← back to report

huggingface/transformers

// all events — 2026-10-25

Type AI Score Description Actor Reason Date
commit 1.00 Match a skip-list entry from the start only up to a module b Commit message contains explicit AI assi 2026-10-23
commit 1.00 docs(whisper): cast input features to model dtype in AutoMod Commit message contains explicit AI assi 2026-10-22
commit 1.00 Fine-grained FP8: DeepSeek-V4's expert_dtype in the quantiza Commit message contains explicit AI assi 2026-10-22
commit 1.00 [Fetcher] Fix missing test deps for dynamically-resolved ima Commit message contains explicit AI assi 2026-10-22
commit 1.00 Defer some imports (#49663) Commit message contains explicit AI assi 2026-10-22
commit 1.00 Fix Windows path separators breaking relative imports of Hub Commit message contains explicit AI assi 2026-10-21
commit 1.00 Fix background_color handling across DeepSeek-VL and Janus i Commit message contains explicit AI assi 2026-10-21
commit 1.00 Fix NameError in CI comment workflows: missing section marke Commit message contains explicit AI assi 2026-10-20
commit 1.00 Add GradiumPhononTTS (#49791) Explicitly references Claude sessions, i 2026-10-20
commit 1.00 [CI] Improve run-slow PR comments and remove CI recap (#4978 Commit message contains explicit AI assi 2026-10-20
commit 1.00 Fix two stale hub-kernel tests (#49795) Commit message contains explicit AI assi 2026-10-20
commit 1.00 Fix CohereCompass text RoPE to use Qwen3-VL interleaved M-Ro Commit message contains explicit AI assi 2026-10-20
commit 1.00 Alias thinking -> reasoning_content (#49535) Commit message contains explicit AI assi 2026-10-20
commit 1.00 Fix tests_fetcher backbone step (never worked since introduc Commit message contains explicit AI assi 2026-10-20
commit 1.00 Fix SSH runner: add /opt/venv/bin to PATH for Python 3.11 im Commit message contains explicit AI assi 2026-10-20
pr 1.00 Apertus 1.5: add model PR body explicitly mentions AI collabora 2026-07-31
pr 1.00 Fix `offloaded_static` cache crashing on models with only sl PR body explicitly mentions AI collabora 2026-10-24
pr 1.00 sam3: enhance output boxes for any resolution PR body explicitly mentions AI collabora 2026-10-24
pr 1.00 Feat test PR body explicitly mentions AI collabora 2026-10-23
pr 1.00 Fix local_files_only tokenizer fallback when tokenizer files PR body explicitly mentions AI collabora 2026-04-21
pr 1.00 Fold GQA query heads onto their KV head when decoding instea PR body explicitly mentions AI collabora 2026-10-19
pr 1.00 Keep image inputs on the first chunk of a chunked prefill PR body explicitly mentions AI collabora 2026-10-20
pr 1.00 :rotating_light: Fix gemma4 cross imports resulting in impor PR body explicitly mentions AI collabora 2026-10-19
pr 1.00 Allow untokenized output from `apply_transcription_request` Explicitly states that Claude Code was u 2026-10-23
pr 1.00 Optim: add SST optimizer wrapper and distributed Trainer sup PR body explicitly mentions AI collabora 2026-10-23
pr 1.00 Support minimum-length logits processing in continuous batch Explicitly states that AI assistance was 2026-09-01
pr 1.00 Check for an empty batch in convert_to_tensors without flatt Explicitly discloses using Claude Code t 2026-10-23
pr 1.00 added chronos2 model PR body explicitly mentions AI collabora 2026-10-23
pr 1.00 [dummy] trivial change to vit test to observe CI fetcher beh Explicitly labels the change as generate 2026-10-23
pr 1.00 Forward code_revision to every pipeline loader Explicitly states that Claude Code gener 2026-10-21
pr 1.00 detect an empty token batch without copying it Explicitly states that the change was ma 2026-10-23
pr 1.00 Quantizer API: checkpoint renames, skip-lists in checkpoint Explicitly states it was generated with 2026-10-20
pr 1.00 [i18n-KO] Translated generation_features.md to Korean PR body explicitly mentions AI collabora 2026-10-23
pr 1.00 Keep a capturing group's escapes when reversing a conversion Explicitly states it was generated with 2026-10-20
pr 1.00 [Gemma4] Stop prefill at last KV-owning layer when logits_to PR body explicitly mentions AI collabora 2026-10-15
pr 1.00 Fix typo: essentialy → essentially in KimiLinearDeltaAttenti Explicitly discloses preparation with AI 2026-10-23
pr 1.00 Fix typo: lagacy → legacy in VideoLlavaImageProcessorKwargs Explicitly discloses preparation with AI 2026-10-23
pr 1.00 Fix grammar: force and update → force an update in ProgressB Explicitly discloses preparation with AI 2026-10-23
pr 1.00 Fix wording: add missing 'of' in encodings docstring Explicitly discloses preparation with AI 2026-10-23
pr 1.00 Fix typo: reasonnable → reasonable in resolve_max_memory_per Explicitly discloses preparation with AI 2026-10-23
pr 1.00 Fix typo: detailled → detailed in PagedAttentionCache docstr Explicitly discloses AI assistance in pr 2026-10-23
pr 1.00 Weight loading: conversion ops declare many-to-many support, Explicitly states it was generated with 2026-10-20
pr 1.00 NVFP4: load a checkpoint's FP8 linears as FP8 Explicitly states it was generated with 2026-10-23
pr 1.00 MXFP8 on the fine-grained machinery (`Mxfp8Config`) Explicitly states it was generated with 2026-10-20
pr 1.00 Fine-grained quantization: a host quantizer that mixes metho Explicitly states it was generated with 2026-10-21
pr 1.00 Experts and TP: per-instance experts flags, load-time-only e Explicitly states it was generated with 2026-10-20
pr 1.00 Rename only MiniMax-M3-VL's MoE block, so a save keeps the d Explicitly states it was generated with 2026-10-20
pr 1.00 NVFP4 on the fine-grained machinery Explicitly states it was generated with 2026-10-20
pr 1.00 Load ModelOpt and compressed-tensors exports on the fine-gra Explicitly states it was generated with 2026-10-21
pr 1.00 Fine-grained block-scaled quantization: the shared machinery Explicitly states it was generated with 2026-10-20
pr 1.00 FP8 on the fine-grained machinery (`Fp8Config`) Explicitly states it was generated with 2026-10-20
pr 1.00 🚨 MXFP4 on the fine-grained machinery, and DeepSeek-V4's FP8 Explicitly states it was generated with 2026-10-20
pr 1.00 Shard MuseSpark's experts whole under tensor parallelism Explicitly states it was generated with 2026-10-20
pr 1.00 Match a skip-list entry from the start only up to a module b Explicitly states it was generated with 2026-10-20
pr 1.00 chore: improve typing in modeling_rope_utils.py PR body explicitly mentions AI collabora 2026-10-21
pr 1.00 🚨 NVFP4: MoE experts and ModelOpt exports on finegrained-ker Explicitly states that it was generated 2026-10-22
pr 1.00 [docs] MiniMax-M3.1-Flash snippet PR body explicitly mentions AI collabora 2026-10-23
pr 1.00 Allow saving a loaded generation config with inherited issue PR body explicitly mentions AI collabora 2026-10-23
pr 1.00 Composite transparency in shared RGB conversion and image pr The description explicitly states that A 2026-09-24
pr 1.00 [vLLM] Add Omni token counting and text replacement offsets The description explicitly states that A 2026-10-08
pr 1.00 allow saving a generation config that was loaded and not edi The description explicitly says it was m 2026-10-23
pr 1.00 Add AudioMAE PR body explicitly mentions AI collabora 2026-10-22
pr 0.88 [CB] Fix the CPU memory budget used to size the paged cache Highly polished, exhaustive prose and re 2026-10-12
pr 0.10 [WIP] modular cleanup Chaotic, highly specific planning notes 2026-10-02
pr 0.10 [pocket-tts] follow ups Detailed domain-specific notes, but no e 2026-10-23
pr 0.08 Simplify the quantizer API changes Brief, informal status update with no AI 2026-10-23
pr 0.08 [loading] Allow Renaming and Converter to work on the sume s Terse, informal description with specifi 2026-10-23
pr 0.08 [Improvement] Formalize KV sharing Detailed domain-specific explanation inc 2026-07-13
pr 0.08 fix: prevent empty frame sampling for short videos Detailed domain-specific explanation wit 2026-10-23
commit 0.06 test: add parity check to the CPU-only continuous batching t Polished but highly specific test ration 2026-10-23
pr 0.06 Add tiny model Detailed domain-specific implementation 2026-08-04
pr 0.06 [CB] Reduce warmup time by not computing attention over full Technical metrics and informal phrasing 2026-10-22
commit 0.05 [CB] Reduce the max_memory_percent in case of a CPU model (# Terse, domain-specific commit-style word 2026-10-23
pr 0.05 Fix get_image_size PIL dimension order and GOT-OCR2 bbox nor Detailed domain-specific diagnosis and t 2026-10-24
pr 0.05 Add qwen3 tts Template structure aside, the model-spec 2026-03-07
pr 0.05 [docs] CB cache behavior Brief, repository-specific documentation 2026-09-22
pr 0.05 Update MiniCPM-V 4.7 test expectations Brief, repository-specific update with n 2026-10-23
pr 0.05 [Qwen2-VL] Precompute data-dependent vision tensors on CPU f Highly technical, issue-specific explana 2026-10-23
pr 0.05 Audio from video in jinja Awkward, domain-specific wording and inf 2026-10-07
commit 0.04 [pocket_tts, gradium_phonon_tts] use sub_configs_defaults (# Technical details and shorthand strongly 2026-10-23
commit 0.04 🚨Add an FP8 backend to `QuantizedCache` (#48677) Highly domain-specific technical details 2026-10-22
pr 0.04 Add LatencyStreamer to track per-token latency and TTFT (#39 Polished but contains detailed, reposito 2026-10-24
pr 0.04 fix(tokenizers): stop treating a missing transformers_versio Highly detailed and technically grounded 2026-10-19
pr 0.04 🚨 Move processing defaults to class Technical, project-specific description 2026-09-24
pr 0.04 Build Idefics-style vision position ids with torch.where Clear technical rationale and concrete t 2026-10-23
commit 0.03 [docs] CB cache behavior (#49025) Terse, domain-specific commit notes with 2026-10-23
commit 0.03 docs(optimization): standardize 'Return:' to 'Returns:' in s Structured technical changelog with spec 2026-10-22
commit 0.03 Fix CLVP `c_proj` residual scaled-init silently never applie Detailed domain-specific debugging and r 2026-10-21
commit 0.03 Fix dtype cast in TimesFM 2.5 ResidualBlock for quantized we Contains precise quantization details, m 2026-10-21
commit 0.03 Apply EXIF orientation in load_image_as_tensor with torchvis Detailed implementation and regression-t 2026-10-21
commit 0.03 Fix expert-parallel token dispatch gradients on torch 2.7 to Detailed technical explanation is specif 2026-10-20
pr 0.03 Only treat marimo as a notebook environment when IPython is Specific failure analysis and verificati 2026-10-23
pr 0.03 Restore BLT's 512-byte sliding windows (entropy patcher + lo Detailed domain-specific analysis and me 2026-09-29
pr 0.03 Use unique request IDs for continuous batching Detailed, specific bug reproduction and 2026-09-18
commit 0.02 [Fix] Remove duplicate assignment in DeepseekV3.2 config (#4 Terse, specific maintenance note with in 2026-10-23
commit 0.02 Make sub-config-defaults easier (#49906) Very terse informal commit notes, not ge 2026-10-23
commit 0.02 [CB] Reduce warmup time by not computing attention over full Concise domain-specific notes with an en 2026-10-23
commit 0.02 Improve clarity and grammar in Auto Classes documentation (# Specific documentation edits and reviewe 2026-10-22
commit 0.02 Upgrade Python Min version to `3.11` (#49251) Terse, typo-containing technical changel 2026-10-22
commit 0.02 [LED][Longformer] Replace for-loop with unfold in _chunk ONN Terse implementation notes, including an 2026-10-22
commit 0.02 logits_to_keep bool mask for all ForXXX classes (#48825) Detailed domain references and informal 2026-10-22
commit 0.02 Fix `base_model_prefix` in the PP-OCR detectors, CLIPSeg, Ti Uses repository-specific terminology and 2026-10-21
commit 0.02 Move the MI300 docker image to ROCm 10.1 and update test exp Specific CI, ROCm, packaging, and Python 2026-10-21
commit 0.02 update some case to add xpu expectation (#49807) Terse, domain-specific commit message wi 2026-10-21
commit 0.02 docs: fix OOM/hang for Qwen dynamic resolution models by set Conventional documentation changelog wit 2026-10-21
commit 0.02 docs: fix prompt formatting and code alignment in Gemma mode Concise documentation fix referencing a 2026-10-21
commit 0.02 Show the actual accelerate version for `transformers env` (# Terse, iterative commit wording and spec 2026-10-20
commit 0.02 Update MIC fleet references and documentation (#49768) Terse, domain-specific change summary wi 2026-10-20
commit 0.02 Deprecate rope_scaling assignment and normalize legacy RoPE Technical commit-style notes and specifi 2026-10-20
pr 0.02 Fix boolean padding in ClapFeatureExtractor (#23648) Concise, issue-specific description with 2026-10-24
pr 0.02 Fix invalid JSON serialization for streaming error events in Concrete bug mechanism, affected files, 2026-10-24
pr 0.02 refactor output tracing in `mobilenet_v1` Terse, informal wording and issue refere 2026-02-14
pr 0.02 refactor output tracing for `vision_encoder_decoder` Informal phrasing, missing space, and br 2026-02-16
pr 0.02 Fix Qwen4-Exp eager attention NaNs with left padding Highly domain-specific explanation and i 2026-10-23
pr 0.02 Fix test pil import leak Terse, specific wording and informal phr 2026-10-23
pr 0.02 Fixes a dtype mismatch in flash_attn test Concrete reproduction steps and casual r 2026-10-23
pr 0.02 Subconfig creation in testers Brief informal bullets and repository-sp 2026-10-12
pr 0.02 fix some error in quantization xpu ci test Terse, informal title and reviewer menti 2026-10-23
pr 0.02 Fix out-of-bounds index bug in musicgen_melody model Technical, concise explanation with a co 2026-10-22
pr 0.02 Simplify using custom resolution for sam3 and sam3_video inf Informal phrasing and minor grammatical 2025-12-10
pr 0.02 Fix assertion diff limit Terse, informal wording and a specific t 2026-10-23
commit 0.01 Fixes a dtype mismatch in flash_attn test (#49912) Brief commit message with a standard sig 2026-10-23
commit 0.01 add XPU Expectations for 4 new model tests (#49815) Short technical message with a standard 2026-10-23
commit 0.01 :rotating_light: Fix gemma4 cross imports resulting in impor Informal fragmented notes and typo-like 2026-10-23
commit 0.01 Fix out-of-bounds index bug in musicgen_melody model (#49842 Repeated sign-offs and terse bug-fix not 2026-10-23
commit 0.01 [gradium_phonon_tts] point to the released Hub repos (#49867 Detailed repository-specific technical n 2026-10-22
commit 0.01 [EmbeddingGemma2] update integration tests (#49576) Terse, informal update typical of a huma 2026-10-22
commit 0.01 Fix a comment (#44872) References specific files and includes o 2026-10-22
commit 0.01 Fix hardcoded torch.float64 crashing on MPS (#46723) (#46730 Brief issue-specific fix description wit 2026-10-22
commit 0.01 Fix Qwen4-Exp indexer position cache lifecycle (#49840) Concise, model-specific engineering summ 2026-10-22
commit 0.01 Remove sdpa from gemma's integration tests (#49849) Very terse fix messages are characterist 2026-10-22
commit 0.01 fixup metadata workflow (#49854) Minimal informal wording strongly sugges 2026-10-22
commit 0.01 Fix torch_float casting to int instead of float (#47468) Specific bug mechanism and relevant func 2026-10-22
commit 0.01 skip `sdpa_can_dispatch_on_flash` tests for 3 models (#49809 Brief, repetitive commit notes and a sta 2026-10-22
commit 0.01 Fix several object detection models/loader fix for module ty Short, informal debugging notes and ters 2026-10-22
commit 0.01 Fix encoder logits processors when precomputed encoder_outpu Informal iterative notes such as “fix th 2026-10-22
commit 0.01 refactor CTC loss (#49833) The extremely terse commit message is un 2026-10-22
commit 0.01 Add Intel XPU CI (#49254) Repetitive test notes, typos, and specif 2026-10-22
commit 0.01 fix(mask2former): register input_projections as nn.ModuleLis Specific PyTorch behavior and issue refe 2026-10-21
commit 0.01 Add tensor parallel plan for Nemotron (#48780) Concise domain-specific change with conc 2026-10-21
commit 0.01 Only tie MM Grounding DINO decoder heads when the config sha Informal iterative commit notes and deta 2026-10-21
commit 0.01 Add UltraVBERT (#49685) Long sequence of terse, project-specific 2026-10-21
commit 0.01 Fix multi-zoom crops in the SAM mask-generation pipeline and Detailed implementation notes use specif 2026-10-21
commit 0.01 Add Flint model support. (#49824) Informal iterative notes and specific ch 2026-10-21
commit 0.01 Remove sdpa support for models using softcapping (#49811) Terse, issue-specific commit bullets and 2026-10-21
commit 0.01 fix failed test cases for unlimited_ocr model (#49806) Brief, informal message with a standard 2026-10-21
commit 0.01 Fix transformers env when requests is not installed (#48412) Terse practical fix description with a n 2026-10-21
commit 0.01 use torchcodec-xpu and upgrade torch to 2.14.1 (#49814) Short infrastructure update with a stand 2026-10-21
commit 0.01 docs: fix 'allows to' grammar in Speech Encoder Decoder docs Brief, targeted grammar-fix message with 2026-10-21
commit 0.01 docs: fix 'allows to' grammar in DPT docs (#48728) Brief, targeted grammar-fix message with 2026-10-21
commit 0.01 docs: fix 'allows to' grammar in LiLT docs (#48731) Brief, targeted grammar-fix message with 2026-10-21
commit 0.01 docs: fix grammar typo in ALBERT model card (#48774) Short duplicated documentation commit wo 2026-10-21
commit 0.01 docs: fix 'allows to' grammar in Vision Encoder Decoder docs Brief, targeted grammar-fix message with 2026-10-21
commit 0.01 Deprecate old input names (#49424) Terse iterative commit notes and informa 2026-10-21
commit 0.01 add PocketTTS (#49589) Highly specific PocketTTS implementation 2026-10-20
commit 0.01 add PocketTTSAudioTokenizer (#49588) Contains detailed repository-specific ch 2026-10-20
commit 0.01 🟡 [mimi] separate encoder and decoder (#49577) Concise, domain-specific Mimi changes in 2026-10-20
commit 0.01 mimi moshi output capturing (#49670) Short, technically specific commit messa 2026-10-20
commit 0.01 [CI] no split modules for SAM (#49766) Extremely brief and informal commit mess 2026-10-20
commit 0.01 Enable tests forgotten in idefics3 (#49758) Contains informal wording and obvious ty 2026-10-20
commit 0.01 [modular] Extend the "del self.xxx" pattern to match typed a Informal incremental notes and typos ind 2026-10-20
commit 0.01 [Cache] Linear attn cache: use new getters (#49375) Fragmentary model-specific notes and mul 2026-10-20
commit 0.01 [Cache] Get conv states (#49309) Brief technical fragments and shorthand 2026-10-20
commit 0.01 [Cache] Add get_recurrent_state (#49306) Short, domain-specific commit notes with 2026-10-20
commit 0.01 [Cache] Add is_recording_past (#49305) Terse technical wording and the typo 'St 2026-10-20
commit 0.01 Xpu add (#49757) Informal implementation notes and signed 2026-10-20
commit 0.01 Make longrope compileable (#49555) Highly specific compiler and caching det 2026-10-20
commit 0.01 [Kernels] Add MPS to supported device types (#49655) Terse wording, typo, and concrete kernel 2026-10-20
commit 0.01 fix idefics3 generation issues (#49738) Informal terse notes and model-specific 2026-10-20
commit 0.01 Fix BLOOM crash with static cache / compiled generate (#4948 Signed-off, iterative commit notes are t 2026-10-20
pr 0.01 Fix FPS sampling for short, non-empty videos Terse summary with concrete testing deta 2026-10-24
pr 0.01 Add SAM 3.1 to Transformers Informal, candid status assessment and d 2026-05-20
pr 0.01 Fix ProcessorMixin.apply_chat_template dropping explicit pro Concise, issue-specific commit-style des 2026-10-21
pr 0.01 Refactor output tracing for `mamba` Specific model details and output-layout 2026-10-24
pr 0.01 Refactor output tracing for `upernet` Concise, repository-specific description 2026-10-24
pr 0.01 Refactor output tracing for `vision_encoder_decoder` Specific comparison to encoder_decoder a 2026-10-24
pr 0.01 Refactor output tracing for `depth_anything` Brief technical description using reposi 2026-10-24
pr 0.01 [pocket_tts, gradium_phonon_tts] use sub_configs_defaults Minimal title-based description with no 2026-10-22
pr 0.01 :rotating_light: Kernels=True ?! Terse informal checklist with typos and 2026-08-04
pr 0.01 Fix IndexError in _init_weights for padded embeddings under Uses specific DeepSpeed details and info 2026-10-23
pr 0.01 [Fix] Remove duplicate assignment in DeepseekV3.2 config Concise, repository-specific explanation 2026-10-23
pr 0.01 Work around BOS decoding regression in AWQ generation tests Informal, narrowly scoped workaround des 2026-10-23
pr 0.01 add XPU Expectations for 4 new model tests Very terse wording and informal reviewer 2026-10-21
pr 0.01 Make sub-config-defaults easier Brief, personal wording and typo-like ph 2026-10-23
commit 0.00 fix(vision): Route the DETR conversion scripts through from_ Short conventional commit message with r 2026-10-21
commit 0.00 [Docs] Add Marathi (मराठी) translation of README (#48746) Brief documentation commit with concrete 2026-10-21
commit 0.00 fix(audio): Route the ALM conversion scripts through from_pr Terse conventional fix message with a no 2026-10-21
commit 0.00 fix model parallel bug for molmo2 model (#49816) Very terse commit with a standard signed 2026-10-21
commit 0.00 Fix Fuyu processor test (#49821) Extremely terse informal message. 2026-10-21
commit 0.00 [serge] Fix 2 integration tests regressed by commit c48824e8 — 2026-10-21
commit 0.00 [CI] fix perceriver old regression (#49792) Brief wording and typo are characteristi 2026-10-20
commit 0.00 Remove the resolved prototype Whisper MIC expected failure ( A terse maintenance commit with no AI-st 2026-10-20
commit 0.00 update speculative decoding docs (#49697) Commit history includes normal human co- 2026-10-20
commit 0.00 [vLLM] Add Qwen-Omni Thinker base models and Gemma 3n per-la Specific refactoring details and standar 2026-10-20
commit 0.00 Fix stack_frames > 1 wiring bug in MiniCPMV4_6VideoProcessor A concise, specific bug-fix title with n 2026-10-20
pr 0.00 Add default values for downsample_mode parameter and move it Contains only a title and image placehol 2026-10-24
pr 0.00 Explicitly set thinking off in MiniCPM integration tests and Contains only a title and image/code pla 2026-10-24
pr 0.00 [serge] Fix 2 integration tests for model `mamba2` failing w — 2026-10-24
pr 0.00 [serge] Fix 1 integration test for model `deepseek_vl_hybrid — 2026-10-14
pr 0.00 [serge] Fix 8 integration tests for model `grounding_dino` f — 2026-10-22