EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

[CI] Follow vLLM FusedMoEParallelConfig interface change and clean up unused config (#1625)

Browse Source 
This commit
78fe77534b
from vllm reverted the change for FusedMoEParallelConfig

This PR do the same to fix the CI error

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
This commit is contained in:
wangxiyuan
2025-07-04 17:54:33 +08:00
committed by GitHub
parent 4e910186de
commit 343955c7ac
4 changed files with 13 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
vllm_ascend/ascend_config.py
View File

@@ -21,7 +21,7 @@ from vllm.logger import logger
TORCHAIR_MODEL_LIST = ["deepseek", "pangu"]
def check_torchair_supported(model_type: str):
def _check_torchair_supported(model_type: str):
for supported_model in TORCHAIR_MODEL_LIST:
if supported_model in model_type.lower():
return True
@@ -147,10 +147,10 @@ def check_ascend_config(vllm_config, enforce_eager):
else:
# torchair_graph case
if ascend_config.torchair_graph_config.enabled:
# torchair_graph is supported for deepseek model only currently.
# torchair_graph is supported for deepseek/pangu model only.
if vllm_config.model_config:
model_type = vllm_config.model_config.hf_config.model_type
if not check_torchair_supported(model_type):
if not _check_torchair_supported(model_type):
raise NotImplementedError(
"Torchair graph mode only works with following model types:"
f"{TORCHAIR_MODEL_LIST}.")