Instructions to use replicate/trimul_gpumode with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Kernels
How to use replicate/trimul_gpumode with Kernels:
# !pip install kernels from kernels import get_kernel kernel = get_kernel("replicate/trimul_gpumode") - Notebooks
- Google Colab
- Kaggle
File size: 897 Bytes
61aea62 c076916 61aea62 4d53041 9fb37a6 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 | ---
tags:
- kernels
- universal
---
> [!CAUTION]
> Starting from September 13, 2026, we will be removing the "model" type repositories of kernels (e.g., kernels-community/flash-attn3). Make sure you're using a latest version of kernels. If you face any disruption, please report them here: https://github.com/huggingface/kernels/issues/new.
These are kernels from the [GPUMODE](https://www.gpumode.com/v2/news) trimul competition written in Triton
The `kernel_mi300` ranked first for MI300, `kernel_a100` ranked first for A100, `kernel_b100` ranked second for B200, and `kernel_h100` ranked third for H100. These kernels are copied from the [trimul](https://github.com/arseniivanov/trimul) repository.
The `kernel_global` has the highest average ranking across all architectures, and was copied and adapted from the [gpumode-trimul](https://github.com/davidberard98/gpumode-trimul) repository. |