From 5c6a41facfacdc80d01015a65be5b8a5ec8eb91e Mon Sep 17 00:00:00 2001
From: Jerry Zhang <jerryzh168@gmail.com>
Date: Wed, 20 Nov 2024 17:37:28 -0800
Subject: [PATCH] Error out when torchao-config option is not recognized
 (#2107)

---
 python/sglang/srt/layers/torchao_utils.py | 2 ++
 python/sglang/srt/server_args.py          | 2 +-
 2 files changed, 3 insertions(+), 1 deletion(-)

diff --git a/python/sglang/srt/layers/torchao_utils.py b/python/sglang/srt/layers/torchao_utils.py
index 46b082401..9395cdf27 100644
--- a/python/sglang/srt/layers/torchao_utils.py
+++ b/python/sglang/srt/layers/torchao_utils.py
@@ -62,6 +62,8 @@ def torchao_quantize_param_data(param: torch.Tensor, torchao_config: str):
                 granularity=GRANULARITY_MAP[granularity]
             ),
         )
+    else:
+        raise ValueError(f"Unexpected config: {torchao_config}")
 
     return dummy_linear.weight
 
diff --git a/python/sglang/srt/server_args.py b/python/sglang/srt/server_args.py
index 4a1cad89e..5487f772f 100644
--- a/python/sglang/srt/server_args.py
+++ b/python/sglang/srt/server_args.py
@@ -661,7 +661,7 @@ class ServerArgs:
             "--torchao-config",
             type=str,
             default=ServerArgs.torchao_config,
-            help="Optimize the model with torchao. Experimental feature. Current choices are: int8dq, int8wo, int4wo-<group_size>, fp8wo",
+            help="Optimize the model with torchao. Experimental feature. Current choices are: int8dq, int8wo, int4wo-<group_size>, fp8wo, fp8dq-per_tensor, fp8dq-per_row",
         )
         parser.add_argument(
             "--enable-nan-detection",