diff --git a/vllm/transformers_utils/configs/mistral.py b/vllm/transformers_utils/configs/mistral.py index 1a0e25021c..aea990b07a 100644 --- a/vllm/transformers_utils/configs/mistral.py +++ b/vllm/transformers_utils/configs/mistral.py @@ -198,6 +198,14 @@ def _remap_mistral_quantization_args(config: dict) -> dict: "quant_method": "fp8", "activation_scheme": "dynamic" if is_dynamic else "static", } + elif ( + str(quantization.get("quant_method", "")).lower().replace("_", "-") + == "compressed-tensors" + ): + # Pass through compressed-tensors config, while normalizing + # quant_method to the canonical community spelling. + quantization["quant_method"] = "compressed-tensors" + config["quantization_config"] = quantization else: raise ValueError(f"Found unknown quantization='{quantization}' in config")