Skip to content

GitLab

Unverified Commit 53cd7f86 authored by Zhengxu Chen's avatar Zhengxu Chen Committed by GitHub
Browse files

[compile] Recompile graph module during Dynamo cache loading. (#30743) 


Signed-off-by: default avatarZhengxu Chen <zhxchen17@fb.com>
parent 7b966ae2
Show whitespace changes
Inline Side-by-side
Showing with 1 addition and 0 deletions
vllm/compilation/caching.py
View file @ 53cd7f86
...@@ -104,6 +104,7 @@ class VllmSerializableFunction(SerializableCallable): ...@@ -104,6 +104,7 @@ class VllmSerializableFunction(SerializableCallable):
state = pickle.loads(data) state = pickle.loads(data)
fake_mode = FakeTensorMode(shape_env=ShapeEnv()) fake_mode = FakeTensorMode(shape_env=ShapeEnv())
state["graph_module"] = GraphPickler.loads(state["graph_module"], fake_mode) state["graph_module"] = GraphPickler.loads(state["graph_module"], fake_mode)
state["graph_module"].recompile()
state["example_inputs"] = GraphPickler.loads(state["example_inputs"], fake_mode) state["example_inputs"] = GraphPickler.loads(state["example_inputs"], fake_mode)
vllm_backend = VllmBackend(get_current_vllm_config(), state["prefix"]) vllm_backend = VllmBackend(get_current_vllm_config(), state["prefix"])
......
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