[Kernel] Support sliding window in flash attention backend (#9403)

This commit is contained in:
Chen Zhang
2024-10-20 10:57:52 -07:00
committed by GitHub
parent 962d2c6349
commit 4fa3e33349
13 changed files with 41 additions and 61 deletions

View File

@@ -53,7 +53,6 @@ class CacheEngine:
# Get attention backend.
self.attn_backend = get_attn_backend(self.head_size,
model_config.get_sliding_window(),
model_config.dtype,
cache_config.cache_dtype,
self.block_size,