apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-11-02 20:54:03 +08:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

support c4 attn && fix cache

Browse Source
This commit is contained in:
lizhenyun01
2025-07-23 23:51:28 +08:00
parent 832d25334a
commit 29c3292f02
16 changed files with 198 additions and 65 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
fastdeploy/model_executor/layers/backends/gcu/attention/flash_attn_backend.py
View File

@@ -211,6 +211,7 @@ class GCUFlashAttnBackend(AttentionBackend):
def get_kv_cache_shape(
self,
max_num_blocks: int,
kv_cache_quant_type: str = None,
):
"""
Caculate kv cache shape

1
fastdeploy/model_executor/layers/backends/gcu/attention/mem_efficient_attn_backend.py
View File

@@ -222,6 +222,7 @@ class GCUMemEfficientAttnBackend(AttentionBackend):
def get_kv_cache_shape(
self,
max_num_blocks: int,
kv_cache_quant_type: str = None,
):
"""
Caculate kv cache shape