-
Notifications
You must be signed in to change notification settings - Fork 34.3k
Pull requests: huggingface/transformers
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Fix stale expected values in integration tests (cuda sm_86 / Aug04 regressions)
#48171
opened Aug 21, 2026 by
ydshieh
Collaborator
Loading…
Fix Apr 05 integration test regressions (cuda sm_86)
#48170
opened Aug 21, 2026 by
ydshieh
Collaborator
Loading…
Fix integration test expected values for cuda sm_86 (Mar 15 regressions)
#48168
opened Aug 21, 2026 by
ydshieh
Collaborator
Loading…
Apply context parallelism to the evaluation path
#48167
opened Aug 20, 2026 by
qgallouedec
Member
Loading…
deterministic layer_types buffer registration in multiple models
#48162
opened Aug 20, 2026 by
mowoe
Loading…
Deprecate backend="torchaudio" for
load_audio
#48158
opened Aug 20, 2026 by
eustlb
Collaborator
Loading…
force_accelerate_hooks should not hide the signature it wraps
#48156
opened Aug 20, 2026 by
SunMarc
Member
Loading…
Respect config.initializer_range for Conv1D weights in DecisionTransformer, ImageGPT and OpenAI GPT
#48152
opened Aug 20, 2026 by
DavidJohnQuinlan
Contributor
Loading…
2 tasks done
fix: make auto_docstring import-time performance test robust to timing noise.
#48149
opened Aug 20, 2026 by
deepresearcher08
Loading…
3 of 6 tasks
Feat/laguna quantized checkpoints support
#48146
opened Aug 20, 2026 by
lulmer
Loading…
2 of 5 tasks
QA: Fall back to the import name when a distribution is missing
#48144
opened Aug 20, 2026 by
tarekziade
Collaborator
Loading…
Persist the Whisper spelling normalizer across save_pretrained
#48143
opened Aug 20, 2026 by
Bluear7878
Contributor
Loading…
Fix
StaticCache for Mllama and enable torch.compile
#48141
opened Aug 20, 2026 by
jiqing-feng
Contributor
Loading…
Honor per-param
_is_hf_initialized flags for built-in models in _initialize_weights
#48140
opened Aug 20, 2026 by
hungnnvidia
Contributor
Loading…
4 tasks done
Don't fail model loading when the accelerator can't report free memory
#48136
opened Aug 20, 2026 by
studioego
Loading…
3 of 5 tasks
[serge] Fix 2 integration tests regressed by commit 16780c86b20a (PR #47622)
#48134
opened Aug 19, 2026 by
sergereview
Bot
Loading…
[serge] Fix 4 integration tests for model
generation failing with output_mismatch (list output differs (4))
#48133
opened Aug 19, 2026 by
sergereview
Bot
Loading…
Compute MoE load-balancing loss per layer to avoid giant one-hot materialization −99.7% @ 128k
#48131
opened Aug 19, 2026 by
qgallouedec
Member
Loading…
Skip caching allocator warmup when the device cannot report free memory
#48130
opened Aug 19, 2026 by
2sumtech
Loading…
Previous Next
ProTip!
Exclude everything labeled
bug with -label:bug.