Skip to content

GitLab

Commit 8364249c authored by zhuwenwen's avatar zhuwenwen
Browse files

set VLLM_USE_OPT_RESHAPE_AND_CACHE=1 for qwen3-30b

parent d533c559
Show whitespace changes
Inline Side-by-side
Showing with 4 additions and 4 deletions
vllm/model_executor/model_loader/utils.py
View file @ 8364249c
......@@ -269,8 +269,8 @@ def get_model_architecture(
os.environ['VLLM_USE_LIGHTOP_MOE_SUM'] = '1'
if not envs.is_set("VLLM_USE_FUSE_SILU_AND_MUL"):
os.environ['VLLM_USE_FUSE_SILU_AND_MUL'] = '1'
# if not envs.is_set("VLLM_USE_OPT_RESHAPE_AND_CACHE"):
# os.environ['VLLM_USE_OPT_RESHAPE_AND_CACHE'] = '1'
if not envs.is_set("VLLM_USE_OPT_RESHAPE_AND_CACHE"):
os.environ['VLLM_USE_OPT_RESHAPE_AND_CACHE'] = '1'
if architectures in [['DeepseekV32ForCausalLM']]:
if not envs.is_set("VLLM_USE_V32_ENCODE"):
......@@ -305,8 +305,8 @@ def get_model_architecture(
os.environ['VLLM_USE_LIGHTOP_MOE_SUM'] = '1'
if not envs.is_set("VLLM_USE_FUSE_SILU_AND_MUL"):
os.environ['VLLM_USE_FUSE_SILU_AND_MUL'] = '1'
# if not envs.is_set("VLLM_USE_OPT_RESHAPE_AND_CACHE"):
# os.environ['VLLM_USE_OPT_RESHAPE_AND_CACHE'] = '1'
if not envs.is_set("VLLM_USE_OPT_RESHAPE_AND_CACHE"):
os.environ['VLLM_USE_OPT_RESHAPE_AND_CACHE'] = '1'
if architectures in [['DeepseekV32ForCausalLM']]:
if not envs.is_set("VLLM_USE_V32_ENCODE"):
......
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