[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>

vllm/entrypoints/api_server.py

@@ -26,7 +26,8 @@ from vllm.entrypoints.utils import with_cancellation
 from vllm.logger import init_logger
 from vllm.sampling_params import SamplingParams
 from vllm.usage.usage_lib import UsageContext
-from vllm.utils import FlexibleArgumentParser, random_uuid, set_ulimit
+from vllm.utils import random_uuid, set_ulimit
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 from vllm.version import __version__ as VLLM_VERSION
 
 logger = init_logger("vllm.entrypoints.api_server")

vllm/entrypoints/cli/benchmark/main.py

@@ -9,7 +9,7 @@ from vllm.entrypoints.cli.types import CLISubcommand
 from vllm.entrypoints.utils import VLLM_SUBCMD_PARSER_EPILOG
 
 if typing.TYPE_CHECKING:
-    from vllm.utils import FlexibleArgumentParser
+    from vllm.utils.argparse_utils import FlexibleArgumentParser
 else:
     FlexibleArgumentParser = argparse.ArgumentParser
 

vllm/entrypoints/cli/collect_env.py

@@ -8,7 +8,7 @@ from vllm.collect_env import main as collect_env_main
 from vllm.entrypoints.cli.types import CLISubcommand
 
 if typing.TYPE_CHECKING:
-    from vllm.utils import FlexibleArgumentParser
+    from vllm.utils.argparse_utils import FlexibleArgumentParser
 else:
     FlexibleArgumentParser = argparse.ArgumentParser
 

vllm/entrypoints/cli/main.py

@@ -20,7 +20,7 @@ def main():
     import vllm.entrypoints.cli.run_batch
     import vllm.entrypoints.cli.serve
     from vllm.entrypoints.utils import VLLM_SUBCMD_PARSER_EPILOG, cli_env_setup
-    from vllm.utils import FlexibleArgumentParser
+    from vllm.utils.argparse_utils import FlexibleArgumentParser
 
     CMD_MODULES = [
         vllm.entrypoints.cli.openai,

vllm/entrypoints/cli/openai.py

@@ -13,7 +13,7 @@ from openai.types.chat import ChatCompletionMessageParam
 from vllm.entrypoints.cli.types import CLISubcommand
 
 if TYPE_CHECKING:
-    from vllm.utils import FlexibleArgumentParser
+    from vllm.utils.argparse_utils import FlexibleArgumentParser
 else:
     FlexibleArgumentParser = argparse.ArgumentParser
 

vllm/entrypoints/cli/run_batch.py

@@ -11,7 +11,7 @@ from vllm.entrypoints.utils import VLLM_SUBCMD_PARSER_EPILOG
 from vllm.logger import init_logger
 
 if typing.TYPE_CHECKING:
-    from vllm.utils import FlexibleArgumentParser
+    from vllm.utils.argparse_utils import FlexibleArgumentParser
 else:
     FlexibleArgumentParser = argparse.ArgumentParser
 

vllm/entrypoints/cli/types.py

@@ -5,7 +5,7 @@ import argparse
 import typing
 
 if typing.TYPE_CHECKING:
-    from vllm.utils import FlexibleArgumentParser
+    from vllm.utils.argparse_utils import FlexibleArgumentParser
 else:
     FlexibleArgumentParser = argparse.ArgumentParser
 

vllm/entrypoints/openai/cli_args.py

@@ -29,7 +29,7 @@ from vllm.entrypoints.constants import (
 from vllm.entrypoints.openai.serving_models import LoRAModulePath
 from vllm.entrypoints.openai.tool_parsers import ToolParserManager
 from vllm.logger import init_logger
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 logger = init_logger(__name__)
 

vllm/entrypoints/openai/run_batch.py

@@ -32,7 +32,8 @@ from vllm.entrypoints.openai.serving_models import BaseModelPath, OpenAIServingM
 from vllm.entrypoints.openai.serving_score import ServingScores
 from vllm.logger import init_logger
 from vllm.reasoning import ReasoningParserManager
-from vllm.utils import FlexibleArgumentParser, random_uuid
+from vllm.utils import random_uuid
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 from vllm.version import __version__ as VLLM_VERSION
 
 logger = init_logger(__name__)

vllm/entrypoints/utils.py

@@ -31,7 +31,7 @@ from vllm.entrypoints.openai.serving_models import LoRAModulePath
 from vllm.logger import init_logger
 from vllm.platforms import current_platform
 from vllm.transformers_utils.tokenizers import MistralTokenizer
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 
 logger = init_logger(__name__)
 

vllm/model_executor/layers/fla/ops/layernorm_guard.py

@@ -21,7 +21,7 @@ import torch.nn.functional as F
 from einops import rearrange
 
 from vllm.triton_utils import tl, triton
-from vllm.utils import cdiv, next_power_of_2
+from vllm.utils.math_utils import cdiv, next_power_of_2
 
 from .utils import input_guard
 

vllm/model_executor/layers/fused_moe/config.py

@@ -14,9 +14,9 @@ from vllm.model_executor.layers.quantization.utils.ocp_mx_utils import (
     OCP_MX_Scheme,
 )
 from vllm.model_executor.layers.quantization.utils.quant_utils import GroupShape
-from vllm.utils import cdiv
 from vllm.utils.flashinfer import has_flashinfer_cutlass_fused_moe
 from vllm.utils.import_utils import has_triton_kernels
+from vllm.utils.math_utils import cdiv
 
 logger = init_logger(__name__)
 

vllm/model_executor/layers/fused_moe/deep_gemm_utils.py

@@ -10,8 +10,8 @@ import torch
 import vllm.model_executor.layers.fused_moe.modular_kernel as mk
 from vllm.model_executor.layers.fused_moe.utils import count_expert_num_tokens
 from vllm.triton_utils import tl, triton
-from vllm.utils import round_up
 from vllm.utils.deep_gemm import get_mk_alignment_for_contiguous_layout
+from vllm.utils.math_utils import round_up
 
 
 def expert_num_tokens_round_up_and_sum(

vllm/model_executor/layers/fused_moe/deepep_ht_prepare_finalize.py

@@ -12,7 +12,7 @@ from vllm.model_executor.layers.fused_moe.topk_weight_and_reduce import (
     TopKWeightAndReduceDelegate,
 )
 from vllm.model_executor.layers.fused_moe.utils import moe_kernel_quantize_input
-from vllm.utils import round_up
+from vllm.utils.math_utils import round_up
 from vllm.v1.worker.ubatching import (
     dbo_current_ubatch_id,
     dbo_enabled,

vllm/model_executor/layers/fused_moe/layer.py

@@ -55,9 +55,9 @@ from vllm.model_executor.layers.quantization.utils.flashinfer_utils import (
 from vllm.model_executor.utils import set_weight_attrs
 from vllm.platforms import current_platform
 from vllm.platforms.interface import CpuArchEnum
-from vllm.utils import cdiv, round_up
 from vllm.utils.flashinfer import has_flashinfer_cutlass_fused_moe
 from vllm.utils.import_utils import has_deep_ep, has_pplx
+from vllm.utils.math_utils import cdiv, round_up
 from vllm.utils.torch_utils import current_stream, direct_register_custom_op
 from vllm.v1.worker.ubatching import dbo_current_ubatch_id
 

vllm/model_executor/layers/fused_moe/modular_kernel.py

@@ -16,7 +16,7 @@ from vllm.model_executor.layers.fused_moe.utils import (
     count_expert_num_tokens,
     disable_inplace,
 )
-from vllm.utils import cdiv
+from vllm.utils.math_utils import cdiv
 from vllm.v1.worker.ubatching import (
     dbo_current_ubatch_id,
     dbo_enabled,

vllm/model_executor/layers/fused_moe/moe_align_block_size.py

@@ -5,7 +5,7 @@ import torch
 
 from vllm import _custom_ops as ops
 from vllm.triton_utils import triton
-from vllm.utils import round_up
+from vllm.utils.math_utils import round_up
 
 
 def moe_align_block_size(

vllm/model_executor/layers/fused_moe/pplx_prepare_finalize.py

@@ -15,7 +15,7 @@ from vllm.model_executor.layers.fused_moe.utils import (
     _validate_scale_shape,
     moe_kernel_quantize_input,
 )
-from vllm.utils import cdiv, round_up
+from vllm.utils.math_utils import cdiv, round_up
 
 logger = init_logger(__name__)
 

vllm/model_executor/layers/fused_moe/utils.py

@@ -23,8 +23,8 @@ from vllm.model_executor.layers.quantization.utils.mxfp8_utils import (
     mxfp8_e4m3_quantize,
 )
 from vllm.triton_utils import tl, triton
-from vllm.utils import cdiv
 from vllm.utils.flashinfer import flashinfer_fp4_quantize
+from vllm.utils.math_utils import cdiv
 from vllm.utils.torch_utils import is_torch_equal_or_newer
 
 

vllm/model_executor/model_loader/tensorizer.py

@@ -26,7 +26,7 @@ from vllm.config import ModelConfig, ParallelConfig, VllmConfig, set_current_vll
 from vllm.logger import init_logger
 from vllm.model_executor.layers.vocab_parallel_embedding import VocabParallelEmbedding
 from vllm.platforms import current_platform
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils.argparse_utils import FlexibleArgumentParser
 from vllm.utils.import_utils import PlaceholderModule
 
 if TYPE_CHECKING: