Remove all references to yapf as it's no longer used (#26251)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
@@ -14,8 +14,6 @@ import vllm.envs as envs
|
||||
import vllm.model_executor.layers.fused_moe.modular_kernel as mk
|
||||
from vllm import _custom_ops as ops
|
||||
from vllm.logger import init_logger
|
||||
|
||||
# yapf: disable
|
||||
from vllm.model_executor.layers.fused_moe.config import (
|
||||
FUSED_MOE_UNQUANTIZED_CONFIG,
|
||||
FusedMoEQuantConfig,
|
||||
@@ -25,8 +23,6 @@ from vllm.model_executor.layers.fused_moe.cutlass_moe import (
|
||||
_valid_cutlass_block_scaled_grouped_gemm,
|
||||
run_cutlass_block_scaled_fused_experts,
|
||||
)
|
||||
|
||||
# yapf: enable
|
||||
from vllm.model_executor.layers.fused_moe.deep_gemm_moe import (
|
||||
_valid_deep_gemm,
|
||||
deep_gemm_moe_fp8,
|
||||
|
||||
@@ -24,8 +24,6 @@ from vllm.distributed.eplb.eplb_state import EplbState
|
||||
from vllm.forward_context import ForwardContext, get_forward_context
|
||||
from vllm.logger import init_logger
|
||||
from vllm.model_executor.custom_op import CustomOp
|
||||
|
||||
# yapf: disable
|
||||
from vllm.model_executor.layers.fused_moe.config import (
|
||||
FUSED_MOE_UNQUANTIZED_CONFIG,
|
||||
FusedMoEConfig,
|
||||
@@ -34,8 +32,6 @@ from vllm.model_executor.layers.fused_moe.config import (
|
||||
biased_moe_quant_config,
|
||||
)
|
||||
from vllm.model_executor.layers.fused_moe.fused_moe import zero_experts_compute_triton
|
||||
|
||||
# yapf: enable
|
||||
from vllm.model_executor.layers.fused_moe.modular_kernel import (
|
||||
FusedMoEActivationFormat,
|
||||
FusedMoEModularKernel,
|
||||
|
||||
@@ -10,7 +10,7 @@ import torch
|
||||
|
||||
import vllm.envs as envs
|
||||
from vllm.model_executor.layers.fused_moe.config import FusedMoEQuantConfig
|
||||
from vllm.model_executor.layers.fused_moe.utils import ( # yapf: disable
|
||||
from vllm.model_executor.layers.fused_moe.utils import (
|
||||
_resize_cache,
|
||||
count_expert_num_tokens,
|
||||
)
|
||||
|
||||
@@ -24,8 +24,6 @@ from vllm.model_executor.layers.quantization.base_config import (
|
||||
QuantizeMethodBase,
|
||||
)
|
||||
from vllm.model_executor.layers.utils import dispatch_unquantized_gemm
|
||||
|
||||
# yapf: disable
|
||||
from vllm.model_executor.parameter import (
|
||||
BasevLLMParameter,
|
||||
BlockQuantScaleParameter,
|
||||
@@ -35,8 +33,6 @@ from vllm.model_executor.parameter import (
|
||||
PerTensorScaleParameter,
|
||||
RowvLLMParameter,
|
||||
)
|
||||
|
||||
# yapf: enable
|
||||
from vllm.model_executor.utils import set_weight_attrs
|
||||
from vllm.platforms import current_platform
|
||||
from vllm.utils import GiB_bytes
|
||||
|
||||
-5
@@ -17,17 +17,12 @@ from vllm.model_executor.layers.quantization.kernels.mixed_precision import (
|
||||
from vllm.model_executor.layers.quantization.utils.marlin_utils import (
|
||||
marlin_repeat_scales_on_all_ranks,
|
||||
)
|
||||
|
||||
# yapf conflicts with isort for this block
|
||||
# yapf: disable
|
||||
from vllm.model_executor.parameter import (
|
||||
BasevLLMParameter,
|
||||
ChannelQuantScaleParameter,
|
||||
GroupQuantScaleParameter,
|
||||
PackedvLLMParameter,
|
||||
)
|
||||
|
||||
# yapf: enable
|
||||
from vllm.scalar_type import scalar_types
|
||||
|
||||
logger = init_logger(__name__)
|
||||
|
||||
-5
@@ -17,9 +17,6 @@ from vllm.model_executor.layers.quantization.kernels.mixed_precision import (
|
||||
from vllm.model_executor.layers.quantization.utils.marlin_utils import (
|
||||
marlin_repeat_scales_on_all_ranks,
|
||||
)
|
||||
|
||||
# yapf conflicts with isort for this block
|
||||
# yapf: disable
|
||||
from vllm.model_executor.parameter import (
|
||||
BasevLLMParameter,
|
||||
ChannelQuantScaleParameter,
|
||||
@@ -28,8 +25,6 @@ from vllm.model_executor.parameter import (
|
||||
PackedvLLMParameter,
|
||||
RowvLLMParameter,
|
||||
)
|
||||
|
||||
# yapf: enable
|
||||
from vllm.scalar_type import scalar_types
|
||||
|
||||
logger = init_logger(__name__)
|
||||
|
||||
@@ -22,8 +22,6 @@ from vllm.distributed import (
|
||||
get_tensor_model_parallel_rank,
|
||||
get_tensor_model_parallel_world_size,
|
||||
)
|
||||
|
||||
# yapf: enable
|
||||
from vllm.logger import init_logger
|
||||
from vllm.model_executor.layers.fused_moe import FusedMoE
|
||||
from vllm.model_executor.layers.linear import (
|
||||
@@ -51,8 +49,6 @@ from vllm.model_executor.utils import (
|
||||
)
|
||||
from vllm.platforms import current_platform
|
||||
|
||||
# yapf conflicts with isort for this block
|
||||
|
||||
logger = init_logger(__name__)
|
||||
|
||||
|
||||
|
||||
@@ -39,13 +39,10 @@ from vllm.multimodal.profiling import BaseDummyInputsBuilder
|
||||
from vllm.sequence import IntermediateTensors
|
||||
from vllm.utils.tensor_schema import TensorSchema, TensorShape
|
||||
|
||||
# yapf: disable
|
||||
from .idefics2_vision_model import Idefics2VisionConfig
|
||||
from .idefics2_vision_model import (
|
||||
Idefics2VisionTransformer as Idefics3VisionTransformer,
|
||||
)
|
||||
|
||||
# yapf: enable
|
||||
from .interfaces import MultiModalEmbeddings, SupportsMultiModal, SupportsQuant
|
||||
from .llama import LlamaDecoderLayer, LlamaMLP, LlamaModel
|
||||
from .utils import (
|
||||
|
||||
@@ -22,8 +22,6 @@ from vllm.multimodal.inputs import (
|
||||
MultiModalKwargsItems,
|
||||
)
|
||||
from vllm.multimodal.parse import ImageProcessorItems, ImageSize, MultiModalDataItems
|
||||
|
||||
# yapf: disable
|
||||
from vllm.multimodal.processing import (
|
||||
BaseMultiModalProcessor,
|
||||
BaseProcessingInfo,
|
||||
@@ -35,8 +33,6 @@ from vllm.multimodal.processing import (
|
||||
PromptUpdateDetails,
|
||||
replace_token_matches,
|
||||
)
|
||||
|
||||
# yapf: enable
|
||||
from vllm.multimodal.profiling import BaseDummyInputsBuilder
|
||||
from vllm.sequence import IntermediateTensors
|
||||
from vllm.utils.tensor_schema import TensorSchema, TensorShape
|
||||
|
||||
@@ -6,14 +6,16 @@ from typing import Annotated, Any, Literal, Optional, Union, cast
|
||||
import numpy as np
|
||||
import torch
|
||||
|
||||
# yapf: disable
|
||||
from torch import nn
|
||||
from transformers import AutoModel, BatchFeature
|
||||
from transformers.models.gemma3n import (Gemma3nAudioConfig,
|
||||
Gemma3nAudioFeatureExtractor,
|
||||
Gemma3nConfig, Gemma3nProcessor,
|
||||
Gemma3nTextConfig,
|
||||
Gemma3nVisionConfig)
|
||||
from transformers.models.gemma3n import (
|
||||
Gemma3nAudioConfig,
|
||||
Gemma3nAudioFeatureExtractor,
|
||||
Gemma3nConfig,
|
||||
Gemma3nProcessor,
|
||||
Gemma3nTextConfig,
|
||||
Gemma3nVisionConfig,
|
||||
)
|
||||
from transformers.models.siglip import SiglipImageProcessorFast
|
||||
|
||||
from vllm.config import ModelConfig, SpeechToTextConfig, VllmConfig
|
||||
@@ -22,25 +24,32 @@ from vllm.inputs.data import PromptType
|
||||
from vllm.logger import init_logger
|
||||
from vllm.model_executor.layers.layernorm import RMSNorm
|
||||
from vllm.model_executor.layers.linear import RowParallelLinear
|
||||
from vllm.model_executor.layers.vocab_parallel_embedding import (
|
||||
VocabParallelEmbedding)
|
||||
from vllm.model_executor.layers.vocab_parallel_embedding import VocabParallelEmbedding
|
||||
from vllm.model_executor.models.gemma3n import Gemma3nForCausalLM
|
||||
from vllm.model_executor.models.module_mapping import MultiModelKeys
|
||||
from vllm.model_executor.models.whisper import ISO639_1_SUPPORTED_LANGS
|
||||
from vllm.multimodal import MULTIMODAL_REGISTRY
|
||||
from vllm.multimodal.inputs import (MultiModalDataDict, MultiModalFieldConfig,
|
||||
MultiModalKwargsItems)
|
||||
from vllm.multimodal.parse import (ImageProcessorItems, MultiModalDataItems,
|
||||
MultiModalDataParser)
|
||||
from vllm.multimodal.processing import (BaseMultiModalProcessor,
|
||||
BaseProcessingInfo,
|
||||
MultiModalPromptUpdates,
|
||||
MultiModalPromptUpdatesApplyResult,
|
||||
PlaceholderFeaturesInfo,
|
||||
PromptReplacement, PromptUpdate,
|
||||
PromptUpdateDetails,
|
||||
replace_token_matches)
|
||||
# yapf: enable
|
||||
from vllm.multimodal.inputs import (
|
||||
MultiModalDataDict,
|
||||
MultiModalFieldConfig,
|
||||
MultiModalKwargsItems,
|
||||
)
|
||||
from vllm.multimodal.parse import (
|
||||
ImageProcessorItems,
|
||||
MultiModalDataItems,
|
||||
MultiModalDataParser,
|
||||
)
|
||||
from vllm.multimodal.processing import (
|
||||
BaseMultiModalProcessor,
|
||||
BaseProcessingInfo,
|
||||
MultiModalPromptUpdates,
|
||||
MultiModalPromptUpdatesApplyResult,
|
||||
PlaceholderFeaturesInfo,
|
||||
PromptReplacement,
|
||||
PromptUpdate,
|
||||
PromptUpdateDetails,
|
||||
replace_token_matches,
|
||||
)
|
||||
from vllm.multimodal.profiling import BaseDummyInputsBuilder
|
||||
from vllm.sequence import IntermediateTensors
|
||||
from vllm.utils.tensor_schema import TensorSchema, TensorShape
|
||||
|
||||
@@ -43,9 +43,6 @@ from vllm.multimodal.inputs import (
|
||||
MultiModalKwargsItems,
|
||||
)
|
||||
from vllm.multimodal.parse import ImageProcessorItems, ImageSize
|
||||
|
||||
# yapf conflicts with isort for this block
|
||||
# yapf: disable
|
||||
from vllm.multimodal.processing import (
|
||||
BaseMultiModalProcessor,
|
||||
BaseProcessingInfo,
|
||||
@@ -54,18 +51,13 @@ from vllm.multimodal.processing import (
|
||||
PromptUpdate,
|
||||
PromptUpdateDetails,
|
||||
)
|
||||
|
||||
# yapf: enable
|
||||
from vllm.multimodal.profiling import BaseDummyInputsBuilder
|
||||
from vllm.sequence import IntermediateTensors
|
||||
from vllm.utils.tensor_schema import TensorSchema, TensorShape
|
||||
|
||||
# yapf: disable
|
||||
from .idefics2_vision_model import (
|
||||
Idefics2VisionTransformer as Idefics3VisionTransformer,
|
||||
)
|
||||
|
||||
# yapf: enable
|
||||
from .interfaces import MultiModalEmbeddings, SupportsLoRA, SupportsMultiModal
|
||||
from .llama import LlamaModel
|
||||
from .utils import AutoWeightsLoader, maybe_prefix
|
||||
|
||||
@@ -45,9 +45,6 @@ from vllm.multimodal.parse import (
|
||||
ImageSize,
|
||||
MultiModalDataItems,
|
||||
)
|
||||
|
||||
# yapf conflicts with isort for this block
|
||||
# yapf: disable
|
||||
from vllm.multimodal.processing import (
|
||||
BaseMultiModalProcessor,
|
||||
BaseProcessingInfo,
|
||||
@@ -57,8 +54,6 @@ from vllm.multimodal.processing import (
|
||||
PromptUpdate,
|
||||
ResolvedPromptUpdate,
|
||||
)
|
||||
|
||||
# yapf: enable
|
||||
from vllm.multimodal.profiling import BaseDummyInputsBuilder
|
||||
from vllm.sequence import IntermediateTensors
|
||||
from vllm.utils import is_list_of
|
||||
|
||||
@@ -52,16 +52,12 @@ from vllm.distributed import utils as dist_utils
|
||||
from vllm.logger import init_logger
|
||||
from vllm.model_executor.layers.activation import get_act_and_mul_fn
|
||||
from vllm.model_executor.layers.layernorm import RMSNorm
|
||||
|
||||
# yapf: disable
|
||||
from vllm.model_executor.layers.linear import (
|
||||
ColumnParallelLinear,
|
||||
MergedColumnParallelLinear,
|
||||
QKVParallelLinear,
|
||||
RowParallelLinear,
|
||||
)
|
||||
|
||||
# yapf: enable
|
||||
from vllm.model_executor.layers.quantization import QuantizationConfig
|
||||
from vllm.model_executor.model_loader.weight_utils import default_weight_loader
|
||||
from vllm.model_executor.models.module_mapping import MultiModelKeys
|
||||
|
||||
@@ -37,12 +37,7 @@ from vllm.model_executor.layers.fla.ops import (
|
||||
fused_recurrent_gated_delta_rule,
|
||||
)
|
||||
from vllm.model_executor.layers.fused_moe import FusedMoE
|
||||
|
||||
# yapf conflicts with isort for this block
|
||||
# yapf: disable
|
||||
from vllm.model_executor.layers.layernorm import GemmaRMSNorm as Qwen3NextRMSNorm
|
||||
|
||||
# yapf: enable
|
||||
from vllm.model_executor.layers.linear import (
|
||||
ColumnParallelLinear,
|
||||
QKVParallelLinear,
|
||||
|
||||
@@ -54,7 +54,6 @@ from .interfaces_base import (
|
||||
|
||||
logger = init_logger(__name__)
|
||||
|
||||
# yapf: disable
|
||||
_TEXT_GENERATION_MODELS = {
|
||||
# [Decoder-only]
|
||||
"ApertusForCausalLM": ("apertus", "ApertusForCausalLM"),
|
||||
@@ -106,8 +105,8 @@ _TEXT_GENERATION_MODELS = {
|
||||
"GPTNeoXForCausalLM": ("gpt_neox", "GPTNeoXForCausalLM"),
|
||||
"GraniteForCausalLM": ("granite", "GraniteForCausalLM"),
|
||||
"GraniteMoeForCausalLM": ("granitemoe", "GraniteMoeForCausalLM"),
|
||||
"GraniteMoeHybridForCausalLM": ("granitemoehybrid", "GraniteMoeHybridForCausalLM"), # noqa: E501
|
||||
"GraniteMoeSharedForCausalLM": ("granitemoeshared", "GraniteMoeSharedForCausalLM"), # noqa: E501
|
||||
"GraniteMoeHybridForCausalLM": ("granitemoehybrid", "GraniteMoeHybridForCausalLM"), # noqa: E501
|
||||
"GraniteMoeSharedForCausalLM": ("granitemoeshared", "GraniteMoeSharedForCausalLM"), # noqa: E501
|
||||
"GritLM": ("gritlm", "GritLM"),
|
||||
"Grok1ModelForCausalLM": ("grok1", "Grok1ForCausalLM"),
|
||||
"HunYuanMoEV1ForCausalLM": ("hunyuan_v1", "HunYuanMoEV1ForCausalLM"),
|
||||
@@ -127,7 +126,7 @@ _TEXT_GENERATION_MODELS = {
|
||||
"LongcatFlashForCausalLM": ("longcat_flash", "LongcatFlashForCausalLM"),
|
||||
"MambaForCausalLM": ("mamba", "MambaForCausalLM"),
|
||||
"FalconMambaForCausalLM": ("mamba", "MambaForCausalLM"),
|
||||
"FalconH1ForCausalLM":("falcon_h1", "FalconH1ForCausalLM"),
|
||||
"FalconH1ForCausalLM": ("falcon_h1", "FalconH1ForCausalLM"),
|
||||
"Mamba2ForCausalLM": ("mamba2", "Mamba2ForCausalLM"),
|
||||
"MiniCPMForCausalLM": ("minicpm", "MiniCPMForCausalLM"),
|
||||
"MiniCPM3ForCausalLM": ("minicpm3", "MiniCPM3ForCausalLM"),
|
||||
@@ -184,7 +183,8 @@ _EMBEDDING_MODELS = {
|
||||
"LlamaModel": ("llama", "LlamaForCausalLM"),
|
||||
**{
|
||||
# Multiple models share the same architecture, so we include them all
|
||||
k: (mod, arch) for k, (mod, arch) in _TEXT_GENERATION_MODELS.items()
|
||||
k: (mod, arch)
|
||||
for k, (mod, arch) in _TEXT_GENERATION_MODELS.items()
|
||||
if arch == "LlamaForCausalLM"
|
||||
},
|
||||
"MistralModel": ("llama", "LlamaForCausalLM"),
|
||||
@@ -201,7 +201,10 @@ _EMBEDDING_MODELS = {
|
||||
"XLMRobertaModel": ("roberta", "RobertaEmbeddingModel"),
|
||||
# [Multimodal]
|
||||
"CLIPModel": ("clip", "CLIPEmbeddingModel"),
|
||||
"LlavaNextForConditionalGeneration": ("llava_next", "LlavaNextForConditionalGeneration"), # noqa: E501
|
||||
"LlavaNextForConditionalGeneration": (
|
||||
"llava_next",
|
||||
"LlavaNextForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"Phi3VForCausalLM": ("phi3v", "Phi3VForCausalLM"),
|
||||
"Qwen2VLForConditionalGeneration": ("qwen2_vl", "Qwen2VLForConditionalGeneration"), # noqa: E501
|
||||
# Technically Terratorch models work on images, both in
|
||||
@@ -214,79 +217,150 @@ _EMBEDDING_MODELS = {
|
||||
_CROSS_ENCODER_MODELS = {
|
||||
"BertForSequenceClassification": ("bert", "BertForSequenceClassification"),
|
||||
"BertForTokenClassification": ("bert", "BertForTokenClassification"),
|
||||
"GteNewForSequenceClassification": ("bert_with_rope",
|
||||
"GteNewForSequenceClassification"),
|
||||
"ModernBertForSequenceClassification": ("modernbert",
|
||||
"ModernBertForSequenceClassification"),
|
||||
"RobertaForSequenceClassification": ("roberta",
|
||||
"RobertaForSequenceClassification"),
|
||||
"XLMRobertaForSequenceClassification": ("roberta",
|
||||
"RobertaForSequenceClassification"),
|
||||
"GteNewForSequenceClassification": (
|
||||
"bert_with_rope",
|
||||
"GteNewForSequenceClassification",
|
||||
),
|
||||
"ModernBertForSequenceClassification": (
|
||||
"modernbert",
|
||||
"ModernBertForSequenceClassification",
|
||||
),
|
||||
"RobertaForSequenceClassification": ("roberta", "RobertaForSequenceClassification"),
|
||||
"XLMRobertaForSequenceClassification": (
|
||||
"roberta",
|
||||
"RobertaForSequenceClassification",
|
||||
),
|
||||
# [Auto-converted (see adapters.py)]
|
||||
"JinaVLForRanking": ("jina_vl", "JinaVLForSequenceClassification"), # noqa: E501,
|
||||
"JinaVLForRanking": ("jina_vl", "JinaVLForSequenceClassification"), # noqa: E501,
|
||||
}
|
||||
|
||||
_MULTIMODAL_MODELS = {
|
||||
# [Decoder-only]
|
||||
"AriaForConditionalGeneration": ("aria", "AriaForConditionalGeneration"),
|
||||
"AyaVisionForConditionalGeneration": ("aya_vision", "AyaVisionForConditionalGeneration"), # noqa: E501
|
||||
"AyaVisionForConditionalGeneration": (
|
||||
"aya_vision",
|
||||
"AyaVisionForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"Blip2ForConditionalGeneration": ("blip2", "Blip2ForConditionalGeneration"),
|
||||
"ChameleonForConditionalGeneration": ("chameleon", "ChameleonForConditionalGeneration"), # noqa: E501
|
||||
"Cohere2VisionForConditionalGeneration": ("cohere2_vision", "Cohere2VisionForConditionalGeneration"), # noqa: E501
|
||||
"ChameleonForConditionalGeneration": (
|
||||
"chameleon",
|
||||
"ChameleonForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"Cohere2VisionForConditionalGeneration": (
|
||||
"cohere2_vision",
|
||||
"Cohere2VisionForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"DeepseekVLV2ForCausalLM": ("deepseek_vl2", "DeepseekVLV2ForCausalLM"),
|
||||
"DotsOCRForCausalLM": ("dots_ocr", "DotsOCRForCausalLM"),
|
||||
"Ernie4_5_VLMoeForConditionalGeneration": ("ernie45_vl", "Ernie4_5_VLMoeForConditionalGeneration"), # noqa: E501
|
||||
"Ernie4_5_VLMoeForConditionalGeneration": (
|
||||
"ernie45_vl",
|
||||
"Ernie4_5_VLMoeForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"FuyuForCausalLM": ("fuyu", "FuyuForCausalLM"),
|
||||
"Gemma3ForConditionalGeneration": ("gemma3_mm", "Gemma3ForConditionalGeneration"), # noqa: E501
|
||||
"Gemma3nForConditionalGeneration": ("gemma3n_mm", "Gemma3nForConditionalGeneration"), # noqa: E501
|
||||
"Gemma3nForConditionalGeneration": (
|
||||
"gemma3n_mm",
|
||||
"Gemma3nForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"GLM4VForCausalLM": ("glm4v", "GLM4VForCausalLM"),
|
||||
"Glm4vForConditionalGeneration": ("glm4_1v", "Glm4vForConditionalGeneration"), # noqa: E501
|
||||
"Glm4vMoeForConditionalGeneration": ("glm4_1v", "Glm4vMoeForConditionalGeneration"), # noqa: E501
|
||||
"GraniteSpeechForConditionalGeneration": ("granite_speech", "GraniteSpeechForConditionalGeneration"), # noqa: E501
|
||||
"GraniteSpeechForConditionalGeneration": (
|
||||
"granite_speech",
|
||||
"GraniteSpeechForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"H2OVLChatModel": ("h2ovl", "H2OVLChatModel"),
|
||||
"InternVLChatModel": ("internvl", "InternVLChatModel"),
|
||||
"NemotronH_Nano_VL_V2": ("nano_nemotron_vl", "NemotronH_Nano_VL_V2"),
|
||||
"InternS1ForConditionalGeneration": ("interns1", "InternS1ForConditionalGeneration"), # noqa: E501
|
||||
"InternVLForConditionalGeneration": ("interns1", "InternS1ForConditionalGeneration"), # noqa: E501
|
||||
"Idefics3ForConditionalGeneration":("idefics3","Idefics3ForConditionalGeneration"),
|
||||
"SmolVLMForConditionalGeneration": ("smolvlm","SmolVLMForConditionalGeneration"), # noqa: E501
|
||||
"InternS1ForConditionalGeneration": (
|
||||
"interns1",
|
||||
"InternS1ForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"InternVLForConditionalGeneration": (
|
||||
"interns1",
|
||||
"InternS1ForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"Idefics3ForConditionalGeneration": (
|
||||
"idefics3",
|
||||
"Idefics3ForConditionalGeneration",
|
||||
),
|
||||
"SmolVLMForConditionalGeneration": ("smolvlm", "SmolVLMForConditionalGeneration"), # noqa: E501
|
||||
"KeyeForConditionalGeneration": ("keye", "KeyeForConditionalGeneration"),
|
||||
"KeyeVL1_5ForConditionalGeneration": ("keye_vl1_5", "KeyeVL1_5ForConditionalGeneration"), # noqa: E501
|
||||
"KeyeVL1_5ForConditionalGeneration": (
|
||||
"keye_vl1_5",
|
||||
"KeyeVL1_5ForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"RForConditionalGeneration": ("rvl", "RForConditionalGeneration"),
|
||||
"KimiVLForConditionalGeneration": ("kimi_vl", "KimiVLForConditionalGeneration"), # noqa: E501
|
||||
"Llama_Nemotron_Nano_VL": ("nemotron_vl", "LlamaNemotronVLChatModel"),
|
||||
"Llama4ForConditionalGeneration": ("mllama4", "Llama4ForConditionalGeneration"), # noqa: E501
|
||||
"LlavaForConditionalGeneration": ("llava", "LlavaForConditionalGeneration"),
|
||||
"LlavaNextForConditionalGeneration": ("llava_next", "LlavaNextForConditionalGeneration"), # noqa: E501
|
||||
"LlavaNextVideoForConditionalGeneration": ("llava_next_video", "LlavaNextVideoForConditionalGeneration"), # noqa: E501
|
||||
"LlavaOnevisionForConditionalGeneration": ("llava_onevision", "LlavaOnevisionForConditionalGeneration"), # noqa: E501
|
||||
"LlavaNextForConditionalGeneration": (
|
||||
"llava_next",
|
||||
"LlavaNextForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"LlavaNextVideoForConditionalGeneration": (
|
||||
"llava_next_video",
|
||||
"LlavaNextVideoForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"LlavaOnevisionForConditionalGeneration": (
|
||||
"llava_onevision",
|
||||
"LlavaOnevisionForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"MantisForConditionalGeneration": ("llava", "MantisForConditionalGeneration"), # noqa: E501
|
||||
"MiDashengLMModel": ("midashenglm", "MiDashengLMModel"),
|
||||
"MiniMaxVL01ForConditionalGeneration": ("minimax_vl_01", "MiniMaxVL01ForConditionalGeneration"), # noqa: E501
|
||||
"MiniMaxVL01ForConditionalGeneration": (
|
||||
"minimax_vl_01",
|
||||
"MiniMaxVL01ForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"MiniCPMO": ("minicpmo", "MiniCPMO"),
|
||||
"MiniCPMV": ("minicpmv", "MiniCPMV"),
|
||||
"Mistral3ForConditionalGeneration": ("mistral3", "Mistral3ForConditionalGeneration"), # noqa: E501
|
||||
"Mistral3ForConditionalGeneration": (
|
||||
"mistral3",
|
||||
"Mistral3ForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"MolmoForCausalLM": ("molmo", "MolmoForCausalLM"),
|
||||
"NVLM_D": ("nvlm_d", "NVLM_D_Model"),
|
||||
"Ovis": ("ovis", "Ovis"),
|
||||
"Ovis2_5": ("ovis2_5", "Ovis2_5"),
|
||||
"PaliGemmaForConditionalGeneration": ("paligemma", "PaliGemmaForConditionalGeneration"), # noqa: E501
|
||||
"PaliGemmaForConditionalGeneration": (
|
||||
"paligemma",
|
||||
"PaliGemmaForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"Phi3VForCausalLM": ("phi3v", "Phi3VForCausalLM"),
|
||||
"Phi4MMForCausalLM": ("phi4mm", "Phi4MMForCausalLM"),
|
||||
"Phi4MultimodalForCausalLM": ("phi4_multimodal", "Phi4MultimodalForCausalLM"), # noqa: E501
|
||||
"PixtralForConditionalGeneration": ("pixtral", "PixtralForConditionalGeneration"), # noqa: E501
|
||||
"QwenVLForConditionalGeneration": ("qwen_vl", "QwenVLForConditionalGeneration"), # noqa: E501
|
||||
"Qwen2VLForConditionalGeneration": ("qwen2_vl", "Qwen2VLForConditionalGeneration"), # noqa: E501
|
||||
"Qwen2_5_VLForConditionalGeneration": ("qwen2_5_vl", "Qwen2_5_VLForConditionalGeneration"), # noqa: E501
|
||||
"Qwen2AudioForConditionalGeneration": ("qwen2_audio", "Qwen2AudioForConditionalGeneration"), # noqa: E501
|
||||
"Qwen2_5OmniModel": ("qwen2_5_omni_thinker", "Qwen2_5OmniThinkerForConditionalGeneration"), # noqa: E501
|
||||
"Qwen2_5OmniForConditionalGeneration": ("qwen2_5_omni_thinker", "Qwen2_5OmniThinkerForConditionalGeneration"), # noqa: E501
|
||||
"Qwen2_5_VLForConditionalGeneration": (
|
||||
"qwen2_5_vl",
|
||||
"Qwen2_5_VLForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"Qwen2AudioForConditionalGeneration": (
|
||||
"qwen2_audio",
|
||||
"Qwen2AudioForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"Qwen2_5OmniModel": (
|
||||
"qwen2_5_omni_thinker",
|
||||
"Qwen2_5OmniThinkerForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"Qwen2_5OmniForConditionalGeneration": (
|
||||
"qwen2_5_omni_thinker",
|
||||
"Qwen2_5OmniThinkerForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"Qwen3VLForConditionalGeneration": ("qwen3_vl", "Qwen3VLForConditionalGeneration"), # noqa: E501
|
||||
"Qwen3VLMoeForConditionalGeneration": ("qwen3_vl_moe", "Qwen3VLMoeForConditionalGeneration"), # noqa: E501
|
||||
"Qwen3VLMoeForConditionalGeneration": (
|
||||
"qwen3_vl_moe",
|
||||
"Qwen3VLMoeForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"SkyworkR1VChatModel": ("skyworkr1v", "SkyworkR1VChatModel"),
|
||||
"Step3VLForConditionalGeneration": ("step3_vl", "Step3VLForConditionalGeneration"), # noqa: E501
|
||||
"TarsierForConditionalGeneration": ("tarsier", "TarsierForConditionalGeneration"), # noqa: E501
|
||||
"Tarsier2ForConditionalGeneration": ("qwen2_vl", "Tarsier2ForConditionalGeneration"), # noqa: E501
|
||||
"Tarsier2ForConditionalGeneration": (
|
||||
"qwen2_vl",
|
||||
"Tarsier2ForConditionalGeneration",
|
||||
), # noqa: E501
|
||||
"UltravoxModel": ("ultravox", "UltravoxModel"),
|
||||
"VoxtralForConditionalGeneration": ("voxtral", "VoxtralForConditionalGeneration"), # noqa: E501
|
||||
# [Encoder-decoder]
|
||||
@@ -324,13 +398,27 @@ _TRANSFORMERS_BACKEND_MODELS = {
|
||||
"TransformersForCausalLM": ("transformers", "TransformersForCausalLM"),
|
||||
"TransformersForMultimodalLM": ("transformers", "TransformersForMultimodalLM"), # noqa: E501
|
||||
"TransformersMoEForCausalLM": ("transformers_moe", "TransformersMoEForCausalLM"), # noqa: E501
|
||||
"TransformersMoEForMultimodalLM": ("transformers_moe", "TransformersMoEForMultimodalLM"), # noqa: E501
|
||||
"TransformersEmbeddingModel": ("transformers_pooling", "TransformersEmbeddingModel"), # noqa: E501
|
||||
"TransformersForSequenceClassification": ("transformers_pooling", "TransformersForSequenceClassification"), # noqa: E501
|
||||
"TransformersMoEForSequenceClassification": ("transformers_pooling", "TransformersMoEForSequenceClassification"), # noqa: E501
|
||||
"TransformersMoEEmbeddingModel": ("transformers_pooling", "TransformersMoEEmbeddingModel"), # noqa: E501
|
||||
"TransformersMoEForMultimodalLM": (
|
||||
"transformers_moe",
|
||||
"TransformersMoEForMultimodalLM",
|
||||
), # noqa: E501
|
||||
"TransformersEmbeddingModel": (
|
||||
"transformers_pooling",
|
||||
"TransformersEmbeddingModel",
|
||||
), # noqa: E501
|
||||
"TransformersForSequenceClassification": (
|
||||
"transformers_pooling",
|
||||
"TransformersForSequenceClassification",
|
||||
), # noqa: E501
|
||||
"TransformersMoEForSequenceClassification": (
|
||||
"transformers_pooling",
|
||||
"TransformersMoEForSequenceClassification",
|
||||
), # noqa: E501
|
||||
"TransformersMoEEmbeddingModel": (
|
||||
"transformers_pooling",
|
||||
"TransformersMoEEmbeddingModel",
|
||||
), # noqa: E501
|
||||
}
|
||||
# yapf: enable
|
||||
|
||||
_VLLM_MODELS = {
|
||||
**_TEXT_GENERATION_MODELS,
|
||||
|
||||
@@ -8,13 +8,10 @@ from transformers import SmolVLMProcessor
|
||||
from vllm.config import VllmConfig
|
||||
from vllm.multimodal import MULTIMODAL_REGISTRY
|
||||
|
||||
# yapf: disable
|
||||
from .idefics3 import Idefics3DummyInputsBuilder as SmolVLMDummyInputsBuilder
|
||||
from .idefics3 import Idefics3ForConditionalGeneration, Idefics3ProcessingInfo
|
||||
from .idefics3 import Idefics3MultiModalProcessor as SmolVLMMultiModalProcessor
|
||||
|
||||
# yapf: enable
|
||||
|
||||
|
||||
class SmolVLMProcessingInfo(Idefics3ProcessingInfo):
|
||||
def get_hf_processor(self, **kwargs: object) -> SmolVLMProcessor:
|
||||
|
||||
@@ -32,11 +32,7 @@ from vllm.model_executor.layers.quantization import QuantizationConfig
|
||||
from vllm.model_executor.model_loader.weight_utils import default_weight_loader
|
||||
from vllm.model_executor.models import SupportsPP
|
||||
from vllm.model_executor.models.module_mapping import MultiModelKeys
|
||||
|
||||
# yapf: disable
|
||||
from vllm.model_executor.models.whisper import WhisperEncoder
|
||||
|
||||
# yapf: enable
|
||||
from vllm.multimodal import MULTIMODAL_REGISTRY
|
||||
from vllm.multimodal.inputs import (
|
||||
MultiModalDataDict,
|
||||
|
||||
Reference in New Issue
Block a user