Kernels:
Trusted publisher
| library_name: kernels | |
| license: mit | |
| This is the repository card of kernels-community/tokenspeed-attention that has been pushed on the Hub. It was built to be used with the [`kernels` library](https://github.com/huggingface/kernels). This card was automatically generated. | |
| ## How to use | |
| ```python | |
| # make sure `kernels` is installed: `pip install -U kernels` | |
| from kernels import get_kernel | |
| kernel_module = get_kernel("kernels-community/tokenspeed-attention", version=1) | |
| __version__ = kernel_module.__version__ | |
| __version__(...) | |
| ``` | |
| ## Available functions | |
| - `__version__` | |
| - `mha_prefill` | |
| - `mha_extend_with_kvcache` | |
| - `mha_decode_with_kvcache` | |
| - `mla_prefill` | |
| - `mla_decode_with_kvcache` | |
| - `attn_merge_state` | |
| - `triton_mha_prefill` | |
| - `triton_mha_extend_with_kvcache` | |
| - `triton_mha_decode_with_kvcache` | |
| - `triton_mla_prefill` | |
| - `triton_mla_decode_with_kvcache` | |
| - `triton_attn_merge_state` | |
| ## Benchmarks | |
| No benchmark available yet. | |