[diffusion] fix: fix bug about pin memory when offloading (#14472)

2025-12-04 23:26:30 -08:00
parent beec8eed6a
commit 46b05ef58f
1 changed files with 1 additions and 1 deletions
--- a/python/sglang/multimodal_gen/runtime/loader/fsdp_load.py
+++ b/python/sglang/multimodal_gen/runtime/loader/fsdp_load.py
@@ -269,7 +269,7 @@ def load_model_from_full_model_state_dict(
                meta_sharded_param.placements,
            )
            if cpu_offload:
-                sharded_tensor = sharded_tensor.to("cpu", pin_memory=True)
+                sharded_tensor = sharded_tensor.to("cpu")
        sharded_sd[target_param_name] = nn.Parameter(sharded_tensor)

    model.reverse_param_names_mapping = reverse_param_names_mapping