Default VLLM_ASCEND_ENABLE_ASYNC_EXPONENTIAL switch to OFF

Signed-off-by: YuhanBai <[email protected]>

Author: YuhanBai

vllm_ascend/envs.py

@@ -176,8 +176,9 @@
     # Whether to anbale dynamic EPLB
     "DYNAMIC_EPLB":
     lambda: os.getenv("DYNAMIC_EPLB", "false").lower(),
+    # Whether to enable exponential overlap with model executing.
     "VLLM_ASCEND_ENABLE_ASYNC_EXPONENTIAL":
-    lambda: int(os.getenv("VLLM_ASCEND_ENABLE_ASYNC_EXPONENTIAL", 1)),
+    lambda: bool(int(os.getenv("VLLM_ASCEND_ENABLE_ASYNC_EXPONENTIAL", '0'))),
 }
 
 # end-env-vars-definition