From cb0c2cbfdb4e1f2cd75977f9653214e3ab0f3431 Mon Sep 17 00:00:00 2001 From: Brayden Zhong Date: Thu, 2 Apr 2026 00:27:20 -0400 Subject: [PATCH] Enable multi-thread weight loading by default (#20289) --- python/sglang/srt/model_loader/loader.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/python/sglang/srt/model_loader/loader.py b/python/sglang/srt/model_loader/loader.py index 3fa320e06..b0884c681 100644 --- a/python/sglang/srt/model_loader/loader.py +++ b/python/sglang/srt/model_loader/loader.py @@ -479,7 +479,7 @@ class DefaultModelLoader(BaseModelLoader): ) -> Generator[Tuple[str, torch.Tensor], None, None]: """Get an iterator for the model weights based on the load format.""" extra_config = self.load_config.model_loader_extra_config - use_multithread = extra_config.get("enable_multithread_load", False) + use_multithread = extra_config.get("enable_multithread_load", True) hf_folder, hf_weights_files, use_safetensors = self._prepare_weights( source.model_or_path, source.revision, source.fall_back_to_pt )