|
|
|
|
|
|
|
|
|
|
|
|
|
import torch |
|
from torch import Tensor, nn |
|
|
|
import math |
|
from typing import Tuple, Type |
|
|
|
from .common import MLPBlock |
|
|
|
|
|
class TwoWayTransformer(nn.Module): |
|
def __init__( |
|
self, |
|
depth: int, |
|
embedding_dim: int, |
|
num_heads: int, |
|
mlp_dim: int, |
|
activation: Type[nn.Module] = nn.ReLU, |
|
attention_downsample_rate: int = 2, |
|
) -> None: |
|
""" |
|
A transformer decoder that attends to an input image using |
|
queries whose positional embedding is supplied. |
|
|
|
Args: |
|
depth (int): number of layers in the transformer |
|
embedding_dim (int): the channel dimension for the input embeddings |
|
num_heads (int): the number of heads for multihead attention. Must |
|
divide embedding_dim |
|
mlp_dim (int): the channel dimension internal to the MLP block |
|
activation (nn.Module): the activation to use in the MLP block |
|
""" |
|
super().__init__() |
|
self.depth = depth |
|
self.embedding_dim = embedding_dim |
|
self.num_heads = num_heads |
|
self.mlp_dim = mlp_dim |
|
self.layers = nn.ModuleList() |
|
|
|
for i in range(depth): |
|
self.layers.append( |
|
TwoWayAttentionBlock( |
|
embedding_dim=embedding_dim, |
|
num_heads=num_heads, |
|
mlp_dim=mlp_dim, |
|
activation=activation, |
|
attention_downsample_rate=attention_downsample_rate, |
|
skip_first_layer_pe=(i == 0), |
|
) |
|
) |
|
|
|
self.final_attn_token_to_image = Attention( |
|
embedding_dim, num_heads, downsample_rate=attention_downsample_rate |
|
) |
|
self.norm_final_attn = nn.LayerNorm(embedding_dim) |
|
|
|
def forward( |
|
self, |
|
image_embedding: Tensor, |
|
image_pe: Tensor, |
|
point_embedding: Tensor, |
|
) -> Tuple[Tensor, Tensor]: |
|
""" |
|
Args: |
|
image_embedding (torch.Tensor): image to attend to. Should be shape |
|
B x embedding_dim x h x w for any h and w. |
|
image_pe (torch.Tensor): the positional encoding to add to the image. Must |
|
have the same shape as image_embedding. |
|
point_embedding (torch.Tensor): the embedding to add to the query points. |
|
Must have shape B x N_points x embedding_dim for any N_points. |
|
|
|
Returns: |
|
torch.Tensor: the processed point_embedding |
|
torch.Tensor: the processed image_embedding |
|
""" |
|
|
|
bs, c, h, w = image_embedding.shape |
|
image_embedding = image_embedding.flatten(2).permute(0, 2, 1) |
|
image_pe = image_pe.flatten(2).permute(0, 2, 1) |
|
|
|
|
|
queries = point_embedding |
|
keys = image_embedding |
|
|
|
|
|
for layer in self.layers: |
|
queries, keys = layer( |
|
queries=queries, |
|
keys=keys, |
|
query_pe=point_embedding, |
|
key_pe=image_pe, |
|
) |
|
|
|
|
|
q = queries + point_embedding |
|
k = keys + image_pe |
|
attn_out = self.final_attn_token_to_image(q=q, k=k, v=keys) |
|
queries = queries + attn_out |
|
queries = self.norm_final_attn(queries) |
|
|
|
return queries, keys |
|
|
|
|
|
class TwoWayAttentionBlock(nn.Module): |
|
def __init__( |
|
self, |
|
embedding_dim: int, |
|
num_heads: int, |
|
mlp_dim: int = 2048, |
|
activation: Type[nn.Module] = nn.ReLU, |
|
attention_downsample_rate: int = 2, |
|
skip_first_layer_pe: bool = False, |
|
) -> None: |
|
""" |
|
A transformer block with four layers: (1) self-attention of sparse |
|
inputs, (2) cross attention of sparse inputs to dense inputs, (3) mlp |
|
block on sparse inputs, and (4) cross attention of dense inputs to sparse |
|
inputs. |
|
|
|
Arguments: |
|
embedding_dim (int): the channel dimension of the embeddings |
|
num_heads (int): the number of heads in the attention layers |
|
mlp_dim (int): the hidden dimension of the mlp block |
|
activation (nn.Module): the activation of the mlp block |
|
skip_first_layer_pe (bool): skip the PE on the first layer |
|
""" |
|
super().__init__() |
|
self.self_attn = Attention(embedding_dim, num_heads) |
|
self.norm1 = nn.LayerNorm(embedding_dim) |
|
|
|
self.cross_attn_token_to_image = Attention( |
|
embedding_dim, num_heads, downsample_rate=attention_downsample_rate |
|
) |
|
self.norm2 = nn.LayerNorm(embedding_dim) |
|
|
|
self.mlp = MLPBlock(embedding_dim, mlp_dim, activation) |
|
self.norm3 = nn.LayerNorm(embedding_dim) |
|
|
|
self.norm4 = nn.LayerNorm(embedding_dim) |
|
self.cross_attn_image_to_token = Attention( |
|
embedding_dim, num_heads, downsample_rate=attention_downsample_rate |
|
) |
|
|
|
self.skip_first_layer_pe = skip_first_layer_pe |
|
|
|
def forward( |
|
self, queries: Tensor, keys: Tensor, query_pe: Tensor, key_pe: Tensor |
|
) -> Tuple[Tensor, Tensor]: |
|
|
|
if self.skip_first_layer_pe: |
|
queries = self.self_attn(q=queries, k=queries, v=queries) |
|
else: |
|
q = queries + query_pe |
|
attn_out = self.self_attn(q=q, k=q, v=queries) |
|
queries = queries + attn_out |
|
queries = self.norm1(queries) |
|
|
|
|
|
q = queries + query_pe |
|
k = keys + key_pe |
|
attn_out = self.cross_attn_token_to_image(q=q, k=k, v=keys) |
|
queries = queries + attn_out |
|
queries = self.norm2(queries) |
|
|
|
|
|
mlp_out = self.mlp(queries) |
|
queries = queries + mlp_out |
|
queries = self.norm3(queries) |
|
|
|
|
|
q = queries + query_pe |
|
k = keys + key_pe |
|
attn_out = self.cross_attn_image_to_token(q=k, k=q, v=queries) |
|
keys = keys + attn_out |
|
keys = self.norm4(keys) |
|
|
|
return queries, keys |
|
|
|
|
|
class Attention(nn.Module): |
|
""" |
|
An attention layer that allows for downscaling the size of the embedding |
|
after projection to queries, keys, and values. |
|
""" |
|
|
|
def __init__( |
|
self, |
|
embedding_dim: int, |
|
num_heads: int, |
|
downsample_rate: int = 1, |
|
) -> None: |
|
super().__init__() |
|
self.embedding_dim = embedding_dim |
|
self.internal_dim = embedding_dim // downsample_rate |
|
self.num_heads = num_heads |
|
assert self.internal_dim % num_heads == 0, "num_heads must divide embedding_dim." |
|
|
|
self.q_proj = nn.Linear(embedding_dim, self.internal_dim) |
|
self.k_proj = nn.Linear(embedding_dim, self.internal_dim) |
|
self.v_proj = nn.Linear(embedding_dim, self.internal_dim) |
|
self.out_proj = nn.Linear(self.internal_dim, embedding_dim) |
|
|
|
def _separate_heads(self, x: Tensor, num_heads: int) -> Tensor: |
|
b, n, c = x.shape |
|
x = x.reshape(b, n, num_heads, c // num_heads) |
|
return x.transpose(1, 2) |
|
|
|
def _recombine_heads(self, x: Tensor) -> Tensor: |
|
b, n_heads, n_tokens, c_per_head = x.shape |
|
x = x.transpose(1, 2) |
|
return x.reshape(b, n_tokens, n_heads * c_per_head) |
|
|
|
def forward(self, q: Tensor, k: Tensor, v: Tensor) -> Tensor: |
|
|
|
q = self.q_proj(q) |
|
k = self.k_proj(k) |
|
v = self.v_proj(v) |
|
|
|
|
|
q = self._separate_heads(q, self.num_heads) |
|
k = self._separate_heads(k, self.num_heads) |
|
v = self._separate_heads(v, self.num_heads) |
|
|
|
|
|
_, _, _, c_per_head = q.shape |
|
attn = q @ k.permute(0, 1, 3, 2) |
|
attn = attn / math.sqrt(c_per_head) |
|
attn = torch.softmax(attn, dim=-1) |
|
|
|
|
|
out = attn @ v |
|
out = self._recombine_heads(out) |
|
out = self.out_proj(out) |
|
|
|
return out |
|
|