File size: 22,969 Bytes
99aee7a |
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 |
# Copyright 2023 Haotian Liu
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from abc import ABC, abstractmethod
import torch
import torch.nn as nn
import torch.nn.functional as F
from .multimodal_encoder.builder import build_vision_tower, build_gen_vision_tower
from .multimodal_projector.builder import build_vision_projector, build_down_projector, build_gen_vision_projector
from llava.constants import IGNORE_INDEX, DEFAULT_IMAGE_TOKEN, DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN, IMAGE_TOKEN_IDX, DEFAULT_IM_START_TOKEN_IDX, DEFAULT_IM_END_TOKEN_IDX
class LlavaMetaModel:
def __init__(self, config):
super(LlavaMetaModel, self).__init__(config)
if hasattr(config, "mm_vision_tower"):
self.vision_tower = build_vision_tower(config, delay_load=True)
self.mm_projector = build_vision_projector(config)
self.down_projector = build_down_projector(config)
if 'unpad' in getattr(config, 'mm_patch_merge_type', ''):
self.image_newline = nn.Parameter(
torch.empty(config.hidden_size, dtype=self.dtype)
)
if hasattr(config, "gen_vision_tower"):
self.gen_vision_tower = build_gen_vision_tower(config, delay_load=True)
self.gen_projector = build_gen_vision_projector(config)
if 'unpad' in getattr(config, 'mm_patch_merge_type', ''):
self.image_newline = nn.Parameter(
torch.empty(config.hidden_size, dtype=self.dtype)
)
def get_vision_tower(self):
vision_tower = getattr(self, 'vision_tower', None)
if type(vision_tower) is list:
vision_tower = vision_tower[0]
return vision_tower
def get_gen_vision_tower(self):
gen_vision_tower = getattr(self, 'gen_vision_tower', None)
if type(gen_vision_tower) is list:
gen_vision_tower = gen_vision_tower[0]
return gen_vision_tower
def initialize_vision_modules(self, model_args, fsdp=None):
vision_tower = model_args.vision_tower
gen_vision_tower = model_args.gen_vision_tower
mm_vision_select_layer = model_args.mm_vision_select_layer
mm_vision_select_feature = model_args.mm_vision_select_feature
pretrain_mm_mlp_adapter = model_args.pretrain_mm_mlp_adapter
pretrain_gen_mlp_adapter = model_args.pretrain_gen_mlp_adapter
mm_patch_merge_type = model_args.mm_patch_merge_type
self.config.mm_vision_tower = vision_tower
self.config.gen_vision_tower = gen_vision_tower
self.config.vision_tower_pretrained = getattr(model_args, "vision_tower_pretrained", "")
if self.get_vision_tower() is None:
vision_tower = build_vision_tower(model_args)
if fsdp is not None and len(fsdp) > 0:
self.vision_tower = [vision_tower]
else:
self.vision_tower = vision_tower
else:
if fsdp is not None and len(fsdp) > 0:
vision_tower = self.vision_tower[0]
else:
vision_tower = self.vision_tower
vision_tower.load_model()
if self.get_gen_vision_tower() is None:
gen_vision_tower = build_gen_vision_tower(model_args)
if fsdp is not None and len(fsdp) > 0:
self.gen_vision_tower = [gen_vision_tower]
else:
self.gen_vision_tower = gen_vision_tower
else:
if fsdp is not None and len(fsdp) > 0:
gen_vision_tower = self.gen_vision_tower[0]
else:
gen_vision_tower = self.gen_vision_tower
gen_vision_tower.load_model()
self.config.use_mm_proj = True
self.config.mm_projector_type = getattr(model_args, 'mm_projector_type', 'linear')
self.config.gen_projector_type = getattr(model_args, 'gen_projector_type', 'linear')
self.config.mm_hidden_size = vision_tower.hidden_size
self.config.gen_hidden_size = gen_vision_tower.hidden_size
self.config.mm_vision_select_layer = mm_vision_select_layer
self.config.mm_vision_select_feature = mm_vision_select_feature
self.config.mm_patch_merge_type = mm_patch_merge_type
self.config.n_query = model_args.n_query
self.config.gen_pooling = model_args.gen_pooling
if getattr(self, 'mm_projector', None) is None:
print("random initiation the mm_project !!!")
self.mm_projector = build_vision_projector(self.config)
if 'unpad' in mm_patch_merge_type:
embed_std = 1 / torch.sqrt(torch.tensor(self.config.hidden_size, dtype=self.dtype))
self.image_newline = nn.Parameter(
torch.randn(self.config.hidden_size, dtype=self.dtype) * embed_std
)
else:
# In case it is frozen by LoRA
for p in self.mm_projector.parameters():
p.requires_grad = True
if getattr(self, 'gen_projector', None) is None:
print("random initiation the gen_projector !!!")
self.gen_projector = build_gen_vision_projector(self.config)
if 'unpad' in mm_patch_merge_type:
embed_std = 1 / torch.sqrt(torch.tensor(self.config.hidden_size, dtype=self.dtype))
self.image_newline = nn.Parameter(
torch.randn(self.config.hidden_size, dtype=self.dtype) * embed_std
)
else:
# In case it is frozen by LoRA
for p in self.gen_projector.parameters():
p.requires_grad = True
if getattr(self, 'down_projector', None) is None:
print("random initiation the down_projector !!!")
self.down_projector = build_down_projector(self.config)
else:
# In case it is frozen by LoRA
for p in self.down_projector.parameters():
p.requires_grad = True
if pretrain_mm_mlp_adapter is not None:
mm_projector_weights = torch.load(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'))
if pretrain_gen_mlp_adapter is not None:
gen_projector_weights = torch.load(pretrain_gen_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.gen_projector.load_state_dict(get_w(gen_projector_weights, 'mm_projector'))
def unpad_image(tensor, original_size):
"""
Unpads a PyTorch tensor of a padded and resized image.
Args:
tensor (torch.Tensor): The image tensor, assumed to be in CxHxW format.
original_size (tuple): The original size of PIL image (width, height).
Returns:
torch.Tensor: The unpadded image tensor.
"""
original_width, original_height = original_size
current_height, current_width = tensor.shape[1:]
original_aspect_ratio = original_width / original_height
current_aspect_ratio = current_width / current_height
if original_aspect_ratio > current_aspect_ratio:
scale_factor = current_width / original_width
new_height = int(original_height * scale_factor)
padding = (current_height - new_height) // 2
unpadded_tensor = tensor[:, padding:current_height - padding, :]
else:
scale_factor = current_height / original_height
new_width = int(original_width * scale_factor)
padding = (current_width - new_width) // 2
unpadded_tensor = tensor[:, :, padding:current_width - padding]
return unpadded_tensor
class LlavaMetaForCausalLM(ABC):
@abstractmethod
def get_model(self):
pass
def get_vision_tower(self):
return self.get_model().get_vision_tower()
def get_gen_vision_tower(self):
return self.get_model().get_gen_vision_tower()
def encode_images(self, images):
device = self.get_vision_tower().device
images = images.to(device)
image_features = self.get_model().get_vision_tower()(images)
num_img, _, c = image_features.shape
gen_pooling = self.get_gen_pooling()
n_query = self.get_n_query() if not 'early' in gen_pooling else 729
if 'pool2d' in gen_pooling:
stride = int(gen_pooling.split('_')[-1])
sqrt_n = int(n_query**0.5)
image_features = image_features.permute(0, 2, 1).view(num_img, -1, sqrt_n, sqrt_n)
image_features = F.avg_pool2d(image_features, kernel_size=(stride, stride), stride=stride)
image_features = image_features.reshape(num_img, c, -1).permute(0,2,1)
# image_features = image_features.contiguous().view(-1, c)
# image_features = self.get_model().mm_projector(image_features)
return image_features
def get_mm_projector(self):
return self.get_model().mm_projector
def get_gen_projector(self):
return self.get_model().gen_projector
def get_n_query(self):
return self.get_model().config.n_query
def get_gen_pooling(self):
return self.get_model().config.gen_pooling
def pool_img(self, image_features):
num_img, n, c = image_features.shape
gen_pooling = self.get_gen_pooling()
# n_query = self.get_n_query()
stride = int(gen_pooling.split('_')[-1])
sqrt_n = int(n**0.5)
image_features = image_features.permute(0, 2, 1).view(num_img, c, sqrt_n, sqrt_n)
image_features = F.avg_pool2d(image_features, kernel_size=(stride, stride), stride=stride)
image_features = image_features.view(num_img, c, -1).permute(0,2,1).contiguous()
return image_features
def prepare_inputs_labels_for_multimodal(
self, input_ids, position_ids, attention_mask, past_key_values, labels,
gen_images, und_images, image_sizes=None
):
vision_tower = self.get_vision_tower()
mm_projector = self.get_mm_projector()
gen_vision_tower = self.get_gen_vision_tower()
gen_projector = self.get_gen_projector()
if (gen_images is None and und_images is None) or input_ids.shape[1] == 1:
return input_ids, position_ids, attention_mask, past_key_values, None, labels, None, None, None
if not gen_images is None:
# print(f"gen_images {gen_images.shape}")
prompt_image_embeds = gen_vision_tower(gen_images) # TODO: check dimension
# print(f"prompt_image_embeds {prompt_image_embeds.shape}")
if 'early' in self.get_gen_pooling():
prompt_image_embeds = self.pool_img(prompt_image_embeds)
num_img, _, c = prompt_image_embeds.shape # [batch, 729, 1152]
# all_image_embeds = torch.clone(prompt_image_embeds).detach()
prompt_image_embeds = prompt_image_embeds.contiguous().view(-1, c)
target_image_embeds = torch.clone(prompt_image_embeds).detach()
prompt_image_embeds = gen_projector(prompt_image_embeds)
else:
# print(f"warning !!!!!!!!!!!!!")
target_image_embeds = None
# quick fix
# change und_images dim so gen_vision_tower process
# und_images torch.Size([2, 3, 336, 336])
# gen_images torch.Size([2, 3, 384, 384])
num_img = und_images.shape[0]
dummy = torch.zeros(num_img, 3, 448, 448 , dtype=und_images.dtype, device=und_images.device) # TODO
temp = gen_vision_tower(dummy)[:,:729,:]
num_img, _, c = temp.shape
temp = temp.contiguous().view(-1, c)
temp = gen_projector(temp) * 1e-9
# print(f"gen temp {temp.sum()}")
if not und_images is None:
# print(f"und_images {und_images.shape}")
und_image_embeds = vision_tower(und_images)
num_img, _, c = und_image_embeds.shape
und_image_embeds = und_image_embeds.contiguous().view(-1, c)
und_image_embeds = mm_projector(und_image_embeds)
if gen_images is None:
und_image_embeds += temp
else:
# print(f"warning !!!!!!!!!!!!!")
num_img = gen_images.shape[0]
dummy = torch.zeros(num_img, 3, 384, 384 , dtype=gen_images.dtype, device=gen_images.device) # clip (3, 336, 336)
temp = vision_tower(dummy)
if 'early' in self.get_gen_pooling():
temp = temp[:,:64,:]
num_img, _, c = temp.shape
temp = temp.contiguous().view(-1, c)
temp = mm_projector(temp) * 1e-9
# print(f"und temp {temp.sum()}")
prompt_image_embeds += temp
image_idx = (input_ids == IMAGE_TOKEN_IDX)
img_indicator = torch.clone(image_idx)
output_indicator = labels != -100
# print(f"### output_indicator {output_indicator.tolist()}")
input_indicator = labels == -100
# print(f"### input_indicator {input_indicator.tolist()}")
# print(f"output_indicator {output_indicator[0]}")
img_loss_indicator = torch.logical_and(output_indicator, img_indicator)
img_loss_indicator = torch.cat(
[img_loss_indicator[:, 1:], img_loss_indicator[:, :1]], dim=1)
img_indicator = torch.cat(
[img_indicator[:, 1:], img_indicator[:, :1]], dim=1)
# num_output_img = img_loss_indicator.sum().item()//self.model.n_query
# print(f"img_loss_indicator {img_loss_indicator[0]}")
# print(f"img_loss_indicator.sum() {img_loss_indicator.sum()}")
if not target_image_embeds is None:
target_image_embeds = target_image_embeds[-img_loss_indicator.sum():,:]
# print(f"target_image_embeds {target_image_embeds}")
# print(f"before embed input ids")
# print(f"image_idx.sum() {image_idx.sum()}")
# print(f"input_ids {input_ids[0,:]}")
# print(f"self.model.decoder.lm.model.emb {self.model.decoder.lm.get_input_embeddings().weight.data.shape}")
text_embeds = self.get_model().embed_tokens(input_ids)
# print(f"text_embeds {text_embeds}")
# print(f"break 1")
N_QUERY = self.get_n_query()
# if not image_idx.sum()/N_QUERY == image_idx.sum()//N_QUERY:
# print('warning half image: ', image_idx.sum()/N_QUERY, image_idx.sum()//N_QUERY)
# breakpoint()
# print(f"image_idx {image_idx}")
# print(f"text_embeds {text_embeds}, prompt_image_embeds {prompt_image_embeds}")
# print(f"prompt_image_embeds {prompt_image_embeds}")
gen_img_idx = torch.logical_and(output_indicator, image_idx)
if not target_image_embeds is None:
text_embeds[gen_img_idx] = prompt_image_embeds.to(text_embeds.device)[:gen_img_idx.sum(),:]
target_image_embeds = target_image_embeds.to(text_embeds.device)[:gen_img_idx.sum(),:]
und_img_idx = torch.logical_and(input_indicator, image_idx)
if not und_images is None:
# text_embeds[und_img_idx] = und_image_embeds.to(text_embeds.device)[:und_img_idx.sum(),:]
# try:
text_embeds[und_img_idx] = und_image_embeds.to(text_embeds.device)[:und_img_idx.sum(), :]
# except RuntimeError as e:
# print(f"RuntimeError: {e}")
# print(f"text_embeds shape: {text_embeds.shape}")
# print(f"und_images: {und_images.shape}")
# print(f"und_image_embeds shape: {und_image_embeds.shape}")
# print(f"und_img_idx sum: {und_img_idx.sum()} (should match number of rows in und_image_embeds)")
# print("Continuing without modifying text_embeds.")
# # Get the shapes involved
# expected_shape = und_img_idx.sum() # Number of True values or indices
# actual_shape = und_image_embeds.shape[0] # Number of rows in und_image_embeds
# if expected_shape > actual_shape:
# # If more indices than embeddings, truncate und_img_idx to match und_image_embeds
# print(f"Shape mismatch: expected {expected_shape} rows, but only {actual_shape} embeddings available.")
# adjusted_idx = und_img_idx.nonzero(as_tuple=True)[0][:actual_shape] # Get the first `actual_shape` indices
# text_embeds[adjusted_idx] = und_image_embeds.to(text_embeds.device)
# print(f"Truncated indices from {expected_shape} to {actual_shape}.")
# else:
# # If more embeddings than indices, trim und_image_embeds to match und_img_idx
# print(f"Shape mismatch: expected {expected_shape} rows, but got {actual_shape}. Using first {expected_shape} embeddings.")
# text_embeds[und_img_idx] = und_image_embeds[:expected_shape, :].to(text_embeds.device)
# print(f"target_image_embeds {target_image_embeds}")
# print(f"break 4")
labels[image_idx] = -100
# print(f"labels[0] {labels[0]}")
# print(f"break 5")
# print({'all_image_embeds':all_image_embeds.shape, 'num_output_img':num_output_img, 'num_img': num_img})
return None, position_ids, attention_mask, past_key_values, text_embeds, labels, img_loss_indicator, img_indicator, target_image_embeds
def prepare_inputs_labels_for_understanding(
self, input_ids, position_ids, attention_mask, past_key_values, labels,
batch_images, image_sizes=None
):
vision_tower = self.get_vision_tower()
mm_projector = self.get_mm_projector()
# pdb.set_trace()
prompt_image_embeds = vision_tower(batch_images) # TODO: check dimension
# print(f"prompt_image_embeds.shape: {prompt_image_embeds.shape}")
num_img, _, c = prompt_image_embeds.shape # [batch, 576, 1024]
all_image_embeds = torch.clone(prompt_image_embeds).detach()
prompt_image_embeds = prompt_image_embeds.contiguous().view(-1, c)
prompt_image_embeds = mm_projector(prompt_image_embeds)
# print(f"prompt_image_embeds {prompt_image_embeds.shape}")
# print(f"input_ids {input_ids}")
# IMAGE = 128259
image_idx = (input_ids == IMAGE_TOKEN_IDX)
# print(f"image_idx {image_idx[0]}")
img_indicator = torch.clone(image_idx)
img_indicator = torch.cat(
[img_indicator[:, 1:], img_indicator[:, :1]], dim=1)
# print(f"before embed input ids")
# print(f"image_idx.sum() {image_idx.sum()}")
# print(f"input_ids {input_ids[0,:]}")
# print(f"self.model.decoder.lm.model.emb {self.model.decoder.lm.get_input_embeddings().weight.data.shape}")
text_embeds = self.get_model().embed_tokens(input_ids)
# print(f"text_embeds {text_embeds}")
# print(f"break 1")
N_QUERY = self.get_n_query()
# if not image_idx.sum()/N_QUERY == image_idx.sum()//N_QUERY:
# print('warning half image: ', image_idx.sum()/N_QUERY, image_idx.sum()//N_QUERY)
# print(f"break 1.5")
# print(f"image_idx {image_idx}")
# print(f"text_embeds {text_embeds}, prompt_image_embeds {prompt_image_embeds}")
text_embeds[image_idx] = prompt_image_embeds.to(text_embeds.device)[:image_idx.sum(),:]
# print({'all_image_embeds':all_image_embeds.shape, 'num_output_img':num_output_img, 'num_img': num_img})
return None, position_ids, attention_mask, past_key_values, text_embeds, img_indicator, labels
def initialize_vision_tokenizer(self, model_args, tokenizer):
if model_args.mm_use_im_patch_token:
tokenizer.add_tokens([DEFAULT_IMAGE_PATCH_TOKEN], special_tokens=True)
self.resize_token_embeddings(len(tokenizer))
if model_args.mm_use_im_start_end:
num_new_tokens = tokenizer.add_tokens([DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN], special_tokens=True)
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
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']
assert num_new_tokens == 2
if input_embeddings.shape == embed_tokens_weight.shape:
input_embeddings[-num_new_tokens:] = embed_tokens_weight[-num_new_tokens:]
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}.")
elif model_args.mm_use_im_patch_token:
if model_args.tune_mm_mlp_adapter:
for p in self.get_input_embeddings().parameters():
p.requires_grad = False
for p in self.get_output_embeddings().parameters():
p.requires_grad = False
|