AIbase
Product LibraryTool Navigation

FlexPrefill

Public

Code for paper: [ICLR2025 Oral] FlexPrefill: A Context-Aware Sparse Attention Mechanism for Efficient Long-Sequence Inference

Creat2025-02-18T15:02:28
Update2025-03-26T16:55:18
https://arxiv.org/abs/2502.20766
80
Stars
0
Stars Increase

Related projects