From a2424068eca21d2b9c0fdfcfdccd588ab8dc87a3 Mon Sep 17 00:00:00 2001
From: gongwei-130 <56567052+gongwei-130@users.noreply.github.com>
Date: Thu, 11 Sep 2025 15:00:21 -0700
Subject: [PATCH] add try catch for quant config hf download (#10340)

---
 python/sglang/srt/configs/model_config.py | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)

diff --git a/python/sglang/srt/configs/model_config.py b/python/sglang/srt/configs/model_config.py
index f16442e4d..0dbe37aa0 100644
--- a/python/sglang/srt/configs/model_config.py
+++ b/python/sglang/srt/configs/model_config.py
@@ -420,11 +420,20 @@ class ModelConfig:
             is_local = os.path.exists(self.model_path)
             modelopt_quant_config = {"quant_method": "modelopt"}
             if not is_local:
-                from huggingface_hub import HfApi
+                import huggingface_hub
+
+                try:
+                    from huggingface_hub import HfApi
+
+                    hf_api = HfApi()
+                    if hf_api.file_exists(self.model_path, "hf_quant_config.json"):
+                        quant_cfg = modelopt_quant_config
+                except huggingface_hub.errors.OfflineModeIsEnabled:
+                    logger.warning(
+                        "Offline mode is enabled, skipping hf_quant_config.json check"
+                    )
+                    pass
 
-                hf_api = HfApi()
-                if hf_api.file_exists(self.model_path, "hf_quant_config.json"):
-                    quant_cfg = modelopt_quant_config
             elif os.path.exists(os.path.join(self.model_path, "hf_quant_config.json")):
                 quant_config_file = os.path.join(
                     self.model_path, "hf_quant_config.json"