| commit |
1.00 |
DeepGEMM BF16 + mixed FP8/FP4 + MegaMoE + refactor (#45634) |
|
Commit message contains explicit AI assi |
2026-06-02 |
| commit |
1.00 |
[Qwen3VL] Fix video token placeholder: use self.video_token |
|
Commit message contains explicit AI assi |
2026-06-02 |
| commit |
1.00 |
Fix fp8 moe reverse substring (#46265) |
|
Commit message contains explicit AI assi |
2026-06-01 |
| commit |
1.00 |
fix: harden check_bad_commit.py against GitHub API rate limi |
|
Commit message contains explicit AI assi |
2026-06-01 |
| commit |
1.00 |
Fix TAPAS tokenizer crash on pandas 3.x string-dtype tables |
|
Commit message contains explicit AI assi |
2026-05-28 |
| commit |
1.00 |
[Gemma4] Replace one-hot matmul with F.embedding in position |
|
Explicit AI-assisted statement: 'AI-assi |
2026-05-28 |
| commit |
1.00 |
Fix `num_items_in_batch` over-counting for causal LM losses |
|
Commit message contains explicit AI assi |
2026-05-27 |
| commit |
1.00 |
Fix is_last off-by-one in MaskGenerationPipeline for partial |
|
Commit message contains explicit AI assi |
2026-05-25 |
| commit |
1.00 |
Add new cohere2_moe model (#46115) |
|
Commit message contains explicit AI assi |
2026-05-20 |
| pr |
1.00 |
check |
|
PR body explicitly mentions AI collabora |
2026-05-30 |
| pr |
1.00 |
Test models instantiate from default config on meta + fix fa |
|
PR body explicitly mentions AI collabora |
2026-06-02 |
| pr |
1.00 |
DeepGEMM BF16 + mixed FP8/FP4 + MegaMoE + refactor |
|
PR body explicitly mentions AI collabora |
2026-04-24 |
| pr |
1.00 |
Fix flip_back graph break |
|
PR body explicitly mentions AI collabora |
2026-06-02 |
| pr |
1.00 |
Add Sapiens2 Model |
|
PR body explicitly mentions AI collabora |
2026-05-12 |
| pr |
1.00 |
fix(llama4): align MoE interface for EP/TP compatibility |
|
PR body explicitly mentions AI collabora |
2026-05-18 |
| pr |
1.00 |
Fix grammar typos in Whisper documentation |
|
PR body explicitly mentions AI collabora |
2026-06-02 |
| pr |
1.00 |
Propagation of output_hidden_states in sub_configs |
|
PR body explicitly mentions AI collabora |
2026-06-02 |
| pr |
1.00 |
docs: customizing models.md, better comments for resnetconfi |
|
PR body explicitly mentions AI collabora |
2026-06-01 |
| pr |
1.00 |
Return logits from semantic segmentation post-process |
|
PR body explicitly mentions AI collabora |
2026-05-22 |
| pr |
1.00 |
Fix path traversal when saving Bark voice preset embeddings |
|
PR body explicitly mentions AI collabora |
2026-05-27 |
| pr |
1.00 |
fix: include transitive relative imports when loading from l |
|
PR body explicitly mentions AI collabora |
2026-05-18 |
| pr |
1.00 |
perf(feature_extraction_sequence): skip re-splitting already |
|
PR body explicitly mentions AI collabora |
2026-06-01 |
| pr |
1.00 |
Fix use_cache with seq_len > 1 ( #46032) |
|
PR body explicitly mentions AI collabora |
2026-05-19 |
| pr |
1.00 |
Added cosmos3 model |
|
PR body explicitly mentions AI collabora |
2026-05-21 |
| pr |
1.00 |
Add ctsm model |
|
PR body explicitly mentions AI collabora |
2026-04-17 |
| pr |
0.50 |
add HyperClovaX Vision |
|
The free-text starts with a formal greet |
2026-02-27 |
| pr |
0.30 |
docs: update ACL Anthology URL in CITATION.cff |
|
Polite full sentence, but not typical AI |
2026-06-02 |
| pr |
0.20 |
Feat/executorch vision audio export |
|
Minimal free-text, some formal tone but |
2026-06-02 |
| pr |
0.20 |
fix(tokenization): avoid O(T·N·logN) performance regression |
|
Minimal human-written summary, not overl |
2026-06-02 |
| pr |
0.20 |
[docs] Romanian translation of `fast_tokenizers.md`, `custom |
|
Standard summary, slightly formal, but n |
2026-06-02 |
| pr |
0.20 |
Adding Support for Qwen3.5 |
|
Slightly formal thanks but overall conci |
2026-02-08 |
| pr |
0.20 |
Add Fun-ASR-Nano model |
|
Well-structured but domain-heavy; still |
2026-05-24 |
| pr |
0.20 |
Support for BharatGen's Param2MoE model architecture |
|
Technical, concise description with doma |
2026-02-10 |
| commit |
0.10 |
fix: include transitive relative imports when loading from l |
|
Terse, domain-specific; no AI hallmarks. |
2026-06-02 |
| commit |
0.10 |
perf(feature_extraction_sequence): skip re-splitting already |
|
Technical, concise, uses domain terminol |
2026-06-02 |
| commit |
0.10 |
Fix TestAppRoutes test failures caused by deprecated asyncio |
|
Detailed technical explanation, not over |
2026-06-02 |
| pr |
0.10 |
[docs] gguf mps |
|
Casual, domain-specific with brief expla |
2026-06-02 |
| pr |
0.10 |
GgufLinear: inference-time GGUF matmul on Apple Silicon — ll |
|
Technical abbreviations, informal struct |
2026-05-14 |
| pr |
0.10 |
[docs] distributed training |
|
Concise with domain terms, terse bullet |
2026-03-03 |
| pr |
0.10 |
Fix convert_tokens_to_ids performance regression for slow to |
|
Uses domain-specific phrasing, reference |
2026-06-01 |
| pr |
0.10 |
Fix incorrect attribute mapping relationships in GLM MoE DSA |
|
Technical description, some markdown, bu |
2026-06-02 |
| pr |
0.10 |
Add FP8 kernel acceleration for compressed-tensors quantized |
|
Technical, some structure but no AI cue |
2026-04-29 |
| pr |
0.10 |
Update compressed tensors minimum version |
|
Domain details, minor typos; normal huma |
2026-06-02 |
| pr |
0.10 |
Fix _is_package_available reporting available without a vers |
|
Technical, to the point; structure and w |
2026-05-20 |
| pr |
0.10 |
Streamable chat parsing |
|
Somewhat explanatory, but informal and u |
2026-05-08 |
| pr |
0.10 |
Fix models for which we don't have a dedicated tokenizer cla |
|
Brief, informal note; clear domain refer |
2026-05-13 |
| pr |
0.10 |
Export is_falcon_mamba_ssm_available from utils |
|
Domain-specific, concise, and directly r |
2026-05-29 |
| pr |
0.10 |
Fix: strip EOS token when continue_final_message=True in Mis |
|
Mentions specific internal logic, not ge |
2026-06-01 |
| pr |
0.10 |
[Zamba] Support attn_implementation dispatch |
|
Technical troubleshooting, realistic phr |
2026-06-01 |
| pr |
0.10 |
Validate Bark speaker embedding save paths |
|
Technical, specific to the codebase, and |
2026-06-01 |
| pr |
0.10 |
docs: add docstrings to create_sinusoidal_embeddings in Dist |
|
Simple, direct summary of docstring chan |
2026-05-30 |
| pr |
0.10 |
fix(moonshine_streaming): include full right window in slidi |
|
Module-specific, focused, and uses code |
2026-05-31 |
| pr |
0.10 |
feat(generation): add GenerationActivations utility for stac |
|
Concise, technical summary; references l |
2026-05-31 |
| pr |
0.10 |
DOC Fix help for chat and serve commands |
|
Correction about argument documentation |
2025-09-15 |
| pr |
0.10 |
Improve dinos |
|
Direct update, technical, references PRs |
2026-05-28 |
| pr |
0.10 |
[CB] [Major] Rework manager to have clearer control flow + h |
|
Direct technical changelog, clear domain |
2026-05-19 |
| pr |
0.10 |
Fix hub kernels compatibility with kernels>=0.15.0 |
|
Includes etiquette from template but con |
2026-05-29 |
| pr |
0.10 |
Add Deepseek-OCR-2 model |
|
Standard, factual domain summary; no AI |
2026-03-27 |
| pr |
0.10 |
[`Kernels`] Sync to latest version and add new kernels (SwiG |
|
WIP, technical, with realistic casual bu |
2026-05-18 |
| pr |
0.10 |
Normalize CUDA OOM errors when comparing commit failures in |
|
Technical, specific, and domain-relevant |
2026-06-01 |
| pr |
0.10 |
Fix unhandled exception noise from background safetensors co |
|
Contains error names and code context; n |
2026-05-03 |
| pr |
0.10 |
clip_grad_norm avoid O(global_params) replication |
|
Technical followup with domain abbreviat |
2026-05-20 |
| pr |
0.10 |
Add MTP speculative decoding via MTPCandidateGenerator |
|
Domain-specific context and function nam |
2026-04-24 |
| pr |
0.10 |
fix: cache warmup `RuntimeError` on mps |
|
Explains platform-specific issue informa |
2026-05-27 |
| pr |
0.05 |
FIX Restore LoRA hotswapping functionality |
|
Brief, issue numbers, domain-specific—hu |
2026-04-28 |
| pr |
0.05 |
Fix missing f-string prefixes in error messages |
|
Technical, specific details, no AI phras |
2026-06-02 |
| pr |
0.05 |
[Mistral] Raise ValueError when continue_final_message=True |
|
Uses domain terms, concise, no AI marker |
2026-06-02 |
| pr |
0.05 |
fix(moe): normalize auxiliary loss by top_k for correct load |
|
Technical focus, technical vocab, not ge |
2026-02-05 |
| pr |
0.05 |
Fix noisy logging from image_processing module aliases issue |
|
Problem-specific detail, natural tone, t |
2026-06-02 |
| pr |
0.05 |
Add some AITER kernel routing for ROCm |
|
Technical, plenty of domain abbreviation |
2026-05-28 |
| pr |
0.05 |
feat(quantizers): support non-weight param names in TorchAo |
|
Technical and terse; no AI phrasing or u |
2026-06-01 |
| pr |
0.05 |
fix: convert to fp32 first when doing image resize |
|
Technical content with specific context; |
2026-05-28 |
| pr |
0.05 |
Quantize models on `__init__` / `from_config` |
|
Domain-specific terminology and technica |
2026-06-01 |
| pr |
0.05 |
Pass library_name/version to Hub calls via a shared HfApi |
|
Direct language, references code and iss |
2026-06-01 |
| pr |
0.05 |
fix: remove dynamic per-actor Slack ID lookup in ssh-runner |
|
Structured commit, technical with specif |
2026-06-01 |
| pr |
0.05 |
Add cache-aware streaming and RNN-T head for Parakeet |
|
Domain jargon and structured points, no |
2026-05-20 |
| pr |
0.05 |
fix: harden check_bad_commit.py against GitHub API rate limi |
|
Technical breakdown, casual intro and sp |
2026-06-01 |
| commit |
0.00 |
docs: update ACL Anthology URL in CITATION.cff (#46352) |
|
Concise commit, human style, no AI signa |
2026-06-02 |
| commit |
0.00 |
[docs] contributing (#45465) |
|
Shows iterative commits and short notes, |
2026-06-02 |
| commit |
0.00 |
[docs] Romanian translation of `contributing.md`, `modular_t |
|
List of translation tasks with terse com |
2026-06-02 |
| commit |
0.00 |
[docs] xpu continuous batching (#46334) |
|
Brief and domain-specific; simple human |
2026-06-02 |
| commit |
0.00 |
Fix incorrect attribute mapping relationships in GLM MoE DSA |
|
Includes human signature, specific conte |
2026-06-02 |
| commit |
0.00 |
Fix grammar typos in Whisper documentation (#46336) |
|
Co-authored commit, minimal human style |
2026-06-02 |
| commit |
0.00 |
[docs] update num_items_in_batch for causal LMs (#46335) |
|
Short, domain-specific, lacks AI text tr |
2026-06-02 |
| commit |
0.00 |
Update compressed tensors minimum version (#46342) |
|
Terse, task-focused, co-author listed, n |
2026-06-02 |
| commit |
0.00 |
Fix _is_package_available reporting available without a vers |
|
Detailed technical explanation, domain j |
2026-06-02 |
| commit |
0.00 |
remove sec (#46346) |
|
Very terse commit, co-author, no AI indi |
2026-06-02 |
| commit |
0.00 |
[Zamba] Support attn_implementation dispatch (#46317) |
|
Fragmented, informal commit summaries. |
2026-06-02 |
| commit |
0.00 |
[CB] [Major] Rework manager to have clearer control flow + h |
|
Very terse, informal fragment entries. |
2026-06-02 |
| commit |
0.00 |
Add Deepseek-OCR-2 model (#45075) |
|
Changelog-only PR with terse, technical |
2026-06-01 |
| commit |
0.00 |
chore(linter): fixes for rule 16 (#46023) |
|
Terse, incomplete commit messages; domai |
2026-06-01 |
| commit |
0.00 |
[docs] Romanian translation of `weightconverter.md`, `mode |
|
Concise changelog entries; no AI-like ph |
2026-06-01 |
| commit |
0.00 |
Normalize CUDA OOM errors when comparing commit failures in |
|
Single word, terse commit; human signed. |
2026-06-01 |
| commit |
0.00 |
Fix unhandled exception noise from background safetensors co |
|
Concise technical fix description, co-au |
2026-06-01 |
| commit |
0.00 |
fix: cache warmup `RuntimeError` on mps (#46239) |
|
Minimal commit summary with inline code, |
2026-06-01 |
| commit |
0.00 |
Add Expectations for pipeline token classification tests (#4 |
|
Terse changelog entries and valid email |
2026-06-01 |
| commit |
0.00 |
[docs] fix auto-add release dates (#46283) |
|
Terse commit messages and domain jargon; |
2026-06-01 |
| commit |
0.00 |
🚨 Fix float16 overflow in Gemma4 vision pooler (#46277) |
|
Highly technical, domain-specific and de |
2026-06-01 |
| commit |
0.00 |
Make more tests work with read-only cache (#46299) |
|
— |
2026-05-30 |
| commit |
0.00 |
Separate pip command syntax for notebook and CLI tabs in Qui |
|
— |
2026-05-29 |
| commit |
0.00 |
Romanian translation of README.md, index.md, installation.md |
|
— |
2026-05-29 |
| commit |
0.00 |
fix series of bugs for model parallel beam search (#46280) |
|
— |
2026-05-29 |
| commit |
0.00 |
Fall back to flat kwarg when modality dict is passed without |
|
— |
2026-05-29 |
| commit |
0.00 |
Fix load_adapter OOM caused by full-model warmup sizing (#46 |
|
— |
2026-05-29 |
| commit |
0.00 |
Replace assert with raise ImportError for optuna/ray depende |
|
— |
2026-05-29 |
| commit |
0.00 |
chore(linter): respect TRF017 modeling rule (#46260) |
|
— |
2026-05-29 |
| commit |
0.00 |
Delete dead code in qwen-vl series (#45827) |
|
— |
2026-05-29 |
| commit |
0.00 |
qa: fix ty caching and align CI with local run (#46278) |
|
— |
2026-05-29 |
| commit |
0.00 |
Guard DeviceMesh import in continuous batching (#46205) |
|
— |
2026-05-29 |
| commit |
0.00 |
Processor compatibility with vLLM (#46258) |
|
— |
2026-05-29 |
| commit |
0.00 |
Fix PR CI workflow cancellation condition (#46276) |
|
— |
2026-05-29 |
| commit |
0.00 |
[fix] toctree (#46106) |
|
— |
2026-05-29 |
| commit |
0.00 |
add more generic support for distributed trainer tests (#461 |
|
— |
2026-05-29 |
| commit |
0.00 |
add XPU Expectations for florence2 and lfm2_vl model test (# |
|
— |
2026-05-29 |
| commit |
0.00 |
Fix model parallel issue for altclip model and ChineseClip m |
|
All commit messages are terse, technical |
2026-05-28 |
| commit |
0.00 |
feat: Add support for JetBrains' `Mellum` v2 code generation |
|
Commit log has topic-focused, concise, i |
2026-05-28 |
| commit |
0.00 |
Fix `StaticCache` building an empty layer list when `num_kv_ |
|
Clear problem description with domain-sp |
2026-05-28 |
| commit |
0.00 |
Model parallel fix (#46230) |
|
Terse changelog, minimal free-text, doma |
2026-05-28 |
| commit |
0.00 |
Enable kernels-community/metal-flash-sdpa on MPS (#45974) |
|
Highly technical commit, with informal t |
2026-05-28 |
| commit |
0.00 |
Fix inverted assertion in remove_handler (#46227) |
|
Human-like technical language, detailed |
2026-05-28 |
| commit |
0.00 |
[ShieldGemma2] Support attn_implementation dispatch (#46069) |
|
Terse list of commit actions, with domai |
2026-05-28 |
| commit |
0.00 |
fix: kosmos2.5: properly expand embeddings table (#45835) |
|
Terse changelog, technical, uses informa |
2026-05-28 |
| commit |
0.00 |
find pytest launch error in torch 2.13.0.dev20260526 (#46252 |
|
Informal tone, direct purpose, human-sou |
2026-05-28 |
| commit |
0.00 |
Update a test to avoid writing to the default xet cache (#46 |
|
Brevity, editing artifacts, and casual t |
2026-05-28 |
| commit |
0.00 |
[Test][Kosmos2.5] Add XPU expectations for integration tests |
|
Concise, uses project-specific notations |
2026-05-28 |
| commit |
0.00 |
[`Revert`] FSDP+Dtensor refactor related changes (#46246) |
|
Revert commits with informal, terse note |
2026-05-28 |
| commit |
0.00 |
Support FA2 flash_attn_with_kvcache for XPU continuous batch |
|
PR and commit messages use terse, techni |
2026-05-27 |
| commit |
0.00 |
[`Configs`] Fix layer type validation to include its mlp cou |
|
Brief, domain-specific message; informal |
2026-05-27 |
| commit |
0.00 |
RF-DETR doc fixes (#46244) |
|
Very terse and technical commit messages |
2026-05-27 |
| commit |
0.00 |
Use `main` instead of commit SHA for now (#46241) |
|
Single-word message; co-authored Git tra |
2026-05-27 |
| commit |
0.00 |
Enable push event (to main) for PR CI workflow (#46240) |
|
Minimal, informal commit; Git trailer in |
2026-05-27 |
| commit |
0.00 |
fix(hrm_text): Add XPU Expectations for tests (#46214) |
|
Human style with domain-specific jargon |
2026-05-27 |
| commit |
0.00 |
Fix model parallel bugs for Gemma4 (#45817) |
|
Technical fix with signed-off trailer, h |
2026-05-27 |
| commit |
0.00 |
Fix bnb 4bit/8bit quantization drop chunked tensors bug (#46 |
|
Technical bug fix, domain-specific, info |
2026-05-27 |
| commit |
0.00 |
[deepseek_v4] keep hc_head / sinks / position_bias in fp32 ( |
|
Issue reference, domain jargon, informal |
2026-05-27 |
| commit |
0.00 |
init FSDP through from_pretrained (#46102) |
|
Changelog-style commit message, terse an |
2026-05-26 |
| commit |
0.00 |
Fix FSDP2 and distributed checkpointing imports for older Py |
|
Standard technical commit; includes huma |
2026-05-26 |
| commit |
0.00 |
Fix Gemma4 Array Mask Indexing (#46203) |
|
Contains domain jargon and typical error |
2026-05-26 |
| commit |
0.00 |
utils: handle flash_attn missing from importlib packages_dis |
|
Technical explanation with nuanced trace |
2026-05-26 |
| commit |
0.00 |
fix model parallel device mismatch issue in `create_bidirect |
|
Terse, technical commit, contains no AI- |
2026-05-26 |
| commit |
0.00 |
Fix a regression in encoder-decoder generation cache initial |
|
Standard technical changelog, no AI sign |
2026-05-26 |
| commit |
0.00 |
[AMD CI] revert AMD mi325 hf-workflows ref from SHA back to |
|
Extremely terse; standard revert message |
2026-05-26 |
| commit |
0.00 |
Trainer.compute_loss: fix loss over-counting under TP and EP |
|
Succinct, direct commit message, no AI i |
2026-05-26 |
| commit |
0.00 |
[GLM-4.6V] Update with GLM-GA Processor (#46184) |
|
Choppy, informal commit notes, contains |
2026-05-26 |
| commit |
0.00 |
🚨 [ALM] Add base model without head (#45534) |
|
Technical, iterative commit message sequ |
2026-05-26 |
| commit |
0.00 |
update xpu expectation for falcon mamba (#46086) |
|
Terse commit message, domain-specific re |
2026-05-26 |
| commit |
0.00 |
chore: enable Dependabot weekly GitHub Actions bumps (#46157 |
|
— |
2026-05-26 |
| commit |
0.00 |
Fix Gemma4 use_bidirectional_attention="all" mask behavior ( |
|
Direct and technical, signed-off, no AI |
2026-05-26 |
| commit |
0.00 |
Fix loading with only 1 device or distributed config (#46197 |
|
Very terse commits, no signs of AI gener |
2026-05-26 |
| commit |
0.00 |
Fix TypeError on list-typed ignore_keys_at_rope_validation i |
|
Commit uses standard merge message and c |
2026-05-25 |
| commit |
0.00 |
Support XPU autocast dtype fallback for FlashAttention (#461 |
|
Standard technical PR title, no AI indic |
2026-05-25 |
| commit |
0.00 |
Fix path traversal when saving named chat templates (#46191) |
|
Contains detailed, domain-specific expla |
2026-05-25 |
| commit |
0.00 |
Fix wrong variable in check_model_type isinstance check (#46 |
|
Clear technical summary, references code |
2026-05-25 |
| commit |
0.00 |
Enable passing kwargs through RoFormer models (#46171) |
|
Terse message matching normal commit sty |
2026-05-25 |
| commit |
0.00 |
Update cohere2_moe tp_plan (#46189) |
|
Short, iterative commit messages common |
2026-05-25 |
| commit |
0.00 |
Update release tool (#46193) |
|
Single-word message, very typical of hum |
2026-05-25 |
| commit |
0.00 |
[loading] Fix base_model_prefix issues in conversions (#4606 |
|
Commit uses normal changelog language wi |
2026-05-25 |
| commit |
0.00 |
Fix caching allocator warmup byte estimation for EP model lo |
|
Domain-specific phrasing, includes human |
2026-05-25 |
| commit |
0.00 |
Bump dev version (#46188) |
|
Single-word update, classic human commit |
2026-05-25 |
| commit |
0.00 |
Fix image-segmentation pipeline support for RF-DETR (#46130) |
|
Terse commit message with domain-specifi |
2026-05-21 |
| commit |
0.00 |
Update self-comment-ci (#46137) |
|
Minimal message, no AI-like phrasing or |
2026-05-21 |
| commit |
0.00 |
[ALM] flaky alm tests (#46074) |
|
Commit messages are terse and informal, |
2026-05-20 |
| commit |
0.00 |
[loading] Free up tensors faster inside ConversionOps (#4611 |
|
Single informal word, no AI-like phrasin |
2026-05-20 |
| commit |
0.00 |
Restore test utils fix (#46065) |
|
Terse and typical bugfix summary, no AI |
2026-05-20 |
| commit |
0.00 |
Allow `ydshieh2` for now for testing migration (#46105) |
|
Short messages, no AI verbosity, co-auth |
2026-05-20 |
| pr |
0.00 |
[docs] gemma4 typos |
|
Extremely terse; classic human typo-fix |
2026-06-02 |
| pr |
0.00 |
[docs] contributing |
|
Bullet style, concise, no AI signals; hu |
2026-04-15 |
| pr |
0.00 |
[docs] metal flash attention |
|
Brief, references issue, informal style; |
2026-06-02 |
| pr |
0.00 |
[docs] Romanian translation of `contributing.md`, `modular_t |
|
Title only, no AI signals, most likely h |
2026-06-02 |
| pr |
0.00 |
[docs] tp for continuous batching |
|
Just refers to another PR, no AI markers |
2026-05-18 |
| pr |
0.00 |
[docs] xpu continuous batching |
|
Very terse and domain-specific, no AI si |
2026-06-01 |
| pr |
0.00 |
[docs] padding-free training |
|
Brief, technical doc change, no AI signa |
2026-06-01 |
| pr |
0.00 |
Update tokenizer mappings to use TokenizersBackend for addit |
|
Informal explanation, direct style, feel |
2026-05-20 |
| pr |
0.00 |
Add Nemotron ASR Streaming |
|
Brief, domain-specific; no AI-style phra |
2026-06-01 |
| pr |
0.00 |
[docs] update num_items_in_batch for causal LMs |
|
Minimal, technical doc update with domai |
2026-06-01 |
| pr |
0.00 |
Fix tokenizer keep_accents defaults and update integration t |
|
Very brief, direct, informal; strongly h |
2026-06-02 |
| pr |
0.00 |
chore(linter): add TRF018 modeling rule |
|
Minimal, terse, domain-specific human ph |
2026-05-28 |
| pr |
0.00 |
Extended & simplified n-to-1 kernel fusion via KernelConfig |
|
Technical, structured, and uses domain-s |
2026-06-02 |
| pr |
0.00 |
Add TIPSv2 |
|
Domain-specific and concise, no AI style |
2026-06-02 |
| pr |
0.00 |
remove sec |
|
Extremely brief and colloquial; certainl |
2026-06-02 |
| pr |
0.00 |
[generation] Encode multimodal data only once |
|
Technical reasoning and informal, not AI |
2026-05-05 |
| pr |
0.00 |
fix: Add prefer_auto_map opt-in for registered-class vs auto |
|
PR talks through context of previous iss |
2026-04-30 |
| pr |
0.00 |
fix(falcon_mamba): recommend `pip install kernels` in fast-p |
|
Informal, references human comments; not |
2026-06-02 |
| pr |
0.00 |
Fix arbitrary code execution via compressed-tensors sparsity |
|
Security context with specific and techn |
2026-05-30 |
| pr |
0.00 |
fix(image_processing): don't warn/import on non-Fast attribu |
|
Specific technical fix with repo and iss |
2026-05-31 |
| pr |
0.00 |
cohere2_moe: skip invalid test cases |
|
Brief technical summary with command-lin |
2026-06-02 |
| pr |
0.00 |
Fix TestAppRoutes test failures caused by deprecated asyncio |
|
Human: informal parenthetical, clear fir |
2026-06-02 |
| pr |
0.00 |
[Qwen3VL] Fix video token placeholder: use self.video_token |
|
Mentions GitHub issue and technical deta |
2026-05-30 |
| pr |
0.00 |
fix qwen series VL model's model parallel bug |
|
Informal ping, tangible test output, no |
2026-06-01 |
| pr |
0.00 |
skip deepgemm test except cuda |
|
Terse, informal wording and natural typo |
2026-05-20 |
| pr |
0.00 |
Tp param level |
|
Concise, domain-specific points; informa |
2026-05-29 |
| pr |
0.00 |
docs: 🌐 add Turkish translation for README file |
|
— |
2026-05-31 |
| pr |
0.00 |
chore(linter): fixes for rule 16 |
|
Informal, includes reviewer comment, cle |
2026-05-18 |
| pr |
0.00 |
Add Parakeet-RNNT |
|
Succinct, technical; contains domain-spe |
2026-06-01 |
| pr |
0.00 |
[docs] chat template |
|
Succinct update with domain terms; revie |
2026-05-08 |
| pr |
0.00 |
[docs] Romanian translation of `weightconverter.md`, `mode |
|
Plain feature/translation note; no AI ph |
2026-05-31 |
| pr |
0.00 |
LocateAnything 3B |
|
Technical content with detail; no AI/hal |
2026-05-30 |
| pr |
0.00 |
Fix fp8 moe reverse substring |
|
Domain-specific abbreviations and terse |
2026-05-28 |
| pr |
0.00 |
Add Molmo2 |
|
— |
2026-01-23 |
| pr |
0.00 |
[RecurrentGemma] Support attn_implementation dispatch |
|
Sparse, minimal free-text response, like |
2026-06-01 |
