[Chore] Clean up pytorch helper functions in vllm.utils (#26908)
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn> Signed-off-by: isotr0py <2037008807@qq.com>
This commit is contained in:
@@ -6,7 +6,8 @@ from typing import TYPE_CHECKING
|
||||
import vllm.envs as envs
|
||||
from vllm.logger import init_logger
|
||||
from vllm.model_executor.models import ModelRegistry
|
||||
from vllm.utils import STR_DTYPE_TO_TORCH_DTYPE, cdiv, round_up
|
||||
from vllm.utils import cdiv, round_up
|
||||
from vllm.utils.torch_utils import STR_DTYPE_TO_TORCH_DTYPE
|
||||
from vllm.v1.kv_cache_interface import FullAttentionSpec, MambaSpec
|
||||
|
||||
if TYPE_CHECKING:
|
||||
|
||||
@@ -79,8 +79,8 @@ from vllm.model_executor.model_loader.weight_utils import (
|
||||
from vllm.model_executor.models.utils import sequence_parallel_chunk
|
||||
from vllm.platforms import current_platform
|
||||
from vllm.sequence import IntermediateTensors
|
||||
from vllm.utils import direct_register_custom_op
|
||||
from vllm.utils.deep_gemm import fp8_mqa_logits, fp8_paged_mqa_logits
|
||||
from vllm.utils.torch_utils import direct_register_custom_op
|
||||
from vllm.v1.attention.backends.mla.indexer import (
|
||||
DeepseekV32IndexerBackend,
|
||||
DeepseekV32IndexerMetadata,
|
||||
|
||||
@@ -18,7 +18,6 @@ from vllm.config import VllmConfig
|
||||
from vllm.config.multimodal import BaseDummyOptions
|
||||
from vllm.distributed import get_tensor_model_parallel_world_size
|
||||
from vllm.model_executor.layers.quantization import QuantizationConfig
|
||||
from vllm.model_executor.model_loader.utils import set_default_torch_dtype
|
||||
from vllm.model_executor.models.transformers.utils import replace_linear_class
|
||||
from vllm.multimodal import MULTIMODAL_REGISTRY
|
||||
from vllm.multimodal.inputs import (
|
||||
@@ -51,6 +50,7 @@ from vllm.transformers_utils.processors.deepseek_vl2 import DeepseekVLV2Processo
|
||||
from vllm.transformers_utils.tokenizer import cached_tokenizer_from_config
|
||||
from vllm.utils.collections import is_list_of
|
||||
from vllm.utils.tensor_schema import TensorSchema, TensorShape
|
||||
from vllm.utils.torch_utils import set_default_torch_dtype
|
||||
|
||||
from .interfaces import MultiModalEmbeddings, SupportsMultiModal, SupportsPP
|
||||
from .utils import (
|
||||
|
||||
@@ -51,8 +51,8 @@ from vllm.multimodal.processing import (
|
||||
from vllm.multimodal.profiling import BaseDummyInputsBuilder
|
||||
from vllm.sequence import IntermediateTensors
|
||||
from vllm.transformers_utils.tokenizer import AnyTokenizer
|
||||
from vllm.utils import set_default_torch_num_threads
|
||||
from vllm.utils.tensor_schema import TensorSchema, TensorShape
|
||||
from vllm.utils.torch_utils import set_default_torch_num_threads
|
||||
|
||||
from .interfaces import (
|
||||
MultiModalEmbeddings,
|
||||
|
||||
@@ -49,7 +49,6 @@ from vllm.model_executor.layers.resampler import (
|
||||
Resampler2,
|
||||
get_2d_sincos_pos_embed,
|
||||
)
|
||||
from vllm.model_executor.model_loader.utils import set_default_torch_dtype
|
||||
from vllm.model_executor.models.llama import LlamaForCausalLM
|
||||
from vllm.model_executor.models.minicpm import MiniCPMForCausalLM
|
||||
from vllm.model_executor.models.module_mapping import MultiModelKeys
|
||||
@@ -88,6 +87,7 @@ from vllm.platforms import current_platform
|
||||
from vllm.sequence import IntermediateTensors
|
||||
from vllm.utils.collections import flatten_2d_lists
|
||||
from vllm.utils.tensor_schema import TensorSchema, TensorShape
|
||||
from vllm.utils.torch_utils import set_default_torch_dtype
|
||||
|
||||
from .idefics2_vision_model import Idefics2VisionTransformer
|
||||
from .interfaces import (
|
||||
|
||||
@@ -64,7 +64,7 @@ from vllm.model_executor.models.utils import (
|
||||
)
|
||||
from vllm.model_executor.utils import set_weight_attrs
|
||||
from vllm.sequence import IntermediateTensors
|
||||
from vllm.utils import direct_register_custom_op
|
||||
from vllm.utils.torch_utils import direct_register_custom_op
|
||||
from vllm.v1.attention.backends.mamba2_attn import Mamba2AttentionMetadata
|
||||
|
||||
|
||||
|
||||
@@ -71,7 +71,7 @@ from vllm.platforms import current_platform
|
||||
from vllm.sequence import IntermediateTensors
|
||||
from vllm.transformers_utils.configs import Qwen3NextConfig
|
||||
from vllm.triton_utils import tl, triton
|
||||
from vllm.utils import direct_register_custom_op
|
||||
from vllm.utils.torch_utils import direct_register_custom_op
|
||||
from vllm.v1.attention.backends.gdn_attn import GDNAttentionMetadata
|
||||
|
||||
from .interfaces import (
|
||||
|
||||
@@ -29,7 +29,7 @@ from vllm.model_executor.layers.fused_moe import FusedMoE
|
||||
from vllm.model_executor.models.interfaces import MixtureOfExperts
|
||||
from vllm.model_executor.models.utils import maybe_prefix
|
||||
from vllm.platforms import current_platform
|
||||
from vllm.utils import direct_register_custom_op
|
||||
from vllm.utils.torch_utils import direct_register_custom_op
|
||||
|
||||
from .utils import log_replacement
|
||||
|
||||
|
||||
@@ -24,11 +24,13 @@ from vllm.multimodal import NestedTensors
|
||||
from vllm.sequence import IntermediateTensors
|
||||
from vllm.utils import (
|
||||
cdiv,
|
||||
direct_register_custom_op,
|
||||
get_cuda_view_from_cpu_tensor,
|
||||
is_pin_memory_available,
|
||||
is_uva_available,
|
||||
)
|
||||
from vllm.utils.torch_utils import (
|
||||
direct_register_custom_op,
|
||||
get_cuda_view_from_cpu_tensor,
|
||||
)
|
||||
|
||||
logger = init_logger(__name__)
|
||||
|
||||
|
||||
@@ -34,7 +34,6 @@ from vllm.model_executor.layers.linear import (
|
||||
from vllm.model_executor.layers.logits_processor import LogitsProcessor
|
||||
from vllm.model_executor.layers.quantization import QuantizationConfig
|
||||
from vllm.model_executor.layers.vocab_parallel_embedding import ParallelLMHead
|
||||
from vllm.model_executor.model_loader.utils import set_default_torch_dtype
|
||||
from vllm.model_executor.model_loader.weight_utils import default_weight_loader
|
||||
from vllm.multimodal import MULTIMODAL_REGISTRY
|
||||
from vllm.multimodal.inputs import (
|
||||
@@ -53,6 +52,7 @@ from vllm.multimodal.profiling import BaseDummyInputsBuilder
|
||||
from vllm.transformers_utils.processor import cached_get_processor
|
||||
from vllm.utils.jsontree import json_map_leaves
|
||||
from vllm.utils.tensor_schema import TensorSchema, TensorShape
|
||||
from vllm.utils.torch_utils import set_default_torch_dtype
|
||||
|
||||
from .interfaces import MultiModalEmbeddings, SupportsMultiModal, SupportsTranscription
|
||||
from .utils import (
|
||||
|
||||
Reference in New Issue
Block a user