| commit |
1.00 |
CI: use a single function for GH calls (#47474) |
|
Commit message contains explicit AI assi |
2026-07-28 |
| commit |
1.00 |
Kernels and loaders robustification (#47334) |
|
Commit message contains explicit AI assi |
2026-07-28 |
| commit |
1.00 |
Fix Gemma4 audio feature dtype mismatch in masked_scatter (# |
|
Commit message contains explicit AI assi |
2026-07-23 |
| commit |
1.00 |
[DiffusionGemma] Support gradient checkpointing (#46572) |
|
Commit message contains explicit AI assi |
2026-07-23 |
| commit |
1.00 |
Fix some failed test cases related with XPU Expectations (#4 |
|
Commit message contains explicit AI assi |
2026-07-23 |
| commit |
1.00 |
Fix MoE expert decompression for non-32-divisible bit widths |
|
Commit message contains explicit AI assi |
2026-07-22 |
| commit |
1.00 |
Fix double-shifted training loss in GitForCausalLM (#47395) |
|
Commit message contains explicit AI assi |
2026-07-21 |
| commit |
1.00 |
Fix `TrackioCallback` fails to log evaluation metrics after |
|
Commit message contains explicit AI assi |
2026-07-21 |
| commit |
1.00 |
Fix orthogonal_ init for low-precision dtypes (bf16/fp16) (# |
|
Commit message contains explicit AI assi |
2026-07-17 |
| pr |
1.00 |
[Mistral] Add native tekken tokenizer support to AutoTokeniz |
|
PR body explicitly mentions AI collabora |
2026-07-23 |
| pr |
1.00 |
Fix CodeLlama tokenizer dropping leading whitespace on decod |
|
PR body explicitly mentions AI collabora |
2026-07-22 |
| pr |
1.00 |
Improve Trainer DataLoader Controls for Streaming and Multip |
|
PR body explicitly mentions AI collabora |
2026-07-08 |
| pr |
1.00 |
Fix model parallel device mismatch in `create_bidirectional_ |
|
PR body explicitly mentions AI collabora |
2026-07-26 |
| pr |
1.00 |
Add Unlimited OCR |
|
PR body explicitly mentions AI collabora |
2026-06-23 |
| pr |
1.00 |
[Experimental / Do not Review] Add multi-method export + QNN |
|
PR body explicitly mentions AI collabora |
2026-07-27 |
| pr |
1.00 |
Exportable kimi |
|
PR body explicitly mentions AI collabora |
2026-07-06 |
| pr |
1.00 |
Batch Rebalance Data Sampler |
|
PR body explicitly mentions AI collabora |
2026-07-15 |
| pr |
1.00 |
[DON'T MERGE] check workflow |
|
PR body explicitly mentions AI collabora |
2026-06-18 |
| pr |
1.00 |
Kernels and loaders robustification |
|
PR body explicitly mentions AI collabora |
2026-07-15 |
| pr |
1.00 |
Executorch exporter fixes |
|
PR body explicitly mentions AI collabora |
2026-07-10 |
| pr |
1.00 |
Allow position_ids_start=2 on DataCollatorWithFlattening for |
|
PR body explicitly mentions AI collabora |
2026-07-24 |
| pr |
1.00 |
Add new model `EfficientVit-SAM` to transformers |
|
PR body explicitly mentions AI collabora |
2026-07-12 |
| pr |
1.00 |
Make _added_tokens_decoder the single source of truth for ad |
|
PR body explicitly mentions AI collabora |
2026-07-20 |
| pr |
1.00 |
Fix SentencePieceBackend decoding for byte-fallback tokens |
|
PR body explicitly mentions AI collabora |
2026-07-27 |
| pr |
0.30 |
fix minor docs, testing, and npu detection issues |
|
Slightly more verbose, but still domain- |
2026-07-25 |
| pr |
0.20 |
Add Audio-Visual Flamingo model |
|
More detailed, formal, but mentions doma |
2026-04-22 |
| commit |
0.10 |
Add AXK2 from SKT (#47528) |
|
Commit uses technical terms, informal to |
2026-07-24 |
| commit |
0.10 |
Fix failing tests for mimo_v2_flash (#47284) |
|
Informal and technical; lists specific t |
2026-07-24 |
| commit |
0.10 |
CPU can incur a slow path on non-contiguous magnitudes (#473 |
|
Technical, domain-specific, concise; no |
2026-07-22 |
| commit |
0.10 |
Route chat management calls to the service root (#47138) (#4 |
|
Informative, detailed, with technical ja |
2026-07-22 |
| commit |
0.10 |
Fix gradient inflation when combining label smoothing with g |
|
Analysis is detailed and technical with |
2026-07-22 |
| commit |
0.10 |
[Qwen3ASR] Add hotword parsing, and fix language parsing and |
|
Domain-specific, with informal amendment |
2026-07-22 |
| commit |
0.10 |
Fix multi-device mxfp4 dequantization race in `_convert_moe_ |
|
Technical, signed, multi-author, and dom |
2026-07-22 |
| commit |
0.10 |
Fix CohereASR training-loss double-shift (same as Moonshine |
|
Informal yet technical commit style; spe |
2026-07-21 |
| commit |
0.10 |
Add distributed runtime utils and DistributedMixin (#47352) |
|
Commit history with terse notes suggests |
2026-07-21 |
| pr |
0.10 |
[`FA`] Native torch integration |
|
Free-text uses abbreviations and technic |
2026-03-31 |
| pr |
0.10 |
feat[vLLM]: Add return_text_replacement_offsets to TextKwarg |
|
Technical references and terse summariza |
2026-07-28 |
| pr |
0.10 |
Also size the device_map buffer from the largest leaf module |
|
Contains commit context and technical de |
2026-07-09 |
| pr |
0.10 |
[`Kernels`] Refactor function handling |
|
Contains domain context and shorthand, i |
2026-06-25 |
| pr |
0.10 |
Resolve the Hub revision once per load instead of passing a |
|
References issue numbers and uses domain |
2026-07-28 |
| pr |
0.10 |
[generation] Encode multimodal data only once |
|
Brief and domain-relevant, with project- |
2026-05-05 |
| pr |
0.10 |
Modularize qwen-format vision processors |
|
Technical title, minimal free-text, and |
2026-07-27 |
| pr |
0.10 |
Add max_thinking_tokens for reasoning models (issue #42111) |
|
Direct, domain-specific phrasing with mi |
2025-11-09 |
| pr |
0.10 |
Support Granite Speech NAR (NLE) |
|
Project-specific acronym usage and conci |
2026-05-18 |
| pr |
0.10 |
[nit] use requires_backends |
|
Brief, domain-relevant change, human-lik |
2026-07-27 |
| pr |
0.10 |
CI: Add serge review relay workflow and review rules |
|
Contains systems context and human troub |
2026-07-28 |
| pr |
0.10 |
[Quantization] Fix CT `apply_quantization_config` |
|
Contains domain-specific terms and brief |
2026-07-23 |
| pr |
0.10 |
fix(cli): Fix TypeAdapter NameError when pydantic is not ins |
|
Short technical description; minimal AI |
2026-02-08 |
| commit |
0.05 |
Fix the HunyuanVL's torchvision backend (#47499) |
|
Terse technical explanation with specifi |
2026-07-23 |
| commit |
0.05 |
Use new `per_layer_config` for Gemma 4 so that heterogeneous |
|
Technical language, direct phrasing, cle |
2026-07-23 |
| commit |
0.05 |
Fix failing tests for zaya (#47268) |
|
Concise commit style, lists failures, st |
2026-07-23 |
| commit |
0.05 |
tipsv2_dpt: fix failing tests for XPU (#47292) |
|
Structured, terse, and domain-specific l |
2026-07-23 |
| commit |
0.05 |
add paged attention tests support for XPU (#47163) |
|
Standard commit format, domain terms, na |
2026-07-23 |
| commit |
0.05 |
Add FSDP CI and end-to-end FSDP tests + save fsdp (#47357) |
|
Highly technical detail, natural human c |
2026-07-23 |
| commit |
0.05 |
fix: guard DTensor import in sharding_utils.py for PyTorch < |
|
Direct, technical fix description with d |
2026-07-23 |
| commit |
0.05 |
[Quantization]: Refactor is_quantization_compressed for form |
|
Standard commit format, technical and co |
2026-07-22 |
| commit |
0.05 |
CI: add serge verify (GPU) caller workflow (#47381) |
|
Highly technical, terse, natural domain |
2026-07-17 |
| commit |
0.05 |
ci: harden daily CI reporting against GitHub API rate limits |
|
Informal, domain-specific, includes erro |
2026-07-17 |
| commit |
0.00 |
Fix model parallel device mismatch in `create_bidirectional_ |
|
Technical, concise, uses domain-specific |
2026-07-28 |
| commit |
0.00 |
Fix A.X-K2 fp8 modules_to_not_convert normalization for the |
|
Detailed technical changelog, clear doma |
2026-07-28 |
| commit |
0.00 |
Remove redundant guarding for distributed (#47570) |
|
Terse, domain jargon, and contains typos |
2026-07-28 |
| commit |
0.00 |
Add FSDP plans to all models (#47165) |
|
Terse, technical edits, includes typos/a |
2026-07-28 |
| commit |
0.00 |
Fix modular for mamba packages (#47494) |
|
Very brief, informal commit messages lac |
2026-07-28 |
| commit |
0.00 |
Remove deprecated conversion in Kimi (#47581) |
|
Very terse, informal commit message; cle |
2026-07-27 |
| commit |
0.00 |
CI: use transformers-ci daily workflow with OTEL (#47360) |
|
Terse, informal style with domain-specif |
2026-07-27 |
| commit |
0.00 |
Fix slow tensor path in _check_special_mm_tokens (#47580) |
|
Terse, technical; no AI hallmarks presen |
2026-07-27 |
| commit |
0.00 |
CI: let's run integration failure cron at 10pm (#47537) |
|
Casual language and repetition; human st |
2026-07-27 |
| commit |
0.00 |
Better and more extensive tests for RoPE (#46912) |
|
Informal, fragmented lines; typical huma |
2026-07-27 |
| commit |
0.00 |
Fix mamba2 family decode and simplify all reshape ops (#4756 |
|
Minimal, vague wording; likely human. |
2026-07-27 |
| commit |
0.00 |
Speed up image preprocessing for vision-language models (#47 |
|
Contains domain-specific detail, no AI p |
2026-07-27 |
| commit |
0.00 |
[DiffusionGemma] Cast the decoder padding mask to bool (#472 |
|
Terse, technical; fixes described in a h |
2026-07-27 |
| commit |
0.00 |
Fix NemotronH: Register `"mlp"` in the cache layer-type mapp |
|
Terse, technical, human co-author traile |
2026-07-27 |
| commit |
0.00 |
Fix `config.per_layer_config[layer_type]` lookup speed (#475 |
|
Short, technical; signed off by a human. |
2026-07-27 |
| commit |
0.00 |
General maintenance (#47517) |
|
Brief, informal, lacks AI hallmarks; cle |
2026-07-27 |
| commit |
0.00 |
:rotating_light: Processors update the rest (#46556) |
|
Very informal language, typos, and perso |
2026-07-27 |
| commit |
0.00 |
fixed the benchmark script with DistributedConfig (#47568) |
|
Direct, brief phrasing; contains a typo, |
2026-07-27 |
| commit |
0.00 |
Fix TP inference for tied embedding (#47503) |
|
Terse commit messages, technical jargon, |
2026-07-27 |
| commit |
0.00 |
Fix `causal_conv1d_fn` positional `activation` colliding wit |
|
Short, technical fix message; no AI sign |
2026-07-24 |
| commit |
0.00 |
Refactor all linear attention models to latest best standard |
|
Fragmented, terse commit messages with t |
2026-07-24 |
| commit |
0.00 |
Make tokenization_mistral_common importable without mistral_ |
|
Very short, purely technical; no AI indi |
2026-07-24 |
| commit |
0.00 |
Use --flake-runs=1 in check_bad_commit.py for PR comment CI |
|
Terse, technical; email in trailer, but |
2026-07-24 |
| commit |
0.00 |
Deprecate the old response_schema (#47320) |
|
Direct, technical, no boilerplate or AI- |
2026-07-24 |
| commit |
0.00 |
fix: add pickle support to _LazyConfigMapping for spawn mult |
|
Contains personal signature, technical p |
2026-07-24 |
| commit |
0.00 |
Fix potential ReDoS by escaping tokenizer filename used as r |
|
Technical language, specific context, no |
2026-07-24 |
| commit |
0.00 |
Fix vision position-embedding init width fallback in Phi4Mul |
|
Very brief, technical, no signals of AI- |
2026-07-24 |
| commit |
0.00 |
Delete old deprecations (#47518) |
|
Commit messages are short, informal, and |
2026-07-24 |
| commit |
0.00 |
Run only @slow tests in PR comment CI (#47521) |
|
Terse commit messages, clear human domai |
2026-07-24 |
| commit |
0.00 |
Fix incorrect type hint (#47519) |
|
Simple, human-authored commit message an |
2026-07-24 |
| commit |
0.00 |
Deprecate CB config in gen configuration (#47291) |
|
Quick, shorthand language and typos indi |
2026-07-24 |
| commit |
0.00 |
Allow metal-flash-sdpa for OpenAIPrivacyFilter on MPS (#4674 |
|
Casual, abbreviated style is clearly hum |
2026-07-24 |
| commit |
0.00 |
[Offloading] [Bugfix] Fix fully offloaded model saving (#473 |
|
Concise, domain-specific commit messages |
2026-07-23 |
| commit |
0.00 |
CI: fix torchaudio pinning +proper break in rnnt (#47422) |
|
Informal, terse commit messages with tec |
2026-07-23 |
| commit |
0.00 |
add_axk1 (#46867) |
|
Technical, domain-focused, with typical |
2026-07-23 |
| commit |
0.00 |
Fix Qwen2.5-Omni Token2Wav DiT rotary embedding layout (inte |
|
Detailed, technical explanation with jar |
2026-07-23 |
| commit |
0.00 |
CI: add reproduce mode to serge verify caller (#47492) |
|
Technical writing, straightforward expla |
2026-07-23 |
| commit |
0.00 |
feat[vLLM x v5]: Make audio optional and support multiple au |
|
Typical commit subject/body, no AI phras |
2026-07-23 |
| commit |
0.00 |
[fix][whisper]: fix max_new_tokens handling (#46795) |
|
Brief, domain-specific fix; no sign of A |
2026-07-23 |
| commit |
0.00 |
Isolate MLA KV expansion to make it easier to bypass (#47460 |
|
Technical commit history with human styl |
2026-07-23 |
| commit |
0.00 |
Update serve chat parsing (#46267) |
|
Terse, domain-internal phrasing; no form |
2026-07-23 |
| commit |
0.00 |
Fix loss alignment and Trainer token counting for encoder de |
|
Technical, detailed, full of context and |
2026-07-23 |
| commit |
0.00 |
[fix] fix requirements audio feature and proc (#47113) |
|
Very terse, telegraphic, and domain-like |
2026-07-22 |
| commit |
0.00 |
fix: liger unnecessarily materializes logits in VRAM during |
|
Single line, domain-specific, terse; no |
2026-07-22 |
| commit |
0.00 |
Remove deprecated training args and `is_fast` property (#469 |
|
Terse, domain-standard, includes sign-of |
2026-07-22 |
| commit |
0.00 |
Consistent output shape from `get_image_features` (#46405) |
|
One-word lines and shorthand, highly inf |
2026-07-22 |
| commit |
0.00 |
fix failed test cases for qwen3_omni_moe model (#47449) |
|
Commit messages are brief, specific, and |
2026-07-22 |
| commit |
0.00 |
Fix Hunyuan-VL PIL image resize parity with reference prepro |
|
Very terse human-like commit messages wi |
2026-07-22 |
| commit |
0.00 |
Move `value` padding into the attention interfaces that need |
|
Human-typical, terse, domain-specific co |
2026-07-22 |
| commit |
0.00 |
Simplify function dispatch for linear attention (#47450) |
|
Short, informal, domain-specific commit |
2026-07-22 |
| commit |
0.00 |
[cache] Allow sliding window layers to be roll-backed for sp |
|
Informal, specific, and human-style comm |
2026-07-22 |
| commit |
0.00 |
Warn when `group_by_length` is silently ignored for iterable |
|
Concise, terse commit message typical of |
2026-07-21 |
| commit |
0.00 |
Update bug report list (#46607) |
|
Short, human-like commit message and che |
2026-07-21 |
| commit |
0.00 |
Fix shape mismatch in KyutaiSpeechToText `generate()` last w |
|
Terse commit messages and use of signed- |
2026-07-21 |
| commit |
0.00 |
Optimize flash attention max seqlen computation in vision at |
|
Terse changelog messages with human-like |
2026-07-21 |
| commit |
0.00 |
fix: remove unreachable return in special token builder (#47 |
|
Minimal, clear, informal message as is c |
2026-07-21 |
| commit |
0.00 |
Add Harry to slow CI (#47454) |
|
Single, brief, informal comment typical |
2026-07-21 |
| commit |
0.00 |
BLT: vectorize patch length processing (#47385) |
|
Changelog-style terse message with domai |
2026-07-21 |
| commit |
0.00 |
[Kimi] add integration tests (#47383) |
|
Terse style with non-standard capitaliza |
2026-07-21 |
| commit |
0.00 |
Fix typo in `MusicgenForCausalLM.generate()` (#46974) |
|
Minimal, concise changelog entries with |
2026-07-21 |
| commit |
0.00 |
Fix Cosmos 3 Edge Patch packing order (#47399) |
|
Commit messages are terse, technical, an |
2026-07-20 |
| commit |
0.00 |
Fix yarn `mscale_all_dim` for DeepSeek v2 and Mistral 4 (#47 |
|
Commit messages are concise, domain-spec |
2026-07-20 |
| commit |
0.00 |
Fix recurrent-layer padding mask being skipped on continued |
|
Technical explanations with natural edit |
2026-07-20 |
| commit |
0.00 |
Hoist special-token lookups out of per-token loops in six sl |
|
Succinct, domain-jargon usage, no signal |
2026-07-20 |
| commit |
0.00 |
fix typos and variable naming in quicktour.md (#47418) |
|
Typical human edit messages and casual l |
2026-07-20 |
| commit |
0.00 |
Normalize multimodal input keys in AnyToAnyPipeline (#47074) |
|
Domain-specific, includes typical commit |
2026-07-20 |
| commit |
0.00 |
Fix Aria checkpoint key conversion mapping (#47151) |
|
Detailed technical note, domain-specific |
2026-07-20 |
| commit |
0.00 |
extend tests/models/qwen3_next/test_modeling_qwen3_next.py:: |
|
Commit is technical, includes email, no |
2026-07-20 |
| commit |
0.00 |
Fix `BlockMask` crash in CTRL flex-attention generation (#46 |
|
Extremely terse, cryptic commit message; |
2026-07-17 |
| commit |
0.00 |
[docs] Fix decode examples and expected output in fast_token |
|
Minimal, typical doc fix, no sign of AI |
2026-07-17 |
| commit |
0.00 |
Fix image-text-to-text stop_sequence handling (#47032) |
|
Standard fix, includes human co-author, |
2026-07-17 |
| commit |
0.00 |
[serge] Fix 20 integration tests for model `whisper` failing |
|
— |
2026-07-17 |
| pr |
0.00 |
[serge] Fix 2 integration tests for model `regnet` failing w |
|
— |
2026-07-28 |
| pr |
0.00 |
Fix CANINE dynamic sequence length export (torch.export/ONNX |
|
Free-text is technical, uses domain jarg |
2026-07-28 |
| pr |
0.00 |
Don't warn about class defaults when diffing a config |
|
Uses specific technical explanation, nat |
2026-07-28 |
| pr |
0.00 |
Gate eager distributed imports on torch.distributed.is_avail |
|
Technical explanation, concise and natur |
2026-07-28 |
| pr |
0.00 |
Add chat input sanitization |
|
No free-text; only template and minimal |
2026-07-17 |
| pr |
0.00 |
Add Granite-swa and Granitemoe-swa model support |
|
Technical, domain-specific, incomplete t |
2026-07-08 |
| pr |
0.00 |
wip step 3 7 flash |
|
Casual, domain-specific and minimal; no |
2026-06-15 |
| pr |
0.00 |
fix(modeling_utils): remove `is_custom_code` gate from per-p |
|
Short, technical, and informal; typical |
2026-07-28 |
| pr |
0.00 |
[Chat Parsing] Type inline tool-call arguments from the call |
|
Direct technical phrasing, no AI-style f |
2026-07-24 |
| pr |
0.00 |
🚨 Enable SDPA (and other attention backends) for T5 and prop |
|
Emoji and directness; casual, not AI-sty |
2026-07-02 |
| pr |
0.00 |
Fix CohereAsr training-loss double-shift (train against labe |
|
Technical and detailed, terse and human- |
2026-07-05 |
| pr |
0.00 |
Add Fun-ASR-Nano model |
|
Minimal, direct, and typical of human re |
2026-05-24 |
| pr |
0.00 |
Port ESMC and ESMFold2 to Transformers |
|
Only technical terms in the title; no AI |
2026-06-04 |
| pr |
0.00 |
Compressed tensors fp8 |
|
Very brief and technical, not AI-formal |
2026-07-09 |
| pr |
0.00 |
byebye torch 2.4 |
|
Extremely terse and informal, clear huma |
2026-07-28 |
| pr |
0.00 |
add HyperClovaX Vision |
|
Title only, no free-text and typical ter |
2026-02-27 |
| pr |
0.00 |
[DO NOT merge!] test migration |
|
Explicit test PR, minimal and informal. |
2026-05-20 |
| pr |
0.00 |
Fix some processors |
|
Only a terse title, no signs of AI gener |
2026-07-28 |
| pr |
0.00 |
Fix dtype casting when dequantizing |
|
Single technical title, no AI indicators |
2026-07-28 |
| pr |
0.00 |
CI: use a single function for GH calls |
|
Brief technical title, clearly human-mad |
2026-07-22 |
| pr |
0.00 |
QA: mlinter rule 20 |
|
Minimal, domain-specific message; not AI |
2026-07-25 |
| pr |
0.00 |
Fix CUDA Graph breaking host to device copy from scalar tens |
|
Short, domain-focused title; not AI-gene |
2026-07-25 |
| pr |
0.00 |
Fix A.X-K2 fp8 modules_to_not_convert normalization for the |
|
Technical and concise, no AI signals. |
2026-07-27 |
| pr |
0.00 |
[serge] Fix 2 integration tests for model `granite4_vision` |
|
— |
2026-07-28 |
| pr |
0.00 |
🚨 Simplify all Rotary modules |
|
Title uses emoji and domain jargon, huma |
2026-07-28 |
| pr |
0.00 |
🚨🚧 FeatureExtractor → AudioProcessor |
|
Human-like checklist with terse, domain |
2026-03-02 |
| pr |
0.00 |
[serge] Fix 4 integration tests for model `hunyuan_vl` faili |
|
— |
2026-07-28 |
| pr |
0.00 |
Align OlmoHybrid to use a native cache |
|
Very terse technical title, human-author |
2026-07-28 |
| pr |
0.00 |
Support solar-open2 model |
|
Very brief and technical title, not AI-g |
2026-07-27 |
| pr |
0.00 |
Enable MTP support for Qwen3.5 |
|
One-line technical title, typical human |
2026-04-24 |
| pr |
0.00 |
Fix MistralCommonBackend autodoc source path when mistral-co |
|
Direct technical fix, lacks AI-like form |
2026-07-24 |
| pr |
0.00 |
skip fsdp tests when backend is mps |
|
Only a technical title present; no AI ha |
2026-07-28 |
| pr |
0.00 |
Remove `is_custom_code` gate from per-param `_is_hf_initiali |
|
Technical, terse title; no free-text or |
2026-07-15 |
| pr |
0.00 |
[Olmo3] different RoPE per layer type |
|
Brief, colloquial description with inlin |
2026-06-26 |
| pr |
0.00 |
Potential fix for code scanning alert no. 265: Environment v |
|
Technical, short title with no free-text |
2026-07-28 |
| pr |
0.00 |
Fix prompt lookup decoding length cap |
|
Concise title only, no AI-style language |
2026-07-09 |
| pr |
0.00 |
Fix incorrect documented default for downsample_in_first_sta |
|
Technical correction title, clearly huma |
2026-07-27 |
| pr |
0.00 |
model: Add NVIDIA Canary-1B-v2 to Transformers |
|
Domain-specific, factual title; lacks AI |
2026-06-22 |
| pr |
0.00 |
Fix position misalignment when past_key_values is externally |
|
Terse, technical title, no generic phras |
2026-07-28 |
| pr |
0.00 |
Add LoMa keypoint matching model |
|
Template and technical title only; free |
2026-07-18 |
| pr |
0.00 |
Vectorize NoRepeatNGramLogitsProcessor and remove its host s |
|
Technical, succinct, with clear domain l |
2026-07-27 |
| pr |
0.00 |
feat[vLLM]: Make VoxtralProcessor compatible with the Transf |
|
Concise, domain-specific and informal, n |
2026-07-27 |
| pr |
0.00 |
TP dtensor inference |
|
Technical abbreviation in title, no AI h |
2026-07-27 |
| pr |
0.00 |
feat[vLLM]: Make VoxtralProcessor compatible with the Transf |
|
Concise, domain-specific and informal, n |
2026-07-27 |
| pr |
0.00 |
Fix Pix2StructTextAttention init using hidden_size instead o |
|
Technical fix title only, no AI phrasing |
2026-07-26 |
| pr |
0.00 |
Remove redundant guarding for distributed |
|
Brief technical change title, no AI hall |
2026-07-27 |
| pr |
0.00 |
Pipeline parallel naive inference |
|
Succinct title; informal and technical l |
2026-07-13 |
| pr |
0.00 |
Add FSDP plans to all models |
|
Concise technical addition, domain-speci |
2026-07-08 |
| pr |
0.00 |
TP dtensor |
|
Technical abbreviation in title; clearly |
2026-07-09 |
| pr |
0.00 |
Fix modular for mamba packages |
|
Minimal domain-specific title, not indic |
2026-07-23 |
| pr |
0.00 |
Fix batched Qwen2.5/3-Omni audio generation |
|
Only PR template boilerplate and a techn |
2026-07-09 |
| pr |
0.00 |
feat(data_collator): add position_ids_start parameter to Dat |
|
PR template and technical title, free te |
2026-07-27 |
| pr |
0.00 |
fix(masking_utils): build bidirectional sliding window mask |
|
Title and template only; content is tech |
2026-07-27 |
| pr |
0.00 |
fix(models/wav2vec2): optimize special tokens lookups in tok |
|
Purely technical title within a template |
2026-07-27 |
| pr |
0.00 |
fix(models/phi3,phimoe): default logits_to_keep to 1 when No |
|
Technical title and template only, no AI |
2026-07-27 |
| pr |
0.00 |
fix(models/mamba2,nemotron_h): squeeze sequence dimension in |
|
Message is a terse technical fix with no |
2026-07-27 |
| pr |
0.00 |
Add heterogeneous model support (per-layer config and modeli |
|
Technical PR with only a succinct title, |
2026-04-09 |
| pr |
0.00 |
Return feature attention masks from Wav2Vec2-derived models |
|
PR title only; domain-specific and conci |
2026-07-25 |
| pr |
0.00 |
Fix GPT-2 c_proj depth scaling initialization |
|
Technical, brief title; no signs of AI-g |
2026-07-21 |
| pr |
0.00 |
fix npu check |
|
Extremely terse, technical style indicat |
2026-07-27 |
| pr |
0.00 |
Add MXFP4 training support via a dequantize-on-the-fly fallb |
|
Only contains a technical PR title, no A |
2026-07-21 |
| pr |
0.00 |
[DiffusionGemma] Avoid recomputing token entropy in the stop |
|
Short technical title, no AI markers pre |
2026-07-27 |
| pr |
0.00 |
Clean up deprecated legacy code for v5 |
|
Brief, domain-specific title, no AI hall |
2026-07-27 |
| pr |
0.00 |
Align the MXFP4 `kernels` version message with the version a |
|
Human-style, concise, with technical spe |
2026-07-27 |
| pr |
0.00 |
regression test covering the mlx-community/MiniCPM5-1B-8bit |
|
Repository-specific regression test titl |
2026-07-26 |
| pr |
0.00 |
Remove deprecated conversion in Kimi |
|
Succinct title about code cleanup, no si |
2026-07-27 |
| pr |
0.00 |
Reuse resolved Hub commits throughout loading |
|
Short technical summary title, no AI ind |
2026-07-27 |
| pr |
0.00 |
docs: fix typo in image_processors.md |
|
Very terse human-written PR title, no AI |
2026-07-25 |
| pr |
0.00 |
docs: fix typo in kimi_k25.md |
|
Terse changelog-style doc fix, fully hum |
2026-07-25 |
| pr |
0.00 |
docs: fix typo in gptq.md |
|
Short doc typo fix title, standard human |
2026-07-25 |
