fixable bug." + }, + { + "left": "issue:44589", + "right": "issue:45362", + "accept": false, + "reason": "Float8 storage import error and Qwen3.5-35B chat crash are unrelated." + }, + { + "left": "issue:41628", + "right": "issue:43976", + "accept": false, + "reason": "AutoImageProcessor import error and Python-version compatibility regression are different issues." + }, + { + "left": "issue:43901", + "right": "issue:44246", + "accept": false, + "reason": "Documentation mismatch vs import-time slowness; no overlap beyond being transformers issues." + }, + { + "left": "issue:44336", + "right": "issue:44373", + "accept": false, + "reason": "ANSI output in loading_report and a wrong docstring are unrelated." + }, + { + "left": "issue:44242", + "right": "issue:45593", + "accept": false, + "reason": "Both mention losses, but load-balancing loss in MoE and D-FINE auxiliary losses are distinct mechanisms." + }, + { + "left": "issue:16998", + "right": "issue:22355", + "accept": false, + "reason": "A model_max_length question and a missing transformers.onnx module are unrelated." + }, + { + "left": "issue:43352", + "right": "issue:45250", + "accept": false, + "reason": "45250 is too generic to be the same concrete bug as NemotronH not supporting Flash Attention 2.0 yet." + }, + { + "left": "issue:44295", + "right": "issue:45397", + "accept": false, + "reason": "Reading a buffered position_ids tensor and gemma-4 zero3 loading are unrelated." + }, + { + "left": "issue:44246", + "right": "issue:44496", + "accept": false, + "reason": "Import latency and an unrecognized model/config error are different problems." + }, + { + "left": "issue:41950", + "right": "issue:44220", + "accept": false, + "reason": "Video-classification looking for image processors and audio fbank extraction are different code paths." + }, + { + "left": "issue:36246", + "right": "issue:42757", + "accept": false, + "reason": "Different missing imports from different packages; same symptom class, not the same bug." + }, + { + "left": "issue:36246", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but for different missing symbols and likely different export fixes." + }, + { + "left": "issue:44336", + "right": "issue:44485", + "accept": false, + "reason": "Terminal ANSI output bug and GLM-5 RoPE discussion are unrelated." + }, + { + "left": "issue:43673", + "right": "issue:44561", + "accept": false, + "reason": "Generation cache issue and removal of is_torch_fx_available are different regression sources." + }, + { + "left": "issue:43824", + "right": "issue:43976", + "accept": false, + "reason": "Qwen2.5-VL import error and Python-version support regression are not the same issue." + }, + { + "left": "issue:43976", + "right": "issue:44964", + "accept": false, + "reason": "Broad packaging/version-compatibility failure vs a specific model-loading problem; different scope and fix." + }, + { + "left": "issue:44246", + "right": "issue:44871", + "accept": false, + "reason": "Import slowness and eos_token_id config inconsistency are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 33, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:36246", + "issue:42886", + "issue:43065", + "issue:43232", + "issue:43257", + "issue:43262", + "issue:43408", + "issue:43475", + "issue:43526", + "issue:43540", + "issue:43596", + "issue:43653", + "issue:43901", + "issue:44060", + "issue:44077", + "issue:44162", + "issue:44242", + "issue:44246", + "issue:44295", + "issue:44336", + "issue:44368", + "issue:44466", + "issue:44496", + "issue:44749", + "issue:44779", + "issue:44857", + "issue:44877", + "issue:44977", + "issue:44987", + "issue:44995", + "issue:45003", + "issue:45137", + "issue:45200", + "issue:45229", + "issue:45292", + "issue:45335", + "issue:45362", + "issue:45375", + "issue:45397", + "issue:45459", + "issue:45561", + "issue:45563", + "issue:45584", + "issue:45588", + "issue:45656", + "issue:45701", + "issue:45759" + ], + "soft_pairs": [ + "issue:44242|issue:44857", + "issue:45362|issue:45375", + "issue:44162|issue:45200", + "issue:44779|issue:45292", + "issue:44295|issue:44749", + "issue:44877|issue:45375", + "issue:45292|issue:45584", + "issue:45292|issue:45701", + "issue:45561|issue:45563", + "issue:43901|issue:45003", + "issue:44987|issue:45459", + "issue:45375|issue:45759", + "issue:43232|issue:44336", + "issue:36246|issue:43596", + "issue:43526|issue:45003", + "issue:44496|issue:45003", + "issue:43257|issue:45137", + "issue:43065|issue:43408", + "issue:44995|issue:45200", + "issue:43262|issue:44246", + "issue:42886|issue:45229", + "issue:43475|issue:43540", + "issue:44466|issue:45335", + "issue:44995|issue:45588", + "issue:43653|issue:44162", + "issue:44368|issue:44977", + "issue:44368|issue:45563", + "issue:45397|issue:45656", + "issue:44060|issue:44077" + ], + "prepared_review_unit_hash": "f78480df2eb971d0dd7e3f6ff3e0bffd1f0070bc89c79e2362119cb9dd08cce3", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35948, + "estimated_input_tokens": 8987, + "estimated_eval_tokens": 18230 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is mostly heterogeneous: several issues are superficially similar around configs, embeddings, tokenizers, and generation, but they do not describe the same underlying bug. No soft duplicate pair is convincing enough to merge.", + "confidence": 0.91, + "canonical_issue_reason": "issue:45292 is the clearest concrete bug report in the set for a model-agnostic embedding-resize failure, and it best serves as an anchor for the smaller related resize/embed reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45292 is the best overall issue to represent the cluster because it describes a specific, reproducible code-path problem with broad applicability, unlike the many model-specific or orthogonal reports here.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44242", + "right": "issue:44857", + "accept": false, + "reason": "Different bugs: MoE load-balancing loss gating vs AMP/CUDA loss crash." + }, + { + "left": "issue:45362", + "right": "issue:45375", + "accept": false, + "reason": "Both involve Qwen models, but one is a chat-generation crash and the other is a config field being dropped; not the same fix." + }, + { + "left": "issue:44162", + "right": "issue:45200", + "accept": false, + "reason": "ESM2 breakage and Gemma mm_token_type_ids defaulting are unrelated model families and code paths." + }, + { + "left": "issue:44779", + "right": "issue:45292", + "accept": false, + "reason": "Tokenizer correctness regression vs embedding-resize propagation bug; different subsystems." + }, + { + "left": "issue:44295", + "right": "issue:44749", + "accept": false, + "reason": "Position_ids buffer access error is unrelated to the transformers upgrade performance slowdown." + }, + { + "left": "issue:44877", + "right": "issue:45375", + "accept": false, + "reason": "Both are strict-config complaints, but they concern different model configs and different missing fields." + }, + { + "left": "issue:45292", + "right": "issue:45584", + "accept": false, + "reason": "Resize-token-embeddings/output-head mismatch is unrelated to Whisper's empty-transcription failure after align_special_tokens." + }, + { + "left": "issue:45292", + "right": "issue:45701", + "accept": false, + "reason": "Embedding resize propagation and version-dependent tokenization are different problems." + }, + { + "left": "issue:45561", + "right": "issue:45563", + "accept": false, + "reason": "pytest-xdist file race vs a stale generation warning; no shared failure mode." + }, + { + "left": "issue:43901", + "right": "issue:45003", + "accept": false, + "reason": "Docs mismatch about return_all_scores vs unsafe sys.modules access in modeling_utils." + }, + { + "left": "issue:44987", + "right": "issue:45459", + "accept": false, + "reason": "Loading a package under transformers>=5.1.0 and protobuf decode error handling are different tokenizer-loading issues." + }, + { + "left": "issue:45375", + "right": "issue:45759", + "accept": false, + "reason": "Both are Qwen multimodal config issues, but one is a missing config field and the other is from_config not unwrapping text_config." + }, + { + "left": "issue:43232", + "right": "issue:44336", + "accept": false, + "reason": "Generation kwargs sync_gpus bug and ANSI-code logging output are unrelated." + }, + { + "left": "issue:36246", + "right": "issue:43596", + "accept": false, + "reason": "ImportError for a missing processor export is unrelated to the DeepSpeed/BertModel index error." + }, + { + "left": "issue:43526", + "right": "issue:45003", + "accept": false, + "reason": "Beit label reduction bug and sys.modules access bug do not share the same code path." + }, + { + "left": "issue:44496", + "right": "issue:45003", + "accept": false, + "reason": "Unrecognized-model config loading failure is unrelated to sys.modules indexing in modeling_utils." + }, + { + "left": "issue:43257", + "right": "issue:45137", + "accept": false, + "reason": "Qwen3 MoE weight-conversion under accelerate+deepspeed and ZeRO3 deque underflow are different distributed-loading failures." + }, + { + "left": "issue:43065", + "right": "issue:43408", + "accept": false, + "reason": "Sam3PixelDecoder dummy layer issue and sam3_video/sam3_tracker warning are different SAM3 model problems." + }, + { + "left": "issue:44995", + "right": "issue:45200", + "accept": false, + "reason": "GlmMoeDsa stale indexer cache and Gemma 4 mm_token_type_ids defaulting are unrelated." + }, + { + "left": "issue:43262", + "right": "issue:44246", + "accept": false, + "reason": "Audio chat-template sample-rate defaulting and import-time slowness are not the same bug." + }, + { + "left": "issue:42886", + "right": "issue:45229", + "accept": false, + "reason": "Offline tokenizer cache loading failure and Gemma multi-GPU OOM are unrelated." + }, + { + "left": "issue:43475", + "right": "issue:43540", + "accept": false, + "reason": "SAM3 vision encoder missing attribute vs Qwen3OmniMoe video-input ValueError; different failures." + }, + { + "left": "issue:44466", + "right": "issue:45335", + "accept": false, + "reason": "Tied-weights serialization inconsistency and resize_token_embeddings not updating decoder.embed_tokens are related by theme but not the same concrete bug." + }, + { + "left": "issue:44995", + "right": "issue:45588", + "accept": false, + "reason": "Stale indexer cache and flash_attention s_aux=None crash are different runtime errors." + }, + { + "left": "issue:43653", + "right": "issue:44162", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and ESM2 breakage are unrelated." + }, + { + "left": "issue:44368", + "right": "issue:44977", + "accept": false, + "reason": "Tie-word-embeddings warning in Qwen3.5 and flash-attention generation failure are distinct issues." + }, + { + "left": "issue:44368", + "right": "issue:45563", + "accept": false, + "reason": "Qwen tied-weights warning and stale num_return_sequences warning are unrelated generation issues." + }, + { + "left": "issue:45397", + "right": "issue:45656", + "accept": false, + "reason": "Gemma zero3 from_pretrained and double optimizer.step under deepspeed are different distributed-training bugs." + }, + { + "left": "issue:44060", + "right": "issue:44077", + "accept": false, + "reason": "Incorrect tied-weights warning in Qwen3-Next and patchtsmixer post_init support are not the same underlying change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 34, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:28282", + "issue:30990", + "issue:41628", + "issue:42371", + "issue:42673", + "issue:42831", + "issue:42994", + "issue:43232", + "issue:43452", + "issue:43475", + "issue:43482", + "issue:43525", + "issue:43643", + "issue:43646", + "issue:43673", + "issue:43704", + "issue:43827", + "issue:43931", + "issue:44038", + "issue:44162", + "issue:44222", + "issue:44297", + "issue:44351", + "issue:44361", + "issue:44368", + "issue:44623", + "issue:44661", + "issue:44734", + "issue:44749", + "issue:44869", + "issue:44933", + "issue:45003", + "issue:45081", + "issue:45092", + "issue:45103", + "issue:45137", + "issue:45230", + "issue:45310", + "issue:45356", + "issue:45375", + "issue:45381", + "issue:45397", + "issue:45459", + "issue:45478", + "issue:45561", + "issue:45600", + "issue:45656" + ], + "soft_pairs": [ + "issue:43673|issue:44734", + "issue:45003|issue:45561", + "issue:45137|issue:45600", + "issue:43931|issue:44368", + "issue:44869|issue:45356", + "issue:44933|issue:45459", + "issue:44297|issue:44623", + "issue:43646|issue:44162", + "issue:44361|issue:44749", + "issue:43232|issue:43673", + "issue:43673|issue:43827", + "issue:43452|issue:45003", + "issue:28282|issue:30990", + "issue:44623|issue:45656", + "issue:42994|issue:43482", + "issue:43475|issue:43643", + "issue:44749|issue:45310", + "issue:44749|issue:45656", + "issue:42371|issue:45092", + "issue:43475|issue:45381", + "issue:44038|issue:44623", + "issue:45230|issue:45397", + "issue:41628|issue:44933", + "issue:45081|issue:45103", + "issue:42994|issue:43525", + "issue:45375|issue:45478", + "issue:42673|issue:43704", + "issue:43475|issue:44661", + "issue:45310|issue:45375", + "issue:42831|issue:44222", + "issue:41628|issue:44351" + ], + "prepared_review_unit_hash": "216b54cc7c38d55e3bc044214ef41e861aa802f7e6c67ea5894cf81f0f2fad85", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36249, + "estimated_input_tokens": 9063, + "estimated_eval_tokens": 18382 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly heterogeneous issue set. The only clear duplicate cluster is the Qwen3.5-Moe v5 from_pretrained/config regression, centered on the missing-field report; there is also one exact duplicate pair for the Qwen3ForCausalLM VRAM leak.", + "confidence": 0.74, + "canonical_issue_reason": "issue:45375 is the most root-cause-like and central report in the Qwen3.5-Moe cluster: it names the missing `deepstack_visual_indexes` field being dropped by `@strict`, which plausibly explains the related from_pretrained failures.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45375 is the strongest issue anchor overall because it is specific, technically actionable, and sits at the center of the only credible multi-issue duplicate cluster in this set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43673", + "right": "issue:44734", + "accept": false, + "reason": "Both involve generation/cache failures, but they describe different APIs and failure modes (`chunked_prefill` cache lookup vs `/v1/responses` tensor indexing)." + }, + { + "left": "issue:45003", + "right": "issue:45561", + "accept": false, + "reason": "`sys.modules` access safety and an xdist `captured_info.txt` race are unrelated bugs in different code paths." + }, + { + "left": "issue:45137", + "right": "issue:45600", + "accept": false, + "reason": "DeepSpeed ZeRO3 deque underflow and a removed-Sam3LiteText config reference are unrelated regressions." + }, + { + "left": "issue:43931", + "right": "issue:44368", + "accept": false, + "reason": "A Qwen3-VL weight-shape mismatch is not the same bug as a Qwen3.5 LoRA warning about `tie_word_embeddings`." + }, + { + "left": "issue:44869", + "right": "issue:45356", + "accept": false, + "reason": "Whisper timestamp decoding and Kimi-K2.5 tokenizer codec handling are different tokenizer bugs." + }, + { + "left": "issue:44933", + "right": "issue:45459", + "accept": false, + "reason": "Missing import from `image_utils` and protobuf-related tokenizer error masking are different failures." + }, + { + "left": "issue:44297", + "right": "issue:44623", + "accept": false, + "reason": "Both are save_pretrained-related, but one is tokenizer metadata mismatch while the other is processor file omission; not the same concrete bug." + }, + { + "left": "issue:43646", + "right": "issue:44162", + "accept": false, + "reason": "A broad custom-model init regression and an ESM2-specific break are too different to treat as the same underlying bug." + }, + { + "left": "issue:44361", + "right": "issue:44749", + "accept": false, + "reason": "`MLukeTokenizer` task-attribute errors and a general slowdown after upgrading Transformers are unrelated." + }, + { + "left": "issue:43232", + "right": "issue:43673", + "accept": false, + "reason": "These are both generation-related, but one is a post-`sync_gpus` kwargs update issue and the other is a missing generation cache in chunked prefill." + }, + { + "left": "issue:43673", + "right": "issue:43827", + "accept": false, + "reason": "A runtime cache bug and docs still mentioning removed `pipeline()` APIs are not the same issue." + }, + { + "left": "issue:43452", + "right": "issue:45003", + "accept": false, + "reason": "`gguf_file`/from_pretrained loading failure is unrelated to unsafe `sys.modules` access in modeling_utils." + }, + { + "left": "issue:28282", + "right": "issue:30990", + "accept": false, + "reason": "PyTorch-missing import errors and Sentence Transformers hanging during load are different problem reports." + }, + { + "left": "issue:44623", + "right": "issue:45656", + "accept": false, + "reason": "Processor save_pretrained file output and a deepspeed double-step optimizer bug are unrelated." + }, + { + "left": "issue:42994", + "right": "issue:43482", + "accept": false, + "reason": "Quantized model saving and Qwen2.5-GGUF loading under v5 are different IO paths and failure modes." + }, + { + "left": "issue:43475", + "right": "issue:43643", + "accept": false, + "reason": "SAM 3 video output attribute errors and missing fields from `trust_remote_code=True` config loading are unrelated." + }, + { + "left": "issue:44749", + "right": "issue:45310", + "accept": false, + "reason": "A performance regression is not the same concrete bug as a Qwen3.5-Moe from_pretrained failure." + }, + { + "left": "issue:44749", + "right": "issue:45656", + "accept": false, + "reason": "Performance slowdown and optimizer stepping twice under deepspeed are different issues." + }, + { + "left": "issue:42371", + "right": "issue:45092", + "accept": false, + "reason": "TF32 settings docs are unrelated to old InternVL2 remote-code checkpoints failing meta initialization." + }, + { + "left": "issue:43475", + "right": "issue:45381", + "accept": false, + "reason": "These are distinct model-specific vision bugs with different missing/incorrect fields." + }, + { + "left": "issue:44038", + "right": "issue:44623", + "accept": false, + "reason": "Qwen3-VL-Moe model loading and processor.save_pretrained file omission are unrelated." + }, + { + "left": "issue:45230", + "right": "issue:45397", + "accept": false, + "reason": "A placeholder bug report is not the same as the Gemma-4 ZeRO3 from_pretrained failure." + }, + { + "left": "issue:41628", + "right": "issue:44933", + "accept": false, + "reason": "Different import problems: `AutoImageProcessor` missing vs a nonexistent import from `image_utils`." + }, + { + "left": "issue:45081", + "right": "issue:45103", + "accept": false, + "reason": "A Mistral regex/tokenizer backend crash and an auto-docstring annotations crash are unrelated." + }, + { + "left": "issue:42994", + "right": "issue:43525", + "accept": false, + "reason": "Quantized model saving failure and missing `pad_token_id` on Llama4Config are different bugs." + }, + { + "left": "issue:45375", + "right": "issue:45478", + "accept": true, + "reason": "Both describe the same Qwen3.5-Moe v5 regression: a config-field handling problem that surfaces as from_pretrained failure." + }, + { + "left": "issue:42673", + "right": "issue:43704", + "accept": true, + "reason": "Exact duplicate title and same VRAM leak in Qwen3ForCausalLM with multiple dataloader threads." + }, + { + "left": "issue:43475", + "right": "issue:44661", + "accept": false, + "reason": "SAM3 vision output attribute errors and `TOKENIZER_MAPPING_NAMES`-related add-new-model-like failure are unrelated." + }, + { + "left": "issue:45310", + "right": "issue:45375", + "accept": true, + "reason": "Same Qwen3.5-Moe v5 loading regression; 45375 appears to capture the underlying missing-field cause behind the from_pretrained error in 45310." + }, + { + "left": "issue:42831", + "right": "issue:44222", + "accept": false, + "reason": "FineGrainedFP8 accuracy loss and FP8 moe save_pretrained failure are different bug classes." + }, + { + "left": "issue:41628", + "right": "issue:44351", + "accept": false, + "reason": "Different import targets and causes (`AutoImageProcessor` vs `HybridCache`)." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded overall. The summary stays conservative, and the two accepted clusters are well supported by the packet: the Qwen3.5-Moe v5 from_pretrained/config regression centered on issue:45375, and the exact duplicate Qwen3ForCausalLM VRAM-leak pair. The rejected soft pairs are also framed cautiously and do not overclaim duplication." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 35, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:30333", + "issue:30990", + "issue:33290", + "issue:35141", + "issue:36010", + "issue:41628", + "issue:42175", + "issue:42222", + "issue:42898", + "issue:42994", + "issue:43232", + "issue:43278", + "issue:43377", + "issue:43479", + "issue:43723", + "issue:43784", + "issue:43824", + "issue:43874", + "issue:43901", + "issue:44038", + "issue:44060", + "issue:44079", + "issue:44188", + "issue:44222", + "issue:44242", + "issue:44295", + "issue:44373", + "issue:44448", + "issue:44496", + "issue:44568", + "issue:44623", + "issue:44734", + "issue:44779", + "issue:44792", + "issue:44964", + "issue:45092", + "issue:45103", + "issue:45125", + "issue:45137", + "issue:45200", + "issue:45310", + "issue:45356", + "issue:45405", + "issue:45561", + "issue:45600", + "issue:45656", + "issue:45712", + "issue:45812" + ], + "soft_pairs": [ + "issue:42222|issue:45200", + "issue:44222|issue:45310", + "issue:42994|issue:43824", + "issue:43377|issue:44242", + "issue:36010|issue:41628", + "issue:42175|issue:45712", + "issue:44038|issue:45310", + "issue:44496|issue:45561", + "issue:44964|issue:45137", + "issue:45561|issue:45600", + "issue:43901|issue:45561", + "issue:30990|issue:35141", + "issue:42898|issue:44568", + "issue:43232|issue:44792", + "issue:43232|issue:44188", + "issue:42898|issue:45812", + "issue:44188|issue:44373", + "issue:43784|issue:43824", + "issue:43278|issue:45137", + "issue:44623|issue:45310", + "issue:44734|issue:45103", + "issue:30333|issue:33290", + "issue:43723|issue:44568", + "issue:44060|issue:45125", + "issue:42898|issue:45356", + "issue:43874|issue:45092", + "issue:42898|issue:44779", + "issue:45405|issue:45712", + "issue:42898|issue:44448", + "issue:44188|issue:44964", + "issue:43479|issue:44079", + "issue:44295|issue:45656" + ], + "prepared_review_unit_hash": "fdc3a090b1e51a48529e1ff51ebda9742a0f215b08dcfe4f95bec49c9b290962", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37368, + "estimated_input_tokens": 9342, + "estimated_eval_tokens": 18940 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: most items are unrelated regressions across tokenizers, multimodal configs, generation, packaging, DeepSpeed, and docs. The soft pairs are mostly superficial keyword matches and should not be merged as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": "No safe canonical issue: the items do not describe one underlying bug or change, so choosing a single issue as the cluster representative would be misleading.", + "canonical_pr_reason": null, + "best_issue_reason": "No single issue is a good global representative; even the closest textual matches point to different code paths, models, or failure modes.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42222 - All vitpose model were brokentransformers/models/vitpose_", + "right": "issue:45200 - [Gemma 4] mm_token_type_ids required for text-only fine-tuning - should default to zeros", + "accept": false, + "reason": "Both are multimodal/model issues, but one is a broken VitPose model path and the other is Gemma 4 text-only mm_token_type_ids handling; different bugs and fixes." + }, + { + "left": "issue:44222 - [Bug] FP8 save_pretrained moe", + "right": "issue:45310 - [BUG] transformers>=5.4.0, Qwen3.5 Moe from_pretrained error", + "accept": false, + "reason": "Save-pretrained FP8/MoE serialization is a different failure path from a Qwen3.5 MoE load-time from_pretrained error." + }, + { + "left": "issue:42994 - quantized model saving failed", + "right": "issue:43824 - ImportError: cannot import name 'Qwen2_5_VLForConditionalGeneration' from 'transformers'", + "accept": false, + "reason": "Quantized save failures and missing class imports are unrelated problems in different subsystems." + }, + { + "left": "issue:43377 - [BUG] MIMI Encoder produces different outputs for batched vs single inputs due to missing padding mask support", + "right": "issue:44242 - Load balancing loss not added when output_router_logits=False", + "accept": false, + "reason": "One is an encoder batching/padding-mask correctness bug; the other is MoE loss computation gated by router logits." + }, + { + "left": "issue:36010 - ImportError: cannot import name 'GenerationMixin' from 'transformers.generation'", + "right": "issue:41628 - Cannot import name 'AutoImageProcessor' from 'transformers'", + "accept": false, + "reason": "Both are import errors, but they affect different symbols and likely different export/re-export fixes." + }, + { + "left": "issue:42175 - Tensorflow not include in the backend when using pip install '.[torch]'", + "right": "issue:45712 - Six leftover dummy classes in `dummy_pt_objects.py` fail `check_repo.py` and leak into `dir(transformers)` without torch", + "accept": false, + "reason": "Packaging dependency selection and leftover dummy class cleanup are different repository hygiene issues." + }, + { + "left": "issue:44038 - [bug] transformers 5.0 & Qwen3-VL-Moe", + "right": "issue:45310 - [BUG] transformers>=5.4.0, Qwen3.5 Moe from_pretrained error", + "accept": false, + "reason": "Same broad model family, but different versions/models and different load/runtime failures; too coarse to be one duplicate." + }, + { + "left": "issue:44496 - ValueError: Unrecognized model in allenai/Olmo-Hybrid-Instruct-SFT-7B. Should have a `model_type` key in its config.json.", + "right": "issue:45561 - [Bug] pytest-xdist workers race on captured_info.txt in patched testing utils", + "accept": false, + "reason": "A model config loading error is unrelated to a parallel test-file race." + }, + { + "left": "issue:44964 - Cannot load `microsoft/Phi-4-multimodal-instruct` model with latest transformers.", + "right": "issue:45137 - IndexError: pop from an empty deque with DeepSpeed ZeRO3", + "accept": false, + "reason": "Multimodal model loading and DeepSpeed ZeRO3 deque underflow are unrelated failures." + }, + { + "left": "issue:45561 - [Bug] pytest-xdist workers race on captured_info.txt in patched testing utils", + "right": "issue:45600 - auto_mappings.py references removed Sam3LiteText configs, breaking CI", + "accept": false, + "reason": "A pytest-xdist file race and stale auto-mapping references are different CI problems." + }, + { + "left": "issue:43901 - TextClassificationPipeline docs still mention return_all_scores, but behavior differs", + "right": "issue:45561 - [Bug] pytest-xdist workers race on captured_info.txt in patched testing utils", + "accept": false, + "reason": "Documentation mismatch for a pipeline option is unrelated to a test utility race." + }, + { + "left": "issue:30990 - Sentence Transformers Gets Stuck loading", + "right": "issue:35141 - resizing token embeddings causes output embedding to be reinitialized in `post_init` when `tie_word_embedding` is False", + "accept": false, + "reason": "Loading hangs and embedding reinitialization are different code paths and different bugs." + }, + { + "left": "issue:42898 - `clean_up_tokenization_spaces` behavior changes in v5", + "right": "issue:44568 - [BUG] add_special_tokens=True doesn't add BOS/EOS tokens for microsoft/mdeberta-v3-base tokenizer in transformers >=5.0", + "accept": false, + "reason": "Both touch tokenizer behavior, but one is cleanup-space semantics and the other is special-token insertion for a specific tokenizer." + }, + { + "left": "issue:43232 - _update_model_kwargs_for_generation after sync_gpus when generation", + "right": "issue:44792 - Failed test case `test_model_generate_images` for janus model", + "accept": false, + "reason": "Generation kwargs update after sync_gpus is not the same as a Janus image-generation test failure." + }, + { + "left": "issue:43232 - _update_model_kwargs_for_generation after sync_gpus when generation", + "right": "issue:44188 - Diverging attention kernels due to `allow_is_bidirectional_skip` branching on torch.compile", + "accept": false, + "reason": "Generation bookkeeping and attention-kernel divergence are different runtime bugs." + }, + { + "left": "issue:42898 - `clean_up_tokenization_spaces` behavior changes in v5", + "right": "issue:45812 - `AutoTokenizer` produces wrong token IDs for all Granite models (silent v4\u2192v5 regression)", + "accept": false, + "reason": "Both are tokenizer regressions, but one is a space-cleanup behavior change and the other is model-specific token ID generation." + }, + { + "left": "issue:44188 - Diverging attention kernels due to `allow_is_bidirectional_skip` branching on torch.compile", + "right": "issue:44373 - Wrong docstring for position_ids", + "accept": false, + "reason": "A runtime kernel divergence bug does not match a documentation-only position_ids issue." + }, + { + "left": "issue:43784 - NameError: name 'nn' is not defined when importing sentence-transformers with latest transformers", + "right": "issue:43824 - ImportError: cannot import name 'Qwen2_5_VLForConditionalGeneration' from 'transformers'", + "accept": false, + "reason": "Both are import failures, but they are for different packages/symbols and likely different missing exports." + }, + { + "left": "issue:43278 - Embedding layer dtype changed from BF16 in training to FP32 in evaluate.", + "right": "issue:45137 - IndexError: pop from an empty deque with DeepSpeed ZeRO3", + "accept": false, + "reason": "A dtype mismatch between train/eval is unrelated to a DeepSpeed ZeRO3 deque error." + }, + { + "left": "issue:44623 - [BUG] processor.save_pretrained(...) missing files", + "right": "issue:45310 - [BUG] transformers>=5.4.0, Qwen3.5 Moe from_pretrained error", + "accept": false, + "reason": "Save-pretrained missing files and load-time from_pretrained errors are opposite sides of the API but not the same bug." + }, + { + "left": "issue:44734 - `transformers serve`: /v1/responses crashes on KV cache continuation due to wrong tensor indexing", + "right": "issue:45103 - [auto_docstring] _process_kwargs_parameters crashes with AttributeError when module uses from __future__ import annotations", + "accept": false, + "reason": "Serving KV-cache tensor indexing and auto-docstring annotation handling are unrelated failures." + }, + { + "left": "issue:30333 - If a training job job failed MLFlow will not be reported and MLFlow shows job still running", + "right": "issue:33290 - oom when using adafactor optimizer in deepspeed", + "accept": false, + "reason": "MLflow job-state reporting and optimizer OOM in DeepSpeed are unrelated issues." + }, + { + "left": "issue:43723 - Issue loading tokenizer using AutoTokenizer.from_pretrained in v5", + "right": "issue:44568 - [BUG] add_special_tokens=True doesn't add BOS/EOS tokens for microsoft/mdeberta-v3-base tokenizer in transformers >=5.0", + "accept": false, + "reason": "A generic tokenizer loading issue is not the same as a specific BOS/EOS insertion regression." + }, + { + "left": "issue:44060 - Qwen3-Next: Incorrect tied weights warning ties embed_tokens.weight to linear_attn.dt_bias across all layers", + "right": "issue:45125 - Qwen3_5MoeForConditionalGeneration missing _tp_plan for tensor parallelism", + "accept": false, + "reason": "These are different Qwen model classes with different bugs: tied-weights warning vs missing tensor-parallel plan." + }, + { + "left": "issue:42898 - `clean_up_tokenization_spaces` behavior changes in v5", + "right": "issue:45356 - Regression in Kimi-K2.5 tokenizer from 5.3.0 to 5.4.0: incorrect codec handling and misleading fix_mistral_regex warning", + "accept": false, + "reason": "Both are tokenizer regressions, but the root causes and affected models differ." + }, + { + "left": "issue:43874 - [GLM46V] Glm46VImageProcessorFast missing get_number_of_image_patches breaks _get_num_multimodal_tokens (AttributeError)", + "right": "issue:45092 - [Bug] Old InternVL2 remote-code checkpoints are incompatible with Transformers v5 meta initialization", + "accept": false, + "reason": "Both are multimodal-model issues, but one is a missing image-processor method and the other is v5 meta-init incompatibility for a different model family." + }, + { + "left": "issue:42898 - `clean_up_tokenization_spaces` behavior changes in v5", + "right": "issue:44779 - Deepseek tokenizer produces incorrect results as of v5 (works in v4)", + "accept": false, + "reason": "Same area, but different tokenizer regressions with different model-specific causes." + }, + { + "left": "issue:45405 - `MIN_PEFT_VERSION` bumped to 0.18.2 which is not yet released on PyPI", + "right": "issue:45712 - Six leftover dummy classes in `dummy_pt_objects.py` fail `check_repo.py` and leak into `dir(transformers)` without torch", + "accept": false, + "reason": "A dependency version bump is unrelated to dummy class cleanup and symbol leakage." + }, + { + "left": "issue:42898 - `clean_up_tokenization_spaces` behavior changes in v5", + "right": "issue:44448 - [BUG] Different output for google/pegasus-cnn_dailymail between Transformers v4 and v5", + "accept": false, + "reason": "Both mention v4\u2192v5 behavior changes, but one is a tokenizer cleanup flag and the other is model output drift." + }, + { + "left": "issue:44188 - Diverging attention kernels due to `allow_is_bidirectional_skip` branching on torch.compile", + "right": "issue:44964 - Cannot load `microsoft/Phi-4-multimodal-instruct` model with latest transformers.", + "accept": false, + "reason": "Attention-kernel divergence and Phi-4 model loading are different problems." + }, + { + "left": "issue:43479 - Phi4MultimodalConfig incorrectly initializes default vision/audio configs when passed as None", + "right": "issue:44079 - `ModelOutput` keys aren't correctly assigned if key was previously None", + "accept": false, + "reason": "A multimodal config defaulting bug is unrelated to ModelOutput key assignment semantics." + }, + { + "left": "issue:44295 - An error occurs when reading position_ids after registering it as a buffer.", + "right": "issue:45656 - Optimizer step being called 2 times when using deepspeed", + "accept": false, + "reason": "Position_ids buffer access and double optimizer stepping in DeepSpeed are different runtime bugs." + } + ] + }, + "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": 16, + "review_unit_index": 36, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:22355", + "issue:28282", + "issue:30333", + "issue:30990", + "issue:34689", + "issue:35141", + "issue:36331", + "issue:38175", + "issue:41628", + "issue:42175", + "issue:42757", + "issue:42831", + "issue:42886", + "issue:42994", + "issue:43232", + "issue:43479", + "issue:43644", + "issue:43723", + "issue:43824", + "issue:43873", + "issue:43976", + "issue:44206", + "issue:44222", + "issue:44261", + "issue:44263", + "issue:44295", + "issue:44297", + "issue:44360", + "issue:44373", + "issue:44483", + "issue:44484", + "issue:44485", + "issue:44514", + "issue:44623", + "issue:44869", + "issue:44871", + "issue:44913", + "issue:44933", + "issue:44945", + "issue:44977", + "issue:45092", + "issue:45103", + "issue:45310", + "issue:45357", + "issue:45406", + "issue:45656", + "issue:45712", + "issue:45834" + ], + "soft_pairs": [ + "issue:44933|issue:45712", + "issue:42175|issue:44933", + "issue:44222|issue:44297", + "issue:43976|issue:44933", + "issue:43479|issue:44869", + "issue:44297|issue:45656", + "issue:44206|issue:45103", + "issue:42831|issue:44623", + "issue:44222|issue:44623", + "issue:30333|issue:36331", + "issue:41628|issue:42757", + "issue:43232|issue:45834", + "issue:30333|issue:34689", + "issue:43723|issue:44977", + "issue:43873|issue:44945", + "issue:44222|issue:44484", + "issue:44297|issue:45310", + "issue:28282|issue:30333", + "issue:30333|issue:35141", + "issue:44263|issue:45092", + "issue:44483|issue:45406", + "issue:44263|issue:44360", + "issue:44373|issue:44485", + "issue:44514|issue:45103", + "issue:43479|issue:43644", + "issue:44295|issue:44623", + "issue:44261|issue:45092", + "issue:45092|issue:45357", + "issue:43723|issue:44871", + "issue:30990|issue:38175", + "issue:43479|issue:44913", + "issue:42886|issue:42994", + "issue:41628|issue:43824", + "issue:22355|issue:28282" + ], + "prepared_review_unit_hash": "f15cab0f109701929c1a9eeb009949df7d9f0b155f2b0c99ebb25a7e2b45aaee", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37581, + "estimated_input_tokens": 9396, + "estimated_eval_tokens": 19048 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These soft pairs are mostly false positives: they mix unrelated import errors, save/load regressions, config persistence bugs, training issues, and serving/runtime crashes. I found no pair that clearly reflects the same underlying bug or change, so there is no real duplicate cluster here.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue is appropriate; the items do not converge on one underlying bug and the strongest similarities are only broad subsystem overlap.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue stands out as a global best representative because the set is heterogeneous and the soft links are not true duplicates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44933", + "right": "issue:45712", + "accept": false, + "reason": "Both involve imports/symbol exposure, but one is about image_utils and the other about dummy PT objects leaking into dir(transformers); different code paths and fixes." + }, + { + "left": "issue:42175", + "right": "issue:44933", + "accept": false, + "reason": "Backend-install packaging problem vs missing image_utils import; unrelated symptoms and remediation." + }, + { + "left": "issue:44222", + "right": "issue:44297", + "accept": false, + "reason": "Both mention save_pretrained, but one is FP8 MoE saving and the other is tokenizer metadata mismatch; not the same bug." + }, + { + "left": "issue:43976", + "right": "issue:44933", + "accept": false, + "reason": "Python version compatibility regression vs import exposure issue; no shared underlying failure." + }, + { + "left": "issue:43479", + "right": "issue:44869", + "accept": false, + "reason": "Config default initialization bug vs Whisper timestamp decode crash; completely different subsystems." + }, + { + "left": "issue:44297", + "right": "issue:45656", + "accept": false, + "reason": "Tokenizer save metadata issue vs deepspeed optimizer double-step; unrelated." + }, + { + "left": "issue:44206", + "right": "issue:45103", + "accept": false, + "reason": "Feature extractor argument regression vs auto-docstring crash from future annotations; different layers." + }, + { + "left": "issue:42831", + "right": "issue:44623", + "accept": false, + "reason": "FP8 accuracy issue vs processor.save_pretrained missing files; not the same code path." + }, + { + "left": "issue:44222", + "right": "issue:44623", + "accept": false, + "reason": "Both touch saving, but model FP8 saving failure and processor artifact omission are distinct bugs." + }, + { + "left": "issue:30333", + "right": "issue:36331", + "accept": false, + "reason": "MLflow reporting on failed training vs CustomTrainer signature change; different behaviors and fixes." + }, + { + "left": "issue:41628", + "right": "issue:42757", + "accept": false, + "reason": "Different missing imports from different packages and likely different dependency/version causes." + }, + { + "left": "issue:43232", + "right": "issue:45834", + "accept": false, + "reason": "Generation kwarg update bug vs long OCR input index error; unrelated model paths." + }, + { + "left": "issue:30333", + "right": "issue:34689", + "accept": false, + "reason": "Training failure reporting vs model loading regression for a specific vision model; not duplicates." + }, + { + "left": "issue:43723", + "right": "issue:44977", + "accept": false, + "reason": "Tokenizer loading regression vs flash-attention generation issue; different failure modes." + }, + { + "left": "issue:43873", + "right": "issue:44945", + "accept": false, + "reason": "Quantization offloading behavior vs pipeline parallelism output correctness; too different to merge." + }, + { + "left": "issue:44222", + "right": "issue:44484", + "accept": false, + "reason": "FP8 save_pretrained failure vs a question about shard size defaults; not the same bug." + }, + { + "left": "issue:44297", + "right": "issue:45310", + "accept": false, + "reason": "Tokenizer config saved incorrectly vs Qwen3.5 MoE from_pretrained error; related broadly to load/save, but not the same underlying defect." + }, + { + "left": "issue:28282", + "right": "issue:30333", + "accept": false, + "reason": "Missing PyTorch import dependency vs failed-job MLflow reporting; unrelated." + }, + { + "left": "issue:30333", + "right": "issue:35141", + "accept": false, + "reason": "Training-job reporting bug vs embedding reinitialization in post_init; different subsystems." + }, + { + "left": "issue:44263", + "right": "issue:45092", + "accept": false, + "reason": "DSA indexer split behavior vs remote-code checkpoint meta-init incompatibility; distinct bugs." + }, + { + "left": "issue:44483", + "right": "issue:45406", + "accept": false, + "reason": "Chat completions request handling vs Gemma4Processor attribute error in serve; different code paths." + }, + { + "left": "issue:44263", + "right": "issue:44360", + "accept": false, + "reason": "Both mention the DSA indexer, but one is about split outputs and the other about missing ReLU; not clearly the same defect." + }, + { + "left": "issue:44373", + "right": "issue:44485", + "accept": false, + "reason": "Docstring mismatch vs RoPE implementation discussion; one is documentation, the other model behavior." + }, + { + "left": "issue:44514", + "right": "issue:45103", + "accept": false, + "reason": "Processor batched chat-template crash vs auto-docstring annotation crash; unrelated." + }, + { + "left": "issue:43479", + "right": "issue:43644", + "accept": false, + "reason": "Specific multimodal config defaulting bug vs general non-persistent buffer initialization regression; only broad init-theme overlap." + }, + { + "left": "issue:44295", + "right": "issue:44623", + "accept": false, + "reason": "Reading position_ids after buffer registration vs processor.save_pretrained missing files; different issues." + }, + { + "left": "issue:44261", + "right": "issue:45092", + "accept": false, + "reason": "Missing rms_norm_eps precision issue vs v5 meta-initialization incompatibility for old checkpoints; not the same root cause." + }, + { + "left": "issue:45092", + "right": "issue:45357", + "accept": false, + "reason": "Both concern Qwen/remote-code model handling, but one is InternVL2 meta-init compatibility and the other is incorrect visual encoder keys on save_pretrained." + }, + { + "left": "issue:43723", + "right": "issue:44871", + "accept": false, + "reason": "Tokenizer loading regression vs eos_token_id config mismatch; related to model/tokenizer config but not duplicate bugs." + }, + { + "left": "issue:30990", + "right": "issue:38175", + "accept": false, + "reason": "Sentence Transformers loading stall vs zero-probability outputs in a vision model; unrelated symptoms." + }, + { + "left": "issue:43479", + "right": "issue:44913", + "accept": false, + "reason": "Different config persistence bugs in different model classes; too broad to treat as duplicates." + }, + { + "left": "issue:42886", + "right": "issue:42994", + "accept": false, + "reason": "Offline tokenizer cache loading vs quantized model saving failure; unrelated." + }, + { + "left": "issue:41628", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but for different exported classes and likely different fixes." + }, + { + "left": "issue:22355", + "right": "issue:28282", + "accept": false, + "reason": "Missing submodule import vs missing backend library dependency; different causes and scopes." + } + ] + }, + "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": 16, + "review_unit_index": 37, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:29127", + "issue:30990", + "issue:33357", + "issue:39401", + "issue:41628", + "issue:42371", + "issue:42757", + "issue:42831", + "issue:42886", + "issue:42907", + "issue:43066", + "issue:43116", + "issue:43122", + "issue:43232", + "issue:43408", + "issue:43475", + "issue:43576", + "issue:43723", + "issue:43756", + "issue:43784", + "issue:43937", + "issue:44038", + "issue:44060", + "issue:44117", + "issue:44261", + "issue:44263", + "issue:44295", + "issue:44297", + "issue:44360", + "issue:44373", + "issue:44442", + "issue:44485", + "issue:44514", + "issue:44521", + "issue:44568", + "issue:44625", + "issue:44704", + "issue:44779", + "issue:44829", + "issue:44945", + "issue:45200", + "issue:45245", + "issue:45278", + "issue:45290", + "issue:45310", + "issue:45593", + "issue:45600", + "issue:45753" + ], + "soft_pairs": [ + "issue:39401|issue:44568", + "issue:42831|issue:45310", + "issue:44263|issue:44373", + "issue:44945|issue:45753", + "issue:42831|issue:44038", + "issue:42757|issue:45278", + "issue:43576|issue:43784", + "issue:42886|issue:43066", + "issue:43232|issue:45593", + "issue:42371|issue:44263", + "issue:43756|issue:44060", + "issue:44117|issue:44704", + "issue:42907|issue:44945", + "issue:30990|issue:33357", + "issue:44704|issue:45290", + "issue:43122|issue:44514", + "issue:43232|issue:45245", + "issue:44295|issue:44297", + "issue:43784|issue:45600", + "issue:42907|issue:43937", + "issue:43723|issue:44779", + "issue:41628|issue:45278", + "issue:43723|issue:44442", + "issue:44261|issue:44373", + "issue:43408|issue:43475", + "issue:44360|issue:44485", + "issue:43122|issue:44521", + "issue:44625|issue:44829", + "issue:43116|issue:45200", + "issue:29127|issue:30990" + ], + "prepared_review_unit_hash": "e2dc9b6a0d0a29e3f5b8bac9ba6ded16e4c1156fc4b350741a151b5e5a50290b", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36340, + "estimated_input_tokens": 9085, + "estimated_eval_tokens": 18426 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a mixed cluster of distinct Transformers issues, mostly around v5 import/tokenizer/processor regressions, plus several unrelated model-specific bugs. No soft pair looks like the same underlying change, so none should be merged as duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the broadest umbrella for the repeated v5 import/loading regressions, but the cluster is too heterogeneous for a true duplicate canonical.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:39401", + "right": "issue:44568", + "accept": false, + "reason": "Different tokenizer bugs: Qwen3 offset mapping vs mdeberta-v3 BOS/EOS handling." + }, + { + "left": "issue:42831", + "right": "issue:45310", + "accept": false, + "reason": "Different failures: FineGrainedFP8 accuracy vs Qwen3.5-Moe from_pretrained error." + }, + { + "left": "issue:44263", + "right": "issue:44373", + "accept": false, + "reason": "One is a runtime torch.split issue, the other is a docstring problem." + }, + { + "left": "issue:44945", + "right": "issue:45753", + "accept": false, + "reason": "Both are output problems, but one is pipeline parallelism and the other is an image-specific infinite loop." + }, + { + "left": "issue:42831", + "right": "issue:44038", + "accept": false, + "reason": "Unrelated model bugs: FineGrainedFP8 accuracy vs Qwen3-VL-Moe compatibility." + }, + { + "left": "issue:42757", + "right": "issue:45278", + "accept": false, + "reason": "Specific import failure for one symbol vs a broad bundle of import errors; not the same bug." + }, + { + "left": "issue:43576", + "right": "issue:43784", + "accept": false, + "reason": "Different code paths: broken env command vs NameError importing sentence-transformers." + }, + { + "left": "issue:42886", + "right": "issue:43066", + "accept": false, + "reason": "Offline cache loading failure is distinct from wrong tokenizer decoder type." + }, + { + "left": "issue:43232", + "right": "issue:45593", + "accept": false, + "reason": "Generation kwargs bug vs missing auxiliary losses in D-FINE training." + }, + { + "left": "issue:42371", + "right": "issue:44263", + "accept": false, + "reason": "TF32 settings guidance is unrelated to the torch.split indexer bug." + }, + { + "left": "issue:43756", + "right": "issue:44060", + "accept": false, + "reason": "Different model-specific regressions: Smollm3 RoPE layers vs Qwen3-Next tied-weights warning." + }, + { + "left": "issue:44117", + "right": "issue:44704", + "accept": false, + "reason": "Tokenizer mapping None handling and AutoProcessor kwargs passing are separate loading issues." + }, + { + "left": "issue:42907", + "right": "issue:44945", + "accept": false, + "reason": "Saving dequantized checkpoints is unrelated to pipeline-parallel inference output correctness." + }, + { + "left": "issue:30990", + "right": "issue:33357", + "accept": false, + "reason": "Loading hang in Sentence Transformers vs MacOS bus error on a CLIP model." + }, + { + "left": "issue:44704", + "right": "issue:45290", + "accept": false, + "reason": "AutoProcessor file loading and chat-template batching crashes are different code paths." + }, + { + "left": "issue:43122", + "right": "issue:44514", + "accept": false, + "reason": "Different tokenizer/processor regressions: tokenization changes vs batched apply_chat_template crash." + }, + { + "left": "issue:43232", + "right": "issue:45245", + "accept": false, + "reason": "Generation state update bug vs a categorical-cardinality runtime error." + }, + { + "left": "issue:44295", + "right": "issue:44297", + "accept": false, + "reason": "Position_ids buffer access and tokenizer_class mismatch are unrelated serialization issues." + }, + { + "left": "issue:43784", + "right": "issue:45600", + "accept": false, + "reason": "Importing sentence-transformers failure is unrelated to auto_mappings references to removed configs." + }, + { + "left": "issue:42907", + "right": "issue:43937", + "accept": false, + "reason": "Dequantized save failure vs invalid GenerationConfig are separate bugs." + }, + { + "left": "issue:43723", + "right": "issue:44779", + "accept": false, + "reason": "Tokenizer loading failure and incorrect Deepseek tokenization are not the same underlying defect." + }, + { + "left": "issue:41628", + "right": "issue:45278", + "accept": false, + "reason": "One specific missing export vs many import errors overall; too broad to treat as the same bug." + }, + { + "left": "issue:43723", + "right": "issue:44442", + "accept": false, + "reason": "FastSpeech2ConformerTokenizer loading failure is distinct from Deepseek tokenization correctness." + }, + { + "left": "issue:44261", + "right": "issue:44373", + "accept": false, + "reason": "MLA layernorm precision/config issue is unrelated to a position_ids docstring." + }, + { + "left": "issue:43408", + "right": "issue:43475", + "accept": false, + "reason": "Same SAM 3 family, but different concrete bugs: model-type warning vs missing attribute crash." + }, + { + "left": "issue:44360", + "right": "issue:44485", + "accept": false, + "reason": "DSA indexer ReLU discussion and GLM-5 RoPE implementation are unrelated." + }, + { + "left": "issue:43122", + "right": "issue:44521", + "accept": false, + "reason": "Both touch templating/tokenization, but one is tokenization drift and the other is multimodal assistant masks." + }, + { + "left": "issue:44625", + "right": "issue:44829", + "accept": false, + "reason": "Qwen3.5 num_labels propagation and flash_attention_3 degenerate training are unrelated." + }, + { + "left": "issue:43116", + "right": "issue:45200", + "accept": false, + "reason": "Example-script multi-label empty outputs and Gemma 4 mm_token_type_ids defaults are different bugs." + }, + { + "left": "issue:29127", + "right": "issue:30990", + "accept": false, + "reason": "LayoutLMv3 error messaging and Sentence Transformers loading hang are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 38, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:22355", + "issue:28218", + "issue:29127", + "issue:33453", + "issue:36296", + "issue:38617", + "issue:39401", + "issue:42222", + "issue:42371", + "issue:42757", + "issue:42831", + "issue:42907", + "issue:43064", + "issue:43066", + "issue:43116", + "issue:43122", + "issue:43441", + "issue:43452", + "issue:43493", + "issue:43673", + "issue:43784", + "issue:43824", + "issue:43825", + "issue:43874", + "issue:43937", + "issue:43976", + "issue:43992", + "issue:44038", + "issue:44075", + "issue:44222", + "issue:44295", + "issue:44297", + "issue:44351", + "issue:44485", + "issue:44488", + "issue:44521", + "issue:44704", + "issue:44779", + "issue:44936", + "issue:44964", + "issue:44977", + "issue:45083", + "issue:45216", + "issue:45278", + "issue:45446", + "issue:45478", + "issue:45563", + "issue:45753" + ], + "soft_pairs": [ + "issue:42907|issue:44964", + "issue:43122|issue:44779", + "issue:44038|issue:44297", + "issue:42907|issue:43452", + "issue:39401|issue:44488", + "issue:44038|issue:44295", + "issue:36296|issue:43116", + "issue:43824|issue:44351", + "issue:44038|issue:44222", + "issue:39401|issue:44521", + "issue:45216|issue:45478", + "issue:42757|issue:44351", + "issue:43066|issue:44779", + "issue:22355|issue:28218", + "issue:42907|issue:44222", + "issue:43122|issue:44977", + "issue:43992|issue:44704", + "issue:43784|issue:45446", + "issue:43937|issue:44485", + "issue:42222|issue:43441", + "issue:42371|issue:43874", + "issue:43976|issue:45446", + "issue:42831|issue:45753", + "issue:43493|issue:45753", + "issue:44075|issue:44936", + "issue:43116|issue:43825", + "issue:44038|issue:45478", + "issue:44222|issue:45478", + "issue:43116|issue:43493", + "issue:43937|issue:45563", + "issue:43064|issue:43673", + "issue:45083|issue:45446", + "issue:38617|issue:45278", + "issue:39401|issue:44779", + "issue:33453|issue:39401", + "issue:22355|issue:29127" + ], + "prepared_review_unit_hash": "34b30488a9487adc13de1a0557d21f9670f726578b7780ee1eeac3d6b00d9d86", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 36, + "serialized_chars": 37929, + "estimated_input_tokens": 9483, + "estimated_eval_tokens": 19222 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is highly heterogeneous: tokenizer regressions, import errors, training bugs, multimodal issues, and model-specific failures are all mixed together. None of the soft pairs look like the same underlying bug/change, so I would not collapse this cluster.", + "confidence": 0.94, + "canonical_issue_reason": "No single canonical issue stands out because the items are mostly unrelated bugs across different code paths and models; this does not look like a true duplicate cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44779 is the most self-contained concrete bug report, but only as a standalone issue; it does not represent the rest of the mixed cluster well.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42907", + "right": "issue:44964", + "accept": false, + "reason": "Both involve model loading/saving, but they are different models and different failure modes: dequantized save vs multimodal load." + }, + { + "left": "issue:43122", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different models and symptoms; not the same concrete bug." + }, + { + "left": "issue:44038", + "right": "issue:44297", + "accept": false, + "reason": "Qwen3-VL-Moe bug vs tokenizer save_pretrained metadata mismatch; unrelated code paths." + }, + { + "left": "issue:42907", + "right": "issue:43452", + "accept": false, + "reason": "Different operations and paths: saving dequantized models versus loading GGUF through AutoTokenizer/AutoModel." + }, + { + "left": "issue:39401", + "right": "issue:44488", + "accept": false, + "reason": "Qwen3 tokenizer offset mapping bug and sleng-bert load failure are unrelated." + }, + { + "left": "issue:44038", + "right": "issue:44295", + "accept": false, + "reason": "Different model bug versus a position_ids buffer read error; no shared concrete failure." + }, + { + "left": "issue:36296", + "right": "issue:43116", + "accept": false, + "reason": "Tensor-parallel training bug and example-script multi-label output bug are unrelated." + }, + { + "left": "issue:43824", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but for different symbols and likely different fixes." + }, + { + "left": "issue:44038", + "right": "issue:44222", + "accept": false, + "reason": "Both mention MoE, but one is a Qwen3-VL-Moe issue and the other is FP8 save_pretrained; not the same bug." + }, + { + "left": "issue:39401", + "right": "issue:44521", + "accept": false, + "reason": "Tokenizer offset mapping and multimodal assistant mask generation are different behaviors and code paths." + }, + { + "left": "issue:45216", + "right": "issue:45478", + "accept": false, + "reason": "Both mention Qwen3.5 and v5.4+, but one is save_pretrained checkpoint correctness and the other is from_pretrained loading failure." + }, + { + "left": "issue:42757", + "right": "issue:44351", + "accept": false, + "reason": "Different missing imports from different packages; these are separate dependency/export issues." + }, + { + "left": "issue:43066", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer regressions in v5, but they concern different tokenizers and different symptoms." + }, + { + "left": "issue:22355", + "right": "issue:28218", + "accept": false, + "reason": "No module transformers.onnx and extra-space tokenization are unrelated." + }, + { + "left": "issue:42907", + "right": "issue:44222", + "accept": false, + "reason": "Both are save-related, but dequantized save failure and FP8 save_pretrained failure are not the same concrete problem." + }, + { + "left": "issue:43122", + "right": "issue:44977", + "accept": false, + "reason": "Tokenizer regression versus flash-attention generation failure; unrelated bugs." + }, + { + "left": "issue:43992", + "right": "issue:44704", + "accept": false, + "reason": "Missing embed_tokens.weight in from_pretrained and kwargs not forwarded to cached_file are different load-path issues." + }, + { + "left": "issue:43784", + "right": "issue:45446", + "accept": false, + "reason": "Sentence-transformers import NameError and flex_attention version-check bug are unrelated." + }, + { + "left": "issue:43937", + "right": "issue:44485", + "accept": false, + "reason": "Both touch GLM-5, but generation config validation and RoPE implementation are distinct issues." + }, + { + "left": "issue:42222", + "right": "issue:43441", + "accept": false, + "reason": "Different models and different failures: vitpose model breakage versus Ministral-3 FlashAttention failure." + }, + { + "left": "issue:42371", + "right": "issue:43874", + "accept": false, + "reason": "TF32 API messaging and a missing image-patch helper are unrelated." + }, + { + "left": "issue:43976", + "right": "issue:45446", + "accept": false, + "reason": "Python-version compatibility and flex_attention import version checks are separate issues." + }, + { + "left": "issue:42831", + "right": "issue:45753", + "accept": false, + "reason": "FP8 accuracy issue and Qwen3_5 infinite-loop-on-image are different bugs." + }, + { + "left": "issue:43493", + "right": "issue:45753", + "accept": false, + "reason": "SigLIP2 implementation discrepancy and Qwen3_5 generation loop are unrelated." + }, + { + "left": "issue:44075", + "right": "issue:44936", + "accept": false, + "reason": "SGD args ignored and train/evaluate interaction failure are different training bugs." + }, + { + "left": "issue:43116", + "right": "issue:43825", + "accept": false, + "reason": "Example script output bug and pipeline error-message bug are unrelated." + }, + { + "left": "issue:44038", + "right": "issue:45478", + "accept": false, + "reason": "Different Qwen3.5/Qwen3-VL-Moe failures: one is a model-specific bug, the other is a from_pretrained error." + }, + { + "left": "issue:44222", + "right": "issue:45478", + "accept": false, + "reason": "FP8 save_pretrained and Qwen3.5 Moe from_pretrained are distinct code paths." + }, + { + "left": "issue:43116", + "right": "issue:43493", + "accept": false, + "reason": "Multi-label classification example bug and SigLIP2 discrepancy are unrelated." + }, + { + "left": "issue:43937", + "right": "issue:45563", + "accept": false, + "reason": "GenerationConfig validation and stale num_return_sequences warning are different generation-related issues." + }, + { + "left": "issue:43064", + "right": "issue:43673", + "accept": false, + "reason": "Wrong optimizer states in FSDP2/PEFT training and missing GenerationMixin cache in chunked_prefill are unrelated." + }, + { + "left": "issue:45083", + "right": "issue:45446", + "accept": false, + "reason": "Qwen3_omni_moe helper behavior and flex_attention import version check are unrelated." + }, + { + "left": "issue:38617", + "right": "issue:45278", + "accept": false, + "reason": "Specific missing layer_type_validation import and a broader upgrade-related import-error report are not the same bug." + }, + { + "left": "issue:39401", + "right": "issue:44779", + "accept": false, + "reason": "Different tokenizers and different manifestations; not a duplicate." + }, + { + "left": "issue:33453", + "right": "issue:39401", + "accept": false, + "reason": "Generic tokenizer loading regression and Qwen3 offset_mapping bug are too different to merge." + }, + { + "left": "issue:22355", + "right": "issue:29127", + "accept": false, + "reason": "ONNX import failure and LayoutLMv3 error-message clarity are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 39, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:38617", + "issue:39401", + "issue:41628", + "issue:42175", + "issue:42222", + "issue:42371", + "issue:42673", + "issue:43012", + "issue:43066", + "issue:43122", + "issue:43381", + "issue:43408", + "issue:43421", + "issue:43450", + "issue:43577", + "issue:43742", + "issue:43827", + "issue:43874", + "issue:43937", + "issue:43986", + "issue:44075", + "issue:44117", + "issue:44220", + "issue:44261", + "issue:44291", + "issue:44410", + "issue:44483", + "issue:44484", + "issue:44496", + "issue:44829", + "issue:44945", + "issue:44964", + "issue:44977", + "issue:45071", + "issue:45072", + "issue:45081", + "issue:45216", + "issue:45278", + "issue:45310", + "issue:45405", + "issue:45446", + "issue:45459", + "issue:45464", + "issue:45478", + "issue:45542", + "issue:45588", + "issue:45656" + ], + "soft_pairs": [ + "issue:44117|issue:45588", + "issue:43381|issue:44945", + "issue:43827|issue:45446", + "issue:43937|issue:44977", + "issue:45405|issue:45446", + "issue:43450|issue:43986", + "issue:42673|issue:45072", + "issue:44117|issue:45081", + "issue:39401|issue:43421", + "issue:38617|issue:41628", + "issue:44829|issue:45216", + "issue:44410|issue:45310", + "issue:44291|issue:45071", + "issue:42222|issue:45542", + "issue:44484|issue:45310", + "issue:42175|issue:42222", + "issue:43937|issue:44496", + "issue:43408|issue:43742", + "issue:44483|issue:45464", + "issue:44075|issue:45656", + "issue:44964|issue:45278", + "issue:43012|issue:43577", + "issue:43066|issue:43122", + "issue:42371|issue:44261", + "issue:38617|issue:44220", + "issue:44410|issue:45478", + "issue:43874|issue:44261", + "issue:44484|issue:45478", + "issue:42222|issue:45459" + ], + "prepared_review_unit_hash": "cfce38af58a126a060bf3e50d1efb1d5d3eeb96eeae4a4206584ff85da56303a", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35813, + "estimated_input_tokens": 8954, + "estimated_eval_tokens": 18164 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly unrelated transformer bug reports, with only one plausible duplicate-style tokenizer regression pair. The broadest representative issue is the upgrade-related regression report 45278.", + "confidence": 0.42, + "canonical_issue_reason": "45278 is the broadest umbrella report in the set: a general breakage after upgrading to 5.5.0, which makes it the best single anchor for this mixed cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "45278 best represents the cluster because it captures the widest user-facing regression scope and can subsume narrower follow-up reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44117", + "right": "issue:45588", + "accept": false, + "reason": "Different subsystems: tokenizer mapping fallback vs flash-attention sinkless-model crash." + }, + { + "left": "issue:43381", + "right": "issue:44945", + "accept": false, + "reason": "Unrelated bugs: grad checkpointing in eval mode vs wrong outputs under pipeline parallelism." + }, + { + "left": "issue:43827", + "right": "issue:45446", + "accept": false, + "reason": "Docs cleanup about pipeline() vs a flex_attention import/version check bug." + }, + { + "left": "issue:43937", + "right": "issue:44977", + "accept": false, + "reason": "Different model-specific generation failures with no clear shared code path." + }, + { + "left": "issue:45405", + "right": "issue:45446", + "accept": false, + "reason": "PEFT version pinning issue is unrelated to the flex_attention import check." + }, + { + "left": "issue:43450", + "right": "issue:43986", + "accept": false, + "reason": "Batched video processor shape bug vs missing torchvision crash during AutoProcessor loading." + }, + { + "left": "issue:42673", + "right": "issue:45072", + "accept": false, + "reason": "VRAM leak in Qwen3ForCausalLM threads is unrelated to dtype mismatch in inference." + }, + { + "left": "issue:44117", + "right": "issue:45081", + "accept": false, + "reason": "Both involve tokenizer loading, but the concrete failures are different code paths and symptoms." + }, + { + "left": "issue:39401", + "right": "issue:43421", + "accept": false, + "reason": "Offset_mapping regression and runtime special-token post-processing are related only broadly, not the same bug." + }, + { + "left": "issue:38617", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors, but they involve different missing symbols and different breakages." + }, + { + "left": "issue:44829", + "right": "issue:45216", + "accept": false, + "reason": "Flash-attention training degeneration is unrelated to save_pretrained checkpoint corruption." + }, + { + "left": "issue:44410", + "right": "issue:45310", + "accept": false, + "reason": "Different Qwen model families and different loading failures; not the same underlying defect." + }, + { + "left": "issue:44291", + "right": "issue:45071", + "accept": false, + "reason": "init_empty_weights argument handling and PretrainedConfig type checking are separate loading regressions." + }, + { + "left": "issue:42222", + "right": "issue:45542", + "accept": false, + "reason": "VitPose model breakage is unrelated to the tensorboard/TensorFlow backend import issue." + }, + { + "left": "issue:44484", + "right": "issue:45464", + "accept": false, + "reason": "Chat/completions request validation and streaming inference failure are different API problems." + }, + { + "left": "issue:42175", + "right": "issue:42222", + "accept": false, + "reason": "Package dependency/backend installation issue vs a broken model implementation." + }, + { + "left": "issue:43937", + "right": "issue:44496", + "accept": false, + "reason": "GenerationConfig validation in GLM-5 is unrelated to unrecognized-model config loading." + }, + { + "left": "issue:43408", + "right": "issue:43742", + "accept": false, + "reason": "Model-type warning for sam3_tracker is not the same as MobileLLM config loading failure." + }, + { + "left": "issue:44483", + "right": "issue:45464", + "accept": false, + "reason": "Both mention chat/completions, but one is endpoint request acceptance and the other is streaming generation failure on a specific model." + }, + { + "left": "issue:44075", + "right": "issue:45656", + "accept": false, + "reason": "SGD argument handling and double optimizer steps under deepspeed are different optimizer bugs." + }, + { + "left": "issue:44964", + "right": "issue:45278", + "accept": false, + "reason": "A specific model load failure is not the same as the broad import-error regression report." + }, + { + "left": "issue:43012", + "right": "issue:43577", + "accept": false, + "reason": "Both touch dtype/precision, but one is a compile-time warning and the other is incorrect loaded dtypes." + }, + { + "left": "issue:43066", + "right": "issue:43122", + "accept": true, + "reason": "These look like the same tokenizer v5 regression: a wrong decoder type can directly explain changed tokenization behavior across versions." + }, + { + "left": "issue:42371", + "right": "issue:44261", + "accept": false, + "reason": "TF32 behavior API guidance is unrelated to the missing rms_norm_eps precision issue." + }, + { + "left": "issue:38617", + "right": "issue:44220", + "accept": false, + "reason": "ImportError for a config helper and an audio feature-extraction issue are unrelated." + }, + { + "left": "issue:44410", + "right": "issue:45478", + "accept": false, + "reason": "Different Qwen model-loading failures; no evidence they share the same fix." + }, + { + "left": "issue:43874", + "right": "issue:44261", + "accept": false, + "reason": "Multimodal token counting bug vs MLA layer norm epsilon precision issue." + }, + { + "left": "issue:44484", + "right": "issue:45478", + "accept": false, + "reason": "Request acceptance at chat/completions and Qwen3.5 MoE loading are separate bugs." + }, + { + "left": "issue:42222", + "right": "issue:45459", + "accept": false, + "reason": "VitPose model breakage is unrelated to tokenizer protobuf decode-error masking." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative about the mostly unrelated set of issues. The chosen anchor issue 45278 is a reasonable broad representative for the upgrade/regression subset, and the only accepted soft pair (43066/43122) is plausibly the same tokenizer regression." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 40, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:38617", + "issue:41628", + "issue:42371", + "issue:42673", + "issue:42757", + "issue:42831", + "issue:43012", + "issue:43064", + "issue:43066", + "issue:43097", + "issue:43335", + "issue:43408", + "issue:43441", + "issue:43452", + "issue:43502", + "issue:43577", + "issue:43873", + "issue:43874", + "issue:43906", + "issue:43931", + "issue:43976", + "issue:43986", + "issue:43992", + "issue:44038", + "issue:44062", + "issue:44117", + "issue:44261", + "issue:44263", + "issue:44315", + "issue:44351", + "issue:44368", + "issue:44410", + "issue:44484", + "issue:44485", + "issue:44493", + "issue:44704", + "issue:44829", + "issue:44945", + "issue:44964", + "issue:44977", + "issue:44987", + "issue:45071", + "issue:45103", + "issue:45245", + "issue:45341", + "issue:45375", + "issue:45542", + "issue:45588" + ], + "soft_pairs": [ + "issue:43097|issue:44493", + "issue:44263|issue:44485", + "issue:43441|issue:44829", + "issue:43012|issue:43408", + "issue:41628|issue:45071", + "issue:45071|issue:45542", + "issue:42673|issue:44977", + "issue:42371|issue:42831", + "issue:42673|issue:43906", + "issue:43931|issue:44038", + "issue:43986|issue:44484", + "issue:43335|issue:45071", + "issue:44117|issue:44987", + "issue:43097|issue:43452", + "issue:43976|issue:44945", + "issue:43408|issue:44368", + "issue:43874|issue:44263", + "issue:44829|issue:45245", + "issue:38617|issue:42757", + "issue:43064|issue:44315", + "issue:44410|issue:45375", + "issue:38617|issue:44351", + "issue:43097|issue:45588", + "issue:44062|issue:44117", + "issue:44704|issue:45103", + "issue:44038|issue:44410", + "issue:42673|issue:43577", + "issue:44964|issue:45071", + "issue:43931|issue:44410", + "issue:44261|issue:44485", + "issue:44964|issue:45341", + "issue:43502|issue:43992", + "issue:43986|issue:44117", + "issue:43097|issue:43873", + "issue:43066|issue:43906" + ], + "prepared_review_unit_hash": "d6bf840f7d22994a4c9df8ce1ec1ee71e76a279fbf855087293664fefc274f73", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 38130, + "estimated_input_tokens": 9533, + "estimated_eval_tokens": 19322 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly false-positive similarity across disparate Transformers issues. The items share broad v5/API/model-regression language, but none of the soft pairs look like the same underlying bug or change.", + "confidence": 0.23, + "canonical_issue_reason": "Issue 43097 is the cleanest representative of the cluster\u2019s broad v5 compatibility-breakage theme: it is concrete, user-facing, and about a specific API removal. That said, the set does not contain a strong true-duplicate group.", + "canonical_pr_reason": null, + "best_issue_reason": "43097 is the best issue to anchor on if this cluster must be collapsed, because it is the most general and actionable regression report among the candidates. The rest are mostly model-specific, import-specific, or unrelated behavioral bugs.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43097", + "right": "issue:44493", + "accept": false, + "reason": "One is a removed embedding-tie API complaint; the other is an unexpected position-id key warning. Different failure modes and code paths." + }, + { + "left": "issue:44263", + "right": "issue:44485", + "accept": false, + "reason": "Torch split return-value issue in a MoE indexer is not the same as a RoPE implementation discussion." + }, + { + "left": "issue:43441", + "right": "issue:44829", + "accept": false, + "reason": "Both mention flash attention, but one is a Ministral-3 load failure and the other is degenerate training with FA3. Different bugs." + }, + { + "left": "issue:43012", + "right": "issue:43408", + "accept": false, + "reason": "bfloat16 compilation warning is unrelated to the SAM3 model-type warning during AutoProcessor/model loading." + }, + { + "left": "issue:41628", + "right": "issue:45071", + "accept": false, + "reason": "AutoImageProcessor import failure and PretrainedConfig type-checking breakage are unrelated API regressions." + }, + { + "left": "issue:45071", + "right": "issue:45542", + "accept": false, + "reason": "Type-checking breakage in PretrainedConfig is not the same as the TensorFlow-backend detection error." + }, + { + "left": "issue:42673", + "right": "issue:44977", + "accept": false, + "reason": "VRAM leak in multithreaded Qwen3ForCausalLM and flash-attention generation corruption are different problems." + }, + { + "left": "issue:42371", + "right": "issue:42831", + "accept": false, + "reason": "TF32 settings warning is not the same as FineGrainedFP8 accuracy issues." + }, + { + "left": "issue:42673", + "right": "issue:43906", + "accept": false, + "reason": "A Qwen3 VRAM leak is not the same as an isolated reproduction of an unrelated earlier issue." + }, + { + "left": "issue:43931", + "right": "issue:44038", + "accept": false, + "reason": "Weight-shape mismatch on one Qwen3-VL model is distinct from a broader transformers-5/Qwen3-VL-Moe failure report." + }, + { + "left": "issue:43986", + "right": "issue:44484", + "accept": false, + "reason": "Missing torchvision crash for video models is unrelated to max_shard_size behavior." + }, + { + "left": "issue:43335", + "right": "issue:45071", + "accept": false, + "reason": "SwitchTransformers sparse-layer construction bug is unrelated to PretrainedConfig type checking." + }, + { + "left": "issue:44117", + "right": "issue:44987", + "accept": false, + "reason": "Tokenizer mapping None-handling is not the same as a model-loading failure for physical-intelligence/fast." + }, + { + "left": "issue:43097", + "right": "issue:43452", + "accept": false, + "reason": "Removed tie-embeddings API and gguf_file failing in tokenizer/model loading are different issues." + }, + { + "left": "issue:43976", + "right": "issue:44945", + "accept": false, + "reason": "Python version compatibility problem and incorrect output under pipeline parallelism are unrelated." + }, + { + "left": "issue:43408", + "right": "issue:44368", + "accept": false, + "reason": "SAM3 model-type warning and tie_word_embeddings warning are different configuration warnings with different causes." + }, + { + "left": "issue:43874", + "right": "issue:44263", + "accept": false, + "reason": "Missing image-patch method in a processor is not the same as a torch.split return-value issue." + }, + { + "left": "issue:44829", + "right": "issue:45245", + "accept": false, + "reason": "FlashAttention-3 training degeneration is unrelated to the 2^24 category limit runtime error." + }, + { + "left": "issue:38617", + "right": "issue:42757", + "accept": false, + "reason": "These are different import errors from different modules and symbols." + }, + { + "left": "issue:43064", + "right": "issue:44315", + "accept": false, + "reason": "Wrong optimizer states under FSDP2/PEFT and Liger kernel not applied with model_init are different training-stack bugs." + }, + { + "left": "issue:44410", + "right": "issue:45375", + "accept": false, + "reason": "Missing qkv/gate projections in qwen3next is unrelated to a missing vision config field." + }, + { + "left": "issue:38617", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they concern different missing names and likely different compatibility fixes." + }, + { + "left": "issue:43097", + "right": "issue:45588", + "accept": false, + "reason": "Embedding-tie API removal is unrelated to a flash_attention.py crash on s_aux=None." + }, + { + "left": "issue:44062", + "right": "issue:44117", + "accept": false, + "reason": "AddedToken constructor argument collision is not the same as tokenizer-mapping None handling." + }, + { + "left": "issue:44704", + "right": "issue:45103", + "accept": false, + "reason": "AutoProcessor kwargs forwarding and auto_docstring crashes with future annotations are unrelated." + }, + { + "left": "issue:44038", + "right": "issue:44410", + "accept": false, + "reason": "A Qwen3-VL-Moe loading bug is different from a qwen3next layer-structure bug." + }, + { + "left": "issue:42673", + "right": "issue:43577", + "accept": false, + "reason": "VRAM leak under dataloader threads is unrelated to BLIP2 dtype propagation problems." + }, + { + "left": "issue:44964", + "right": "issue:45071", + "accept": false, + "reason": "Phi-4 multimodal load failure and PretrainedConfig type-checking breakage are unrelated." + }, + { + "left": "issue:43931", + "right": "issue:44410", + "accept": false, + "reason": "Weight-shape mismatch in Qwen3-VL is not the same as missing projections in qwen3next." + }, + { + "left": "issue:44261", + "right": "issue:44485", + "accept": false, + "reason": "Missing rms_norm_eps in MLA q_a_layernorm and GLM-5 RoPE implementation are separate model-specific issues." + }, + { + "left": "issue:44964", + "right": "issue:45341", + "accept": false, + "reason": "Model loading failure and a testing_utils bug are unrelated." + }, + { + "left": "issue:43502", + "right": "issue:43992", + "accept": false, + "reason": "local_files_only still making API requests is unrelated to missing embed_tokens weights on UMT5Encoder load." + }, + { + "left": "issue:43986", + "right": "issue:44117", + "accept": false, + "reason": "Video-model crashes without torchvision and tokenizer mapping None assumptions are unrelated." + }, + { + "left": "issue:43097", + "right": "issue:43873", + "accept": false, + "reason": "Removed tie-embeddings API and offloading/quantization behavior are different concerns." + }, + { + "left": "issue:43066", + "right": "issue:43906", + "accept": false, + "reason": "Wrong tokenizer decoder type in v5 and an isolated reproduction issue are not the same bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 41, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:36010", + "issue:36683", + "issue:38617", + "issue:42617", + "issue:42757", + "issue:42886", + "issue:42907", + "issue:42994", + "issue:43012", + "issue:43064", + "issue:43066", + "issue:43097", + "issue:43208", + "issue:43299", + "issue:43352", + "issue:43381", + "issue:43475", + "issue:43502", + "issue:43526", + "issue:43531", + "issue:43550", + "issue:43618", + "issue:43761", + "issue:43824", + "issue:43825", + "issue:43867", + "issue:43901", + "issue:43937", + "issue:43986", + "issue:43992", + "issue:44038", + "issue:44220", + "issue:44230", + "issue:44368", + "issue:44380", + "issue:44483", + "issue:44484", + "issue:44514", + "issue:44655", + "issue:44683", + "issue:45216", + "issue:45290", + "issue:45341", + "issue:45372", + "issue:45381", + "issue:45561", + "issue:45588", + "issue:45753" + ], + "soft_pairs": [ + "issue:43986|issue:45290", + "issue:45341|issue:45561", + "issue:43986|issue:45372", + "issue:43097|issue:43901", + "issue:42617|issue:45341", + "issue:43986|issue:45341", + "issue:43825|issue:44655", + "issue:45341|issue:45588", + "issue:43992|issue:44484", + "issue:43012|issue:43550", + "issue:43208|issue:43550", + "issue:43012|issue:44368", + "issue:42886|issue:43381", + "issue:42907|issue:45753", + "issue:42907|issue:44655", + "issue:43531|issue:45341", + "issue:43208|issue:43526", + "issue:44483|issue:44514", + "issue:43012|issue:44683", + "issue:42757|issue:43824", + "issue:44038|issue:44230", + "issue:38617|issue:43502", + "issue:43618|issue:43761", + "issue:44483|issue:45290", + "issue:43064|issue:44380", + "issue:43299|issue:43867", + "issue:44483|issue:45381", + "issue:43901|issue:44655", + "issue:36683|issue:43475", + "issue:43867|issue:45216", + "issue:42994|issue:44655", + "issue:43352|issue:43937", + "issue:43066|issue:45381", + "issue:36010|issue:42757", + "issue:44220|issue:44655" + ], + "prepared_review_unit_hash": "0ca3a5f689b20593d0ab6ff1bd7f9c259a6555fed6fecca380470a70314df04d", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 37884, + "estimated_input_tokens": 9471, + "estimated_eval_tokens": 19198 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous Transformers bugs that only superficially overlap by subsystem or version. I found one likely true duplicate pair in CLIP vision outputs; the rest should stay separate.", + "confidence": 0.71, + "canonical_issue_reason": "issue:43761 is the best anchor because it clearly describes the CLIPVisionModel.forward regression and is broad enough to cover the paired CLIP attentions-output omission in issue:43618.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43761 is the strongest standalone report: it has a concrete symptom, a regression framing, and a specific code path/model component, making it the most useful issue to keep as the representative bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43986", + "right": "issue:45290", + "accept": false, + "reason": "Different failures: AutoProcessor loading without torchvision vs chat template tokenization on assistant tool-call messages." + }, + { + "left": "issue:45341", + "right": "issue:45561", + "accept": false, + "reason": "Both involve testing utilities, but one is a small utility bug and the other is a pytest-xdist race; not the same underlying defect." + }, + { + "left": "issue:43986", + "right": "issue:45372", + "accept": false, + "reason": "Unrelated crashes in different areas: video processor loading vs mistral_common ReasoningEffort import failure." + }, + { + "left": "issue:43097", + "right": "issue:43901", + "accept": false, + "reason": "One is a removed API without notice; the other is outdated docs for TextClassificationPipeline. Similar theme, different bug." + }, + { + "left": "issue:42617", + "right": "issue:45341", + "accept": false, + "reason": "3d_parallel.py runtime failure is unrelated to a testing_utils issue." + }, + { + "left": "issue:43986", + "right": "issue:45341", + "accept": false, + "reason": "Different components and symptoms; no shared code-path evidence." + }, + { + "left": "issue:43825", + "right": "issue:44655", + "accept": false, + "reason": "Pipeline error-message wording vs saving Pipeline objects are separate issues." + }, + { + "left": "issue:45341", + "right": "issue:45588", + "accept": false, + "reason": "Testing utils bug vs flash_attention crash on sink-less models are different code paths." + }, + { + "left": "issue:43992", + "right": "issue:44484", + "accept": false, + "reason": "Model weight-loading omission vs save_pretrained shard-size question; not the same bug." + }, + { + "left": "issue:43012", + "right": "issue:43550", + "accept": false, + "reason": "A precision warning during compile is not the same as a torch.compile SDPA failure." + }, + { + "left": "issue:43208", + "right": "issue:43550", + "accept": false, + "reason": "xLSTM training bugs and Bamba SDPA compile failure are unrelated." + }, + { + "left": "issue:43012", + "right": "issue:44368", + "accept": false, + "reason": "Both are warnings, but they concern different behaviors and different configs." + }, + { + "left": "issue:42886", + "right": "issue:43381", + "accept": false, + "reason": "Offline-cache loading and gradient checkpointing in eval mode are separate regressions." + }, + { + "left": "issue:42907", + "right": "issue:45753", + "accept": false, + "reason": "Saving dequantized models vs an infinite-loop generation bug are unrelated." + }, + { + "left": "issue:42907", + "right": "issue:44655", + "accept": false, + "reason": "Both touch saving, but one is quantized model serialization and the other is Pipeline.save_pretrained." + }, + { + "left": "issue:43531", + "right": "issue:45341", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior is unrelated to a testing_utils bug." + }, + { + "left": "issue:43208", + "right": "issue:43526", + "accept": false, + "reason": "xLSTM training issues and BeitImageProcessorFast reduce_labels behavior do not share a concrete bug." + }, + { + "left": "issue:44483", + "right": "issue:44514", + "accept": false, + "reason": "Chat/completions API rejection and apply_chat_template batching crash are different layers and different failures." + }, + { + "left": "issue:43012", + "right": "issue:44683", + "accept": false, + "reason": "Precision warning during bfloat16 compile is not the same as flex_attention compile failure on torch>=2.9." + }, + { + "left": "issue:42757", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but for different symbols, different packages, and different breakages." + }, + { + "left": "issue:44038", + "right": "issue:44230", + "accept": false, + "reason": "Related model family, but one is a generic Qwen3-VL-Moe regression and the other is fp8 inference support; not the same issue." + }, + { + "left": "issue:38617", + "right": "issue:43502", + "accept": false, + "reason": "Importing layer_type_validation and ignoring local_files_only are unrelated bugs." + }, + { + "left": "issue:43618", + "right": "issue:43761", + "accept": true, + "reason": "Both describe the same CLIP vision output regression in the forward/output path: missing attentions and missing hidden_states under v5-style behavior." + }, + { + "left": "issue:44483", + "right": "issue:45290", + "accept": false, + "reason": "An API endpoint compatibility issue is not the same as a chat-template tokenization crash." + }, + { + "left": "issue:43064", + "right": "issue:44380", + "accept": false, + "reason": "FSDP2/PEFT optimizer-state corruption and GPT2 attention-scaling being ignored are different training/runtime bugs." + }, + { + "left": "issue:43299", + "right": "issue:43867", + "accept": false, + "reason": "Qwen3VL MoE loading breakage and sorted state_dict load errors are distinct model-loading problems." + }, + { + "left": "issue:44483", + "right": "issue:45381", + "accept": false, + "reason": "Chat completions API behavior and Qwen2.5-VL video vision_position_ids are unrelated." + }, + { + "left": "issue:43901", + "right": "issue:44655", + "accept": false, + "reason": "Documentation drift for a pipeline flag is not the same as inability to save Pipeline objects." + }, + { + "left": "issue:36683", + "right": "issue:43475", + "accept": false, + "reason": "Gemma3Config missing vocab_size and SAM3 output missing fpn_position_embeddings are unrelated attribute errors." + }, + { + "left": "issue:43867", + "right": "issue:45216", + "accept": false, + "reason": "Both concern loading/saving state, but one is a load error and the other is a save_pretrained regression for Qwen3.5 checkpoints." + }, + { + "left": "issue:42994", + "right": "issue:44655", + "accept": false, + "reason": "Quantized model saving failure and Pipeline.save_pretrained failure are different serialization paths." + }, + { + "left": "issue:43352", + "right": "issue:43937", + "accept": false, + "reason": "Flash Attention support error for Nemotron differs from GLM generation-config validation failure." + }, + { + "left": "issue:43066", + "right": "issue:45381", + "accept": false, + "reason": "Tokenizer decoder type in v5 and Qwen2.5-VL video position IDs are unrelated regressions." + }, + { + "left": "issue:36010", + "right": "issue:42757", + "accept": false, + "reason": "Different missing imports from different modules; no common code-path bug." + }, + { + "left": "issue:44220", + "right": "issue:44655", + "accept": false, + "reason": "Audio feature extraction and Pipeline.save_pretrained are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and appropriately conservative overall. The only accepted duplicate pair (issue:43618 \u2194 issue:43761) is supported by the packet as the same CLIP vision output regression, and the other soft-edge verdicts stay on the cautious side without overclaiming shared causes." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 42, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:28282", + "issue:33453", + "issue:36296", + "issue:36683", + "issue:41669", + "issue:42890", + "issue:43064", + "issue:43097", + "issue:43299", + "issue:43352", + "issue:43441", + "issue:43525", + "issue:43575", + "issue:43618", + "issue:43756", + "issue:43867", + "issue:43881", + "issue:43901", + "issue:43906", + "issue:43976", + "issue:43992", + "issue:44077", + "issue:44117", + "issue:44242", + "issue:44246", + "issue:44263", + "issue:44355", + "issue:44360", + "issue:44380", + "issue:44484", + "issue:44485", + "issue:44492", + "issue:44655", + "issue:44855", + "issue:44857", + "issue:44938", + "issue:44964", + "issue:44977", + "issue:45216", + "issue:45310", + "issue:45357", + "issue:45375", + "issue:45440", + "issue:45478", + "issue:45479", + "issue:45561", + "issue:45588", + "issue:45753" + ], + "soft_pairs": [ + "issue:43756|issue:44485", + "issue:44263|issue:45440", + "issue:36683|issue:43525", + "issue:43618|issue:45216", + "issue:43906|issue:45753", + "issue:36683|issue:44242", + "issue:43299|issue:45753", + "issue:36683|issue:43881", + "issue:43352|issue:43976", + "issue:43441|issue:44380", + "issue:43352|issue:44380", + "issue:43992|issue:45310", + "issue:44380|issue:45479", + "issue:43867|issue:45375", + "issue:41669|issue:44246", + "issue:43064|issue:44492", + "issue:43352|issue:45478", + "issue:43618|issue:44380", + "issue:44355|issue:44855", + "issue:36683|issue:44117", + "issue:42890|issue:45561", + "issue:43097|issue:43352", + "issue:43867|issue:45357", + "issue:43976|issue:44938", + "issue:43992|issue:45478", + "issue:44360|issue:45440", + "issue:43064|issue:44655", + "issue:36296|issue:43575", + "issue:44077|issue:44380", + "issue:43867|issue:44484", + "issue:28282|issue:33453", + "issue:43618|issue:44857", + "issue:43618|issue:45588", + "issue:43618|issue:44977", + "issue:43901|issue:44938", + "issue:44655|issue:44938", + "issue:42890|issue:44964" + ], + "prepared_review_unit_hash": "7413898c4948b771711287688f34d6e1e1af7664dc13b0d950935fb75eb255df", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 37, + "serialized_chars": 38412, + "estimated_input_tokens": 9603, + "estimated_eval_tokens": 19462 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a noisy cluster of mostly unrelated Transformers regression reports (model config/loading, flash-attention, save/load, Python-version issues). I only found one convincing duplicate-like pair: import-time slowdown reports in issue:41669 and issue:44246. All other soft pairs look like different bugs that merely share a broad subsystem or symptom class.", + "confidence": 0.77, + "canonical_issue_reason": "issue:41669 is the best canonical issue because it most concretely describes the underlying import-time regression and cleanly subsumes issue:44246's vaguer 'import transformers takes long sometimes' report.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:41669 is the strongest representative of the only clear duplicate-like topic in this set and is specific enough to serve as the anchor for deduping the import slowdown reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43756", + "right": "issue:44485", + "accept": false, + "reason": "Different model-specific bugs: Smollm3 RoPE-layer count vs GLM-5 RoPE implementation." + }, + { + "left": "issue:44263", + "right": "issue:45440", + "accept": false, + "reason": "Unrelated failures: a torch.split DSA indexer issue vs DeepseekV3MoE divergence from remote implementation." + }, + { + "left": "issue:36683", + "right": "issue:43525", + "accept": false, + "reason": "Similar shape (missing config attribute) but different models and missing fields; not the same bug." + }, + { + "left": "issue:43618", + "right": "issue:45216", + "accept": false, + "reason": "CLIPOutput attentions regression is unrelated to Qwen3.5 save_pretrained checkpoint corruption." + }, + { + "left": "issue:43906", + "right": "issue:45753", + "accept": false, + "reason": "A reproduction of an old issue vs a Qwen3_5 image-specific generation loop; no clear shared code-path." + }, + { + "left": "issue:36683", + "right": "issue:44242", + "accept": false, + "reason": "Both are model/config related, but one is missing vocab_size and the other is load-balancing loss logic." + }, + { + "left": "issue:43299", + "right": "issue:45753", + "accept": false, + "reason": "Qwen3VL MoE loading breakage is unrelated to a Qwen3_5 infinite-loop generation bug." + }, + { + "left": "issue:36683", + "right": "issue:43881", + "accept": false, + "reason": "Different model-loading failures with different missing attributes/behaviors." + }, + { + "left": "issue:43352", + "right": "issue:43976", + "accept": false, + "reason": "FlashAttention support error for a model vs Python version compatibility regression; not the same underlying issue." + }, + { + "left": "issue:43441", + "right": "issue:44380", + "accept": false, + "reason": "Both involve attention backends, but one is a specific model unsupported by FlashAttention 2.0 while the other is GPT-2 scaling being ignored." + }, + { + "left": "issue:43352", + "right": "issue:44380", + "accept": false, + "reason": "Different attention-backend bugs affecting different models and code paths." + }, + { + "left": "issue:43992", + "right": "issue:45310", + "accept": false, + "reason": "Different loading failures: missing embed_tokens.weight for UMT5Encoder vs Qwen3.5 MoE from_pretrained error." + }, + { + "left": "issue:44380", + "right": "issue:45479", + "accept": false, + "reason": "Attention-backend scaling bug vs a classification-loss degenerate-zero-loss bug." + }, + { + "left": "issue:43867", + "right": "issue:45375", + "accept": false, + "reason": "Both are loading-related, but one is state_dict sorting and the other is a missing config field silently dropped by strict parsing." + }, + { + "left": "issue:41669", + "right": "issue:44246", + "accept": true, + "reason": "Both report the same underlying import-time slowdown in transformers; one is the more specific root-cause report and the other is the user-facing symptom." + }, + { + "left": "issue:43064", + "right": "issue:44492", + "accept": false, + "reason": "FSDP/PEFT optimizer-state corruption is unrelated to a cache-strategy typo." + }, + { + "left": "issue:43352", + "right": "issue:45478", + "accept": false, + "reason": "Similar FlashAttention/version wording, but the affected models and failure modes are different." + }, + { + "left": "issue:43618", + "right": "issue:44380", + "accept": false, + "reason": "Different attention-related regressions with different symptoms and models." + }, + { + "left": "issue:44355", + "right": "issue:44855", + "accept": false, + "reason": "Compiled Python file import errors are not the same as a torch.jit.script parsing/IndentationError regression." + }, + { + "left": "issue:36683", + "right": "issue:44117", + "accept": false, + "reason": "Both involve missing metadata, but one is a config attribute bug and the other is tokenizer mapping returning None." + }, + { + "left": "issue:42890", + "right": "issue:45561", + "accept": false, + "reason": "Test flakiness from missing set_seed is unrelated to xdist races on captured_info.txt." + }, + { + "left": "issue:43097", + "right": "issue:43352", + "accept": false, + "reason": "Removed tie_embeddings behavior and FlashAttention support are unrelated." + }, + { + "left": "issue:43867", + "right": "issue:45357", + "accept": false, + "reason": "State-dict sorting load failure vs incorrect visual encoder keys in save_pretrained; different serialization paths." + }, + { + "left": "issue:43976", + "right": "issue:44938", + "accept": false, + "reason": "Both are Python-version compatibility reports, but they concern different versions and failure mechanisms." + }, + { + "left": "issue:43992", + "right": "issue:45478", + "accept": false, + "reason": "Different Qwen3.5 loading bugs with different missing/mis-generated weights." + }, + { + "left": "issue:44360", + "right": "issue:45440", + "accept": false, + "reason": "DSA indexer ReLU discussion is unrelated to DeepseekV3MoE implementation divergence." + }, + { + "left": "issue:43064", + "right": "issue:44655", + "accept": false, + "reason": "Training optimizer-state corruption is unrelated to saving Pipeline objects." + }, + { + "left": "issue:36296", + "right": "issue:43575", + "accept": false, + "reason": "Tensor-parallel training bug is not the same as Qwen2-57B-A14B-Instruct OOM on load." + }, + { + "left": "issue:44077", + "right": "issue:44380", + "accept": false, + "reason": "Patchtsmixer optional post_init policy change is unrelated to GPT-2 attention scaling under SDPA/FlashAttention." + }, + { + "left": "issue:43867", + "right": "issue:44484", + "accept": false, + "reason": "A load-model error from sorted state_dict is unrelated to the 50GB max_shard_size question." + }, + { + "left": "issue:28282", + "right": "issue:33453", + "accept": false, + "reason": "Generic import/config errors, but one is an AutoModel PyTorch missing-dependency ImportError and the other is tokenizer loading regression." + }, + { + "left": "issue:43618", + "right": "issue:44857", + "accept": false, + "reason": "CLIP attentions assignment regression is unrelated to LwDetrImageLoss AMP/CUDA crash." + }, + { + "left": "issue:43618", + "right": "issue:45588", + "accept": false, + "reason": "Missing CLIPOutput attentions and flash_attention.py s_aux=None crash are different bugs." + }, + { + "left": "issue:43618", + "right": "issue:44977", + "accept": false, + "reason": "CLIP attentions regression is unrelated to Qwen3.5 flash-attention generation failure." + }, + { + "left": "issue:43901", + "right": "issue:44938", + "accept": false, + "reason": "Docs mismatch about return_all_scores is unrelated to Python 3.14 import/load failure." + }, + { + "left": "issue:44655", + "right": "issue:44938", + "accept": false, + "reason": "Saving Pipeline objects and loading transformers on Python 3.14 are unrelated." + }, + { + "left": "issue:42890", + "right": "issue:44964", + "accept": false, + "reason": "Test seeding flakiness is unrelated to Phi-4 multimodal loading failure." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The 41669/44246 pair is plausibly the only clearly duplicate-like import-slowdown cluster, and the rest of the soft-edge rejections are cautious rather than overstated." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 43, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:30333", + "issue:30990", + "issue:31795", + "issue:33453", + "issue:34689", + "issue:36296", + "issue:38617", + "issue:41628", + "issue:42548", + "issue:42757", + "issue:42831", + "issue:42890", + "issue:43010", + "issue:43299", + "issue:43502", + "issue:43576", + "issue:43723", + "issue:43824", + "issue:43827", + "issue:43874", + "issue:43976", + "issue:44062", + "issue:44075", + "issue:44112", + "issue:44222", + "issue:44230", + "issue:44315", + "issue:44351", + "issue:44355", + "issue:44360", + "issue:44368", + "issue:44387", + "issue:44485", + "issue:44556", + "issue:44655", + "issue:44749", + "issue:44938", + "issue:44961", + "issue:45071", + "issue:45125", + "issue:45278", + "issue:45341", + "issue:45375", + "issue:45468", + "issue:45478", + "issue:45600", + "issue:45656", + "issue:45850" + ], + "soft_pairs": [ + "issue:42890|issue:44112", + "issue:42890|issue:45600", + "issue:43502|issue:44062", + "issue:43010|issue:44075", + "issue:42890|issue:45341", + "issue:44222|issue:45125", + "issue:42548|issue:44351", + "issue:43502|issue:44485", + "issue:44355|issue:44749", + "issue:45125|issue:45478", + "issue:30990|issue:31795", + "issue:45125|issue:45375", + "issue:44387|issue:44556", + "issue:30333|issue:33453", + "issue:43010|issue:45278", + "issue:43723|issue:44556", + "issue:43502|issue:44222", + "issue:43010|issue:44368", + "issue:44315|issue:44360", + "issue:43010|issue:45850", + "issue:43502|issue:44961", + "issue:42831|issue:43874", + "issue:30333|issue:36296", + "issue:44355|issue:44655", + "issue:44222|issue:44230", + "issue:44355|issue:45071", + "issue:43827|issue:44961", + "issue:33453|issue:34689", + "issue:42757|issue:43576", + "issue:43299|issue:44556", + "issue:43010|issue:45656", + "issue:33453|issue:36296", + "issue:43576|issue:45071", + "issue:43576|issue:43824", + "issue:44961|issue:45468", + "issue:31795|issue:38617", + "issue:41628|issue:42548", + "issue:38617|issue:43576", + "issue:43010|issue:43976", + "issue:43576|issue:44938" + ], + "prepared_review_unit_hash": "657b76dce04fcace7e111f8fb932c0e3b9715a0cb0a2754f7230cc0a059087a6", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 40, + "serialized_chars": 39096, + "estimated_input_tokens": 9774, + "estimated_eval_tokens": 19804 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose collection of Transformers v5 regression reports, especially import/loading breakages, but many items are about different symbols, models, or unrelated test/feature bugs. I would not merge the soft pairs; they are too diverse to be true duplicates.", + "confidence": 0.83, + "canonical_issue_reason": "issue:45278 is the best umbrella issue: it broadly captures the recurring v5 import/load regressions that overlap with several other reports, without being as vague as the most generic complaints.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the strongest representative for triage because it summarizes the dominant cluster theme (post-upgrade import failures) and is specific enough to anchor related reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42890", + "right": "issue:44112", + "accept": false, + "reason": "Both mention CI/test flakiness, but one is a seed-related SamHQ integration issue and the other is a GraniteSpeech stale device override test; different failing paths." + }, + { + "left": "issue:42890", + "right": "issue:45600", + "accept": false, + "reason": "Both are test/CI-related, but the bugs are unrelated: missing set_seed in SamHQ tests vs broken auto_mappings references for removed configs." + }, + { + "left": "issue:43502", + "right": "issue:44062", + "accept": false, + "reason": "One is an offline/network regression in from_pretrained; the other is an AddedToken constructor TypeError. No shared concrete code path." + }, + { + "left": "issue:43010", + "right": "issue:44075", + "accept": false, + "reason": "Cache update no_grad and SGD optimizer args are separate runtime behaviors in different subsystems." + }, + { + "left": "issue:42890", + "right": "issue:45341", + "accept": false, + "reason": "Both touch testing utilities, but one is a model test needing seed setup and the other is a generic bug in testing_utils.py; too little evidence of the same defect." + }, + { + "left": "issue:44222", + "right": "issue:45125", + "accept": false, + "reason": "Both involve Qwen3.5/MoE/FP8, but one is FP8 save_pretrained and the other is tensor-parallel _tp_plan loading; not the same bug." + }, + { + "left": "issue:42548", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but for different symbols (PreTrainedModel vs HybridCache) and likely different missing-export paths." + }, + { + "left": "issue:43502", + "right": "issue:44485", + "accept": false, + "reason": "local_files_only network access and GLM-5 RoPE implementation are unrelated." + }, + { + "left": "issue:44355", + "right": "issue:44749", + "accept": false, + "reason": "Compiled Python file errors and a slowdown after upgrading to v5.3 are not the same underlying issue." + }, + { + "left": "issue:45125", + "right": "issue:45478", + "accept": false, + "reason": "Same model family, but one reports missing _tp_plan for tensor parallelism while the other is a broader from_pretrained error; insufficient evidence for one concrete fix." + }, + { + "left": "issue:30990", + "right": "issue:31795", + "accept": false, + "reason": "Loading gets stuck vs documentation confusion; these are not duplicate bugs." + }, + { + "left": "issue:45125", + "right": "issue:45375", + "accept": false, + "reason": "MoE tensor-parallel loading and a missing config field silently dropped by @strict are unrelated." + }, + { + "left": "issue:44387", + "right": "issue:44556", + "accept": false, + "reason": "Both are version-regression complaints, but one is int4 memory/OOM behavior and the other is checkpoint reload compatibility." + }, + { + "left": "issue:30333", + "right": "issue:33453", + "accept": false, + "reason": "MLflow reporting on training failure and tokenizer loading regression are unrelated." + }, + { + "left": "issue:43010", + "right": "issue:45278", + "accept": false, + "reason": "No overlap: no_grad cache update vs broad import errors after upgrade." + }, + { + "left": "issue:43723", + "right": "issue:44556", + "accept": false, + "reason": "Tokenizer loading and checkpoint reload after upgrade are both load-related, but not the same bug or code path." + }, + { + "left": "issue:43502", + "right": "issue:44222", + "accept": false, + "reason": "local_files_only network leakage and FP8 MoE save_pretrained are unrelated." + }, + { + "left": "issue:43010", + "right": "issue:44368", + "accept": false, + "reason": "Cache update semantics and a tie_word_embeddings warning in Qwen3.5 LoRA are unrelated." + }, + { + "left": "issue:44315", + "right": "issue:44360", + "accept": false, + "reason": "Liger kernel application and a DSA indexer ReLU discussion have nothing in common." + }, + { + "left": "issue:43010", + "right": "issue:45850", + "accept": false, + "reason": "A specific no_grad cache bug and a generic 'too many models break' complaint are not duplicates." + }, + { + "left": "issue:43502", + "right": "issue:44961", + "accept": false, + "reason": "No relation between offline file loading and the racoon issue." + }, + { + "left": "issue:42831", + "right": "issue:43874", + "accept": false, + "reason": "FineGrainedFP8 accuracy and a missing image-patch-count method in GLM46V are different model bugs." + }, + { + "left": "issue:30333", + "right": "issue:36296", + "accept": false, + "reason": "MLflow failure reporting and tensor-parallel training bug are unrelated." + }, + { + "left": "issue:44355", + "right": "issue:44655", + "accept": false, + "reason": "Compiled Python file handling and save_pretrained for Pipeline objects are different features." + }, + { + "left": "issue:44222", + "right": "issue:44230", + "accept": false, + "reason": "FP8 save_pretrained for MoE and FP8 inference support for Qwen3-VL/Qwen3.5 are related thematically but not the same concrete change." + }, + { + "left": "issue:44355", + "right": "issue:45071", + "accept": false, + "reason": "Compiled-file errors and PretrainedConfig type-checking are separate regressions." + }, + { + "left": "issue:43827", + "right": "issue:44961", + "accept": false, + "reason": "Docs still referencing pipeline() and the racoon issue are unrelated." + }, + { + "left": "issue:33453", + "right": "issue:34689", + "accept": false, + "reason": "Tokenizer loading regression and Llama 3.2 vision model loading breakage are distinct failures." + }, + { + "left": "issue:42757", + "right": "issue:43576", + "accept": false, + "reason": "Missing is_offline_mode import and broken transformers env command are different v5 breakages." + }, + { + "left": "issue:43299", + "right": "issue:44556", + "accept": false, + "reason": "Qwen3VL MoE loading and checkpoint reload incompatibility are both loading issues, but not the same bug." + }, + { + "left": "issue:43010", + "right": "issue:45656", + "accept": false, + "reason": "No_grad cache updates and optimizer step called twice with deepspeed are separate optimizer/runtime behaviors." + }, + { + "left": "issue:33453", + "right": "issue:36296", + "accept": false, + "reason": "Tokenizer loading regression and tensor-parallel training bug are unrelated." + }, + { + "left": "issue:43576", + "right": "issue:45071", + "accept": false, + "reason": "Broken env command and PretrainedConfig type checking are not the same issue." + }, + { + "left": "issue:43576", + "right": "issue:43824", + "accept": false, + "reason": "CLI/env command breakage and missing Qwen2.5-VL export are distinct import/command problems." + }, + { + "left": "issue:44961", + "right": "issue:45468", + "accept": false, + "reason": "No meaningful overlap between the racoon issue and Gemma-4 audio positional encoding." + }, + { + "left": "issue:31795", + "right": "issue:38617", + "accept": false, + "reason": "Documentation confusion and missing layer_type_validation export are different kinds of issues." + }, + { + "left": "issue:41628", + "right": "issue:42548", + "accept": false, + "reason": "Both are import errors, but for different missing names and likely different modules/exports; too broad to merge." + }, + { + "left": "issue:38617", + "right": "issue:43576", + "accept": false, + "reason": "An import name error and a broken env command are unrelated." + }, + { + "left": "issue:43010", + "right": "issue:43976", + "accept": false, + "reason": "Cache update no_grad and Python 3.9/3.10 compatibility are separate bugs." + }, + { + "left": "issue:43576", + "right": "issue:44938", + "accept": false, + "reason": "Broken env command and Python 3.14 load failure are both compatibility issues, but not the same defect." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 44, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:36296", + "issue:41628", + "issue:41669", + "issue:42890", + "issue:43352", + "issue:43408", + "issue:43525", + "issue:43526", + "issue:43572", + "issue:43576", + "issue:43673", + "issue:43827", + "issue:43874", + "issue:43901", + "issue:43976", + "issue:44038", + "issue:44077", + "issue:44246", + "issue:44261", + "issue:44273", + "issue:44295", + "issue:44297", + "issue:44315", + "issue:44351", + "issue:44355", + "issue:44360", + "issue:44373", + "issue:44393", + "issue:44492", + "issue:44556", + "issue:44623", + "issue:44704", + "issue:44829", + "issue:44861", + "issue:44908", + "issue:44961", + "issue:45003", + "issue:45071", + "issue:45092", + "issue:45125", + "issue:45230", + "issue:45278", + "issue:45341", + "issue:45362", + "issue:45478", + "issue:45542", + "issue:45753", + "issue:45850" + ], + "soft_pairs": [ + "issue:44908|issue:44961", + "issue:36296|issue:43901", + "issue:41628|issue:44355", + "issue:36296|issue:43526", + "issue:44261|issue:44360", + "issue:44273|issue:45003", + "issue:44273|issue:45542", + "issue:44623|issue:45230", + "issue:44393|issue:45753", + "issue:43576|issue:44556", + "issue:44038|issue:45125", + "issue:44393|issue:45071", + "issue:43525|issue:45850", + "issue:41669|issue:43673", + "issue:44556|issue:45092", + "issue:44373|issue:44961", + "issue:43572|issue:45850", + "issue:45362|issue:45850", + "issue:44704|issue:44908", + "issue:42890|issue:44908", + "issue:44246|issue:44273", + "issue:43408|issue:44315", + "issue:43976|issue:44393", + "issue:43352|issue:44393", + "issue:43576|issue:44908", + "issue:44908|issue:45071", + "issue:44297|issue:45230", + "issue:41669|issue:44351", + "issue:44295|issue:45230", + "issue:43673|issue:44492", + "issue:41669|issue:45341", + "issue:45278|issue:45850", + "issue:44393|issue:45478", + "issue:43874|issue:45850", + "issue:44315|issue:44829", + "issue:41669|issue:44492", + "issue:43827|issue:44908", + "issue:44077|issue:44861" + ], + "prepared_review_unit_hash": "2082ef0bb15b724250fda3cd5be71d8f6fc34d4fa67aedd2d90e9c7b0845ffda", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 38, + "serialized_chars": 38444, + "estimated_input_tokens": 9611, + "estimated_eval_tokens": 19478 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a mixed bag of unrelated Transformers issues, with one clear duplicate-like pair around import/lazy-loading slowness. Most other pairs share only broad theme words (v5 regressions, import problems, specific model bugs) rather than the same concrete defect.", + "confidence": 0.73, + "canonical_issue_reason": "issue:44246 is the best anchor for the only clear duplicate subgroup: it is the broadest and most central report about `import transformers` being slow, and it aligns closely with the more specific lazy-loading complaint.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44246 is the strongest representative issue in this set because it captures the import/lazy-loading regression at a higher level and has the clearest overlap with the only accepted soft duplicate.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44908", + "right": "issue:44961", + "accept": false, + "reason": "Different topics entirely: scheduler kwargs vs an unrelated one-word issue." + }, + { + "left": "issue:36296", + "right": "issue:43901", + "accept": false, + "reason": "Tensor-parallel training bug and a docs mismatch for text classification are unrelated." + }, + { + "left": "issue:41628", + "right": "issue:44355", + "accept": false, + "reason": "Both are errors, but one is a missing top-level import and the other is a runtime problem with compiled Python files." + }, + { + "left": "issue:36296", + "right": "issue:43526", + "accept": false, + "reason": "Tensor parallelism vs a BeitImageProcessorFast `reduce_labels` bug; no shared code-path." + }, + { + "left": "issue:44261", + "right": "issue:44360", + "accept": false, + "reason": "Different model families and different failures; not the same underlying bug." + }, + { + "left": "issue:44273", + "right": "issue:45003", + "accept": false, + "reason": "Both touch loading behavior, but one is lazy loading not working and the other is unsafe `sys.modules` access; too different to merge." + }, + { + "left": "issue:44273", + "right": "issue:45542", + "accept": false, + "reason": "Lazy loading/import behavior is unrelated to the TensorBoard-without-TF backend error." + }, + { + "left": "issue:44623", + "right": "issue:45230", + "accept": false, + "reason": "A specific `save_pretrained` file-missing bug and a generic bug report are not duplicates." + }, + { + "left": "issue:44393", + "right": "issue:45753", + "accept": false, + "reason": "Both involve Qwen3-VL/Qwen3_5 vision behavior, but the symptoms and concrete failures differ too much." + }, + { + "left": "issue:43576", + "right": "issue:44556", + "accept": false, + "reason": "Both are v5 regressions, but one is the `transformers env` command and the other is checkpoint reload incompatibility." + }, + { + "left": "issue:44038", + "right": "issue:45125", + "accept": false, + "reason": "General Qwen3-VL-Moe bug vs missing `_tp_plan` for tensor parallelism; different problems." + }, + { + "left": "issue:44393", + "right": "issue:45071", + "accept": false, + "reason": "Vision hallucination/bounding-box failure is not the same as a `PretrainedConfig` type-checking regression." + }, + { + "left": "issue:43525", + "right": "issue:45850", + "accept": false, + "reason": "Specific missing config attribute in Llama4 vs a broad complaint that v5 breaks many models." + }, + { + "left": "issue:41669", + "right": "issue:43673", + "accept": false, + "reason": "Import-star slowdown and missing generation cache in chunked prefill are different bugs." + }, + { + "left": "issue:44556", + "right": "issue:45092", + "accept": false, + "reason": "Both are upgrade compatibility issues, but they affect different model families and failure mechanisms." + }, + { + "left": "issue:44373", + "right": "issue:44961", + "accept": false, + "reason": "A docstring bug and an unrelated issue title are not duplicates." + }, + { + "left": "issue:43572", + "right": "issue:45850", + "accept": false, + "reason": "Missing `pad_token_idx` in StableLmConfig is too specific to be the same as a vague global v5 complaint." + }, + { + "left": "issue:45362", + "right": "issue:45850", + "accept": false, + "reason": "Qwen3.5 chat crash is a concrete model bug, not the same as a broad regression report." + }, + { + "left": "issue:45278", + "right": "issue:45850", + "accept": false, + "reason": "Both are about v5 breakage, but one is specifically many import errors after upgrade and the other is a generic complaint; not the same bug." + }, + { + "left": "issue:43874", + "right": "issue:45850", + "accept": false, + "reason": "Specific GLM46V multimodal token bug vs a broad v5 breakage report." + }, + { + "left": "issue:44704", + "right": "issue:44908", + "accept": false, + "reason": "`AutoProcessor.from_pretrained` kwargs forwarding and scheduler kwargs handling are unrelated." + }, + { + "left": "issue:42890", + "right": "issue:44908", + "accept": false, + "reason": "Seed-related integration flakiness and scheduler kwargs are different problems." + }, + { + "left": "issue:44246", + "right": "issue:44273", + "accept": true, + "reason": "These describe the same import/lazy-loading regression: one says imports are intermittently slow, the other says lazy loading is not working properly." + }, + { + "left": "issue:43408", + "right": "issue:44315", + "accept": false, + "reason": "Model-type warning and Liger Kernel application are different code paths and symptoms." + }, + { + "left": "issue:43976", + "right": "issue:44393", + "accept": false, + "reason": "Python version compatibility is unrelated to the Qwen3-VL bounding-box issue." + }, + { + "left": "issue:43352", + "right": "issue:44393", + "accept": false, + "reason": "Flash Attention support limitations are a different issue from the Qwen3-VL output problem." + }, + { + "left": "issue:43576", + "right": "issue:44908", + "accept": false, + "reason": "The env command regression is unrelated to scheduler kwargs." + }, + { + "left": "issue:44297", + "right": "issue:45230", + "accept": false, + "reason": "Tokenizer class metadata mismatch in `save_pretrained` is not the same as a generic bug report." + }, + { + "left": "issue:41669", + "right": "issue:44351", + "accept": false, + "reason": "Import-star performance cost and missing `HybridCache` import are distinct import-related bugs." + }, + { + "left": "issue:44295", + "right": "issue:45230", + "accept": false, + "reason": "Position-ids buffer reading error and a generic bug report do not match." + }, + { + "left": "issue:43673", + "right": "issue:44492", + "accept": false, + "reason": "Chunked-prefill cache regression and a typo in cache strategies are unrelated." + }, + { + "left": "issue:41669", + "right": "issue:45341", + "accept": false, + "reason": "Import performance regression and a testing utility bug are not the same defect." + }, + { + "left": "issue:45278", + "right": "issue:45850", + "accept": false, + "reason": "Duplicate theme only at a very high level; the specific failure mode differs and should not be merged." + }, + { + "left": "issue:44393", + "right": "issue:45478", + "accept": false, + "reason": "Different Qwen3.5/Qwen3-VL failure modes with no evidence of a shared code-path." + }, + { + "left": "issue:43874", + "right": "issue:45850", + "accept": false, + "reason": "Specific GLM46V multimodal-token bug vs a generic v5 complaint." + }, + { + "left": "issue:44315", + "right": "issue:44829", + "accept": false, + "reason": "Liger Kernel not applied during model init and Flash Attention 3 degenerate training are different issues." + }, + { + "left": "issue:41669", + "right": "issue:44492", + "accept": false, + "reason": "Import slowdown vs a typo in cache strategy text are unrelated." + }, + { + "left": "issue:43827", + "right": "issue:44908", + "accept": false, + "reason": "Docs for removed pipeline APIs and scheduler kwargs handling are unrelated." + }, + { + "left": "issue:44077", + "right": "issue:44861", + "accept": false, + "reason": "PatchTSMixer post-init policy and tied-weight key handling are different model internals, not the same bug." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary stays grounded in the packet and is appropriately conservative: it identifies only one plausible duplicate-like pair (44246/44273) and treats the rest as largely unrelated. I don\u2019t see evidence of overstatement in the soft-edge verdicts." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 45, + "review_unit_count": 45, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-391", + "nodes": [ + "issue:30333", + "issue:31515", + "issue:31795", + "issue:33453", + "issue:34689", + "issue:36296", + "issue:37428", + "issue:38617", + "issue:41669", + "issue:42371", + "issue:42913", + "issue:43066", + "issue:43298", + "issue:43502", + "issue:43519", + "issue:43525", + "issue:43572", + "issue:43576", + "issue:43606", + "issue:43704", + "issue:44230", + "issue:44261", + "issue:44263", + "issue:44273", + "issue:44315", + "issue:44360", + "issue:44485", + "issue:44492", + "issue:44623", + "issue:44655", + "issue:44861", + "issue:44908", + "issue:44998", + "issue:45092", + "issue:45230", + "issue:45290", + "issue:45310", + "issue:45341", + "issue:45468", + "issue:45850" + ], + "soft_pairs": [ + "issue:31795|issue:33453", + "issue:41669|issue:44908", + "issue:42371|issue:44492", + "issue:43576|issue:44273", + "issue:44360|issue:45468", + "issue:44655|issue:44861", + "issue:44492|issue:45850", + "issue:44273|issue:44908", + "issue:45230|issue:45310", + "issue:43519|issue:44485", + "issue:41669|issue:43704", + "issue:43519|issue:44908", + "issue:31795|issue:34689", + "issue:31795|issue:37428", + "issue:44315|issue:45092", + "issue:43704|issue:44485", + "issue:44485|issue:45468", + "issue:42371|issue:43704", + "issue:43502|issue:43519", + "issue:43519|issue:44623", + "issue:43519|issue:43704", + "issue:44263|issue:45468", + "issue:43606|issue:45468", + "issue:44261|issue:45468", + "issue:43704|issue:45341", + "issue:42913|issue:43066", + "issue:44908|issue:44998", + "issue:44861|issue:45290", + "issue:43704|issue:44998", + "issue:41669|issue:44998", + "issue:43525|issue:44861", + "issue:43066|issue:44998", + "issue:43572|issue:44861", + "issue:31515|issue:38617", + "issue:44360|issue:44998", + "issue:44230|issue:44623", + "issue:30333|issue:31515", + "issue:31515|issue:36296", + "issue:31515|issue:31795", + "issue:44230|issue:45310", + "issue:31515|issue:33453", + "issue:43298|issue:44230" + ], + "prepared_review_unit_hash": "74e0136bdbba713ac92fdbe20e7a436c69b67af734cf1a3c48b6088bda4b7351", + "component_budget": { + "node_count": 391, + "item_count": 391, + "soft_pair_count": 1476, + "serialized_chars": 759370, + "estimated_input_tokens": 189843, + "estimated_eval_tokens": 379942 + }, + "budget": { + "node_count": 40, + "item_count": 40, + "soft_pair_count": 42, + "serialized_chars": 34927, + "estimated_input_tokens": 8732, + "estimated_eval_tokens": 17720 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose semantic grouping of unrelated Transformers issues. The only clear common thread is broad v5 regression fallout, but the individual soft pairs do not look like the same underlying bug/change, so I would not merge any of them.", + "confidence": 0.73, + "canonical_issue_reason": "Issue 45850 is the broadest umbrella report here: it explicitly frames Transformers v5 as breaking many models, so it can serve as the top-level tracker for related regression reports.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45850 is the best representative of the cluster because it is broad, current, and could subsume multiple v5-related complaints, even though it is not a precise root-cause report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:31795", + "right": "issue:33453", + "accept": false, + "reason": "Different problem types: documentation wording vs tokenizer loading regression." + }, + { + "left": "issue:41669", + "right": "issue:44908", + "accept": false, + "reason": "Import-star cleanup/perf issue is unrelated to inverse_sqrt scheduler kwargs handling." + }, + { + "left": "issue:42371", + "right": "issue:44492", + "accept": false, + "reason": "TF32 settings behavior and a cache-strategy typo are unrelated." + }, + { + "left": "issue:43576", + "right": "issue:44273", + "accept": false, + "reason": "Broken env command is not the same code path as lazy loading behavior." + }, + { + "left": "issue:44360", + "right": "issue:45468", + "accept": false, + "reason": "DSA indexer/ReLU complaint is unrelated to Gemma-4 audio positional encoding." + }, + { + "left": "issue:44655", + "right": "issue:44861", + "accept": false, + "reason": "Saving pipelines and tied-weight-key crashes are different save-path bugs." + }, + { + "left": "issue:44492", + "right": "issue:45850", + "accept": false, + "reason": "A typo report is not the same as the broad v5 breakage umbrella." + }, + { + "left": "issue:44273", + "right": "issue:44908", + "accept": false, + "reason": "Lazy loading and scheduler kwargs are unrelated subsystems." + }, + { + "left": "issue:45230", + "right": "issue:45310", + "accept": false, + "reason": "The first is too generic to establish the same concrete from_pretrained failure as the Qwen3.5 MoE report." + }, + { + "left": "issue:43519", + "right": "issue:44485", + "accept": false, + "reason": "Different models and different logic: timestamp calculation vs RoPE implementation." + }, + { + "left": "issue:41669", + "right": "issue:43704", + "accept": false, + "reason": "Import cleanup/perf work is unrelated to a VRAM leak in dataloader threads." + }, + { + "left": "issue:43519", + "right": "issue:44908", + "accept": false, + "reason": "Timestamp math in Qwen3VL and scheduler keyword propagation are unrelated." + }, + { + "left": "issue:31795", + "right": "issue:34689", + "accept": false, + "reason": "Documentation confusion and Llama 3.2 model loading regression are not the same bug." + }, + { + "left": "issue:31795", + "right": "issue:37428", + "accept": false, + "reason": "Docs issue vs flash-attention import error; no shared code-path bug." + }, + { + "left": "issue:44315", + "right": "issue:45092", + "accept": false, + "reason": "Both mention loading/init, but they target different models and different failures." + }, + { + "left": "issue:43704", + "right": "issue:44485", + "accept": false, + "reason": "VRAM leak under threading is unrelated to GLM-5 RoPE math." + }, + { + "left": "issue:44485", + "right": "issue:45468", + "accept": false, + "reason": "RoPE implementation and audio positional encoding are different model-specific issues." + }, + { + "left": "issue:42371", + "right": "issue:43704", + "accept": false, + "reason": "TF32 configuration and VRAM leakage are unrelated." + }, + { + "left": "issue:43502", + "right": "issue:43519", + "accept": false, + "reason": "local_files_only network leakage and timestamp calculation are unrelated." + }, + { + "left": "issue:43519", + "right": "issue:44623", + "accept": false, + "reason": "Timestamp calculation bug is unrelated to missing processor save files." + }, + { + "left": "issue:43519", + "right": "issue:43704", + "accept": false, + "reason": "Different failures: timestamp math vs multi-thread VRAM leak." + }, + { + "left": "issue:44263", + "right": "issue:45468", + "accept": false, + "reason": "torch.split return handling and Gemma audio encoding are unrelated." + }, + { + "left": "issue:43606", + "right": "issue:45468", + "accept": false, + "reason": "CPU offload device mismatch is a different bug than Gemma audio positional encoding." + }, + { + "left": "issue:44261", + "right": "issue:45468", + "accept": false, + "reason": "Missing rms_norm_eps in MLA and Gemma audio positional encoding are different implementation issues." + }, + { + "left": "issue:43704", + "right": "issue:45341", + "accept": false, + "reason": "VRAM leak and a testing_utils bug are unrelated." + }, + { + "left": "issue:42913", + "right": "issue:43066", + "accept": false, + "reason": "Both are tokenizer-v5 complaints, but they describe different concrete failures and code paths." + }, + { + "left": "issue:44908", + "right": "issue:44998", + "accept": false, + "reason": "Scheduler kwargs bug is unrelated to the placeholder 'Unemployment' issue." + }, + { + "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:43704", + "right": "issue:44998", + "accept": false, + "reason": "VRAM leak in threading and the unrelated placeholder issue are not duplicates." + }, + { + "left": "issue:41669", + "right": "issue:44998", + "accept": false, + "reason": "Import-star cleanup/perf issue does not match the unrelated placeholder issue." + }, + { + "left": "issue:43525", + "right": "issue:44861", + "accept": false, + "reason": "Missing pad_token_id and tied_weight_key crashes are unrelated." + }, + { + "left": "issue:43066", + "right": "issue:44998", + "accept": false, + "reason": "Wrong tokenizer decoder type and the unrelated placeholder issue are not the same bug." + }, + { + "left": "issue:43572", + "right": "issue:44861", + "accept": false, + "reason": "Missing pad_token_idx and tied-weight-key crashes are separate compatibility bugs." + }, + { + "left": "issue:31515", + "right": "issue:38617", + "accept": false, + "reason": "Both are loading-related, but one is slow checkpoint loading and the other is an ImportError." + }, + { + "left": "issue:44360", + "right": "issue:44998", + "accept": false, + "reason": "DSA/ReLU discussion is unrelated to the placeholder issue." + }, + { + "left": "issue:44230", + "right": "issue:44623", + "accept": false, + "reason": "FP8 inference support for Qwen models is unrelated to processor.save_pretrained missing files." + }, + { + "left": "issue:30333", + "right": "issue:31515", + "accept": false, + "reason": "MLflow failure reporting and slow checkpoint loading are unrelated." + }, + { + "left": "issue:31515", + "right": "issue:36296", + "accept": false, + "reason": "Slow checkpoint loading and tensor-parallel training bugs are different issues." + }, + { + "left": "issue:31515", + "right": "issue:31795", + "accept": false, + "reason": "Checkpoint loading performance and documentation confusion are unrelated." + }, + { + "left": "issue:44230", + "right": "issue:45310", + "accept": false, + "reason": "Both involve Qwen3.5, but FP8 inference support and a from_pretrained error are not the same concrete bug." + }, + { + "left": "issue:31515", + "right": "issue:33453", + "accept": false, + "reason": "Checkpoint load slowness and tokenizer-loading regression are not the same failure." + }, + { + "left": "issue:43298", + "right": "issue:44230", + "accept": false, + "reason": "Continuous batching audio-model support and Qwen fp8 inference support are different feature gaps." + } + ] + }, + "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": 16, + "review_unit_index": 1, + "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": "519d87cc20a230a2a3f40bc380aef1c2475c503c0ce9f47bc203d14abd88334a", + "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 not duplicates: one requests Flash Attention 2 support for more architectures, the other requests GGUF support for more architectures. They concern different features and code paths, so the soft link should be rejected.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "Issue #33260 is the more active and recent enhancement request, with more inbound references; it is the better representative issue in this cluster, though it is not a duplicate of #26350.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:26350", + "right": "issue:33260", + "accept": false, + "reason": "Different underlying feature requests: Flash Attention 2 support vs GGUF support. They target separate functionality and are not the same bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 4, + "component_count": 16, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-27088-4", + "nodes": [ + "issue:27088", + "issue:43834", + "issue:44365", + "issue:44370" + ], + "soft_pairs": [ + "issue:27088|issue:44370", + "issue:27088|issue:44365", + "issue:27088|issue:43834", + "issue:43834|issue:44370", + "issue:43834|issue:44365", + "issue:44365|issue:44370" + ], + "prepared_review_unit_hash": "57e9692166a6300ac44fc31a952b9144d3960b11c7d5cb8da932fc36bb01b983", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4454, + "estimated_input_tokens": 1114, + "estimated_eval_tokens": 2484 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4454, + "estimated_input_tokens": 1114, + "estimated_eval_tokens": 2484 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are all docs translation request issues with highly similar template titles, but they do not provide enough evidence to treat them as the same underlying request. The Turkish-specific issue is concrete; the others look like placeholder/template variants that may refer to different languages.", + "confidence": 0.94, + "canonical_issue_reason": "issue:27088 is the most concrete and specific item: it names Turkish explicitly, while the other issues use generic placeholder titles and do not show a shared target language or identical request.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:27088 is the best representative issue because it is specific, actionable, and not just a template placeholder.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:27088", + "right": "issue:44370", + "accept": false, + "reason": "Same broad i18n-docs theme, but no evidence they target the same language or the same request; one is Turkish-specific and the other is a generic template issue." + }, + { + "left": "issue:27088", + "right": "issue:44365", + "accept": false, + "reason": "The titles are superficially similar, but the generic placeholder issue does not show it is the same Turkish translation request." + }, + { + "left": "issue:27088", + "right": "issue:43834", + "accept": false, + "reason": "No proof these are the same underlying issue; 27088 is explicitly Turkish, while 43834 is a generic language template." + }, + { + "left": "issue:43834", + "right": "issue:44370", + "accept": false, + "reason": "Both are generic translation-template issues, but there is no concrete language target or identical request evidence to merge them." + }, + { + "left": "issue:43834", + "right": "issue:44365", + "accept": false, + "reason": "The near-identical template titles are not enough to confirm the same issue; they could be separate translation requests for different languages." + }, + { + "left": "issue:44365", + "right": "issue:44370", + "accept": false, + "reason": "They look templated and similar, but without explicit language targets or shared unique details, they should not be treated as duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 5, + "component_count": 16, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-36817-15", + "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" + ], + "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:37015|issue:42503", + "issue:36879|issue:37015", + "issue:41115|issue:43837", + "issue:41115|issue:44121", + "issue:41115|issue:42503", + "issue:41115|issue:45027", + "issue:41211|issue:43671", + "issue:42503|issue:43837", + "issue:42503|issue:43671", + "issue:42503|issue:42738", + "issue:42738|issue:45027", + "issue:43671|issue:43837", + "issue:43671|issue:45027", + "issue:43671|issue:45177", + "issue:43671|issue:44121", + "issue:43837|issue:45322", + "issue:43837|issue:43909", + "issue:43909|issue:44205", + "issue:43671|issue:43909", + "issue:44121|issue:45322", + "issue:43837|issue:44121", + "issue:43837|issue:45027", + "issue:37015|issue:45027", + "issue:43837|issue:45177" + ], + "prepared_review_unit_hash": "774ef35ef344ffc8d750041b305d6553f672888c3f09c1fb512514bd140073a8", + "component_budget": { + "node_count": 15, + "item_count": 15, + "soft_pair_count": 35, + "serialized_chars": 20705, + "estimated_input_tokens": 5177, + "estimated_eval_tokens": 10610 + }, + "budget": { + "node_count": 15, + "item_count": 15, + "soft_pair_count": 35, + "serialized_chars": 20705, + "estimated_input_tokens": 5177, + "estimated_eval_tokens": 10610 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are superficially similar model-request issues, but they target different architectures and products (vision, audio, OCR, BERT-like, etc.). They are not duplicates of the same underlying change.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue stands out: the cluster contains distinct requests for different models, not one shared bug or feature.", + "canonical_pr_reason": null, + "best_issue_reason": "No global best issue for deduping; the items are separate model-support requests and should remain independent.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:36817", + "right": "issue:41211", + "accept": false, + "reason": "Different requested models (EuroBert vs DEIMv2); only the generic 'add model' framing matches." + }, + { + "left": "issue:36817", + "right": "issue:43671", + "accept": false, + "reason": "EuroBert config support and Qwen3-TTS support are unrelated model additions with different code paths." + }, + { + "left": "issue:36817", + "right": "issue:36879", + "accept": false, + "reason": "EuroBert and RF-DETR are different model families; not the same feature request." + }, + { + "left": "issue:36879", + "right": "issue:41211", + "accept": false, + "reason": "RF-DETR and DEIMv2 are distinct detection models; no duplicate underlying change." + }, + { + "left": "issue:36879", + "right": "issue:43909", + "accept": false, + "reason": "RF-DETR is a vision model request, while LFM2.5 Audio is an audio-model request." + }, + { + "left": "issue:36879", + "right": "issue:44121", + "accept": false, + "reason": "Different model targets and likely different integration paths; not a single duplicate request." + }, + { + "left": "issue:36879", + "right": "issue:43837", + "accept": false, + "reason": "RF-DETR and Qwen3-ASR are different modalities and support requests." + }, + { + "left": "issue:36879", + "right": "issue:43671", + "accept": false, + "reason": "RF-DETR (vision) and Qwen3-TTS (speech generation) are unrelated model additions." + }, + { + "left": "issue:37015", + "right": "issue:43837", + "accept": false, + "reason": "NeoBERT and Qwen3-ASR are unrelated models with different architecture and modality." + }, + { + "left": "issue:37015", + "right": "issue:44121", + "accept": false, + "reason": "NeoBERT and the weight-sparse transformer request are not the same model or code change." + }, + { + "left": "issue:37015", + "right": "issue:43909", + "accept": false, + "reason": "NeoBERT is a BERT-like NLP model; LFM2.5 Audio is an audio model request." + }, + { + "left": "issue:37015", + "right": "issue:42503", + "accept": false, + "reason": "NeoBERT and ModernVBERT are different model variants; similar naming only." + }, + { + "left": "issue:36879", + "right": "issue:37015", + "accept": false, + "reason": "RF-DETR and NeoBERT are distinct model requests across different domains." + }, + { + "left": "issue:41115", + "right": "issue:43837", + "accept": false, + "reason": "MiniCPM3 model architecture and Qwen3-ASR support are different requests." + }, + { + "left": "issue:41115", + "right": "issue:44121", + "accept": false, + "reason": "MiniCPM3 and the weight-sparse transformer proposal are not the same underlying model change." + }, + { + "left": "issue:41115", + "right": "issue:42503", + "accept": false, + "reason": "MiniCPM3 and ModernVBERT are separate model additions, not duplicates." + }, + { + "left": "issue:41115", + "right": "issue:45027", + "accept": false, + "reason": "MiniCPM3 and Voxtral TTS support are different model support requests with different modalities." + }, + { + "left": "issue:41211", + "right": "issue:43671", + "accept": false, + "reason": "DEIMv2 and Qwen3-TTS are unrelated model requests; only broad similarity exists." + }, + { + "left": "issue:42503", + "right": "issue:43837", + "accept": false, + "reason": "ModernVBERT and Qwen3-ASR are distinct speech/NLP model requests." + }, + { + "left": "issue:42503", + "right": "issue:43671", + "accept": false, + "reason": "ModernVBERT and Qwen3-TTS are different models and different tasks." + }, + { + "left": "issue:42503", + "right": "issue:42738", + "accept": false, + "reason": "ModernVBERT and BERT-like models with RoPE are related only at a naming level, not the same concrete change." + }, + { + "left": "issue:42738", + "right": "issue:45027", + "accept": false, + "reason": "BERT-like RoPE support and Voxtral TTS support are not the same feature." + }, + { + "left": "issue:43671", + "right": "issue:43837", + "accept": false, + "reason": "Qwen3-TTS and Qwen3-ASR are both audio-related but still different model support requests." + }, + { + "left": "issue:43671", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-TTS and Voxtral-4B-TTS are different TTS model requests, not one duplicate fix." + }, + { + "left": "issue:43671", + "right": "issue:45177", + "accept": false, + "reason": "Qwen3-TTS and DeepSeek-OCR2 target different modalities and code paths." + }, + { + "left": "issue:43671", + "right": "issue:44121", + "accept": false, + "reason": "TTS support and the weight-sparse transformer request are unrelated model changes." + }, + { + "left": "issue:43837", + "right": "issue:45322", + "accept": false, + "reason": "Qwen3-ASR support and EUPE are different model requests in different subsystems." + }, + { + "left": "issue:43837", + "right": "issue:43909", + "accept": false, + "reason": "Qwen3-ASR and LFM2.5 Audio are both audio-related but not the same model or implementation." + }, + { + "left": "issue:43909", + "right": "issue:44205", + "accept": false, + "reason": "LFM2.5 Audio and SAM3-LiteText are distinct model requests; no shared duplicate change." + }, + { + "left": "issue:43671", + "right": "issue:43909", + "accept": false, + "reason": "Qwen3-TTS and LFM2.5 Audio are separate audio-model support requests." + }, + { + "left": "issue:44121", + "right": "issue:45322", + "accept": false, + "reason": "Weight-sparse transformer support and EUPE are different model proposals with different semantics." + }, + { + "left": "issue:43837", + "right": "issue:44121", + "accept": false, + "reason": "Qwen3-ASR and the weight-sparse transformer request are unrelated changes." + }, + { + "left": "issue:43837", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-ASR and Voxtral TTS are different speech model support requests." + }, + { + "left": "issue:37015", + "right": "issue:45027", + "accept": false, + "reason": "NeoBERT and Voxtral TTS are unrelated model additions." + }, + { + "left": "issue:43837", + "right": "issue:45177", + "accept": false, + "reason": "Qwen3-ASR and DeepSeek-OCR2 are different modality/model requests." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 6, + "component_count": 16, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-37934-6", + "nodes": [ + "issue:37934", + "issue:42936", + "issue:43240", + "issue:43668", + "issue:43813", + "issue:44031" + ], + "soft_pairs": [ + "issue:37934|issue:43668", + "issue:42936|issue:43240", + "issue:42936|issue:44031", + "issue:42936|issue:43813", + "issue:42936|issue:43668", + "issue:43240|issue:43668", + "issue:43668|issue:44031", + "issue:43668|issue:43813" + ], + "prepared_review_unit_hash": "2465db6d3a1554098e99f145482d26e3767baebc94884ad91c807f8cff6eda83", + "component_budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 8, + "serialized_chars": 6289, + "estimated_input_tokens": 1573, + "estimated_eval_tokens": 3402 + }, + "budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 8, + "serialized_chars": 6289, + "estimated_input_tokens": 1573, + "estimated_eval_tokens": 3402 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is not a duplicate set: the issues span unrelated models, tokenizer warnings, config typing, and a stray typo. No soft edge should be accepted.", + "confidence": 0.98, + "canonical_issue_reason": "No canonical issue: the items are unrelated bugs across different components (Llama4, Mask2Former, ModernBERT, quantization, tokenizer regex warnings).", + "canonical_pr_reason": null, + "best_issue_reason": "No single issue is a good cluster representative; if forced, 43668 is the cleanest standalone bug report, but it does not represent the rest of the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:37934", + "right": "issue:43668", + "accept": false, + "reason": "Different model/component and different bug class: Llama4 norm behavior vs ModernBERT config type hint." + }, + { + "left": "issue:42936", + "right": "issue:43240", + "accept": false, + "reason": "Mask2former ignore_value being unused is unrelated to kwargs not being forwarded to a loss function." + }, + { + "left": "issue:42936", + "right": "issue:44031", + "accept": false, + "reason": "Different subsystems and symptoms: model loss parameter handling vs tokenizer regex warnings." + }, + { + "left": "issue:42936", + "right": "issue:43813", + "accept": false, + "reason": "An unused loss parameter and a typo in a quantization attribute are not the same underlying bug." + }, + { + "left": "issue:42936", + "right": "issue:43668", + "accept": false, + "reason": "These are separate issues in different models: Mask2former loss wiring vs ModernBERT config typing." + }, + { + "left": "issue:43240", + "right": "issue:43668", + "accept": false, + "reason": "Loss kwargs forwarding bug and a config type-hint error are unrelated." + }, + { + "left": "issue:43668", + "right": "issue:44031", + "accept": false, + "reason": "ModernBERTConfig typing issue does not match the tokenizer regex warning regression." + }, + { + "left": "issue:43668", + "right": "issue:43813", + "accept": false, + "reason": "Config type hint incorrectness and a typo in a quantization symbol are different problems." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 7, + "component_count": 16, + "review_unit_index": 1, + "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": "827e6bd2a0944778c7ad390c12ebb457d7fc7e7be8d36660672ebd8ea990a9c2", + "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 are unrelated open issues: one about block decoding behavior and one about adding DeepSeek v3.2 support. No shared targets, filenames, or explicit linkage suggests duplication.", + "confidence": 0.98, + "canonical_issue_reason": "#41084 is the more specific implementation/behavior issue; #43037 is a model-support request and does not appear to be a duplicate of it.", + "canonical_pr_reason": null, + "best_issue_reason": "Neither issue is a strong global canonical for the other because they cover different product areas and different kinds of work.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41084", + "right": "issue:43037", + "accept": false, + "reason": "Different subjects and likely different code paths: block decoding vs DeepSeek v3.2 model support. Soft similarity alone is not enough to treat them as the same underlying bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 8, + "component_count": 16, + "review_unit_index": 1, + "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": "932a3ee1df1b0e3bccc096c5ca5449d3ecc4bcd6ebff5514aacbb40ab393d1fe", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1583, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1583, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are both about Trainer extensibility, but they describe different problems: one is about `remove_unused_column` behavior/customization, the other about FLOPs estimation for custom trainers. They should not be clustered as duplicates.", + "confidence": 0.87, + "canonical_issue_reason": "Issue #43598 is the better representative because it states a specific Trainer customization problem and is more directly about a code-path behavior change.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #43598 is the strongest issue to keep as the cluster representative; #43600 is a separate Trainer accounting problem rather than the same bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43598", + "right": "issue:43600", + "accept": false, + "reason": "They share a broad Trainer/customization theme, but the concrete problems differ: unused-column removal vs FLOPs estimation for custom trainers." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 9, + "component_count": 16, + "review_unit_index": 1, + "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": "f4c92435dc4ee3df0a87e4596eb4d92012908abe7a0885c5b430b3e83c35fb23", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1612, + "estimated_input_tokens": 403, + "estimated_eval_tokens": 1062 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1612, + "estimated_input_tokens": 403, + "estimated_eval_tokens": 1062 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are not duplicates: one request is for Qwen3-Omni registration in AutoModel/AutoModelForConditionalGeneration, while the other is for AutoModelForSequenceClassification support for Qwen3.5. They target different model families and different head/task support.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue fits both; they describe different underlying changes and should remain separate.", + "canonical_pr_reason": null, + "best_issue_reason": "Neither issue is a strong global canonical for the cluster because the requests are unrelated in model family and API surface.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43684", + "right": "issue:44405", + "accept": false, + "reason": "Different underlying requests: Qwen3-Omni model registration vs Qwen3.5 sequence-classification support. They do not fix the same code path 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": 10, + "component_count": 16, + "review_unit_index": 1, + "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": "279365bda713883de957e004ac3ce915d1284e6cf5084ffe538f356ac363dac6", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4265, + "estimated_input_tokens": 1067, + "estimated_eval_tokens": 2390 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4265, + "estimated_input_tokens": 1067, + "estimated_eval_tokens": 2390 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The cluster is mostly low-information spam-like issues. The three \"huggingface\" tickets are superficially similar, but there isn\u2019t enough body detail, repro, or target evidence to safely treat them as confirmed duplicates. \"Fett\" looks unrelated.", + "confidence": 0.61, + "canonical_issue_reason": "issue:43845 is the best available anchor only because it is the earliest of the repeated \"huggingface\" reports; however, all of these issues are too sparse to prove a real duplicate bug.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43845 is the most plausible cluster representative for triage since it matches the repeated title pattern and is the earliest candidate, but the report quality is too poor to call it a strong canonical.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43835", + "right": "issue:43845", + "accept": false, + "reason": "Different titles and no shared bug details, targets, or files; similarity alone is not enough." + }, + { + "left": "issue:43835", + "right": "issue:43846", + "accept": false, + "reason": "Different titles and no evidence they describe the same underlying problem." + }, + { + "left": "issue:43835", + "right": "issue:43859", + "accept": false, + "reason": "No substantive content to connect \"Fett\" with the \"huggingface\" reports." + }, + { + "left": "issue:43845", + "right": "issue:43846", + "accept": false, + "reason": "Identical titles, but the bodies are too thin to confirm the same bug or one concrete duplicate change." + }, + { + "left": "issue:43845", + "right": "issue:43859", + "accept": false, + "reason": "Same broad title only; no repro, target, or code-path evidence tying them together." + }, + { + "left": "issue:43846", + "right": "issue:43859", + "accept": false, + "reason": "Shared title pattern only; insufficient detail to accept as the same issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 11, + "component_count": 16, + "review_unit_index": 1, + "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": "0b5dbb12d7c3895d4bfaab7863562af247e50e9b60c009ad856995977a29fc64", + "component_budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 9, + "serialized_chars": 5788, + "estimated_input_tokens": 1447, + "estimated_eval_tokens": 3150 + }, + "budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 9, + "serialized_chars": 5788, + "estimated_input_tokens": 1447, + "estimated_eval_tokens": 3150 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster looks like low-information or spam-like issues rather than a real duplicate set. The only somewhat descriptive item is the documentation suggestion issue; the others are placeholder/spam titles.", + "confidence": 0.78, + "canonical_issue_reason": "issue:44418 is the most informative and least spam-like title in the set; the others are obvious placeholders or spam-style posts.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44418 is the best representative of the cluster because it has the clearest, non-placeholder title, even though the overall cluster is not a strong duplicate group.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44371", + "right": "issue:44572", + "accept": false, + "reason": "Both are placeholder '' issues, but there is no evidence they describe the same underlying report rather than separate spam posts." + }, + { + "left": "issue:44371", + "right": "issue:44573", + "accept": false, + "reason": "Same title is not enough here; these look like independent spam/placeholder issues with no shared substantive content." + }, + { + "left": "issue:44371", + "right": "issue:44418", + "accept": false, + "reason": "One is a spam placeholder and the other is a documentation suggestion; they do not appear to be the same issue." + }, + { + "left": "issue:44371", + "right": "issue:44937", + "accept": false, + "reason": "A spam placeholder and a specific-sounding title are not enough to conclude a shared underlying bug/change." + }, + { + "left": "issue:44418", + "right": "issue:44572", + "accept": false, + "reason": "Documentation enhancement vs placeholder spam; no concrete overlap in problem statement." + }, + { + "left": "issue:44418", + "right": "issue:44573", + "accept": false, + "reason": "Documentation enhancement vs placeholder spam; they are not the same underlying issue." + }, + { + "left": "issue:44572", + "right": "issue:44573", + "accept": false, + "reason": "Identical spam titles alone are insufficient evidence of the same underlying bug/change." + }, + { + "left": "issue:44572", + "right": "issue:44937", + "accept": false, + "reason": "Placeholder spam and a specific named-check title do not clearly match the same report." + }, + { + "left": "issue:44573", + "right": "issue:44937", + "accept": false, + "reason": "No substantive evidence ties these together as the same issue; titles point to different content." + } + ] + }, + "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": 16, + "review_unit_index": 1, + "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": "f213f635b28561fb70ebcb6d958345f272a008a50a1a99f4297dc33fead21fb4", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1646, + "estimated_input_tokens": 412, + "estimated_eval_tokens": 1080 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1646, + "estimated_input_tokens": 412, + "estimated_eval_tokens": 1080 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two closed issues are very likely duplicates: they share the exact same title, close creation times, and similar body sizes, with soft-similarity support. The evidence is strong enough to treat them as the same issue thread, though not as absolute certainty.", + "confidence": 0.92, + "canonical_issue_reason": "Issue #45145 is the better canonical representative because it was created first and has an inbound reference, making it the slightly more established thread.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #45145 is the best representative for the cluster because it is the earliest instance of the identical issue title and already has some reference activity.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45145", + "right": "issue:45151", + "accept": true, + "reason": "Accept: the issues have the exact same title and similar-sized bodies, and the packet flags soft similarity between them. This is consistent with a duplicate issue rather than a distinct report." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The packet supports a strong similarity/duplicate hypothesis, but the analyst slightly overstates the evidence. In particular, phrases like \"clearly the same underlying proposal\" and the 0.99 confidence go beyond what is directly grounded in the packet, which only shows identical titles, similar body sizes, and soft-similarity evidence. A more conservative wording would be safer." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 13, + "component_count": 16, + "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": 14, + "component_count": 16, + "review_unit_index": 1, + "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": "5f8153d558f6a927653737f9f3a7de72453ba19fdea0a5211284060859ea898e", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1668, + "estimated_input_tokens": 417, + "estimated_eval_tokens": 1090 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1668, + "estimated_input_tokens": 417, + "estimated_eval_tokens": 1090 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are two separate 4-bit inference bugs: one is a Gemma4/audio-module uint8 torch.finfo crash, the other is a BitsAndBytesConfig regression where llm_int8_skip_modules drops the default lm_head exclusion and triggers an AssertionError.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue: the reports describe different failure modes and different root causes/code paths.", + "canonical_pr_reason": null, + "best_issue_reason": "If one report must be preferred as the broader/clearer issue, issue 45674 is more general and actionable; 45672 is model-specific and unrelated to the skip_modules regression.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45672", + "right": "issue:45674", + "accept": false, + "reason": "Reject: different underlying bugs and fixes. 45672 is a Gemma4/audio-module uint8 torch.finfo TypeError during NF4 inference, while 45674 is an AssertionError caused by llm_int8_skip_modules clearing the default lm_head exclusion." + } + ] + }, + "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": 16, + "review_unit_index": 1, + "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": "f92ab211d7336bee30bffa8aa6b52122d78b160533114577dbda37ffd78cc7c8", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1477, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1477, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are not duplicates; they are separate closed items with different personal-name/date titles and no body, links, or shared context beyond superficial similarity.", + "confidence": 0.99, + "canonical_issue_reason": "issue:45762 is only a marginally better representative because it was created first, but neither issue is a meaningful canonical duplicate target.", + "canonical_pr_reason": null, + "best_issue_reason": "No strong best issue exists; both are low-information, unrelated-looking issues and should not be merged as duplicates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45762", + "right": "issue:45763", + "accept": false, + "reason": "Different names and dates with no supporting body or references; superficial resemblance only, not the same underlying issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 16, + "component_count": 16, + "review_unit_index": 1, + "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": "3ee35c90b902a4631387ad8b0e2ecaedffff191938842c628fa80c70723e9311", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1481, + "estimated_input_tokens": 371, + "estimated_eval_tokens": 998 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1481, + "estimated_input_tokens": 371, + "estimated_eval_tokens": 998 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two closed issues are unrelated: they are just distinct name/date strings with no shared topic, code location, or discussion context. No duplicate relationship is evident.", + "confidence": 0.99, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "Neither issue is a good canonical choice because there is no shared underlying bug or request to consolidate.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45764", + "right": "issue:45765", + "accept": false, + "reason": "The titles refer to different people and different dates; there is no sign 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": "pull_request", + "component_index": 1, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-39895-13", + "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:45795" + ], + "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:44827", + "pull_request:43995|pull_request:45213", + "pull_request:44827|pull_request:45170", + "pull_request:44827|pull_request:45795", + "pull_request:44827|pull_request:45655", + "pull_request:44827|pull_request:45213", + "pull_request:43823|pull_request:45213", + "pull_request:43823|pull_request:45795", + "pull_request:43251|pull_request:43254" + ], + "prepared_review_unit_hash": "8e20c9186238d166d1a988d06c7e9922cecb54000799f18c1d40480c07a66b59", + "component_budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 19, + "serialized_chars": 35200, + "estimated_input_tokens": 8800, + "estimated_eval_tokens": 17856 + }, + "budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 19, + "serialized_chars": 35200, + "estimated_input_tokens": 8800, + "estimated_eval_tokens": 17856 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is mostly a set of unrelated PRs that happen to share broad themes like model additions, docs, or refactors. The only strong duplicate-like pair is the two `fixed_cross_entropy` changes, which target the same function and same loss kwargs fix.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 43251 is the better canonical representative: it directly fixes the issue-linked `fixed_cross_entropy` behavior and cleanly forwards the new cross-entropy kwargs. PR 43254 is the same underlying change with a slightly different API wording.", + "best_issue_reason": null, + "best_pr_reason": "PR 43251 is the best global representative because it is the clearest, issue-linked implementation of the only true duplicate pair in the cluster.", + "soft_edge_verdicts": [ + { + "left": "pull_request:39895", + "right": "pull_request:43424", + "accept": false, + "reason": "VideoPrism model addition and an executorch dynamic-shape test are unrelated changes; they do not fix the same bug or implement the same feature." + }, + { + "left": "pull_request:41855", + "right": "pull_request:43823", + "accept": false, + "reason": "Mistral tokenizer methods and adding MobileLLM are unrelated model-level changes." + }, + { + "left": "pull_request:41855", + "right": "pull_request:43424", + "accept": false, + "reason": "Tokenizer API additions and executorch export tests are unrelated." + }, + { + "left": "pull_request:42668", + "right": "pull_request:43823", + "accept": false, + "reason": "A processor-loading robustness fix is not the same underlying change as adding a new model implementation." + }, + { + "left": "pull_request:43251", + "right": "pull_request:43823", + "accept": false, + "reason": "A cross-entropy kwargs fix and a MobileLLM model add are unrelated." + }, + { + "left": "pull_request:43251", + "right": "pull_request:44827", + "accept": false, + "reason": "Both touch training/model code, but one fixes loss kwargs while the other updates Mistral4 tests and MoE behavior; not the same bug/change." + }, + { + "left": "pull_request:43424", + "right": "pull_request:44827", + "accept": false, + "reason": "Executorch export test coverage and Mistral4 test fixes are different test changes for different subsystems." + }, + { + "left": "pull_request:43424", + "right": "pull_request:43823", + "accept": false, + "reason": "Dynamic-shape executorch tests and MobileLLM model addition are unrelated." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45655", + "accept": false, + "reason": "Adding MobileLLM and fixing `cls.config` resolution order are unrelated code paths." + }, + { + "left": "pull_request:43823", + "right": "pull_request:43995", + "accept": false, + "reason": "New model addition and Falcon output-interface refactoring are not the same change." + }, + { + "left": "pull_request:43995", + "right": "pull_request:44827", + "accept": false, + "reason": "Falcon standardized output refactoring and Mistral4 test fixes do not target the same underlying bug." + }, + { + "left": "pull_request:43995", + "right": "pull_request:45213", + "accept": false, + "reason": "A model refactor and the model-creation skill/docs are unrelated." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45170", + "accept": false, + "reason": "Mistral4 test fixes and a `layrnorm`\u2192`layernorm` rename pass are unrelated changes." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45795", + "accept": false, + "reason": "Mistral4 tests and audio/video processor docs are unrelated." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45655", + "accept": false, + "reason": "Mistral4 test changes and `cls.config` resolution order fixes are different issues." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45213", + "accept": false, + "reason": "Test fixes for Mistral4 are unrelated to the model-creation skill draft." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45213", + "accept": false, + "reason": "Adding MobileLLM and introducing the model-creation skill are unrelated." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45795", + "accept": false, + "reason": "MobileLLM model code and docs for audio/video processors are unrelated." + }, + { + "left": "pull_request:43251", + "right": "pull_request:43254", + "accept": true, + "reason": "Both PRs change `src/transformers/loss/loss_utils.py` in the same way: they extend `fixed_cross_entropy` to forward supported `nn.functional.cross_entropy` kwargs for the same underlying loss behavior fix." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded overall. The cluster appears mostly unrelated, and the only clearly mergeable duplicate pair is PR 43251/43254: both touch `src/transformers/loss/loss_utils.py`, share issue target 43240, and implement the same `fixed_cross_entropy` kwargs-forwarding fix with minor API-shape differences. The other soft-edge verdicts stay conservative and do not overstate evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 2, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-40546-2", + "nodes": [ + "pull_request:40546", + "pull_request:42781" + ], + "soft_pairs": [ + "pull_request:40546|pull_request:42781" + ], + "prepared_review_unit_hash": "cb6cc1072a53783e76a994e92ced85a5c190039577b8126d4abc504f3319c677", + "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": "Two open PRs both touch VibeVoice, but they are not the same change: #40546 implements the base VibeVoice integration, while #42781 adds a separate realtime variant plus acoustic-tokenizer work. Shared filenames are mostly the common model scaffold, not evidence of the same concrete fix.", + "confidence": 0.92, + "canonical_issue_reason": null, + "canonical_pr_reason": "#40546 is the better canonical PR because it is the broader, non-draft base VibeVoice implementation and appears to establish the core model package that later work extends.", + "best_issue_reason": null, + "best_pr_reason": "#40546 is the best representative of the cluster since it covers the primary VibeVoice integration rather than a narrower realtime follow-up.", + "soft_edge_verdicts": [ + { + "left": "pull_request:40546", + "right": "pull_request:42781", + "accept": false, + "reason": "Related VibeVoice work, but different scope and code path: base model integration versus realtime/acoustic-tokenizer additions. They share scaffold files, not the same underlying change, so they should not be merged as duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 3, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-40755-2", + "nodes": [ + "pull_request:40755", + "pull_request:41224" + ], + "soft_pairs": [ + "pull_request:40755|pull_request:41224" + ], + "prepared_review_unit_hash": "c1c706f1f74c0978886eddd287a0d82ccfc49f26afe881c79cffafcf910ea7d2", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4675, + "estimated_input_tokens": 1169, + "estimated_eval_tokens": 2594 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4675, + "estimated_input_tokens": 1169, + "estimated_eval_tokens": 2594 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two PRs are unrelated feature additions in different model families: TimesFM covariate forecasting vs DINOv3 image classification support. They do not describe the same bug or change, so the soft duplicate link should be rejected.", + "confidence": 0.99, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #41224 is the more self-contained and concrete change, adding a specific DINOv3 image-classification path; however, it is not a duplicate of the TimesFM covariate PR.", + "best_issue_reason": null, + "best_pr_reason": "PR #41224 is the best representative of the cluster because it is narrowly scoped and clearly tied to one model capability, whereas PR #40755 addresses a separate forecasting feature in a different subsystem.", + "soft_edge_verdicts": [ + { + "left": "pull_request:40755", + "right": "pull_request:41224", + "accept": false, + "reason": "Different models, tasks, and code paths: TimesFM covariate forecasting vs DINOv3 image classification. They are not the same underlying change and should not be merged as duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 4, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-41797-3", + "nodes": [ + "pull_request:41797", + "pull_request:45186", + "pull_request:45630" + ], + "soft_pairs": [ + "pull_request:41797|pull_request:45186", + "pull_request:45186|pull_request:45630" + ], + "prepared_review_unit_hash": "12af8159060eea929c5fcdf1fc4d8ec4a46cb67384cc3c8bd5b9e113e8496773", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 10428, + "estimated_input_tokens": 2607, + "estimated_eval_tokens": 5470 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 10428, + "estimated_input_tokens": 2607, + "estimated_eval_tokens": 5470 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are three unrelated model-addition PRs (DeepSeek OCR, Isaac, Kimi2-6). Their overlaps are mostly boilerplate registration/docs changes, not the same underlying bug or feature, so the soft similarity edges should be rejected.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: the items introduce different models and do not represent duplicate fixes or the same concrete code change.", + "best_issue_reason": null, + "best_pr_reason": "No single best PR for deduplication purposes; each PR targets a different model integration and should stand on its own.", + "soft_edge_verdicts": [ + { + "left": "pull_request:41797", + "right": "pull_request:45186", + "accept": false, + "reason": "Different model additions (DeepSeek OCR vs Isaac). Shared auto-registration/docs files are generic scaffolding, not evidence of the same fix." + }, + { + "left": "pull_request:45186", + "right": "pull_request:45630", + "accept": false, + "reason": "Different model additions (Isaac vs Kimi2-6). The overlap is broad framework wiring, but the concrete changes and model code paths are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 6, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-42975-2", + "nodes": [ + "pull_request:42975", + "pull_request:42976" + ], + "soft_pairs": [ + "pull_request:42975|pull_request:42976" + ], + "prepared_review_unit_hash": "e1b761349bcc15ce2d757565f7bcd60611c9b468b191b5efb180ea36f0922c64", + "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 two separate GitHub Actions maintenance PRs: one upgrades core workflow actions for Node 24 compatibility, the other bumps docker/build-push-action for Docker image workflows. They overlap in subsystem but not in the underlying change or fix.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "PR 42975 is the broader representative of the cluster theme because it updates multiple workflow actions across many CI jobs for Node 24 compatibility; PR 42976 is a narrower Docker action version bump.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42975", + "right": "pull_request:42976", + "accept": false, + "reason": "Different underlying changes: 42975 updates checkout/cache/upload-artifact and related workflow actions for Node 24 compatibility, while 42976 only bumps docker/build-push-action from v5 to v6 in Docker image workflows. Same broad area, but not the same concrete fix." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 7, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-42978-7", + "nodes": [ + "pull_request:42978", + "pull_request:43448", + "pull_request:43451", + "pull_request:43973", + "pull_request:44601", + "pull_request:45490", + "pull_request:45838" + ], + "soft_pairs": [ + "pull_request:42978|pull_request:43448", + "pull_request:42978|pull_request:43451", + "pull_request:42978|pull_request:45490", + "pull_request:43448|pull_request:43451", + "pull_request:43448|pull_request:45490", + "pull_request:43451|pull_request:45490", + "pull_request:43973|pull_request:45838", + "pull_request:44601|pull_request:45838", + "pull_request:45490|pull_request:45838" + ], + "prepared_review_unit_hash": "8ca98e9bb2b8e0745e9789b25f981163a31083dd71daaf1a8cb04b33782cd19f", + "component_budget": { + "node_count": 7, + "item_count": 7, + "soft_pair_count": 9, + "serialized_chars": 23513, + "estimated_input_tokens": 5879, + "estimated_eval_tokens": 12014 + }, + "budget": { + "node_count": 7, + "item_count": 7, + "soft_pair_count": 9, + "serialized_chars": 23513, + "estimated_input_tokens": 5879, + "estimated_eval_tokens": 12014 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is not a duplicate set: it mixes unrelated model-addition PRs (ViT NEPA, Molmo, Molmo2, LFM2 audio, CTSM, PP-OCRv6) plus a distributed pipeline-parallel infrastructure PR. The shared files are mostly boilerplate registration/docs, not the same concrete bug or change.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "pull_request:42978", + "right": "pull_request:43448", + "accept": false, + "reason": "Different model families and different feature scopes: ViT NEPA vs Molmo. Shared docs/auto-registration files are boilerplate, not evidence of the same change." + }, + { + "left": "pull_request:42978", + "right": "pull_request:43451", + "accept": false, + "reason": "ViT NEPA and Molmo2 are separate model additions with distinct architectures and code paths; no same underlying bug or mergeable combined PR." + }, + { + "left": "pull_request:42978", + "right": "pull_request:45490", + "accept": false, + "reason": "ViT NEPA and CTSM are unrelated model additions in different domains; overlap is only in generic transformer model registry/docs." + }, + { + "left": "pull_request:43448", + "right": "pull_request:43451", + "accept": false, + "reason": "Molmo and Molmo2 are related by name and domain, but they are different model families and implementations, not the same concrete change." + }, + { + "left": "pull_request:43448", + "right": "pull_request:45490", + "accept": false, + "reason": "Molmo vs CTSM is unrelated; the shared files are broad auto/model registration updates and do not indicate the same fix." + }, + { + "left": "pull_request:43451", + "right": "pull_request:45490", + "accept": false, + "reason": "Molmo2 and CTSM are different model additions with no shared bug or code path beyond generic framework plumbing." + }, + { + "left": "pull_request:43973", + "right": "pull_request:45838", + "accept": false, + "reason": "LFM2 audio is an audio ASR model, while PP-OCRv6 text recognition is an OCR vision/text model; these are entirely different changes." + }, + { + "left": "pull_request:44601", + "right": "pull_request:45838", + "accept": false, + "reason": "Pipeline-parallel infrastructure support is orthogonal to PP-OCRv6 model support; they do not fix the same concrete code-path problem." + }, + { + "left": "pull_request:45490", + "right": "pull_request:45838", + "accept": false, + "reason": "CTSM time-series model support and PP-OCRv6 text recognition support are unrelated model additions, not a duplicate or combined PR candidate." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 8, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43291-3", + "nodes": [ + "pull_request:43291", + "pull_request:43665", + "pull_request:44815" + ], + "soft_pairs": [ + "pull_request:43291|pull_request:43665", + "pull_request:43291|pull_request:44815", + "pull_request:43665|pull_request:44815" + ], + "prepared_review_unit_hash": "d954f0ef8b843168b9a9434f7154d50cb201987e19db618df067240dd19f6ad7", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 7093, + "estimated_input_tokens": 1774, + "estimated_eval_tokens": 3804 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 7093, + "estimated_input_tokens": 1774, + "estimated_eval_tokens": 3804 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "All three pull requests address different changes: a Whisper tokenizer/test fix, unrelated test edits for CLIP/ViT, and a broader dequantization/fine-grained FP8 loading fix. None look like duplicates of the same underlying bug or patch.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44815 is the strongest canonical PR because it is the most substantive and coherent fix, spanning the dequantization/loading path with a focused change set across implementation and tests.", + "best_issue_reason": null, + "best_pr_reason": "PR 44815 is the best standalone PR: it targets a concrete code-path problem in model loading / FP8 dequantization and has a consistent scope. The other PRs are narrower test-centric changes.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43291", + "right": "pull_request:43665", + "accept": false, + "reason": "Different areas and intents: Whisper tokenization/test cleanup versus unrelated CLIP/ViT test edits. They do not fix the same bug or change the same code path." + }, + { + "left": "pull_request:43291", + "right": "pull_request:44815", + "accept": false, + "reason": "One is a Whisper-specific tokenizer/test fix; the other is a model-loading/FP8 dequantization fix. No shared underlying defect." + }, + { + "left": "pull_request:43665", + "right": "pull_request:44815", + "accept": false, + "reason": "Although both touch tests, 43665 is just arbitrary failing-test assertions, while 44815 fixes dequantization/model-loading behavior. Not the same change and not mergeable as one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 9, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43488-2", + "nodes": [ + "pull_request:43488", + "pull_request:43532" + ], + "soft_pairs": [ + "pull_request:43488|pull_request:43532" + ], + "prepared_review_unit_hash": "fb99b8e13ee84a1402ed33c35fb5a0f8b44af7b635925c8f5dfd883aea8bf684", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4562, + "estimated_input_tokens": 1141, + "estimated_eval_tokens": 2538 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4562, + "estimated_input_tokens": 1141, + "estimated_eval_tokens": 2538 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two PRs are not duplicates: one is a narrow repo-bot/setup/test tweak, the other is a broader CI/workflow/testing utility change. They do not look like the same underlying fix or mergeable into one PR.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 43532 is the more substantial and representative change in this cluster; PR 43488 is a small, unrelated experimental setup/version tweak.", + "best_issue_reason": null, + "best_pr_reason": "PR 43532 is the best representative PR because it has the broader, more substantive code and workflow changes, whereas 43488 is a tiny targeted setup/test adjustment.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43488", + "right": "pull_request:43532", + "accept": false, + "reason": "Different change sets and purposes: 43488 adjusts packaging/versioning and a Vit test for repo-bot formatting, while 43532 rewires CI/workflow and testing utilities. They do not fix the same concrete bug or belong in one merged 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": 10, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43612-2", + "nodes": [ + "pull_request:43612", + "pull_request:43613" + ], + "soft_pairs": [ + "pull_request:43612|pull_request:43613" + ], + "prepared_review_unit_hash": "b211b49304801b3f3678ba231aafe9b5aead179ae042955d4d4d4986adf21403", + "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": "These are two separate pipeline additions for different segmentation tasks: promptable concept segmentation vs promptable visual segmentation. They share some framework/doc scaffolding, but the underlying feature scope and model backends differ, so they are not duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical duplicate PR here: PR 43612 and PR 43613 implement different segmentation pipelines with different backends and test suites.", + "best_issue_reason": null, + "best_pr_reason": "If one must be used as a representative anchor, PR 43613 is the broader pipeline addition, but it is still not a duplicate of PR 43612.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43612", + "right": "pull_request:43613", + "accept": false, + "reason": "Different feature targets and code paths: concept segmentation with SAM3 vs visual segmentation with SAM/SAM2/EdgeTam/tracker. Shared plumbing alone is not enough to merge them as one change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 11, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43651-2", + "nodes": [ + "pull_request:43651", + "pull_request:43663" + ], + "soft_pairs": [ + "pull_request:43651|pull_request:43663" + ], + "prepared_review_unit_hash": "d19c5e7f6d947b65fd0e4d27710b75936b33d2f56a91f5ac5445998f16fa16c1", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4289, + "estimated_input_tokens": 1073, + "estimated_eval_tokens": 2402 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4289, + "estimated_input_tokens": 1073, + "estimated_eval_tokens": 2402 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two PRs both touch `trainer.py`, but they solve different problems: one adds a hook for gradient-accumulation loss scaling, the other refactors signature-column filtering for PEFT/custom trainers. They are not duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: the changes address separate trainer code paths and are not mergeable as one fix.", + "best_issue_reason": null, + "best_pr_reason": "Neither PR is a good cluster representative for duplicate triage because they are unrelated enhancements in the same file, not the same underlying change.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43651", + "right": "pull_request:43663", + "accept": false, + "reason": "Different features and different code paths in `trainer.py` (`training_step` loss scaling vs signature-column filtering); they should not be treated 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": 12, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43656-3", + "nodes": [ + "pull_request:43656", + "pull_request:43836", + "pull_request:43842" + ], + "soft_pairs": [ + "pull_request:43656|pull_request:43842", + "pull_request:43836|pull_request:43842" + ], + "prepared_review_unit_hash": "59435728ce49d97e28a589aa75a7a52bf78e82f6c0457920ba7a99ae060deece", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 5993, + "estimated_input_tokens": 1499, + "estimated_eval_tokens": 3254 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 5993, + "estimated_input_tokens": 1499, + "estimated_eval_tokens": 3254 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "All three PRs appear to address the same underlying import-time TypeAdapter annotation NameError in `src/transformers/cli/serve.py`, with different levels of scope. The most focused fix is 43842.", + "confidence": 0.81, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 43842 is the clearest, most minimal fix: it directly prevents runtime evaluation of `TypeAdapter` by moving the import under `TYPE_CHECKING` and quoting the annotation where used.", + "best_issue_reason": null, + "best_pr_reason": "PR 43842 is the best representative of the cluster because it precisely targets the reported NameError with the smallest blast radius and the clearest description.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43656", + "right": "pull_request:43842", + "accept": true, + "reason": "Both target the same `TypeAdapter` NameError in the CLI serve module; 43656 and 43842 are different implementations of the same import-time annotation problem." + }, + { + "left": "pull_request:43836", + "right": "pull_request:43842", + "accept": true, + "reason": "Both are fixing the same concrete `TypeAdapter` annotation/runtime-evaluation bug in `src/transformers/cli/serve.py`, and their changes could plausibly be combined into one PR." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary stays within the packet: 43842 is clearly the most minimal TypeAdapter NameError fix, and the other PRs are plausibly related changes in the same file. The soft-edge judgments are conservative enough, especially since 43836 and 43842 share the same explicit issue target and 43656 is at least title-aligned with the same bug class." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 13, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43779-2", + "nodes": [ + "pull_request:43779", + "pull_request:43816" + ], + "soft_pairs": [ + "pull_request:43779|pull_request:43816" + ], + "prepared_review_unit_hash": "e7c674adcc031aa8883109a5dfc678b4407fb20b10764302092e384d031ba83a", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4362, + "estimated_input_tokens": 1091, + "estimated_eval_tokens": 2438 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4362, + "estimated_input_tokens": 1091, + "estimated_eval_tokens": 2438 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both pull requests appear to implement the same SwanLab integration change: adding run id/resume support in `integration_utils.py` for issue #43698, with one variant also expanding the docs. They are strong duplicate PR candidates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #43816 is the cleaner representative: it carries the same SwanLab id/resume support plus the accompanying documentation updates, making it the more complete version of the shared change.", + "best_issue_reason": null, + "best_pr_reason": "PR #43816 is the best global representative because it appears to include the full feature change and docs for SwanLab run id/resume support, while #43779 is the narrower implementation-focused variant.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43779", + "right": "pull_request:43816", + "accept": true, + "reason": "Same underlying SwanLab callback enhancement for id/resume support in the same file and same issue target; the doc differences do not change the core code-path change, so these are duplicate PRs." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: both PRs target issue #43698, touch the same file, and the previews indicate the same SwanLab id/resume support with #43816 adding docs. The duplicate/mergeable verdict is conservative enough given the shared target and overlapping code change." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 14, + "component_count": 26, + "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:44071|pull_request:44722", + "pull_request:44018|pull_request:44116", + "pull_request:44018|pull_request:44086", + "pull_request:44018|pull_request:44085", + "pull_request:44068|pull_request:44086", + "pull_request:44071|pull_request:44085", + "pull_request:44059|pull_request:44086", + "pull_request:44085|pull_request:44086", + "pull_request:44059|pull_request:44722", + "pull_request:44098|pull_request:44101", + "pull_request:44018|pull_request:44071", + "pull_request:44017|pull_request:44722", + "pull_request:44017|pull_request:44161", + "pull_request:44017|pull_request:44073", + "pull_request:44068|pull_request:44071", + "pull_request:44071|pull_request:44086", + "pull_request:43997|pull_request:44073", + "pull_request:44017|pull_request:44071", + "pull_request:44073|pull_request:44101", + "pull_request:44059|pull_request:44071", + "pull_request:44059|pull_request:44068", + "pull_request:44056|pull_request:44101", + "pull_request:44018|pull_request:44059", + "pull_request:44073|pull_request:44161", + "pull_request:44101|pull_request:44161", + "pull_request:43997|pull_request:44017", + "pull_request:43997|pull_request:44056", + "pull_request:44071|pull_request:44076", + "pull_request:44074|pull_request:44101", + "pull_request:43997|pull_request:44018", + "pull_request:44056|pull_request:44161", + "pull_request:43997|pull_request:44161", + "pull_request:44074|pull_request:44161", + "pull_request:44018|pull_request:44161", + "pull_request:43997|pull_request:44071", + "pull_request:44010|pull_request:44071", + "pull_request:44010|pull_request:44073", + "pull_request:44018|pull_request:44129", + "pull_request:43997|pull_request:44074", + "pull_request:43997|pull_request:44129", + "pull_request:44010|pull_request:44018", + "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": "d8bd8cec9a2e933dd09b671a715d8bb528789a3b98f071340df4ea6157283ccc", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 78, + "serialized_chars": 84902, + "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 set of model-specific refactors for standardized output tracing. Most pairs share only the umbrella issue target and should not be merged; the true duplicates are the GPT-J/CodeGen family and the GPT-Neo pair.", + "confidence": 0.86, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44722 is the most complete representative of the duplicated GPT-J/CodeGen refactor family: it explicitly targets the standardized output-tracing change and appears to subsume earlier GPT-J-only revisions.", + "best_issue_reason": null, + "best_pr_reason": "PR 44722 is the best cluster representative because it is the newest consolidated version of the concrete output-tracing refactor and covers the same code paths as the earlier GPT-J/CodeGen PRs.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43996", + "right": "pull_request:44085", + "accept": false, + "reason": "Different model families (CVT/FNet vs GPT-J); same umbrella issue target is not enough." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44044", + "accept": false, + "reason": "CVT/FNet and DeBERTa are unrelated implementations; only the tracking issue is shared." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44085", + "accept": true, + "reason": "Both are GPT-J output-tracing refactors on the same code path; 44085 looks like an overlapping/earlier revision of the same change." + }, + { + "left": "pull_request:44007", + "right": "pull_request:44072", + "accept": false, + "reason": "RegNet and EfficientNet are separate model refactors; shared issue target alone is insufficient." + }, + { + "left": "pull_request:44072", + "right": "pull_request:44722", + "accept": false, + "reason": "EfficientNet vs GPT-J/CodeGen are different model changes, not the same underlying bug." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44072", + "accept": false, + "reason": "GPT-J/CodeGen refactor and EfficientNet refactor are unrelated changes." + }, + { + "left": "pull_request:44013", + "right": "pull_request:44044", + "accept": false, + "reason": "MobileNetV2 and DeBERTa are distinct model implementations with different refactors." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-J/CodeGen output tracing and MGP-STR output tracing are different model-specific changes." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-J/CodeGen and MPT are separate code paths; same broad refactor theme is not enough." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44068", + "accept": true, + "reason": "Both PRs refactor GPT-Neo output tracing in the same file and appear to address the same concrete change." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT-J/CodeGen and GPT-Neo are different models; they do not fix the same concrete path." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44066", + "accept": false, + "reason": "GPT-Neo and GPT-J/CodeGen are separate model refactors, not mergeable duplicates." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44722", + "accept": false, + "reason": "ResNet output tracing is unrelated to GPT-J/CodeGen output tracing." + }, + { + "left": "pull_request:44086", + "right": "pull_request:44722", + "accept": false, + "reason": "MGP-STR and GPT-J/CodeGen are different model changes." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44085", + "accept": false, + "reason": "ResNet and GPT-J refactors are unrelated." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44019", + "accept": false, + "reason": "GPT-Neo and ResNet are different models; no shared concrete change." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44071", + "accept": false, + "reason": "ResNet and MPT output tracing are separate refactors." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo and Flaubert are unrelated model changes." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT-Neo refactor is not the same underlying change as GPT-J/CodeGen." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44722", + "accept": false, + "reason": "MPT and GPT-J/CodeGen are different model-specific output-tracing changes." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo and Flaubert are distinct changes." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo and MGP-STR are different code paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44085", + "accept": false, + "reason": "GPT-Neo vs GPT-J are different model refactors." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo and MGP-STR are unrelated model changes." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44086", + "accept": false, + "reason": "MPT and MGP-STR are separate output-tracing refactors." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT2 and MGP-STR are different model-specific changes." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-J and MGP-STR are unrelated changes." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT2 output tracing is not the same concrete change as GPT-J/CodeGen." + }, + { + "left": "pull_request:44098", + "right": "pull_request:44101", + "accept": false, + "reason": "ViLT and Flaubert are separate model refactors." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-Neo and MPT are different models and code paths." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44722", + "accept": false, + "reason": "Segformer and GPT-J/CodeGen are unrelated output-tracing refactors." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44161", + "accept": false, + "reason": "Segformer and LongT5 are different model changes." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44073", + "accept": false, + "reason": "Segformer and VisualBert are unrelated model refactors." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-Neo and MPT are different models." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44076", + "accept": false, + "reason": "MPT and ImageGPT are separate output-tracing changes." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44101", + "accept": false, + "reason": "TextNet and Flaubert are unrelated model refactors." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44017", + "accept": false, + "reason": "RegNet and Segformer are different models; shared umbrella issue is not enough." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44056", + "accept": false, + "reason": "RegNet and MPNet are distinct refactors." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44161", + "accept": false, + "reason": "RegNet and LongT5 are unrelated model changes." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44074", + "accept": false, + "reason": "RegNet and TextNet are different output-tracing implementations." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44129", + "accept": false, + "reason": "RegNet and SpeechT5 are separate model refactors." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44071", + "accept": false, + "reason": "SqueezeBert and MPT are different code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44073", + "accept": false, + "reason": "SqueezeBert and VisualBert are unrelated model changes." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44018", + "accept": false, + "reason": "SqueezeBert and GPT-Neo are different model refactors." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44074", + "accept": false, + "reason": "DeBERTa and TextNet are separate output-tracing changes." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44722", + "accept": true, + "reason": "Same GPT-J/CodeGen standardized output-tracing refactor family; 44722 looks like the consolidated version of the same concrete change." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44722", + "accept": true, + "reason": "Both target GPT-J output tracing on the same code path, and 44722 appears to be the consolidated follow-up." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-J and SpeechT5 are unrelated model refactors." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44007", + "accept": false, + "reason": "CVT/FNet and RegNet are different model changes." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44129", + "accept": false, + "reason": "CVT/FNet and SpeechT5 are unrelated refactors." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44129", + "accept": false, + "reason": "DeBERTa and SpeechT5 are different model-specific changes." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44154", + "accept": false, + "reason": "DeBERTa and VITS are unrelated output-tracing refactors." + }, + { + "left": "pull_request:44098", + "right": "pull_request:44017", + "accept": false, + "reason": "ViLT and Segformer are different models." + }, + { + "left": "pull_request:44098", + "right": "pull_request:44073", + "accept": false, + "reason": "ViLT and VisualBert are distinct changes." + }, + { + "left": "pull_request:44101", + "right": "pull_request:44116", + "accept": false, + "reason": "Flaubert and VITS are unrelated model refactors." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44161", + "accept": false, + "reason": "MPNet and LongT5 are separate output-tracing changes." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44018", + "accept": false, + "reason": "MPNet and GPT-Neo are different model implementations." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative about the soft edges. The accepted pairs all look mergeable as overlapping revisions of the same model-specific output-tracing bugfix, especially the GPT-J/CodeGen family and the GPT-Neo pair. The rejected pairs mostly rely only on the shared umbrella issue and correctly avoid overstating duplication across different model implementations." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 14, + "component_count": 26, + "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": "039519c8a1d555417675dd12f030731906cab5c5e261950b23c524ec8b6eba9e", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 78, + "serialized_chars": 84902, + "estimated_input_tokens": 21226, + "estimated_eval_tokens": 42708 + }, + "budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 6, + "serialized_chars": 17124, + "estimated_input_tokens": 4281, + "estimated_eval_tokens": 8818 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a set of model-specific refactors to standardize output tracing/capture_outputs across different Transformers architectures. They share a common theme, but they are not the same concrete code change or bug fix.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44073 is the clearest representative of the cluster: it explicitly frames the change as output tracing standardization, uses the new decorators, and its scope is easy to understand from the title and diff.", + "best_issue_reason": null, + "best_pr_reason": "PR 44073 is the best cluster representative because it is a well-scoped, clearly titled example of the broader standardized output-capture refactor and is easier to interpret than the more vague or typo-prone alternatives.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44024", + "right": "pull_request:44076", + "accept": false, + "reason": "Both are output-tracing refactors, but they touch different model implementations (FocalNet vs ImageGPT) and are separate architecture-specific changes, not one mergeable code-path fix." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44074", + "accept": false, + "reason": "These are related standardization refactors, but they apply to different models (VisualBert vs TextNet) with different forward paths and test impact; too broad to treat as the same PR." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44073", + "accept": false, + "reason": "Both modernize output capture in distinct models, but the affected code paths are unrelated model-specific implementations, so they do not look like one underlying change." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44074", + "accept": false, + "reason": "MPNet and TextNet are separate model refactors; the similarity is only the shared migration pattern, not the same concrete bug or patch." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44010", + "accept": false, + "reason": "RegNet and SqueezeBert are different architectures and the diffs are separate refactors toward standardized output handling, not the same fix." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44129", + "accept": false, + "reason": "They both relate to output tracing, but SpeechT5 is a different subsystem and includes distinct signature/behavior changes; this is not a single shared change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 15, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": "450f4f3c6c37f0f2b5b662dd094b6405e54cdd1a39da60f8fb1bbbab3fb0eb1b", + "component_budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 40, + "serialized_chars": 40159, + "estimated_input_tokens": 10040, + "estimated_eval_tokens": 20336 + }, + "budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 40, + "serialized_chars": 40159, + "estimated_input_tokens": 10040, + "estimated_eval_tokens": 20336 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These PRs are all variations of the same refactor theme, but they target different model implementations and code paths, so they are not duplicates of one another. The cluster should be treated as related-but-separate model-specific changes, not one mergeable fix.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44003 is the best representative because it is the broadest coherent change in the set, touching both falcon_mamba and mamba with the new output-capture plumbing. It best reflects the underlying refactor pattern without being just a tiny one-model tweak.", + "best_issue_reason": null, + "best_pr_reason": "44003 is the strongest representative PR in this cluster: it spans two closely related model files and introduces the output tracing changes directly, making it more substantive than the single-model refactors.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43998", + "right": "pull_request:44002", + "accept": false, + "reason": "Same high-level refactor style, but timm_backbone and upernet are different models with different forward paths and outputs; not the same concrete change." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44000", + "accept": false, + "reason": "Different model families and different forward signatures; only the tracing refactor theme is shared." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44001", + "accept": false, + "reason": "UnivNet is a separate audio model with a distinct code path, so this is not the same underlying change as timm_backbone." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44003", + "accept": false, + "reason": "timm_backbone and mamba/falcon_mamba are unrelated implementations; the shared wording is just a refactor pattern." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44029", + "accept": false, + "reason": "Different model and different output-capturing context; not mergeable as one PR." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44001", + "accept": false, + "reason": "MobileNetV1 and UnivNet are unrelated models; the overlap is only that both refactor return/output handling." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44002", + "accept": false, + "reason": "MobileNetV1 and UperNet have different forward logic and outputs; this is not the same bug or change." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44025", + "accept": false, + "reason": "MobileNetV1 and Depth Anything are separate vision models with distinct tracing changes." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44026", + "accept": false, + "reason": "VisionEncoderDecoder and MobileNetV1 are unrelated code paths; similar decorator/signature cleanup is not enough to merge." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44027", + "accept": false, + "reason": "SpeechEncoderDecoder is a different model wrapper and does not fix the same concrete issue as MobileNetV1." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44001", + "accept": false, + "reason": "Vision-text dual encoder and UnivNet are completely different model classes; same refactor motif only." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44002", + "accept": false, + "reason": "Different models and different forward contracts; not the same underlying fix." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44003", + "accept": false, + "reason": "Vision-text dual encoder vs mamba/falcon_mamba are unrelated code paths; shared output-tracing language is superficial." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44025", + "accept": false, + "reason": "These are different model families with different outputs and tracing behavior, so they are not duplicates." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44002", + "accept": false, + "reason": "UnivNet and UperNet are unrelated implementations; the PRs are only similar in style." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44003", + "accept": false, + "reason": "Different model families and different changed files; no evidence of a shared underlying bug." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44025", + "accept": false, + "reason": "Depth Anything and UnivNet are separate domains and code paths; not one mergeable fix." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44029", + "accept": false, + "reason": "UperNet and RWKV are unrelated model architectures, so these are not the same change." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44003", + "accept": false, + "reason": "UperNet and mamba/falcon_mamba do not fix the same concrete issue; only the refactor theme matches." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44028", + "accept": false, + "reason": "FalconMamba/Mamba and SuperPoint are unrelated code paths; the shared output-tracing modernization is too broad to merge." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44004", + "accept": false, + "reason": "Mamba/falcon_mamba and CodeGen are different model implementations with different tracing behavior." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44004", + "accept": false, + "reason": "TimmBackbone and CodeGen are unrelated modules; similar signature cleanup does not make them duplicates." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44004", + "accept": false, + "reason": "UnivNet and CodeGen are different models and different code paths; not the same concrete fix." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44004", + "accept": false, + "reason": "UperNet and CodeGen are unrelated implementations; the common theme is only output-tracing refactoring." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44004", + "accept": false, + "reason": "Vision-text dual encoder and CodeGen are unrelated; no shared bug or change beyond generic refactor style." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44026", + "accept": false, + "reason": "Depth Anything and VisionEncoderDecoder are different model wrappers and forward paths; similar decorator cleanup is not enough." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44027", + "accept": false, + "reason": "Depth Anything and SpeechEncoderDecoder are separate models with different code paths, so not the same underlying change." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44028", + "accept": false, + "reason": "Depth Anything and SuperPoint are different vision models; these PRs are not duplicates." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44029", + "accept": false, + "reason": "Depth Anything and RWKV address different architectures and output handling." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44030", + "accept": false, + "reason": "Depth Anything and DPR are unrelated models; the overlap is only that both refactor output tracing." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44029", + "accept": false, + "reason": "VisionEncoderDecoder and RWKV are unrelated code paths and do not appear to fix the same issue." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44027", + "accept": false, + "reason": "VisionEncoderDecoder and SpeechEncoderDecoder are analogous wrappers, but they are still different model paths and changes." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44028", + "accept": false, + "reason": "VisionEncoderDecoder and SuperPoint are separate models; no evidence they are the same concrete change." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44030", + "accept": false, + "reason": "VisionEncoderDecoder and DPR are different architectures and different forward logic." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44028", + "accept": false, + "reason": "SpeechEncoderDecoder and SuperPoint are unrelated modules; only the refactor theme is shared." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44029", + "accept": false, + "reason": "SpeechEncoderDecoder and RWKV are different model families with different output contracts." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44030", + "accept": false, + "reason": "SpeechEncoderDecoder and DPR do not share the same underlying bug or code path." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44029", + "accept": false, + "reason": "SuperPoint and RWKV are unrelated architectures; not a duplicate fix." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44030", + "accept": false, + "reason": "SuperPoint and DPR are different models with different output handling, so they should stay separate." + }, + { + "left": "pull_request:44029", + "right": "pull_request:44030", + "accept": false, + "reason": "RWKV and DPR are unrelated model implementations; similar output-tracing refactors do not make them 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": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-44662-3", + "nodes": [ + "pull_request:44662", + "pull_request:45075", + "pull_request:45401" + ], + "soft_pairs": [ + "pull_request:44662|pull_request:45401", + "pull_request:44662|pull_request:45075", + "pull_request:45075|pull_request:45401" + ], + "prepared_review_unit_hash": "6398e298064659d1e33aa50717ce4f7a9e0ed5ddc16fad98ba731e1af032e828", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 10912, + "estimated_input_tokens": 2728, + "estimated_eval_tokens": 5712 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 10912, + "estimated_input_tokens": 2728, + "estimated_eval_tokens": 5712 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are three separate model-addition PRs (PenguinVL, DeepSeek-OCR-2, Voxtral TTS). They only overlap on common Transformers integration scaffolding and docs, not on the same underlying bug or change, 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": "PR 45075 is the strongest representative of the cluster: it is non-draft, has the highest review activity, and contains a full model integration. The others are for different models, and 45401 is still draft.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44662", + "right": "pull_request:45401", + "accept": false, + "reason": "Different model implementations (PenguinVL vs Voxtral TTS); shared auto/docs files are generic integration boilerplate, not the same fix/change." + }, + { + "left": "pull_request:44662", + "right": "pull_request:45075", + "accept": false, + "reason": "Different model additions (PenguinVL vs DeepSeek-OCR-2) with distinct configs, processors, and model code; overlap is only in common registration/docs plumbing." + }, + { + "left": "pull_request:45075", + "right": "pull_request:45401", + "accept": false, + "reason": "Different concrete changes for different models (DeepSeek-OCR-2 vs Voxtral TTS); they cannot plausibly be merged as one PR for the same code-path problem." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 17, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-44979-2", + "nodes": [ + "pull_request:44979", + "pull_request:45363" + ], + "soft_pairs": [ + "pull_request:44979|pull_request:45363" + ], + "prepared_review_unit_hash": "d49ea25632609450da2ec1e8c905aa7d6a1e7d852ad355c1b1be99fe3fc422ac", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4408, + "estimated_input_tokens": 1102, + "estimated_eval_tokens": 2460 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4408, + "estimated_input_tokens": 1102, + "estimated_eval_tokens": 2460 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These PRs are related in the broad fusion/kernel-optimization area, but they are not the same change: one introduces a generic Module Fusion API, while the other wires up n-to-1 kernel fusion through KernelConfig and integrations. They should not be treated as duplicates.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #44979 is the more foundational and reusable change: it introduces the generic Module Fusion API and test coverage for the abstraction itself. PR #45363 is a narrower, downstream kernel-fusion configuration feature built on top of that idea.", + "best_issue_reason": null, + "best_pr_reason": "PR #44979 is the best cluster anchor because it defines the core fusion abstraction rather than one specific application of it, making it the more representative and durable reference point.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44979", + "right": "pull_request:45363", + "accept": false, + "reason": "Same broad fusion theme, but different concrete implementations and code paths: generic module-fusion API vs KernelConfig-driven n-to-1 kernel fusion. Not close enough to be 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": 18, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45086-2", + "nodes": [ + "pull_request:45086", + "pull_request:45317" + ], + "soft_pairs": [ + "pull_request:45086|pull_request:45317" + ], + "prepared_review_unit_hash": "ddb9a23444c59c77aa4b77173cc343e32886b6bc0f2103fa316931467bc8f721", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4414, + "estimated_input_tokens": 1104, + "estimated_eval_tokens": 2464 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4414, + "estimated_input_tokens": 1104, + "estimated_eval_tokens": 2464 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs target the same Mistral regex patch AttributeError in `src/transformers/tokenization_utils_tokenizers.py` against issue #45081; 45317 is the more complete version because it adds tests in addition to the code fix.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45317 is the stronger canonical PR because it fixes the same concrete bug as 45086 and also includes test coverage, making it the more complete representative of the change.", + "best_issue_reason": null, + "best_pr_reason": "PR 45317 is the best PR candidate: same fix, same target issue, and it adds verification via tests rather than only the code change.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45086", + "right": "pull_request:45317", + "accept": true, + "reason": "Same underlying AttributeError fix in `_patch_mistral_regex`, same shared issue target, and the diffs are the same code-path change; 45317 simply adds tests." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 19, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45115-2", + "nodes": [ + "pull_request:45115", + "pull_request:45569" + ], + "soft_pairs": [ + "pull_request:45115|pull_request:45569" + ], + "prepared_review_unit_hash": "c6e21db8fe39d78ba480a5b3274c9a008cc414857da0c0e8ab80d9a07a7da12d", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6066, + "estimated_input_tokens": 1517, + "estimated_eval_tokens": 3290 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6066, + "estimated_input_tokens": 1517, + "estimated_eval_tokens": 3290 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs are NemotronH-related, but they address different changes: 45115 is a narrower conversion-mapping/refactor for NemotronH compatibility, while 45569 is the broader model-family split introducing nemotron_h_dense and nemotron_h_sparse plus docs, auto-mappings, and tests. They overlap in shared model files, but not in the same underlying change.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "45569 is the better cluster representative because it is the more complete, user-facing NemotronH change set; 45115 is a narrower conversion-mapping/refactor and not the same change.", + "best_issue_reason": null, + "best_pr_reason": "45569 best represents the cluster because it introduces the actual NemotronH dense/sparse model split and updates the surrounding API/docs; 45115 is a smaller compatibility refactor.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45115", + "right": "pull_request:45569", + "accept": false, + "reason": "Related theme and shared files, but different concrete changes: conversion-mapping refactor vs architecture/model split with docs and tests. Not the same underlying PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 20, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45332-2", + "nodes": [ + "pull_request:45332", + "pull_request:45333" + ], + "soft_pairs": [ + "pull_request:45332|pull_request:45333" + ], + "prepared_review_unit_hash": "33e974db553f53d158706a9ad7ed5057c48eb64e2274d9acddd01712edd5ba64", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5398, + "estimated_input_tokens": 1350, + "estimated_eval_tokens": 2956 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5398, + "estimated_input_tokens": 1350, + "estimated_eval_tokens": 2956 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are related but distinct PRs: one adds heterogeneous per-layer config infrastructure, the other adds broader heterogeneous modeling/runtime support that appears to build on that config work. They should not be deduplicated.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45333 is the more foundational, narrowly scoped change: it introduces heterogeneous per-layer configuration support and the shared config helpers/classes. It is a cleaner canonical representative than the broader modeling PR.", + "best_issue_reason": null, + "best_pr_reason": "PR #45333 is the best single PR to represent the cluster because it defines the underlying config mechanism that the other PR seems to consume. PR #45332 is broader and depends on the same heterogeneity groundwork, so it is less suitable as the canonical representative.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45332", + "right": "pull_request:45333", + "accept": false, + "reason": "Different changes: #45333 adds per-layer heterogeneous config support and helpers, while #45332 adds heterogeneous model/runtime behavior across cache, masking, modeling, and GPT-OSS integration. They are related but not the same bug/change and would not be merged as one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 21, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45415-2", + "nodes": [ + "pull_request:45415", + "pull_request:45425" + ], + "soft_pairs": [ + "pull_request:45415|pull_request:45425" + ], + "prepared_review_unit_hash": "dd0c2945a0f477996e6128c1281138bef86a8c9902bc23dc094f012fc87c60d9", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5513, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5513, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs are type-checking/typing-oriented, but they address different scopes: #45415 is a broad repo-wide typing cleanup and mypy/config adjustment, while #45425 adds specific typing protocols/aliases for modeling_utils/peft-related code paths. They are related in theme, not the same change.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "#45415 is the broader, more representative typing initiative across many files and configuration, so it is the better canonical PR for this cluster.", + "best_issue_reason": null, + "best_pr_reason": "#45415 is the best overall representative for the cluster because it spans the widest set of typing changes and subsumes the general theme better than the narrower follow-up typing support in #45425.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45415", + "right": "pull_request:45425", + "accept": false, + "reason": "Do not merge: both are typing-related, but #45415 is a broad type-checking pass over many modules, while #45425 adds focused protocol/alias support for specific modeling_utils and PEFT integrations. They do not appear to fix the same concrete bug or be mergeable as one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 22, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45487-4", + "nodes": [ + "pull_request:45487", + "pull_request:45817", + "pull_request:45836", + "pull_request:45843" + ], + "soft_pairs": [ + "pull_request:45487|pull_request:45843", + "pull_request:45487|pull_request:45836", + "pull_request:45817|pull_request:45843", + "pull_request:45836|pull_request:45843" + ], + "prepared_review_unit_hash": "f22783679188155488146f1254baace45e22287936a170f5ab4f9025f9e1ca71", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 4, + "serialized_chars": 9783, + "estimated_input_tokens": 2446, + "estimated_eval_tokens": 5148 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 4, + "serialized_chars": 9783, + "estimated_input_tokens": 2446, + "estimated_eval_tokens": 5148 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are separate pull requests for different models and different fixes: two model-parallel device-placement bugs, one additional model-parallel bug in Gemma4, and one test skip for MiniCPM-V 4.6. They are related only at a broad theme level, not the same underlying change.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45487 is the broadest concrete code fix in the set: it patches device placement in AltCLIP/ChineseCLIP model-parallel paths and includes targeted test updates. The others are different model-specific fixes or a test-only skip, so they are not better canonical representatives.", + "best_issue_reason": null, + "best_pr_reason": "PR 45487 is the best representative PR because it contains an actual bug fix affecting multiple code paths, not just a test adjustment, and it is more substantial than the other single-model patches.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45487", + "right": "pull_request:45843", + "accept": false, + "reason": "Both are model-parallel fixes, but for different architectures (AltCLIP/ChineseCLIP vs Deimv2) and different code paths. They are not the same concrete bug and would not plausibly merge into one PR." + }, + { + "left": "pull_request:45487", + "right": "pull_request:45836", + "accept": false, + "reason": "45487 fixes runtime device-placement logic; 45836 only skips a failing test for MiniCPM-V 4.6. This is not the same underlying change." + }, + { + "left": "pull_request:45817", + "right": "pull_request:45843", + "accept": false, + "reason": "Gemma4 and Deimv2 have separate model-parallel device-placement fixes in different forward paths. Same broad theme, but distinct bugs." + }, + { + "left": "pull_request:45836", + "right": "pull_request:45843", + "accept": false, + "reason": "One is a test skip, the other is a code fix for Deimv2 model parallelism. They do not describe the same bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 23, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45548-2", + "nodes": [ + "pull_request:45548", + "pull_request:45662" + ], + "soft_pairs": [ + "pull_request:45548|pull_request:45662" + ], + "prepared_review_unit_hash": "89c24f112bb465c101cbeb673bbc7a94f6aa4cf6de993a72e9894adc74d6cf96", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4505, + "estimated_input_tokens": 1127, + "estimated_eval_tokens": 2510 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4505, + "estimated_input_tokens": 1127, + "estimated_eval_tokens": 2510 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are related EP/distributed-loading PRs, but they fix different failure modes: one is about preserving/reading distributed_config during load, the other about expert tensors being overwritten during rank-0 broadcast in FSDP2. They should not be treated as duplicates.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single canonical PR as a duplicate target; 45662 is the broader fix for the rank-0 broadcast overwrite bug, while 45548 addresses a separate accelerate/ZeRO-3 loading path.", + "best_issue_reason": null, + "best_pr_reason": "45548 is a narrower loader/config serialization fix, but 45662 is the more substantive and multi-file change for the concrete expert-overwrite issue. If forced to pick one representative, 45662 is the better umbrella PR for EP distributed-loading problems.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45548", + "right": "pull_request:45662", + "accept": false, + "reason": "Different root causes and code paths: 45548 adds distributed_config serialization and a has_ep helper for EP/ZeRO-3 loading via accelerate, while 45662 fixes experts being clobbered by rank-0 broadcast with FSDP2. Shared EP context is too broad to call them the same change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 24, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45639-2", + "nodes": [ + "pull_request:45639", + "pull_request:45645" + ], + "soft_pairs": [ + "pull_request:45639|pull_request:45645" + ], + "prepared_review_unit_hash": "2ef108318803ee0f07caedd4f46ecaaf576d254b8b8f777b1b656b3ec58411d4", + "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 target the same xdist-related captured_info logging problem; PR 45645 is the broader, more complete fix and subsumes the narrower utility change in 45639.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45645 is the better canonical PR because it addresses the captured_info xdist collision in the code and also updates the CI workflow to consume multiple captured_info files, making it the more complete end-to-end fix.", + "best_issue_reason": null, + "best_pr_reason": "PR 45645 is the strongest representative: it fixes the xdist-safe output handling and preserves CI debug log collection, whereas 45639 only changes the test utility side.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45639", + "right": "pull_request:45645", + "accept": true, + "reason": "Both PRs fix the same underlying captured_info/xdist collision path in testing_utils, and 45645 appears to extend rather than diverge from 45639 by adding workflow support for the same artifact change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 25, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45683-2", + "nodes": [ + "pull_request:45683", + "pull_request:45734" + ], + "soft_pairs": [ + "pull_request:45683|pull_request:45734" + ], + "prepared_review_unit_hash": "eba4ca0f26b2d3be0c00e316270a9341d31af2208b44dbab31b1d3a8e81e7538", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4469, + "estimated_input_tokens": 1118, + "estimated_eval_tokens": 2492 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4469, + "estimated_input_tokens": 1118, + "estimated_eval_tokens": 2492 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two open PRs touch the same quantization helper, but they fix different bugs: one special-cases audio submodules, the other changes how default skip lists are combined with user-supplied skips. They should not be clustered as duplicates.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "45734 is the better representative PR because it fixes the core skip-module composition logic in `get_modules_to_not_convert`, which is broader and more central than the narrow audio-module special case in 45683.", + "best_issue_reason": null, + "best_pr_reason": "45734 is the strongest standalone PR in the set: it addresses a concrete quantization failure mode in the main code path and its fix is directly about the helper\u2019s behavior rather than a model-specific workaround.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45683", + "right": "pull_request:45734", + "accept": false, + "reason": "Both touch quantization skip logic, but 45683 adds audio-tower/embed_audio exclusions while 45734 changes additive handling of `skip_modules`; these are different underlying bugs and would not naturally merge into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 26, + "component_count": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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 + } + ] +} diff --git a/snapshots/20260509T000027Z/analysis-runs/hybrid-model-20260509t000027z/manifest.json b/snapshots/20260509T000027Z/analysis-runs/hybrid-model-20260509t000027z/manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..d7e81a6d015d05e3f6a84d2b254590610e008ee9 --- /dev/null +++ b/snapshots/20260509T000027Z/analysis-runs/hybrid-model-20260509t000027z/manifest.json @@ -0,0 +1,14 @@ +{ + "analysis_id": "hybrid-model-20260509t000027z", + "artifacts": { + "hybrid": "snapshots/20260509T000027Z/analysis-runs/hybrid-model-20260509t000027z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260509T000027Z/analysis-runs/hybrid-model-20260509t000027z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "model": null, + "published_at": "2026-05-09T00:11:03Z", + "repo": "huggingface/transformers", + "schema_version": 1, + "snapshot_id": "20260509T000027Z", + "variant": "hybrid" +} diff --git a/snapshots/20260509T000027Z/manifest.json b/snapshots/20260509T000027Z/manifest.json index ae728cfb044d49201909403f99a1792f1a007384..c5da29564ad85c763058f0c4d12dfb1aec3a3e2e 100644 --- a/snapshots/20260509T000027Z/manifest.json +++ b/snapshots/20260509T000027Z/manifest.json @@ -32,6 +32,24 @@ "timeline_events": 558 }, "extracted_at": "2026-05-09T00:00:27Z", + "published_analysis": { + "canonical_analysis_id": "hybrid-model-20260509t000027z", + "runs": { + "hybrid-model-20260509t000027z": { + "analysis_id": "hybrid-model-20260509t000027z", + "artifacts": { + "hybrid": "snapshots/20260509T000027Z/analysis-runs/hybrid-model-20260509t000027z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260509T000027Z/analysis-runs/hybrid-model-20260509t000027z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "manifest_path": "snapshots/20260509T000027Z/analysis-runs/hybrid-model-20260509t000027z/manifest.json", + "model": null, + "published_at": "2026-05-09T00:11:03Z", + "variant": "hybrid" + } + }, + "schema_version": 1 + }, "repo": "huggingface/transformers", "snapshot_id": "20260509T000027Z", "watermark": {