Support W4A8 quantization for vllm (#5218)

vllm/model_executor/layers/quantization/__init__.py

@@ -19,6 +19,7 @@ from vllm.model_executor.layers.quantization.gptq_marlin import (
 from vllm.model_executor.layers.quantization.gptq_marlin_24 import (
     GPTQMarlin24Config)
 from vllm.model_executor.layers.quantization.marlin import MarlinConfig
+from vllm.model_executor.layers.quantization.qqq import QQQConfig
 from vllm.model_executor.layers.quantization.squeezellm import SqueezeLLMConfig
 
 QUANTIZATION_METHODS: Dict[str, Type[QuantizationConfig]] = {
@@ -37,6 +38,7 @@ QUANTIZATION_METHODS: Dict[str, Type[QuantizationConfig]] = {
     "squeezellm": SqueezeLLMConfig,
     "compressed-tensors": CompressedTensorsConfig,
     "bitsandbytes": BitsAndBytesConfig,
+    "qqq": QQQConfig,
 }