Source codes for BlockFFN, introduced by the paper: BlockFFN: Towards End-Side Acceleration-Friendly Mixture-of-Experts with Chunk-Level Activation Sparsity.

Links: [Paper] [Models]

1. For codes about the architecture and pre-training process of BlockFFN, see the directory pretrain.
2. For codes about the implementation and pre-training process of baseline methods, see the directory baseline.
3. For codes about the inference acceleration (i.e., the efficient acceleration kernels) of BlockFFN, see the directory inference.

If you find our work useful for your research, please kindly cite our paper as follows:

@article{song2025blockffn,
      title={{BlockFFN}: Towards End-Side Acceleration-Friendly Mixture-of-Experts with Chunk-Level Activation Sparsity}, 
      author={Chenyang Song and Weilin Zhao and Xu Han and Chaojun Xiao and Yingfa Chen and Yuxuan Li and Zhiyuan Liu and Maosong Sun},
      journal={arXiv preprint arXiv:2507.08771},
      year={2025},
      url={https://arxiv.org/pdf/2507.08771}, 
}