[ Misc ] fbgemm checkpoints (#6559)

2024-07-20 12:36:57 -04:00
parent 9042d68362
commit 683e3cb9c4
24 changed files with 234 additions and 47 deletions
--- a/vllm/config.py
+++ b/vllm/config.py
@@ -251,7 +251,7 @@ class ModelConfig:
                    f"supported in ROCm.")
            if (self.quantization
                    not in ("fp8", "marlin", "gptq_marlin_24", "gptq_marlin",
-                            "compressed_tensors")):
+                            "fbgemm_fp8", "compressed_tensors")):
                logger.warning(
                    "%s quantization is not fully "
                    "optimized yet. The speed can be slower than "