[Frontend] Add FlexibleArgumentParser to support both underscore and dash in names (#5718)

benchmarks/kernels/benchmark_aqlm.py

@@ -1,4 +1,3 @@
-import argparse
 import os
 import sys
 from typing import Optional
@@ -10,6 +9,7 @@ from vllm import _custom_ops as ops
 from vllm.model_executor.layers.quantization.aqlm import (
     dequantize_weight, generic_dequantize_gemm, get_int_dtype,
     optimized_dequantize_gemm)
+from vllm.utils import FlexibleArgumentParser
 
 os.environ['CUDA_VISIBLE_DEVICES'] = '0'
 
@@ -137,7 +137,7 @@ def dequant_test(k: int, parts: torch.Tensor, nbooks: int, bits: int) -> None:
 
 def main():
 
-    parser = argparse.ArgumentParser(description="Benchmark aqlm performance.")
+    parser = FlexibleArgumentParser(description="Benchmark aqlm performance.")
 
     # Add arguments
     parser.add_argument("--nbooks",

benchmarks/kernels/benchmark_marlin.py

@@ -1,4 +1,3 @@
-import argparse
 from typing import List
 
 import torch
@@ -16,6 +15,7 @@ from vllm.model_executor.layers.quantization.utils.marlin_utils import (
     MarlinWorkspace, marlin_24_quantize, marlin_quantize)
 from vllm.model_executor.layers.quantization.utils.quant_utils import (
     gptq_pack, quantize_weights, sort_weights)
+from vllm.utils import FlexibleArgumentParser
 
 DEFAULT_MODELS = ["meta-llama/Llama-2-7b-hf/TP1"]
 DEFAULT_BATCH_SIZES = [1, 16, 32, 64, 128, 256, 512]
@@ -211,7 +211,7 @@ def main(args):
 #   python benchmark_marlin.py --batch-sizes 1 16 32 --limit-k 4096 --limit-n 4096 --limit-group-size 128 --limit-num-bits 4 --limit-act-order 0 --limit-k-full 1 # noqa E501
 #
 if __name__ == "__main__":
-    parser = argparse.ArgumentParser(
+    parser = FlexibleArgumentParser(
         description="Benchmark Marlin across specified models/shapes/batches")
     parser.add_argument(
         "--models",

benchmarks/kernels/benchmark_moe.py

@@ -10,6 +10,7 @@ from ray.experimental.tqdm_ray import tqdm
 from transformers import AutoConfig
 
 from vllm.model_executor.layers.fused_moe.fused_moe import *
+from vllm.utils import FlexibleArgumentParser
 
 
 class BenchmarkConfig(TypedDict):
@@ -315,7 +316,7 @@ def main(args: argparse.Namespace):
 
 
 if __name__ == "__main__":
-    parser = argparse.ArgumentParser()
+    parser = FlexibleArgumentParser()
     parser.add_argument("--model",
                         type=str,
                         default="mistralai/Mixtral-8x7B-Instruct-v0.1")

benchmarks/kernels/benchmark_paged_attention.py

@@ -1,4 +1,3 @@
-import argparse
 import random
 import time
 from typing import List, Optional
@@ -6,7 +5,8 @@ from typing import List, Optional
 import torch
 
 from vllm import _custom_ops as ops
-from vllm.utils import STR_DTYPE_TO_TORCH_DTYPE, create_kv_caches_with_random
+from vllm.utils import (STR_DTYPE_TO_TORCH_DTYPE, FlexibleArgumentParser,
+                        create_kv_caches_with_random)
 
 NUM_BLOCKS = 1024
 PARTITION_SIZE = 512
@@ -161,7 +161,7 @@ def main(
 
 
 if __name__ == '__main__':
-    parser = argparse.ArgumentParser(
+    parser = FlexibleArgumentParser(
         description="Benchmark the paged attention kernel.")
     parser.add_argument("--version",
                         type=str,

benchmarks/kernels/benchmark_rope.py

@@ -1,4 +1,3 @@
-import argparse
 from itertools import accumulate
 from typing import List, Optional
 
@@ -7,6 +6,7 @@ import torch
 
 from vllm.model_executor.layers.rotary_embedding import (RotaryEmbedding,
                                                          get_rope)
+from vllm.utils import FlexibleArgumentParser
 
 
 def benchmark_rope_kernels_multi_lora(
@@ -86,7 +86,7 @@ def benchmark_rope_kernels_multi_lora(
 
 
 if __name__ == '__main__':
-    parser = argparse.ArgumentParser(
+    parser = FlexibleArgumentParser(
         description="Benchmark the rotary embedding kernels.")
     parser.add_argument("--is-neox-style", type=bool, default=True)
     parser.add_argument("--batch-size", type=int, default=16)