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

vllm/model_executor/model_loader/tensorizer.py

@@ -21,6 +21,7 @@ from vllm.model_executor.layers.quantization.base_config import (
     QuantizationConfig)
 from vllm.model_executor.layers.vocab_parallel_embedding import (
     VocabParallelEmbedding)
+from vllm.utils import FlexibleArgumentParser
 
 tensorizer_error_msg = None
 
@@ -177,8 +178,7 @@ class TensorizerArgs:
                 self.deserializer_params['encryption'] = decryption_params
 
     @staticmethod
-    def add_cli_args(
-            parser: argparse.ArgumentParser) -> argparse.ArgumentParser:
+    def add_cli_args(parser: FlexibleArgumentParser) -> FlexibleArgumentParser:
         """Tensorizer CLI arguments"""
 
         # Tensorizer options arg group

vllm/utils.py

+import argparse
 import asyncio
 import datetime
 import enum
@@ -775,3 +776,21 @@ def run_once(f):
 
     wrapper.has_run = False  # type: ignore[attr-defined]
     return wrapper
+
+
+class FlexibleArgumentParser(argparse.ArgumentParser):
+    """ArgumentParser that allows both underscore and dash in names."""
+
+    def parse_args(self, args=None, namespace=None):
+        if args is None:
+            args = sys.argv[1:]
+
+        # Convert underscores to dashes and vice versa in argument names
+        processed_args = []
+        for arg in args:
+            if arg.startswith('--'):
+                processed_args.append('--' + arg[len('--'):].replace('_', '-'))
+            else:
+                processed_args.append(arg)
+
+        return super().parse_args(processed_args, namespace)