[Attention] Remove imports from `vllm/attention/__init__.py` (#29342)

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

vllm/model_executor/models/mpt.py

@@ -10,7 +10,7 @@ import torch
 import torch.nn as nn
 from transformers import MptConfig
 
-from vllm.attention import Attention
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import (

vllm/model_executor/models/nemotron.py

@@ -30,7 +30,7 @@ from itertools import islice
 import torch
 from torch import nn
 
-from vllm.attention import Attention
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size

vllm/model_executor/models/nemotron_nas.py

@@ -31,7 +31,7 @@ import torch
 from torch import nn
 from transformers import LlamaConfig
 
-from vllm.attention import AttentionType
+from vllm.attention.backends.abstract import AttentionType
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_pp_group

vllm/model_executor/models/olmo.py

@@ -31,7 +31,7 @@ import torch
 from torch import nn
 from transformers import OlmoConfig
 
-from vllm.attention import Attention
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size

vllm/model_executor/models/olmo2.py

@@ -32,7 +32,7 @@ import torch
 from torch import nn
 from transformers import Olmo2Config
 
-from vllm.attention import Attention
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import VllmConfig
 from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size

vllm/model_executor/models/olmoe.py

@@ -21,7 +21,7 @@ from itertools import islice
 import torch
 from torch import nn
 
-from vllm.attention import Attention
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import VllmConfig
 from vllm.distributed import (

vllm/model_executor/models/openpangu.py

@@ -29,7 +29,8 @@ import torch
 from torch import nn
 from transformers import PretrainedConfig
 
-from vllm.attention import Attention, AttentionType
+from vllm.attention.backends.abstract import AttentionType
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, ParallelConfig, VllmConfig
 from vllm.distributed import (

vllm/model_executor/models/opt.py

@@ -27,7 +27,7 @@ import torch
 from torch import nn
 from transformers import OPTConfig
 
-from vllm.attention import Attention
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size

vllm/model_executor/models/orion.py

@@ -15,7 +15,7 @@ import torch
 from torch import nn
 from transformers import PretrainedConfig
 
-from vllm.attention import Attention
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size

vllm/model_executor/models/ouro.py

@@ -33,7 +33,8 @@ import torch
 from torch import nn
 from transformers import PretrainedConfig
 
-from vllm.attention import Attention, AttentionType
+from vllm.attention.backends.abstract import AttentionType
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_tensor_model_parallel_world_size

vllm/model_executor/models/persimmon.py

@@ -30,7 +30,7 @@ import torch
 from torch import nn
 from transformers import PersimmonConfig
 
-from vllm.attention import Attention
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size

vllm/model_executor/models/phi.py

@@ -45,7 +45,7 @@ import torch
 from torch import nn
 from transformers import PhiConfig
 
-from vllm.attention import Attention
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size

vllm/model_executor/models/phimoe.py

@@ -31,7 +31,7 @@ import torch
 from torch import nn
 from transformers.configuration_utils import PretrainedConfig
 
-from vllm.attention import Attention
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size

vllm/model_executor/models/qwen.py

@@ -16,7 +16,7 @@ import torch
 from torch import nn
 from transformers import PretrainedConfig
 
-from vllm.attention import Attention
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size

vllm/model_executor/models/qwen2.py

@@ -33,7 +33,8 @@ import torch
 from torch import nn
 from transformers import Qwen2Config
 
-from vllm.attention import Attention, AttentionType
+from vllm.attention.backends.abstract import AttentionType
+from vllm.attention.layer import Attention
 from vllm.attention.layers.encoder_only_attention import EncoderOnlyAttention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig

vllm/model_executor/models/qwen2_moe.py

@@ -34,7 +34,7 @@ import torch.nn.functional as F
 from torch import nn
 from transformers import Qwen2MoeConfig
 
-from vllm.attention import Attention
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size

vllm/model_executor/models/qwen3.py

@@ -30,7 +30,8 @@ import torch
 from torch import nn
 from transformers import Qwen3Config
 
-from vllm.attention import Attention, AttentionType
+from vllm.attention.backends.abstract import AttentionType
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size

vllm/model_executor/models/qwen3_moe.py

@@ -31,7 +31,7 @@ from typing import Any
 import torch
 from torch import nn
 
-from vllm.attention import Attention
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig, get_current_vllm_config
 from vllm.distributed import (

vllm/model_executor/models/qwen3_next.py

@@ -10,7 +10,8 @@ from einops import rearrange
 from torch import nn
 from transformers.activations import ACT2FN
 
-from vllm.attention import Attention, AttentionMetadata
+from vllm.attention.backends.abstract import AttentionMetadata
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import (
     CacheConfig,

vllm/model_executor/models/seed_oss.py

@@ -30,7 +30,8 @@ import torch
 from torch import nn
 from transformers import PretrainedConfig as SeedOssConfig
 
-from vllm.attention import Attention, AttentionType
+from vllm.attention.backends.abstract import AttentionType
+from vllm.attention.layer import Attention
 from vllm.compilation.decorators import support_torch_compile
 from vllm.config import CacheConfig, VllmConfig
 from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size