-
Notifications
You must be signed in to change notification settings - Fork 33.1k
Pull requests: huggingface/transformers
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Fix import error in moe.py by providing explicit schema to custom_op
#45804
opened May 6, 2026 by
rraghavkaushik
Loading…
Fix
kernelize() crash for gpt_oss: missing @use_kernel_func_from_hub on apply_rotary_pos_emb
#45799
opened May 6, 2026 by
jiqing-feng
Contributor
Loading…
Fix WeightConverter substring match on leaf-style source patterns
#45794
opened May 5, 2026 by
qgallouedec
Member
Loading…
fix: validate special token ids against attribute values
#45791
opened May 5, 2026 by
CharlieKerfoot
Contributor
Loading…
fix: ModuleNotFoundError caused by distributed race condition in ci/cd test
#45787
opened May 5, 2026 by
rigen1048
Contributor
Loading…
1 of 5 tasks
fix: remove upper bound on tokenizers version constraint
#45784
opened May 5, 2026 by
ranjithchinnanan
Loading…
[generation] Encode multimodal data only once
#45783
opened May 5, 2026 by
zucchini-nlp
Member
Loading…
Fix WeightConverter regex incorrectly matching shared_experts as experts in DeepSeek V4
#45782
opened May 5, 2026 by
silencelamb
Loading…
torch.backends.fp32_precision cascade conv/rnn so removing the temp fix
#45777
opened May 4, 2026 by
khushali9
Contributor
Loading…
2 of 6 tasks
Gate enable_gqa=True on actual flash-attention eligibility
#45776
opened May 4, 2026 by
dvdimitrov13
•
Draft
Fix UnboundLocalError for
is_updated in encoder-decoder cross-attention
#45773
opened May 4, 2026 by
SeaL773
Loading…
feat: add bf16_loss training argument for VRAM-efficient QLoRA by keeping loss in BF16 during training to save ~1.4 GB VRAM
#45769
opened May 3, 2026 by
butterwecksolutions
Loading…
End-to-end test of Gemma 3 + FA2 construction
#45760
opened May 3, 2026 by
jamesbraza
Contributor
Loading…
fix attribute access in PermuteForRope._apply
#45756
opened May 3, 2026 by
CharlieKerfoot
Contributor
Loading…
Fix unhandled exception noise from background safetensors conversion thread
#45752
opened May 3, 2026 by
dhruv7477
Loading…
feat: add crop() to StaticCache layers for assisted generation
#45745
opened May 2, 2026 by
ArkaD171717
Loading…
fix(bitsandbytes): implement reverse_op for Bnb4bitDeserialize and Bnb8bitDeserialize
#45743
opened May 2, 2026 by
Kaisan10
Loading…
2 of 3 tasks
feat(llama): add has_weight parameter to LlamaRMSNorm for FlashNorm-folded checkpoints (+12.77% e2e on Llama-3.2-1B at bf16/A100)
#45742
opened May 2, 2026 by
fm1320
Loading…
5 tasks done
fix(musicgen_melody): use DynamicCache instead of EncoderDecoderCache
#45738
opened May 1, 2026 by
adityachoksi2512
Loading…
Add regression test for MusicgenMelody audio conditioning (GH #45647)
#45737
opened May 1, 2026 by
voodoovampire
Loading…
Previous Next
ProTip!
Find all pull requests that aren't related to any open issues with -linked:issue.