Make benchmark_serving part of the library

Signed-off-by: Kaiyu Xie <[email protected]>

Author: kaiyux

tensorrt_llm/serve/scripts/__init__.py

@@ -26,10 +26,11 @@
 
 import numpy as np
 import pandas as pd
-from benchmark_utils import download_and_cache_file
 from datasets import load_dataset
 from transformers import PreTrainedTokenizerBase
 
+from .benchmark_utils import download_and_cache_file
+
 logger = logging.getLogger(__name__)
 
 # -----------------------------------------------------------------------------

tensorrt_llm/serve/scripts/benchmark_dataset.py

@@ -31,18 +31,19 @@
 from typing import Any, Optional
 
 import numpy as np
-from backend_request_func import (ASYNC_REQUEST_FUNCS,
-                                  OPENAI_COMPATIBLE_BACKENDS, RequestFuncInput,
-                                  RequestFuncOutput, get_tokenizer)
-from benchmark_dataset import (AIMODataset, BurstGPTDataset,
-                               ConversationDataset, HuggingFaceDataset,
-                               InstructCoderDataset, RandomDataset,
-                               SampleRequest, ShareGPTDataset, SonnetDataset,
-                               VisionArenaDataset)
-from benchmark_utils import convert_to_pytorch_benchmark_format, write_to_json
 from tqdm.asyncio import tqdm
 from transformers import PreTrainedTokenizerBase
 
+from .backend_request_func import (ASYNC_REQUEST_FUNCS,
+                                   OPENAI_COMPATIBLE_BACKENDS, RequestFuncInput,
+                                   RequestFuncOutput, get_tokenizer)
+from .benchmark_dataset import (AIMODataset, BurstGPTDataset,
+                                ConversationDataset, HuggingFaceDataset,
+                                InstructCoderDataset, RandomDataset,
+                                SampleRequest, ShareGPTDataset, SonnetDataset,
+                                VisionArenaDataset)
+from .benchmark_utils import convert_to_pytorch_benchmark_format, write_to_json
+
 MILLISECONDS_TO_SECONDS_CONVERSION = 1000