[MM Encoder]: Migrate legacy ViT MultiHeadAttention to new MMEncoderAttention interface (#30684)
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
This commit is contained in:
@@ -17,7 +17,8 @@ from transformers import BatchFeature, PretrainedConfig, ProcessorMixin, TensorT
|
||||
from transformers.image_utils import ImageInput
|
||||
from transformers.tokenization_utils_base import TextInput
|
||||
|
||||
from vllm.attention.layer import Attention, MultiHeadAttention
|
||||
from vllm.attention.layer import Attention
|
||||
from vllm.attention.layers.mm_encoder_attention import MMEncoderAttention
|
||||
from vllm.compilation.decorators import support_torch_compile
|
||||
from vllm.config import CacheConfig, VllmConfig
|
||||
from vllm.config.multimodal import BaseDummyOptions
|
||||
@@ -222,7 +223,7 @@ class MultiHeadDotProductAttention(nn.Module):
|
||||
)
|
||||
|
||||
self.scale = self.head_dim**-0.5
|
||||
self.attn = MultiHeadAttention(
|
||||
self.attn = MMEncoderAttention(
|
||||
self.num_heads, self.head_dim, self.scale, num_kv_heads=self.num_kv_heads
|
||||
)
|
||||
|
||||
|
||||
Reference in New Issue
Block a user