Post
1197
Weclome to use MInference, which leverages the dynamic sparse nature of LLMs' attention, which exhibits some static patterns, to speed up the pre-filling for million tokens LLMs. It first determines offline which sparse pattern each head belongs to, then approximates the sparse index online and dynamically computes attention with the optimal custom kernels. This approach achieves up to a 10x speedup for pre-filling on an A100 while maintaining accuracy with 1M tokens.
For more detail please check,
project page: https://aka.ms/MInference
code: https://github.com/microsoft/MInference
paper: MInference 1.0: Accelerating Pre-filling for Long-Context LLMs via Dynamic Sparse Attention (2407.02490)
hf demo: microsoft/MInference
For more detail please check,
project page: https://aka.ms/MInference
code: https://github.com/microsoft/MInference
paper: MInference 1.0: Accelerating Pre-filling for Long-Context LLMs via Dynamic Sparse Attention (2407.02490)
hf demo: microsoft/MInference
