Instructions to use kernels-community/metal-flash-sdpa with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Kernels
How to use kernels-community/metal-flash-sdpa with Kernels:
# !pip install kernels from kernels import get_kernel kernel = get_kernel("kernels-community/metal-flash-sdpa") - Notebooks
- Google Colab
- Kaggle
Add flash_attn_with_kvcache paged-decode kernel (port from huggingface/transformers#45977)
#4 opened 11 days ago
by
ArthurZ
Add flash_attn_func + harden MPS dispatch for transformers compatibility
#3 opened 14 days ago
by
ArthurZ
Test code with transformers
2
#1 opened 7 months ago
by
manueldeprada