Amd test fp8 (#4261)

2025-03-11 01:12:09 +08:00
parent 5a6400eec5
commit 2ac189edc8
6 changed files with 84 additions and 0 deletions
--- a/python/sglang/srt/configs/model_config.py
+++ b/python/sglang/srt/configs/model_config.py
@@ -237,6 +237,7 @@ class ModelConfig:
            "compressed_tensors",
            "compressed-tensors",
            "fbgemm_fp8",
+            "w8a8_fp8",
        ]
        optimized_quantization_methods = [
            "fp8",