[Chore]:Extract math and argparse utilities to separate modules (#27188)

Signed-off-by: Yeshwanth Surya <yeshsurya@gmail.com>
Signed-off-by: Yeshwanth N <yeshsurya@gmail.com>
Signed-off-by: yeshsurya <yeshsurya@gmail.com>

examples/offline_inference/llm_engine_example.py

@@ -8,7 +8,7 @@ for processing prompts with various sampling parameters.
 import argparse
 
 from vllm import EngineArgs, LLMEngine, RequestOutput, SamplingParams
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 
 def create_test_prompts() -> list[tuple[str, SamplingParams]]:

examples/offline_inference/load_sharded_state.py

@@ -25,7 +25,7 @@ python load_sharded_state.py \
 import dataclasses
 
 from vllm import LLM, EngineArgs, SamplingParams
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 
 def parse_args():

examples/offline_inference/pooling/embed_jina_embeddings_v3.py

@@ -4,7 +4,7 @@
 from argparse import Namespace
 
 from vllm import LLM, EngineArgs
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 
 def parse_args():

examples/offline_inference/pooling/embed_matryoshka_fy.py

@@ -4,7 +4,7 @@
 from argparse import Namespace
 
 from vllm import LLM, EngineArgs, PoolingParams
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 
 def parse_args():

examples/offline_inference/pooling/multi_vector_retrieval.py

@@ -4,7 +4,7 @@
 from argparse import Namespace
 
 from vllm import LLM, EngineArgs
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 
 def parse_args():

examples/offline_inference/pooling/ner.py

@@ -5,7 +5,7 @@
 from argparse import Namespace
 
 from vllm import LLM, EngineArgs
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 
 def parse_args():

examples/offline_inference/profiling_tpu/profiling.py

@@ -13,7 +13,7 @@ from tqdm import tqdm
 from vllm import LLM, SamplingParams
 from vllm.engine.arg_utils import EngineArgs
 from vllm.inputs import PromptType
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 DURATION_MS = int(os.getenv("VLLM_TPU_PROFILE_DURATION_MS", 3000))
 DELAY_MS = int(os.getenv("VLLM_TPU_PROFILE_DELAY_MS", 0))

examples/offline_inference/qwen2_5_omni/only_thinker.py

@@ -13,7 +13,7 @@ from vllm.assets.audio import AudioAsset
 from vllm.assets.image import ImageAsset
 from vllm.assets.video import VideoAsset
 from vllm.multimodal.image import convert_image_mode
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 
 class QueryResult(NamedTuple):

examples/offline_inference/save_sharded_state.py

@@ -30,7 +30,7 @@ from pathlib import Path
 
 from vllm import LLM, EngineArgs
 from vllm.model_executor.model_loader import ShardedStateLoader
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 
 def parse_args():

examples/offline_inference/spec_decode.py

@@ -9,7 +9,7 @@ from vllm.inputs import TokensPrompt
 from vllm.v1.metrics.reader import Counter, Vector
 
 try:
-    from vllm.utils import FlexibleArgumentParser
+    from vllm.utils.argparse_utils import FlexibleArgumentParser
 except ImportError:
     from argparse import ArgumentParser as FlexibleArgumentParser
 

examples/offline_inference/vision_language.py

@@ -22,7 +22,7 @@ from vllm.assets.image import ImageAsset
 from vllm.assets.video import VideoAsset
 from vllm.lora.request import LoRARequest
 from vllm.multimodal.image import convert_image_mode
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 
 class ModelRequestData(NamedTuple):

examples/offline_inference/vision_language_multi_image.py

@@ -18,7 +18,7 @@ from transformers import AutoProcessor, AutoTokenizer
 from vllm import LLM, EngineArgs, SamplingParams
 from vllm.lora.request import LoRARequest
 from vllm.multimodal.utils import fetch_image
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 QUESTION = "What is the content of each image?"
 IMAGE_URLS = [

examples/offline_inference/vision_language_pooling.py

@@ -18,7 +18,7 @@ from PIL.Image import Image
 from vllm import LLM, EngineArgs
 from vllm.entrypoints.score_utils import ScoreMultiModalParam
 from vllm.multimodal.utils import fetch_image
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 ROOT_DIR = Path(__file__).parent.parent.parent
 EXAMPLES_DIR = ROOT_DIR / "examples"

examples/online_serving/openai_chat_completion_client_for_multimodal.py

@@ -26,7 +26,7 @@ import requests
 from openai import OpenAI
 from utils import get_first_model
 
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 # Modify OpenAI's API key and API base to use vLLM's API server.
 openai_api_key = "EMPTY"

examples/others/tensorize_vllm_model.py

@@ -16,7 +16,7 @@ from vllm.model_executor.model_loader.tensorizer import (
     tensorize_vllm_model,
     tensorizer_kwargs_arg,
 )
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 logger = logging.getLogger()
 

tests/engine/test_arg_utils.py

@@ -22,7 +22,7 @@ from vllm.engine.arg_utils import (
     optional_type,
     parse_type,
 )
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 
 @pytest.mark.parametrize(

tests/entrypoints/openai/test_cli_args.py

@@ -7,7 +7,7 @@ import pytest
 
 from vllm.entrypoints.openai.cli_args import make_arg_parser, validate_parsed_serve_args
 from vllm.entrypoints.openai.serving_models import LoRAModulePath
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 from ...utils import VLLM_PATH
 

tests/kernels/attention/test_deepgemm_attention.py

@@ -6,7 +6,6 @@ import pytest
 import torch
 
 from vllm.platforms import current_platform
-from vllm.utils import cdiv
 from vllm.utils.deep_gemm import (
     _ceil_to_ue8m0,
     calc_diff,
@@ -16,6 +15,7 @@ from vllm.utils.deep_gemm import (
     get_paged_mqa_logits_metadata,
 )
 from vllm.utils.import_utils import has_deep_gemm
+from vllm.utils.math_utils import cdiv
 
 
 def kv_cache_cast_to_fp8(x: torch.Tensor) -> torch.Tensor:

tests/kernels/attention/test_flashinfer_trtllm_attention.py

@@ -10,7 +10,7 @@ from tests.kernels.quantization.nvfp4_utils import (
     get_nvfp4_global_scale,
 )
 from vllm.platforms import current_platform
-from vllm.utils import round_up
+from vllm.utils.math_utils import round_up
 
 if not current_platform.is_device_capability(100):
     pytest.skip(

tests/kernels/attention/test_mla_decode_cpu.py

@@ -7,7 +7,7 @@ from torch import Tensor
 
 import vllm._custom_ops as ops
 from vllm.platforms import current_platform
-from vllm.utils import cdiv
+from vllm.utils.math_utils import cdiv
 
 
 def ref_mla(