add compile

vllm/worker/model_runner.py

@@ -787,6 +787,16 @@ class GPUModelRunnerBase(ModelRunnerBase[TModelInputForGPU]):
                     "provided. Defaulting to scaling factors of 1.0. "
                     "This may lead to less accurate results!")
 
+        count = 0
+
+        def backend(gm, input):
+            nonlocal count
+            count += 1
+            print(count)
+            return gm.forward
+
+        self.model = torch.compile(self.model, backend=backend, fullgraph=True)
+
     def save_sharded_state(
         self,
         path: str,