File size: 44,758 Bytes
9fa8856 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 |
from __future__ import annotations
from typing import Union
from transformers import LlamaConfig, LlamaModel, LlamaForCausalLM
from transformers.modeling_outputs import CausalLMOutputWithPast
from transformers.generation.utils import GenerateOutput
from .configuration_m3d_lamed import LamedConfig
from abc import ABC, abstractmethod
from torch import Tensor
import math
from typing import Any, Dict, List
import torch
import torch.nn as nn
from typing import Optional, Tuple, Type
from monai.networks.blocks import PatchEmbed
import numpy as np
import torch.nn.functional as F
from einops import rearrange
from einops.layers.torch import Rearrange
from collections.abc import Sequence
from monai.networks.blocks.patchembedding import PatchEmbeddingBlock
from monai.networks.blocks.transformerblock import TransformerBlock
from monai.networks.nets import ViT
class LayerNorm2d(nn.Module):
def __init__(self, num_channels: int, eps: float = 1e-6) -> None:
super().__init__()
self.weight = nn.Parameter(torch.ones(num_channels))
self.bias = nn.Parameter(torch.zeros(num_channels))
self.eps = eps
def forward(self, x: torch.Tensor) -> torch.Tensor:
u = x.mean(1, keepdim=True)
s = (x - u).pow(2).mean(1, keepdim=True)
x = (x - u) / torch.sqrt(s + self.eps)
x = self.weight[:, None, None] * x + self.bias[:, None, None]
return x
class MLPBlock(nn.Module):
def __init__(
self,
embedding_dim: int,
mlp_dim: int,
act: Type[nn.Module] = nn.GELU,
) -> None:
super().__init__()
self.lin1 = nn.Linear(embedding_dim, mlp_dim)
self.lin2 = nn.Linear(mlp_dim, embedding_dim)
self.act = act()
def forward(self, x: torch.Tensor) -> torch.Tensor:
return self.lin2(self.act(self.lin1(x)))
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
"""
# BxCxHxW -> BxHWxC == B x N_image_tokens x C
bs, c, h, w, d = image_embedding.shape
image_embedding = image_embedding.flatten(2).permute(0, 2, 1)
image_pe = image_pe.flatten(2).permute(0, 2, 1)
# Prepare queries
queries = point_embedding
keys = image_embedding
# Apply transformer blocks and final layernorm
for layer in self.layers:
queries, keys = layer(
queries=queries,
keys=keys,
query_pe=point_embedding,
key_pe=image_pe,
)
# Apply the final attention layer from the points to the image
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]:
# Self attention block
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)
# Cross attention block, tokens attending to image embedding
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 block
mlp_out = self.mlp(queries)
queries = queries + mlp_out
queries = self.norm3(queries)
# Cross attention block, image embedding attending to tokens
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) # B x N_heads x N_tokens x C_per_head
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) # B x N_tokens x C
def forward(self, q: Tensor, k: Tensor, v: Tensor) -> Tensor:
# Input projections
q = self.q_proj(q)
k = self.k_proj(k)
v = self.v_proj(v)
# Separate into heads
q = self._separate_heads(q, self.num_heads)
k = self._separate_heads(k, self.num_heads)
v = self._separate_heads(v, self.num_heads)
# Attention
_, _, _, c_per_head = q.shape
attn = q @ k.permute(0, 1, 3, 2) # B x N_heads x N_tokens x N_tokens
attn = attn / math.sqrt(c_per_head)
attn = torch.softmax(attn, dim=-1)
# Get output
out = attn @ v
out = self._recombine_heads(out)
out = self.out_proj(out)
return out
# This class and its supporting functions below lightly adapted from the ViTDet backbone available at: https://github.com/facebookresearch/detectron2/blob/main/detectron2/modeling/backbone/vit.py # noqa
class ImageEncoderViT(nn.Module):
def __init__(
self,
img_size: int = 1024,
patch_size: int = 16,
in_chans: int = 1,
embed_dim: int = 768,
depth: int = 12,
num_heads: int = 12,
mlp_ratio: float = 4.0,
out_chans: int = 256,
qkv_bias: bool = True,
norm_layer: Type[nn.Module] = nn.LayerNorm,
act_layer: Type[nn.Module] = nn.GELU,
use_abs_pos: bool = True,
use_rel_pos: bool = False,
rel_pos_zero_init: bool = True,
window_size: int = 0,
global_attn_indexes: Tuple[int, ...] = (),
) -> None:
"""
Args:
img_size (int): Input image size.
patch_size (int): Patch size.
in_chans (int): Number of input image channels.
embed_dim (int): Patch embedding dimension.
depth (int): Depth of ViT.
num_heads (int): Number of attention heads in each ViT block.
mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
qkv_bias (bool): If True, add a learnable bias to query, key, value.
norm_layer (nn.Module): Normalization layer.
act_layer (nn.Module): Activation layer.
use_abs_pos (bool): If True, use absolute positional embeddings.
use_rel_pos (bool): If True, add relative positional embeddings to the attention map.
rel_pos_zero_init (bool): If True, zero initialize relative positional parameters.
window_size (int): Window size for window attention blocks.
global_attn_indexes (list): Indexes for blocks using global attention.
"""
super().__init__()
self.img_size = img_size
# self.patch_embed = PatchEmbed(
# kernel_size=(patch_size, patch_size),
# stride=(patch_size, patch_size),
# in_chans=in_chans,
# embed_dim=embed_dim,
# )
self.patch_embed = PatchEmbed(
patch_size=patch_size,
in_chans=in_chans,
embed_dim=embed_dim,
spatial_dims=3,
)
self.pos_embed: Optional[nn.Parameter] = None
if use_abs_pos:
# Initialize absolute positional embedding with pretrain image size.
self.pos_embed = nn.Parameter(
torch.zeros(1, img_size // patch_size, img_size // patch_size, img_size // patch_size, embed_dim)
)
self.blocks = nn.ModuleList()
for i in range(depth):
block = Block(
dim=embed_dim,
num_heads=num_heads,
mlp_ratio=mlp_ratio,
qkv_bias=qkv_bias,
norm_layer=norm_layer,
act_layer=act_layer,
use_rel_pos=use_rel_pos,
rel_pos_zero_init=rel_pos_zero_init,
window_size=window_size if i not in global_attn_indexes else 0,
input_size=(img_size // patch_size, img_size // patch_size),
)
self.blocks.append(block)
self.neck = nn.Sequential(
nn.Conv2d(
embed_dim,
out_chans,
kernel_size=1,
bias=False,
),
LayerNorm2d(out_chans),
nn.Conv2d(
out_chans,
out_chans,
kernel_size=3,
padding=1,
bias=False,
),
LayerNorm2d(out_chans),
)
def forward(self, x: torch.Tensor) -> torch.Tensor:
x = self.patch_embed(x)
print('patch embedded shape: ', x.shape) # embedded: [8, 768, 6, 6, 6]
if self.pos_embed is not None:
x = x + self.pos_embed
for blk in self.blocks:
x = blk(x)
x = self.neck(x.permute(0, 3, 1, 2))
return x
class Block(nn.Module):
"""Transformer blocks with support of window attention and residual propagation blocks"""
def __init__(
self,
dim: int,
num_heads: int,
mlp_ratio: float = 4.0,
qkv_bias: bool = True,
norm_layer: Type[nn.Module] = nn.LayerNorm,
act_layer: Type[nn.Module] = nn.GELU,
use_rel_pos: bool = False,
rel_pos_zero_init: bool = True,
window_size: int = 0,
input_size: Optional[Tuple[int, int]] = None,
) -> None:
"""
Args:
dim (int): Number of input channels.
num_heads (int): Number of attention heads in each ViT block.
mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
qkv_bias (bool): If True, add a learnable bias to query, key, value.
norm_layer (nn.Module): Normalization layer.
act_layer (nn.Module): Activation layer.
use_rel_pos (bool): If True, add relative positional embeddings to the attention map.
rel_pos_zero_init (bool): If True, zero initialize relative positional parameters.
window_size (int): Window size for window attention blocks. If it equals 0, then
use global attention.
input_size (tuple(int, int) or None): Input resolution for calculating the relative
positional parameter size.
"""
super().__init__()
self.norm1 = norm_layer(dim)
self.attn = Attention2(
dim,
num_heads=num_heads,
qkv_bias=qkv_bias,
use_rel_pos=use_rel_pos,
rel_pos_zero_init=rel_pos_zero_init,
input_size=input_size if window_size == 0 else (window_size, window_size),
)
self.norm2 = norm_layer(dim)
self.mlp = MLPBlock(embedding_dim=dim, mlp_dim=int(dim * mlp_ratio), act=act_layer)
self.window_size = window_size
def forward(self, x: torch.Tensor) -> torch.Tensor:
shortcut = x
x = self.norm1(x)
# Window partition
if self.window_size > 0:
H, W = x.shape[1], x.shape[2]
x, pad_hw = window_partition(x, self.window_size)
x = self.attn(x)
# Reverse window partition
if self.window_size > 0:
x = window_unpartition(x, self.window_size, pad_hw, (H, W))
x = shortcut + x
x = x + self.mlp(self.norm2(x))
return x
class Attention2(nn.Module):
"""Multi-head Attention block with relative position embeddings."""
def __init__(
self,
dim: int,
num_heads: int = 8,
qkv_bias: bool = True,
use_rel_pos: bool = False,
rel_pos_zero_init: bool = True,
input_size: Optional[Tuple[int, int]] = None,
) -> None:
"""
Args:
dim (int): Number of input channels.
num_heads (int): Number of attention heads.
qkv_bias (bool): If True, add a learnable bias to query, key, value.
rel_pos (bool): If True, add relative positional embeddings to the attention map.
rel_pos_zero_init (bool): If True, zero initialize relative positional parameters.
input_size (tuple(int, int) or None): Input resolution for calculating the relative
positional parameter size.
"""
super().__init__()
self.num_heads = num_heads
head_dim = dim // num_heads
self.scale = head_dim ** -0.5
self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias)
self.proj = nn.Linear(dim, dim)
self.use_rel_pos = use_rel_pos
if self.use_rel_pos:
assert (
input_size is not None
), "Input size must be provided if using relative positional encoding."
# initialize relative positional embeddings
self.rel_pos_h = nn.Parameter(torch.zeros(2 * input_size[0] - 1, head_dim))
self.rel_pos_w = nn.Parameter(torch.zeros(2 * input_size[1] - 1, head_dim))
def forward(self, x: torch.Tensor) -> torch.Tensor:
B, H, W, _ = x.shape
# qkv with shape (3, B, nHead, H * W, C)
qkv = self.qkv(x).reshape(B, H * W, 3, self.num_heads, -1).permute(2, 0, 3, 1, 4)
# q, k, v with shape (B * nHead, H * W, C)
q, k, v = qkv.reshape(3, B * self.num_heads, H * W, -1).unbind(0)
attn = (q * self.scale) @ k.transpose(-2, -1)
if self.use_rel_pos:
attn = add_decomposed_rel_pos(attn, q, self.rel_pos_h, self.rel_pos_w, (H, W), (H, W))
attn = attn.softmax(dim=-1)
x = (attn @ v).view(B, self.num_heads, H, W, -1).permute(0, 2, 3, 1, 4).reshape(B, H, W, -1)
x = self.proj(x)
return x
def window_partition(x: torch.Tensor, window_size: int) -> Tuple[torch.Tensor, Tuple[int, int]]:
"""
Partition into non-overlapping windows with padding if needed.
Args:
x (tensor): input tokens with [B, H, W, C].
window_size (int): window size.
Returns:
windows: windows after partition with [B * num_windows, window_size, window_size, C].
(Hp, Wp): padded height and width before partition
"""
B, H, W, C = x.shape
pad_h = (window_size - H % window_size) % window_size
pad_w = (window_size - W % window_size) % window_size
if pad_h > 0 or pad_w > 0:
x = F.pad(x, (0, 0, 0, pad_w, 0, pad_h))
Hp, Wp = H + pad_h, W + pad_w
x = x.view(B, Hp // window_size, window_size, Wp // window_size, window_size, C)
windows = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(-1, window_size, window_size, C)
return windows, (Hp, Wp)
def window_unpartition(windows: torch.Tensor, window_size: int, pad_hw: Tuple[int, int], hw: Tuple[int, int]) -> torch.Tensor:
"""
Window unpartition into original sequences and removing padding.
Args:
windows (tensor): input tokens with [B * num_windows, window_size, window_size, C].
window_size (int): window size.
pad_hw (Tuple): padded height and width (Hp, Wp).
hw (Tuple): original height and width (H, W) before padding.
Returns:
x: unpartitioned sequences with [B, H, W, C].
"""
Hp, Wp = pad_hw
H, W = hw
B = windows.shape[0] // (Hp * Wp // window_size // window_size)
x = windows.view(B, Hp // window_size, Wp // window_size, window_size, window_size, -1)
x = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(B, Hp, Wp, -1)
if Hp > H or Wp > W:
x = x[:, :H, :W, :].contiguous()
return x
def get_rel_pos(q_size: int, k_size: int, rel_pos: torch.Tensor) -> torch.Tensor:
"""
Get relative positional embeddings according to the relative positions of
query and key sizes.
Args:
q_size (int): size of query q.
k_size (int): size of key k.
rel_pos (Tensor): relative position embeddings (L, C).
Returns:
Extracted positional embeddings according to relative positions.
"""
max_rel_dist = int(2 * max(q_size, k_size) - 1)
# Interpolate rel pos if needed.
if rel_pos.shape[0] != max_rel_dist:
# Interpolate rel pos.
rel_pos_resized = F.interpolate(
rel_pos.reshape(1, rel_pos.shape[0], -1).permute(0, 2, 1),
size=max_rel_dist,
mode="linear",
)
rel_pos_resized = rel_pos_resized.reshape(-1, max_rel_dist).permute(1, 0)
else:
rel_pos_resized = rel_pos
# Scale the coords with short length if shapes for q and k are different.
q_coords = torch.arange(q_size)[:, None] * max(k_size / q_size, 1.0)
k_coords = torch.arange(k_size)[None, :] * max(q_size / k_size, 1.0)
relative_coords = (q_coords - k_coords) + (k_size - 1) * max(q_size / k_size, 1.0)
return rel_pos_resized[relative_coords.long()]
def add_decomposed_rel_pos(
attn: torch.Tensor,
q: torch.Tensor,
rel_pos_h: torch.Tensor,
rel_pos_w: torch.Tensor,
q_size: Tuple[int, int],
k_size: Tuple[int, int],
) -> torch.Tensor:
"""
Calculate decomposed Relative Positional Embeddings from :paper:`mvitv2`.
https://github.com/facebookresearch/mvit/blob/19786631e330df9f3622e5402b4a419a263a2c80/mvit/models/attention.py # noqa B950
Args:
attn (Tensor): attention map.
q (Tensor): query q in the attention layer with shape (B, q_h * q_w, C).
rel_pos_h (Tensor): relative position embeddings (Lh, C) for height axis.
rel_pos_w (Tensor): relative position embeddings (Lw, C) for width axis.
q_size (Tuple): spatial sequence size of query q with (q_h, q_w).
k_size (Tuple): spatial sequence size of key k with (k_h, k_w).
Returns:
attn (Tensor): attention map with added relative positional embeddings.
"""
q_h, q_w = q_size
k_h, k_w = k_size
Rh = get_rel_pos(q_h, k_h, rel_pos_h)
Rw = get_rel_pos(q_w, k_w, rel_pos_w)
B, _, dim = q.shape
r_q = q.reshape(B, q_h, q_w, dim)
rel_h = torch.einsum("bhwc,hkc->bhwk", r_q, Rh)
rel_w = torch.einsum("bhwc,wkc->bhwk", r_q, Rw)
attn = (
attn.view(B, q_h, q_w, k_h, k_w) + rel_h[:, :, :, :, None] + rel_w[:, :, :, None, :]
).view(B, q_h * q_w, k_h * k_w)
return attn
class IdentityMap(nn.Module):
def __init__(self):
super().__init__()
def forward(self, x, *args, **kwargs):
return x
@property
def config(self):
return {"mm_projector_type": 'identity'}
class SpatialPoolingProjector(nn.Module):
def __init__(self, image_size, patch_size, in_dim, out_dim, layer_type, layer_num, pooling_type='spatial', pooling_size=2):
super().__init__()
self.in_dim = in_dim
self.pooling_size = pooling_size
self.num_patches_pre = [img // pch for img, pch in zip(image_size, patch_size)]
self.num_patches_post = [num // pooling_size for num in self.num_patches_pre]
if layer_type == 'linear':
depth = int(layer_num)
modules = [nn.Linear(in_dim, out_dim)]
for _ in range(1, depth):
modules.append(nn.Linear(out_dim, out_dim))
self.projector = nn.Sequential(*modules)
elif layer_type == 'mlp':
depth = int(layer_num)
modules = [nn.Linear(in_dim, out_dim)]
for _ in range(1, depth):
modules.append(nn.GELU())
modules.append(nn.Linear(out_dim, out_dim))
self.projector = nn.Sequential(*modules)
else:
print("Projector error!")
self.pooling_type = pooling_type
def forward(self, x):
B = x.shape[0] # B*N*D
if self.pooling_type == 'spatial':
to_3d = Rearrange("b (p1 p2 p3) d -> b d p1 p2 p3", b=B, d=self.in_dim, p1=self.num_patches_pre[0], p2=self.num_patches_pre[1], p3=self.num_patches_pre[2])
x = to_3d(x)
x = F.avg_pool3d(x, kernel_size=self.pooling_size, stride=self.pooling_size)
to_seq = Rearrange("b d p1 p2 p3 -> b (p1 p2 p3) d", b=B, d=self.in_dim, p1=self.num_patches_post[0], p2=self.num_patches_post[1], p3=self.num_patches_post[2])
x = to_seq(x)
elif self.pooling_type == 'sequence':
x = x.permute(0, 2, 1) #b d n
x = F.avg_pool1d(x, kernel_size=self.pooling_size**3, stride=self.pooling_size**3)
x = x.permute(0, 2, 1) #b n d
x = rearrange(x, "b n d -> (b n) d")
x = self.projector(x)
x = rearrange(x, "(b n) d -> b n d", b=B)
return x
@property
def proj_out_num(self):
num = 1
for n in self.num_patches_post:
num *= n
return num
class Minigpt(nn.Module):
def __init__(self, config=None):
super(Minigpt, self).__init__()
# c*4 is the input size, and c is the output size for the linear layer
inc, ouc = config.mm_hidden_size, config.hidden_size
self.linear = nn.Linear(inc * 4, ouc)
def forward(self, x):
# x is the input tensor with shape [b, num_tokens, c]
b, num_tokens, c = x.shape
# Check if num_tokens is divisible by 4
if num_tokens % 4 != 0:
raise ValueError("num_tokens must be divisible by 4")
# Reshape x to [b, num_tokens/4, c*4]
x = x.view(b, num_tokens // 4, c * 4)
# Apply the linear transformation
x = self.linear(x)
return x
class Vanilla(nn.Module):
def __init__(self, config=None):
super(Vanilla, self).__init__()
# c*4 is the input size, and c is the output size for the linear layer
inc, ouc = config.mm_hidden_size, config.hidden_size
self.linear = nn.Linear(inc * 4, ouc)
def forward(self, x):
b, num_tokens, c = x.shape
# Check if num_tokens is divisible by 4
if num_tokens % 4 != 0:
raise ValueError("num_tokens must be divisible by 4")
# First, reshape to [b, num_tokens//4, 4, c]
x = x.view(b, num_tokens // 4, 4, c)
# Then, permute to interleave the tokens
x = x.permute(0, 1, 3, 2).contiguous()
# Finally, reshape to [b, num_tokens//4, c*4] to interleave features of 4 tokens
x = x.view(b, num_tokens // 4, c * 4)
# Apply the linear transformation
x = self.linear(x)
return x
def build_mm_projector(config, delay_load=False, **kwargs):
projector_type = getattr(config, 'mm_projector_type')
if projector_type == 'linear':
return nn.Linear(config.mm_hidden_size, config.hidden_size)
elif projector_type == 'spp':
return SpatialPoolingProjector(image_size=config.image_size,
patch_size=config.patch_size,
in_dim=config.mm_hidden_size,
out_dim=config.hidden_size,
layer_type=config.proj_layer_type,
layer_num=config.proj_layer_num,
pooling_type=config.proj_pooling_type,
pooling_size=config.proj_pooling_size)
elif projector_type == 'identity':
return IdentityMap()
else:
raise ValueError(f'Unknown projector type: {projector_type}')
class myViT(nn.Module):
"""
Vision Transformer (ViT), based on: "Dosovitskiy et al.,
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale <https://arxiv.org/abs/2010.11929>"
ViT supports Torchscript but only works for Pytorch after 1.8.
"""
def __init__(
self,
in_channels: int,
img_size: Sequence[int] | int,
patch_size: Sequence[int] | int,
hidden_size: int = 768,
mlp_dim: int = 3072,
num_layers: int = 12,
num_heads: int = 12,
pos_embed: str = "conv",
classification: bool = False,
num_classes: int = 2,
dropout_rate: float = 0.0,
spatial_dims: int = 3,
post_activation="Tanh",
qkv_bias: bool = False,
save_attn: bool = False,
) -> None:
"""
Args:
in_channels (int): dimension of input channels.
img_size (Union[Sequence[int], int]): dimension of input image.
patch_size (Union[Sequence[int], int]): dimension of patch size.
hidden_size (int, optional): dimension of hidden layer. Defaults to 768.
mlp_dim (int, optional): dimension of feedforward layer. Defaults to 3072.
num_layers (int, optional): number of transformer blocks. Defaults to 12.
num_heads (int, optional): number of attention heads. Defaults to 12.
pos_embed (str, optional): position embedding layer type. Defaults to "conv".
classification (bool, optional): bool argument to determine if classification is used. Defaults to False.
num_classes (int, optional): number of classes if classification is used. Defaults to 2.
dropout_rate (float, optional): faction of the input units to drop. Defaults to 0.0.
spatial_dims (int, optional): number of spatial dimensions. Defaults to 3.
post_activation (str, optional): add a final acivation function to the classification head
when `classification` is True. Default to "Tanh" for `nn.Tanh()`.
Set to other values to remove this function.
qkv_bias (bool, optional): apply bias to the qkv linear layer in self attention block. Defaults to False.
save_attn (bool, optional): to make accessible the attention in self attention block. Defaults to False.
Examples::
# for single channel input with image size of (96,96,96), conv position embedding and segmentation backbone
>>> net = ViT(in_channels=1, img_size=(96,96,96), pos_embed='conv')
# for 3-channel with image size of (128,128,128), 24 layers and classification backbone
>>> net = ViT(in_channels=3, img_size=(128,128,128), pos_embed='conv', classification=True)
# for 3-channel with image size of (224,224), 12 layers and classification backbone
>>> net = ViT(in_channels=3, img_size=(224,224), pos_embed='conv', classification=True, spatial_dims=2)
"""
super().__init__()
if not (0 <= dropout_rate <= 1):
raise ValueError("dropout_rate should be between 0 and 1.")
if hidden_size % num_heads != 0:
raise ValueError("hidden_size should be divisible by num_heads.")
self.hidden_size = hidden_size
self.classification = classification
self.patch_embedding = PatchEmbeddingBlock(
in_channels=in_channels,
img_size=img_size,
patch_size=patch_size,
hidden_size=hidden_size,
num_heads=num_heads,
pos_embed=pos_embed,
dropout_rate=dropout_rate,
spatial_dims=spatial_dims,
)
self.blocks = nn.ModuleList(
[
TransformerBlock(hidden_size, mlp_dim, num_heads, dropout_rate, qkv_bias, save_attn)
for i in range(num_layers)
]
)
self.norm = nn.LayerNorm(hidden_size)
if self.classification:
self.cls_token = nn.Parameter(torch.zeros(1, 1, hidden_size))
# if post_activation == "Tanh":
# self.classification_head = nn.Sequential(nn.Linear(hidden_size, num_classes), nn.Tanh())
# else:
# self.classification_head = nn.Linear(hidden_size, num_classes) # type: ignore
def forward(self, x):
x = self.patch_embedding(x)
if hasattr(self, "cls_token"):
cls_token = self.cls_token.expand(x.shape[0], -1, -1)
x = torch.cat((cls_token, x), dim=1)
hidden_states_out = []
for blk in self.blocks:
x = blk(x)
hidden_states_out.append(x)
x = self.norm(x)
# if hasattr(self, "classification_head"):
# x = self.classification_head(x[:, 0])
return x, hidden_states_out
class ViT3DTower(nn.Module):
def __init__(self, config):
super().__init__()
self.config = config
self.select_layer = config.vision_select_layer
self.select_feature = config.vision_select_feature
self.vision_tower = myViT(
in_channels=self.config.image_channel,
img_size=self.config.image_size,
patch_size=self.config.patch_size,
pos_embed="perceptron",
spatial_dims=len(self.config.patch_size),
classification=True,
)
def forward(self, images):
last_feature, hidden_states = self.vision_tower(images)
if self.select_layer == -1:
image_features = last_feature
elif self.select_layer < -1:
image_features = hidden_states[self.select_feature]
else:
raise ValueError(f'Unexpected select layer: {self.select_layer}')
if self.select_feature == 'patch':
image_features = image_features[:, 1:]
elif self.select_feature == 'cls_patch':
image_features = image_features
else:
raise ValueError(f'Unexpected select feature: {self.select_feature}')
return image_features
@property
def dtype(self):
return self.vision_tower.dtype
@property
def device(self):
return self.vision_tower.device
@property
def hidden_size(self):
return self.vision_tower.hidden_size
def build_vision_tower(config, **kwargs):
vision_tower = getattr(config, 'vision_tower', None)
if 'vit3d' in vision_tower.lower():
return ViT3DTower(config, **kwargs)
else:
raise ValueError(f'Unknown vision tower: {vision_tower}')
class LamedMetaModel:
def __init__(self, config):
super(LamedMetaModel, self).__init__(config)
self.config = config
if hasattr(config, "vision_tower"):
self.vision_tower = build_vision_tower(config)
self.mm_projector = build_mm_projector(config)
def get_vision_tower(self):
vision_tower = getattr(self, 'vision_tower', None)
return vision_tower
def initialize_vision_modules(self, model_args):
self.config.image_channel = model_args.image_channel
self.config.image_size = model_args.image_size
self.config.patch_size = model_args.patch_size
self.config.vision_tower = model_args.vision_tower
self.config.vision_select_layer = model_args.vision_select_layer
self.config.vision_select_feature = model_args.vision_select_feature
self.config.mm_projector_type = model_args.mm_projector_type
self.config.proj_layer_type = model_args.proj_layer_type
self.config.proj_layer_num = model_args.proj_layer_num
self.config.proj_pooling_type = model_args.proj_pooling_type
self.config.proj_pooling_size = model_args.proj_pooling_size
# vision tower
if self.get_vision_tower() is None:
self.vision_tower = build_vision_tower(self.config)
# If you have a more robust vision encoder, try freezing the vision tower by requires_grad_(False)
if model_args.pretrain_vision_model is not None:
vision_model_weights = torch.load(model_args.pretrain_vision_model, map_location='cpu')
self.vision_tower.vision_tower.load_state_dict(vision_model_weights, strict=True)
self.config.mm_hidden_size = self.vision_tower.hidden_size
# mm_projector
if getattr(self, 'mm_projector', None) is None:
self.mm_projector = build_mm_projector(self.config)
if model_args.pretrain_mm_mlp_adapter is not None:
mm_projector_weights = torch.load(model_args.pretrain_mm_mlp_adapter, map_location='cpu')
def get_w(weights, keyword):
return {k.split(keyword + '.')[1]: v for k, v in weights.items() if keyword in k}
self.mm_projector.load_state_dict(get_w(mm_projector_weights, 'mm_projector'), strict=True)
class LamedMetaForCausalLM(ABC):
@abstractmethod
def get_model(self):
pass
def get_vision_tower(self):
return self.get_model().get_vision_tower()
def encode_images(self, images):
image_features = self.get_model().get_vision_tower()(images)
image_features = self.get_model().mm_projector(image_features)
return image_features
def prepare_inputs_for_multimodal(self, input_ids, position_ids, attention_mask, past_key_values, labels, images):
vision_tower = self.get_vision_tower()
if vision_tower is None or images is None or input_ids.shape[1] == 1:
return input_ids, position_ids, attention_mask, past_key_values, None, labels
else:
image_features = self.encode_images(images)
inputs_embeds = self.get_model().embed_tokens(input_ids)
inputs_embeds = torch.cat((inputs_embeds[:, :1, :], image_features, inputs_embeds[:, (image_features.shape[1] + 1):, :]), dim=1)
return None, position_ids, attention_mask, past_key_values, inputs_embeds, labels
def initialize_vision_tokenizer(self, model_args, tokenizer):
num_new_tokens = model_args.num_new_tokens
self.resize_token_embeddings(len(tokenizer))
if num_new_tokens > 0:
input_embeddings = self.get_input_embeddings().weight.data
output_embeddings = self.get_output_embeddings().weight.data
input_embeddings_avg = input_embeddings[:-num_new_tokens].mean(
dim=0, keepdim=True)
output_embeddings_avg = output_embeddings[:-num_new_tokens].mean(
dim=0, keepdim=True)
input_embeddings[-num_new_tokens:] = input_embeddings_avg
output_embeddings[-num_new_tokens:] = output_embeddings_avg
if model_args.tune_mm_mlp_adapter:
for p in self.get_input_embeddings().parameters():
p.requires_grad = True
for p in self.get_output_embeddings().parameters():
p.requires_grad = False
else:
# we add 4 new tokens
# if new tokens need input, please train input_embeddings
for p in self.get_input_embeddings().parameters():
p.requires_grad = True
# if new tokens need predict, please train output_embeddings
for p in self.get_output_embeddings().parameters():
p.requires_grad = True
if model_args.pretrain_mm_mlp_adapter:
mm_projector_weights = torch.load(model_args.pretrain_mm_mlp_adapter, map_location='cpu')
embed_tokens_weight = mm_projector_weights['model.embed_tokens.weight']
if input_embeddings.shape == embed_tokens_weight.shape:
input_embeddings = embed_tokens_weight
elif embed_tokens_weight.shape[0] == num_new_tokens:
input_embeddings[-num_new_tokens:] = embed_tokens_weight
else:
raise ValueError(f"Unexpected embed_tokens_weight shape. Pretrained: {embed_tokens_weight.shape}. Current: {input_embeddings.shape}. Numer of new tokens: {num_new_tokens}.")
class LamedLlamaModel(LamedMetaModel, LlamaModel):
config_class = LamedConfig
def __init__(self, config: LlamaConfig):
super(LamedLlamaModel, self).__init__(config)
class LamedLlamaForCausalLM(LamedMetaForCausalLM, LlamaForCausalLM):
config_class = LamedConfig
def __init__(self, config):
super(LlamaForCausalLM, self).__init__(config)
self.model = LamedLlamaModel(config)
self.pretraining_tp = config.pretraining_tp
self.vocab_size = config.vocab_size
self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
# Initialize weights and apply final processing
self.post_init()
def get_model(self):
return self.model
def forward(
self,
images: Optional[torch.FloatTensor] = None,
input_ids: torch.LongTensor = None,
labels: Optional[torch.LongTensor] = None,
attention_mask: Optional[torch.Tensor] = None,
position_ids: Optional[torch.LongTensor] = None,
past_key_values: Optional[List[torch.FloatTensor]] = None,
inputs_embeds: Optional[torch.FloatTensor] = None,
use_cache: Optional[bool] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
cache_position: Optional[torch.LongTensor] = None,
) -> Union[Tuple, CausalLMOutputWithPast]:
input_ids_pre = input_ids
if inputs_embeds is None:
(
input_ids,
position_ids,
attention_mask,
past_key_values,
inputs_embeds,
labels
) = self.prepare_inputs_for_multimodal(
input_ids,
position_ids,
attention_mask,
past_key_values,
labels,
images,
)
return super().forward(
input_ids=input_ids,
attention_mask=attention_mask,
position_ids=position_ids,
past_key_values=past_key_values,
inputs_embeds=inputs_embeds,
labels=labels,
use_cache=use_cache,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict
)
@torch.no_grad()
def generate(
self,
images: Optional[torch.Tensor] = None,
inputs: Optional[torch.Tensor] = None,
**kwargs,
) -> Union[GenerateOutput, torch.LongTensor, Any]:
position_ids = kwargs.pop("position_ids", None)
attention_mask = kwargs.pop("attention_mask", None)
if "inputs_embeds" in kwargs:
raise NotImplementedError("`inputs_embeds` is not supported")
if images is not None:
(
inputs,
position_ids,
attention_mask,
_,
inputs_embeds,
_
) = self.prepare_inputs_for_multimodal(
inputs,
position_ids,
attention_mask,
None,
None,
images,
)
else:
inputs_embeds = self.get_model().embed_tokens(inputs)
return super().generate(
position_ids=position_ids,
attention_mask=attention_mask,
inputs_embeds=inputs_embeds,
**kwargs
)
def prepare_inputs_for_generation(self, input_ids, past_key_values=None, inputs_embeds=None, **kwargs):
images = kwargs.pop("images", None)
inputs = super().prepare_inputs_for_generation(input_ids, past_key_values=past_key_values, inputs_embeds=inputs_embeds, **kwargs)
if images is not None:
inputs['images'] = images
return inputs |