diff --git a/fastdeploy/config.py b/fastdeploy/config.py index 3114acc88f3..1623408820d 100644 --- a/fastdeploy/config.py +++ b/fastdeploy/config.py @@ -1545,6 +1545,8 @@ def postprocess(self): self.cache_config.max_encoder_cache = 0 # Adjustment GraphOptConfig + if self.scheduler_config is not None and self.scheduler_config.splitwise_role == "prefill": + self.graph_opt_config.use_cudagraph = self.graph_opt_config.cudagraph_only_prefill if self.load_config is not None and self.load_config.dynamic_load_weight is True: self.graph_opt_config.graph_opt_level = 0 logger.info(