fix bugs

src/bespokelabs/curator/llm/llm.py

@@ -110,20 +110,7 @@ def __init__(
             "gpu_memory_utilization": gpu_memory_utilization,
         }
 
-        if self.backend == "vllm":
-            config_params = {
-                "model": model_name,
-                "generation_params": generation_params,
-                "tensor_parallel_size": tensor_parallel_size,
-                "enforce_eager": enforce_eager,
-                "max_model_length": max_model_length,
-                "max_tokens": max_tokens,
-                "gpu_memory_utilization": gpu_memory_utilization,
-                "batch_size": batch_size if batch_size is not None else 256,
-                "distributed_executor_backend": distributed_executor_backend,
-                "pipeline_parallel_size": pipeline_parallel_size,
-            }
-            config = OfflineRequestProcessorConfig(**_remove_none_values(config_params))
+        self._request_processor = _RequestProcessorFactory.create(backend_params, batch=batch, response_format=response_format, backend=backend)
 
     def _hash_fingerprint(self, fingerprint_str):
         return xxh64(fingerprint_str.encode("utf-8")).hexdigest()