Kernels:
Trusted publisher
| library_name: kernels | |
| license: apache-2.0 | |
| This is the repository card of kernels-community/vllm-flash-attn3 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/vllm-flash-attn3", version=1) | |
| flash_attn_combine = kernel_module.flash_attn_combine | |
| flash_attn_combine(...) | |
| ``` | |
| ## Available functions | |
| - `flash_attn_combine` | |
| - `flash_attn_func` | |
| - `flash_attn_qkvpacked_func` | |
| - `flash_attn_varlen_func` | |
| - `flash_attn_with_kvcache` | |
| - `get_scheduler_metadata` | |
| ## Benchmarks | |
| Benchmarking script is available for this kernel. Run `kernels benchmark kernels-community/vllm-flash-attn3 --version 1`. | |