Disable gemma3 for SWA due to CUDA illegal memory access error (#8895)

Signed-off-by: Xinyuan Tong <xinyuantong.cs@gmail.com>

Disable gemma3 for SWA due to CUDA illegal memory access error (#8895)
Signed-off-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
c36a6693 · Xinyuan Tong · GitHub · 62f8eb48 · c36a6693
Unverified Commit c36a6693 authored Aug 07, 2025 by Xinyuan Tong Committed by GitHub Aug 07, 2025
Hide whitespace changes
Inline Side-by-side

Showing with 2 additions and 0 deletions

python/sglang/srt/server_args.py python/sglang/srt/server_args.py +2 -0

No files found.
--- a/python/sglang/srt/server_args.py
+++ b/python/sglang/srt/server_args.py
@@ -2037,6 +2037,8 @@ class ServerArgs:

        if model_arch in [
            "Gemma2ForCausalLM",
+            "Gemma3ForCausalLM",
+            "Gemma3ForConditionalGeneration",
            "Gemma3nForCausalLM",
            "Gemma3nForConditionalGeneration",
        ]: