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

polish code with new pre-commit rule (#2923)

Browse Source
This commit is contained in:
Zero Rains
2025-07-19 23:19:27 +08:00
committed by GitHub
parent b8676d71a8
commit 25698d56d1
424 changed files with 14307 additions and 13518 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
fastdeploy/platforms/gcu.py
View File

@@ -25,6 +25,7 @@ class GCUPlatform(Platform):
"""
gcu platform class
"""
device_name = "gcu"
@classmethod
@@ -33,7 +34,7 @@ class GCUPlatform(Platform):
Check whether GCU is available.
"""
try:
assert paddle.base.core.get_custom_device_count('gcu') > 0
assert paddle.base.core.get_custom_device_count("gcu") > 0
return True
except Exception as e:
logger.warning(
@@ -50,10 +51,10 @@ class GCUPlatform(Platform):
"""
if selected_backend == _Backend.NATIVE_ATTN:
logger.info("Using GCU mem_efficient ATTN backend.")
return ("fastdeploy.model_executor.layers.backends.gcu.attention.mem_efficient_attn_backend.GCUMemEfficientAttnBackend")
return "fastdeploy.model_executor.layers.backends.gcu.attention.mem_efficient_attn_backend.GCUMemEfficientAttnBackend"
elif selected_backend == _Backend.APPEND_ATTN:
logger.info("Using GCU ATTN backend.")
return ("fastdeploy.model_executor.layers.backends.gcu.attention.flash_attn_backend.GCUFlashAttnBackend")
return "fastdeploy.model_executor.layers.backends.gcu.attention.flash_attn_backend.GCUFlashAttnBackend"
else:
raise ValueError(
"Invalid attention backend you specified.\n"