File size: 14,482 Bytes
2a00960
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
# -*- coding: utf-8 -*-
# Copyright (c) Alibaba, Inc. and its affiliates.
import copy
import random
from contextlib import nullcontext

import torch
import torch.nn.functional as F
from torch import nn

from scepter.modules.model.network.ldm import LatentDiffusion
from scepter.modules.model.registry import MODELS
from scepter.modules.utils.config import dict_to_yaml
from scepter.modules.utils.distribute import we

from ..utils.basic_utils import (
    check_list_of_list,
    pack_imagelist_into_tensor_v2 as pack_imagelist_into_tensor,
    to_device,
    unpack_tensor_into_imagelist
)


class TextEmbedding(nn.Module):
    def __init__(self, embedding_shape):
        super().__init__()
        self.pos = nn.Parameter(data=torch.zeros(embedding_shape))


@MODELS.register_class()
class LdmACE(LatentDiffusion):
    para_dict = LatentDiffusion.para_dict
    para_dict['DECODER_BIAS'] = {'value': 0, 'description': ''}

    def __init__(self, cfg, logger=None):
        super().__init__(cfg, logger=logger)
        self.interpolate_func = lambda x: (F.interpolate(
            x.unsqueeze(0),
            scale_factor=1 / self.size_factor,
            mode='nearest-exact') if x is not None else None)

        self.text_indentifers = cfg.get('TEXT_IDENTIFIER', [])
        self.use_text_pos_embeddings = cfg.get('USE_TEXT_POS_EMBEDDINGS',
                                               False)
        if self.use_text_pos_embeddings:
            self.text_position_embeddings = TextEmbedding(
                (10, 4096)).eval().requires_grad_(False)
        else:
            self.text_position_embeddings = None

        self.logger.info(self.model)

    @torch.no_grad()
    def encode_first_stage(self, x, **kwargs):
        return [
            self.scale_factor *
            self.first_stage_model._encode(i.unsqueeze(0).to(torch.float16))
            for i in x
        ]

    @torch.no_grad()
    def decode_first_stage(self, z):
        return [
            self.first_stage_model._decode(1. / self.scale_factor *
                                           i.to(torch.float16)) for i in z
        ]

    def cond_stage_embeddings(self, prompt, edit_image, cont, cont_mask):
        if self.use_text_pos_embeddings and not torch.sum(
                self.text_position_embeddings.pos) > 0:
            identifier_cont, identifier_cont_mask = getattr(
                self.cond_stage_model, 'encode')(self.text_indentifers,
                                                 return_mask=True)
            self.text_position_embeddings.load_state_dict(
                {'pos': identifier_cont[:, 0, :]})
        cont_, cont_mask_ = [], []
        for pp, edit, c, cm in zip(prompt, edit_image, cont, cont_mask):
            if isinstance(pp, list):
                cont_.append([c[-1], *c] if len(edit) > 0 else [c[-1]])
                cont_mask_.append([cm[-1], *cm] if len(edit) > 0 else [cm[-1]])
            else:
                raise NotImplementedError

        return cont_, cont_mask_

    def limit_batch_data(self, batch_data_list, log_num):
        if log_num and log_num > 0:
            batch_data_list_limited = []
            for sub_data in batch_data_list:
                if sub_data is not None:
                    sub_data = sub_data[:log_num]
                batch_data_list_limited.append(sub_data)
            return batch_data_list_limited
        else:
            return batch_data_list

    def forward_train(self,
                      edit_image=[],
                      edit_image_mask=[],
                      image=None,
                      image_mask=None,
                      noise=None,
                      prompt=[],
                      **kwargs):
        '''
        Args:
            edit_image: list of list of edit_image
            edit_image_mask: list of list of edit_image_mask
            image: target image
            image_mask: target image mask
            noise: default is None, generate automaticly
            prompt: list of list of text
            **kwargs:
        Returns:
        '''
        assert check_list_of_list(prompt) and check_list_of_list(
            edit_image) and check_list_of_list(edit_image_mask)
        assert len(edit_image) == len(edit_image_mask) == len(prompt)
        assert self.cond_stage_model is not None
        gc_seg = kwargs.pop('gc_seg', [])
        gc_seg = int(gc_seg[0]) if len(gc_seg) > 0 else 0
        context = {}

        # process image
        image = to_device(image)
        x_start = self.encode_first_stage(image, **kwargs)
        x_start, x_shapes = pack_imagelist_into_tensor(x_start)  # B, C, L
        n, _, _ = x_start.shape
        t = torch.randint(0, self.num_timesteps, (n, ),
                          device=x_start.device).long()
        context['x_shapes'] = x_shapes

        # process image mask
        image_mask = to_device(image_mask, strict=False)
        context['x_mask'] = [self.interpolate_func(i) for i in image_mask
                             ] if image_mask is not None else [None] * n

        # process text
        # with torch.autocast(device_type="cuda", enabled=True, dtype=torch.bfloat16):
        prompt_ = [[pp] if isinstance(pp, str) else pp for pp in prompt]
        try:
            cont, cont_mask = getattr(self.cond_stage_model,
                                      'encode_list')(prompt_, return_mask=True)
        except Exception as e:
            print(e, prompt_)
        cont, cont_mask = self.cond_stage_embeddings(prompt, edit_image, cont,
                                                     cont_mask)
        context['crossattn'] = cont

        # process edit image & edit image mask
        edit_image = [to_device(i, strict=False) for i in edit_image]
        edit_image_mask = [to_device(i, strict=False) for i in edit_image_mask]
        e_img, e_mask = [], []
        for u, m in zip(edit_image, edit_image_mask):
            if m is None:
                m = [None] * len(u) if u is not None else [None]
            e_img.append(
                self.encode_first_stage(u, **kwargs) if u is not None else u)
            e_mask.append([
                self.interpolate_func(i) if i is not None else None for i in m
            ])
        context['edit'], context['edit_mask'] = e_img, e_mask

        # process loss
        loss = self.diffusion.loss(
            x_0=x_start,
            t=t,
            noise=noise,
            model=self.model,
            model_kwargs={
                'cond':
                context,
                'mask':
                cont_mask,
                'gc_seg':
                gc_seg,
                'text_position_embeddings':
                self.text_position_embeddings.pos if hasattr(
                    self.text_position_embeddings, 'pos') else None
            },
            **kwargs)
        loss = loss.mean()
        ret = {'loss': loss, 'probe_data': {'prompt': prompt}}
        return ret

    @torch.no_grad()
    def forward_test(self,
                     edit_image=[],
                     edit_image_mask=[],
                     image=None,
                     image_mask=None,
                     prompt=[],
                     n_prompt=[],
                     sampler='ddim',
                     sample_steps=20,
                     guide_scale=4.5,
                     guide_rescale=0.5,
                     log_num=-1,
                     seed=2024,
                     **kwargs):

        assert check_list_of_list(prompt) and check_list_of_list(
            edit_image) and check_list_of_list(edit_image_mask)
        assert len(edit_image) == len(edit_image_mask) == len(prompt)
        assert self.cond_stage_model is not None
        # gc_seg is unused
        kwargs.pop('gc_seg', -1)
        # prepare data
        context, null_context = {}, {}

        prompt, n_prompt, image, image_mask, edit_image, edit_image_mask = self.limit_batch_data(
            [prompt, n_prompt, image, image_mask, edit_image, edit_image_mask],
            log_num)
        g = torch.Generator(device=we.device_id)
        seed = seed if seed >= 0 else random.randint(0, 2**32 - 1)
        g.manual_seed(seed)
        n_prompt = copy.deepcopy(prompt)
        # only modify the last prompt to be zero
        for nn_p_id, nn_p in enumerate(n_prompt):
            if isinstance(nn_p, str):
                n_prompt[nn_p_id] = ['']
            elif isinstance(nn_p, list):
                n_prompt[nn_p_id][-1] = ''
            else:
                raise NotImplementedError
        # process image
        image = to_device(image)
        x = self.encode_first_stage(image, **kwargs)
        noise = [
            torch.empty(*i.shape, device=we.device_id).normal_(generator=g)
            for i in x
        ]
        noise, x_shapes = pack_imagelist_into_tensor(noise)
        context['x_shapes'] = null_context['x_shapes'] = x_shapes

        # process image mask
        image_mask = to_device(image_mask, strict=False)
        cond_mask = [self.interpolate_func(i) for i in image_mask
                     ] if image_mask is not None else [None] * len(image)
        context['x_mask'] = null_context['x_mask'] = cond_mask
        # process text
        # with torch.autocast(device_type="cuda", enabled=True, dtype=torch.bfloat16):
        prompt_ = [[pp] if isinstance(pp, str) else pp for pp in prompt]
        cont, cont_mask = getattr(self.cond_stage_model,
                                  'encode_list')(prompt_, return_mask=True)
        cont, cont_mask = self.cond_stage_embeddings(prompt, edit_image, cont,
                                                     cont_mask)
        null_cont, null_cont_mask = getattr(self.cond_stage_model,
                                            'encode_list')(n_prompt,
                                                           return_mask=True)
        null_cont, null_cont_mask = self.cond_stage_embeddings(
            prompt, edit_image, null_cont, null_cont_mask)
        context['crossattn'] = cont
        null_context['crossattn'] = null_cont

        # processe edit image & edit image mask
        edit_image = [to_device(i, strict=False) for i in edit_image]
        edit_image_mask = [to_device(i, strict=False) for i in edit_image_mask]
        e_img, e_mask = [], []
        for u, m in zip(edit_image, edit_image_mask):
            if u is None:
                continue
            if m is None:
                m = [None] * len(u)
            e_img.append(self.encode_first_stage(u, **kwargs))
            e_mask.append([self.interpolate_func(i) for i in m])
        null_context['edit'] = context['edit'] = e_img
        null_context['edit_mask'] = context['edit_mask'] = e_mask

        # process sample
        model = self.model_ema if self.use_ema and self.eval_ema else self.model
        embedding_context = model.no_sync if isinstance(model, torch.distributed.fsdp.FullyShardedDataParallel) \
            else nullcontext
        with embedding_context():
            samples = self.diffusion.sample(
                sampler=sampler,
                noise=noise,
                model=model,
                model_kwargs=[{
                    'cond':
                    context,
                    'mask':
                    cont_mask,
                    'text_position_embeddings':
                    self.text_position_embeddings.pos if hasattr(
                        self.text_position_embeddings, 'pos') else None
                }, {
                    'cond':
                    null_context,
                    'mask':
                    null_cont_mask,
                    'text_position_embeddings':
                    self.text_position_embeddings.pos if hasattr(
                        self.text_position_embeddings, 'pos') else None
                }] if guide_scale is not None and guide_scale > 1 else {
                    'cond':
                    context,
                    'mask':
                    cont_mask,
                    'text_position_embeddings':
                    self.text_position_embeddings.pos if hasattr(
                        self.text_position_embeddings, 'pos') else None
                },
                steps=sample_steps,
                guide_scale=guide_scale,
                guide_rescale=guide_rescale,
                show_progress=True,
                **kwargs)

        samples = unpack_tensor_into_imagelist(samples, x_shapes)
        x_samples = self.decode_first_stage(samples)
        outputs = list()
        for i in range(len(prompt)):
            rec_img = torch.clamp(
                (x_samples[i] + 1.0) / 2.0 + self.decoder_bias / 255,
                min=0.0,
                max=1.0)
            rec_img = rec_img.squeeze(0)
            edit_imgs, edit_img_masks = [], []
            if edit_image is not None and edit_image[i] is not None:
                if edit_image_mask[i] is None:
                    edit_image_mask[i] = [None] * len(edit_image[i])
                for edit_img, edit_mask in zip(edit_image[i],
                                               edit_image_mask[i]):
                    edit_img = torch.clamp((edit_img + 1.0) / 2.0,
                                           min=0.0,
                                           max=1.0)
                    edit_imgs.append(edit_img.squeeze(0))
                    if edit_mask is None:
                        edit_mask = torch.ones_like(edit_img[[0], :, :])
                    edit_img_masks.append(edit_mask)
            one_tup = {
                'reconstruct_image': rec_img,
                'instruction': prompt[i],
                'edit_image': edit_imgs if len(edit_imgs) > 0 else None,
                'edit_mask': edit_img_masks if len(edit_imgs) > 0 else None
            }
            if image is not None:
                if image_mask is None:
                    image_mask = [None] * len(image)
                ori_img = torch.clamp((image[i] + 1.0) / 2.0, min=0.0, max=1.0)
                one_tup['target_image'] = ori_img.squeeze(0)
                one_tup['target_mask'] = image_mask[i] if image_mask[
                    i] is not None else torch.ones_like(ori_img[[0], :, :])
            outputs.append(one_tup)
        return outputs

    @staticmethod
    def get_config_template():
        return dict_to_yaml('MODEL',
                            __class__.__name__,
                            LdmACE.para_dict,
                            set_name=True)