[Feature][Kernel] Support bitsandbytes quantization and QLoRA (#4776)

vllm/model_executor/layers/quantization/__init__.py

@@ -4,6 +4,8 @@ from vllm.model_executor.layers.quantization.aqlm import AQLMConfig
 from vllm.model_executor.layers.quantization.awq import AWQConfig
 from vllm.model_executor.layers.quantization.base_config import (
     QuantizationConfig)
+from vllm.model_executor.layers.quantization.bitsandbytes import (
+    BitsAndBytesConfig)
 from vllm.model_executor.layers.quantization.compressed_tensors.compressed_tensors import (  # noqa: E501
     CompressedTensorsConfig)
 from vllm.model_executor.layers.quantization.deepspeedfp import (
@@ -30,6 +32,7 @@ QUANTIZATION_METHODS: Dict[str, Type[QuantizationConfig]] = {
     "gptq": GPTQConfig,
     "squeezellm": SqueezeLLMConfig,
     "sparseml": CompressedTensorsConfig,
+    "bitsandbytes": BitsAndBytesConfig,
 }