apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2026-04-23 17:11:21 +08:00
Code Issues Actions 10 Packages Projects Releases Wiki Activity

[Others] Remove useless code (#5404)

Browse Source
This commit is contained in:
周周周
2025-12-08 13:59:46 +08:00
committed by GitHub
parent 3066a0c34b
commit 2aea8a3a60
8 changed files with 139 additions and 166 deletions
Download Patch File Download Diff File Expand all files Collapse all files
fastdeploy/model_executor/layers/attention/append_attn_backend.py
+2 -13
View File
@@ -206,20 +206,9 @@ class AppendAttentionBackend(AttentionBackend):
Calculate kv cache shape
"""
key_cache_shape = [max_num_blocks, self.kv_num_heads, self.block_size, self.head_dim]
value_cache_shape = [max_num_blocks, self.kv_num_heads, self.block_size, self.head_dim]
if kv_cache_quant_type is not None and kv_cache_quant_type == "int4_zp":
key_cache_shape = [
max_num_blocks,
self.kv_num_heads,
self.block_size,
self.head_dim // 2,
]
value_cache_shape = [
max_num_blocks,
self.kv_num_heads,
self.block_size,
self.head_dim // 2,
]
key_cache_shape[-1] = self.head_dim // 2
value_cache_shape = key_cache_shape
return key_cache_shape, value_cache_shape
def forward_mixed(