[5/N][Attention] Finish eliminating `vllm/attention` folder (#32064)

Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>

vllm/model_executor/models/step3_text.py

@@ -9,7 +9,6 @@ from typing import Any
 import torch
 from torch import nn
 
-from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, ModelConfig, VllmConfig
 from vllm.distributed import (
@@ -19,6 +18,7 @@ from vllm.distributed import (
 )
 from vllm.logger import init_logger
 from vllm.model_executor.layers.activation import SiluAndMul
+from vllm.model_executor.layers.attention import Attention
 from vllm.model_executor.layers.fused_moe import FusedMoE
 from vllm.model_executor.layers.layernorm import RMSNorm
 from vllm.model_executor.layers.linear import (

vllm/model_executor/models/step3_vl.py

@@ -19,7 +19,7 @@ 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.activation import get_act_fn
-from vllm.model_executor.layers.attention.mm_encoder_attention import MMEncoderAttention
+from vllm.model_executor.layers.attention import MMEncoderAttention
 from vllm.model_executor.layers.conv import Conv2dLayer
 from vllm.model_executor.layers.linear import (
     ColumnParallelLinear,

vllm/model_executor/models/transformers/base.py

@@ -27,12 +27,12 @@ from torch import nn
 from transformers import AutoModel
 from transformers.modeling_utils import ALL_ATTENTION_FUNCTIONS
 
-from vllm.attention.layer import Attention
 from vllm.config.utils import getattr_iter
 from vllm.distributed import get_pp_group, get_tp_group
 from vllm.distributed.utils import get_pp_indices
 from vllm.logger import init_logger
-from vllm.model_executor.layers.attention.encoder_only_attention import (
+from vllm.model_executor.layers.attention import (
+    Attention,
     EncoderOnlyAttention,
 )
 from vllm.model_executor.layers.vocab_parallel_embedding import VocabParallelEmbedding

vllm/model_executor/models/whisper.py

@@ -17,7 +17,6 @@ from transformers import (
 )
 from transformers.models.whisper.modeling_whisper import sinusoids
 
-from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, ModelConfig, SpeechToTextConfig, VllmConfig
 from vllm.config.multimodal import BaseDummyOptions
@@ -25,8 +24,11 @@ from vllm.distributed import get_tensor_model_parallel_world_size
 from vllm.inputs.data import PromptType
 from vllm.logger import init_logger
 from vllm.model_executor.layers.activation import get_act_fn
-from vllm.model_executor.layers.attention.cross_attention import CrossAttention
-from vllm.model_executor.layers.attention.mm_encoder_attention import MMEncoderAttention
+from vllm.model_executor.layers.attention import (
+    Attention,
+    CrossAttention,
+    MMEncoderAttention,
+)
 from vllm.model_executor.layers.linear import (
     ColumnParallelLinear,
     QKVParallelLinear,

vllm/model_executor/models/whisper_causal.py

@@ -10,9 +10,9 @@ import torch
 import torch.nn.functional as F
 from torch import nn
 
-from vllm.attention.layer import Attention
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_tensor_model_parallel_world_size
+from vllm.model_executor.layers.attention import Attention
 from vllm.model_executor.layers.layernorm import RMSNorm
 from vllm.model_executor.layers.linear import (
     QKVParallelLinear,

vllm/model_executor/models/zamba2.py

@@ -16,11 +16,11 @@ import torch
 from torch import nn
 from transformers import Zamba2Config
 
-from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, ModelConfig, VllmConfig
 from vllm.distributed import get_tensor_model_parallel_world_size
 from vllm.model_executor.layers.activation import GeluAndMul
+from vllm.model_executor.layers.attention import Attention
 from vllm.model_executor.layers.layernorm import RMSNorm
 from vllm.model_executor.layers.linear import (
     ColumnParallelLinear,

vllm/v1/attention/backends/flash_attn.py

@@ -9,7 +9,7 @@ from typing import ClassVar
 import numpy as np
 import torch
 
-from vllm.attention.layer import Attention
+from vllm.model_executor.layers.attention import Attention
 from vllm.v1.attention.backend import (
     AttentionBackend,
     AttentionImpl,

vllm/v1/attention/backends/rocm_aiter_fa.py

@@ -8,9 +8,9 @@ from typing import ClassVar
 import torch
 
 from vllm._aiter_ops import rocm_aiter_ops
-from vllm.attention.layer import Attention
 from vllm.config import VllmConfig, get_layers_from_vllm_config
 from vllm.logger import init_logger
+from vllm.model_executor.layers.attention import Attention
 from vllm.platforms import current_platform
 from vllm.utils.math_utils import cdiv
 from vllm.utils.platform_utils import get_cu_count

vllm/v1/spec_decode/draft_model.py

@@ -4,10 +4,10 @@ from typing import Any
 
 import torch
 
-from vllm.attention.layer import Attention
 from vllm.config import VllmConfig, get_layers_from_vllm_config
 from vllm.config.speculative import SpeculativeConfig
 from vllm.logger import init_logger
+from vllm.model_executor.layers.attention import Attention
 from vllm.model_executor.model_loader import get_model
 from vllm.triton_utils import tl, triton
 from vllm.v1.attention.backends.utils import (

vllm/v1/worker/gpu_model_runner.py

@@ -21,7 +21,6 @@ import torch.nn as nn
 from tqdm import tqdm
 
 import vllm.envs as envs
-from vllm.attention.layer import Attention, MLAAttention
 from vllm.compilation.counter import compilation_counter
 from vllm.compilation.cuda_graph import CUDAGraphStat, CUDAGraphWrapper
 from vllm.compilation.monitor import set_cudagraph_capturing_enabled
@@ -50,6 +49,7 @@ from vllm.forward_context import (
 )
 from vllm.logger import init_logger
 from vllm.lora.layers import LoRAMapping, LoRAMappingType
+from vllm.model_executor.layers.attention import Attention, MLAAttention
 from vllm.model_executor.layers.attention_layer_base import AttentionLayerBase
 from vllm.model_executor.layers.fused_moe.routed_experts_capturer import (
     RoutedExpertsCapturer,

vllm/v1/worker/utils.py

@@ -7,9 +7,9 @@ from dataclasses import dataclass, field
 import torch
 from typing_extensions import deprecated
 
-from vllm.attention.layer import Attention
 from vllm.config import CacheConfig, VllmConfig
 from vllm.logger import init_logger
+from vllm.model_executor.layers.attention import Attention
 from vllm.model_executor.models.interfaces import MultiModalEmbeddings
 from vllm.model_executor.models.utils import extract_layer_index
 from vllm.multimodal.registry import MultiModalRegistry