Garbage collector regression in the online server (#10621)

2025-09-18 11:52:35 -07:00
parent 9a5c42f9ad
commit 67073dde85
1 changed files with 1 additions and 0 deletions
--- a/python/sglang/srt/model_executor/cuda_graph_runner.py
+++ b/python/sglang/srt/model_executor/cuda_graph_runner.py
@@ -100,6 +100,7 @@ def freeze_gc(enable_cudagraph_gc: bool):
    finally:
        if should_freeze:
            gc.unfreeze()
+            gc.collect()


 def _to_torch(model: torch.nn.Module, reverse: bool, num_tokens: int):