Commit f8cf43ae authored by zhuwenwen's avatar zhuwenwen
Browse files

update VLLM_USE_LIGHTOP and VLLM_USE_OPT_CAT

parent f70d0ec6
......@@ -190,14 +190,23 @@ def get_model_architecture(
else:
os.environ['LM_NN'] = '1'
if (architectures == ['DeepseekV3ForCausalLM'] or architectures == ['DeepSeekMTPModel']):
if architectures in [['DeepseekV3ForCausalLM'], ['DeepSeekMTPModel']]:
if not envs.is_set("VLLM_USE_LIGHTOP"):
os.environ['VLLM_USE_LIGHTOP'] = '1'
if not envs.is_set("VLLM_USE_OPT_CAT"):
os.environ['VLLM_USE_OPT_CAT'] = '1'
if os.getenv('GEMM_PAD') != '1':
os.environ['GEMM_PAD'] = '0'
if os.getenv('FA_PAD') != '1':
os.environ['FA_PAD'] = '0'
else:
if architectures in [['DeepseekV3ForCausalLM'], ['DeepSeekMTPModel']]:
if not envs.is_set("VLLM_USE_LIGHTOP"):
os.environ['VLLM_USE_LIGHTOP'] = '1'
if not envs.is_set("VLLM_USE_OPT_CAT"):
os.environ['VLLM_USE_OPT_CAT'] = '1'
# awq相关配置
try:
if os.getenv('AWQ_MOE_SZ') == None:
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment