transformers
https://github.com/huggingface/transformers
Python
🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
Triage Issues!
When you volunteer to triage issues, you'll receive an email each day with a link to an open issue that needs help in this project. You'll also receive instructions on how to triage issues.
Triage Docs!
Receive a documented method or class from your favorite GitHub repos in your inbox every day. If you're really pro, receive undocumented methods or classes and supercharge your commit history.
Python not yet supported54 Subscribers
Add a CodeTriage badge to transformers
Help out
- Issues
- AMD CI: Strix Halo (gfx1151) on ROCm 10.1
- AMD CI: add Radeon AI Pro R9700 (gfx1201)
- [Mamba] Fused kernel at inference without cache or padding
- GGUF reader: llama.cpp Mixtral loads as dense Llama with random MLPs; Qwen3 head_dim and Mistral detection not mapped
- DeBERTa: `from_pretrained(dtype=torch.bfloat16)` leaves `q_bias`/`v_bias` in float32, forward fails (`expected m1 and m2 to have the same dtype`)
- Fix DeBERTa q_bias/v_bias staying float32 when loaded in another dtype
- Fix offset mapping padding in `tokenizer.pad`
- [Bug]: resume_from_checkpoint silently skips the root-level (trained) adapter when the checkpoint contains adapter subdirectories — DPO loss stuck at ln(2) ≈ 0.6931, no error or warning
- 5.21.0 regression: end-anchored `WeightRenaming` rules save corrupted keys (`mask_embedding$`) that silently fail to reload
- GGUF: Gemma 3 loses global-layer RoPE linear scaling and query_pre_attn_scalar when loaded from a llama.cpp GGUF
- Docs
- Python not yet supported