Support cuda graph for DP attention (#2061)
This commit is contained in:
@@ -128,9 +128,6 @@ class TpModelWorker:
|
||||
def get_tp_cpu_group(self):
|
||||
return self.model_runner.tp_group.cpu_group
|
||||
|
||||
def get_tp_device_group(self):
|
||||
return self.model_runner.tp_group.device_group
|
||||
|
||||
def get_memory_pool(self):
|
||||
return (
|
||||
self.model_runner.req_to_token_pool,
|
||||
|
||||
Reference in New Issue
Block a user