nuwa pytorch
0.7.8
Implementierung von NÜWA, einem hochmodernen Aufmerksamkeitsnetzwerk für Text-zu-Video-Synthese, in Pytorch. Es enthält auch eine Erweiterung der Video- und Audioerzeugung unter Verwendung eines Dual-Decoder-Ansatzes.
Yannic Kilcher
DeepReader
März 2022 – Lebenszeichen mit einer schwierigen Version des beweglichen Mnisten sehen
April 2022 – Es scheint, als ob eine diffusionsbasierte Methode den neuen Thron für SOTA erobert hat. Ich werde jedoch mit NUWA fortfahren und es erweitern, um mehrköpfige Codes + hierarchischen Kausaltransformator zu verwenden. Ich denke, dass diese Richtung zur Verbesserung dieser Arbeitsrichtung noch ungenutzt ist.
$ pip install nuwa-pytorchTrainieren Sie zunächst die VAE
import torch
from nuwa_pytorch import VQGanVAE
vae = VQGanVAE (
dim = 512 ,
channels = 3 , # default is 3, but can be changed to any value for the training of the segmentation masks (sketches)
image_size = 256 , # image size
num_layers = 4 , # number of downsampling layers
num_resnet_blocks = 2 , # number of resnet blocks
vq_codebook_size = 8192 , # codebook size
vq_decay = 0.8 # codebook exponential decay
)
imgs = torch . randn ( 10 , 3 , 256 , 256 )
# alternate learning for autoencoder ...
loss = vae ( imgs , return_loss = True )
loss . backward ()
# and the discriminator ...
discr_loss = vae ( imgs , return_discr_loss = True )
discr_loss . backward ()
# do above for many steps
# return reconstructed images and make sure they look ok
recon_imgs = vae ( imgs )Dann mit Ihrem erlernten VAE
import torch
from nuwa_pytorch import NUWA , VQGanVAE
# autoencoder
vae = VQGanVAE (
dim = 64 ,
num_layers = 4 ,
image_size = 256 ,
num_conv_blocks = 2 ,
vq_codebook_size = 8192
)
# NUWA transformer
nuwa = NUWA (
vae = vae ,
dim = 512 ,
text_num_tokens = 20000 , # number of text tokens
text_enc_depth = 12 , # text encoder depth
text_enc_heads = 8 , # number of attention heads for encoder
text_max_seq_len = 256 , # max sequence length of text conditioning tokens (keep at 256 as in paper, or shorter, if your text is not that long)
max_video_frames = 10 , # number of video frames
image_size = 256 , # size of each frame of video
dec_depth = 64 , # video decoder depth
dec_heads = 8 , # number of attention heads in decoder
dec_reversible = True , # reversible networks - from reformer, decoupling memory usage from depth
enc_reversible = True , # reversible encoders, if you need it
attn_dropout = 0.05 , # dropout for attention
ff_dropout = 0.05 , # dropout for feedforward
sparse_3dna_kernel_size = ( 5 , 3 , 3 ), # kernel size of the sparse 3dna attention. can be a single value for frame, height, width, or different values (to simulate axial attention, etc)
sparse_3dna_dilation = ( 1 , 2 , 4 ), # cycle dilation of 3d conv attention in decoder, for more range
shift_video_tokens = True # cheap relative positions for sparse 3dna transformer, by shifting along spatial dimensions by one
). cuda ()
# data
text = torch . randint ( 0 , 20000 , ( 1 , 256 )). cuda ()
video = torch . randn ( 1 , 10 , 3 , 256 , 256 ). cuda () # (batch, frames, channels, height, width)
loss = nuwa (
text = text ,
video = video ,
return_loss = True # set this to True, only for training, to return cross entropy loss
)
loss . backward ()
# do above with as much data as possible
# then you can generate a video from text
video = nuwa . generate ( text = text , num_frames = 5 ) # (1, 5, 3, 256, 256) In dem Artikel stellen sie auch eine Möglichkeit vor, die Videogenerierung basierend auf Segmentierungsmaske(n) zu konditionieren. Sie können dies auch problemlos tun, vorausgesetzt, Sie trainieren zuvor einen VQGanVAE anhand der Skizzen.
Dann verwenden Sie NUWASketch anstelle von NUWA , das die Skizzen-VAE als Referenz akzeptieren kann
ex.
import torch
from nuwa_pytorch import NUWASketch , VQGanVAE
# autoencoder, one for main video, the other for the sketch
vae = VQGanVAE (
dim = 64 ,
num_layers = 4 ,
image_size = 256 ,
num_conv_blocks = 2 ,
vq_codebook_size = 8192
)
sketch_vae = VQGanVAE (
dim = 512 ,
channels = 5 , # say the sketch has 5 classes
num_layers = 4 ,
image_size = 256 ,
num_conv_blocks = 2 ,
vq_codebook_size = 8192
)
# NUWA transformer for conditioning with sketches
nuwa = NUWASketch (
vae = vae ,
sketch_vae = sketch_vae ,
dim = 512 , # model dimensions
sketch_enc_depth = 12 , # sketch encoder depth
sketch_enc_heads = 8 , # number of attention heads for sketch encoder
sketch_max_video_frames = 3 , # max number of frames for sketches
sketch_enc_use_sparse_3dna = True , # whether to use 3d-nearby attention (of full attention if False) for sketch encoding transformer
max_video_frames = 10 , # number of video frames
image_size = 256 , # size of each frame of video
dec_depth = 64 , # video decoder depth
dec_heads = 8 , # number of attention heads in decoder
dec_reversible = True , # reversible networks - from reformer, decoupling memory usage from depth
enc_reversible = True , # reversible encoders, if you need it
attn_dropout = 0.05 , # dropout for attention
ff_dropout = 0.05 , # dropout for feedforward
sparse_3dna_kernel_size = ( 5 , 3 , 3 ), # kernel size of the sparse 3dna attention. can be a single value for frame, height, width, or different values (to simulate axial attention, etc)
sparse_3dna_dilation = ( 1 , 2 , 4 ), # cycle dilation of 3d conv attention in decoder, for more range
cross_2dna_kernel_size = 5 , # 2d kernel size of spatial grouping of attention from video frames to sketches
cross_2dna_dilation = 1 , # 2d dilation of spatial attention from video frames to sketches
shift_video_tokens = True # cheap relative positions for sparse 3dna transformer, by shifting along spatial dimensions by one
). cuda ()
# data
sketch = torch . randn ( 2 , 2 , 5 , 256 , 256 ). cuda () # (batch, frames, segmentation classes, height, width)
sketch_mask = torch . ones ( 2 , 2 ). bool (). cuda () # (batch, frames) [Optional]
video = torch . randn ( 2 , 10 , 3 , 256 , 256 ). cuda () # (batch, frames, channels, height, width)
loss = nuwa (
sketch = sketch ,
sketch_mask = sketch_mask ,
video = video ,
return_loss = True # set this to True, only for training, to return cross entropy loss
)
loss . backward ()
# do above with as much data as possible
# then you can generate a video from sketch(es)
video = nuwa . generate ( sketch = sketch , num_frames = 5 ) # (1, 5, 3, 256, 256) Dieses Repository wird auch eine Variante von NUWA anbieten, die sowohl Video als auch Audio produzieren kann. Vorerst muss der Ton manuell kodiert werden.
import torch
from nuwa_pytorch import NUWAVideoAudio , VQGanVAE
# autoencoder
vae = VQGanVAE (
dim = 64 ,
num_layers = 4 ,
image_size = 256 ,
num_conv_blocks = 2 ,
vq_codebook_size = 100
)
# NUWA transformer
nuwa = NUWAVideoAudio (
vae = vae ,
dim = 512 ,
num_audio_tokens = 2048 , # codebook size for audio tokens
num_audio_tokens_per_video_frame = 32 , # number of audio tokens per video frame
cross_modality_attn_every = 3 , # cross modality attention every N layers
text_num_tokens = 20000 , # number of text tokens
text_enc_depth = 1 , # text encoder depth
text_enc_heads = 8 , # number of attention heads for encoder
text_max_seq_len = 256 , # max sequence length of text conditioning tokens (keep at 256 as in paper, or shorter, if your text is not that long)
max_video_frames = 10 , # number of video frames
image_size = 256 , # size of each frame of video
dec_depth = 4 , # video decoder depth
dec_heads = 8 , # number of attention heads in decoder
enc_reversible = True , # reversible encoders, if you need it
dec_reversible = True , # quad-branched reversible network, for making depth of twin video / audio decoder independent of network depth. recommended to be turned on unless you have a ton of memory at your disposal
attn_dropout = 0.05 , # dropout for attention
ff_dropout = 0.05 , # dropout for feedforward
sparse_3dna_kernel_size = ( 5 , 3 , 3 ), # kernel size of the sparse 3dna attention. can be a single value for frame, height, width, or different values (to simulate axial attention, etc)
sparse_3dna_dilation = ( 1 , 2 , 4 ), # cycle dilation of 3d conv attention in decoder, for more range
shift_video_tokens = True # cheap relative positions for sparse 3dna transformer, by shifting along spatial dimensions by one
). cuda ()
# data
text = torch . randint ( 0 , 20000 , ( 1 , 256 )). cuda ()
audio = torch . randint ( 0 , 2048 , ( 1 , 32 * 10 )). cuda () # (batch, audio tokens per frame * max video frames)
video = torch . randn ( 1 , 10 , 3 , 256 , 256 ). cuda () # (batch, frames, channels, height, width)
loss = nuwa (
text = text ,
video = video ,
audio = audio ,
return_loss = True # set this to True, only for training, to return cross entropy loss
)
loss . backward ()
# do above with as much data as possible
# then you can generate a video from text
video , audio = nuwa . generate ( text = text , num_frames = 5 ) # (1, 5, 3, 256, 256), (1, 32 * 5 == 160) Diese Bibliothek bietet einige Dienstprogramme, die das Training erleichtern. Für den Anfang können Sie die VQGanVAETrainer -Klasse verwenden, um sich um das Training des VQGanVAE zu kümmern. Wickeln Sie einfach das Modell ein und übergeben Sie auch den Bildordnerpfad sowie die verschiedenen Trainingshyperparameter.
import torch
from nuwa_pytorch import VQGanVAE , VQGanVAETrainer
vae = VQGanVAE (
dim = 64 ,
image_size = 256 ,
num_layers = 5 ,
vq_codebook_size = 1024 ,
vq_use_cosine_sim = True ,
vq_codebook_dim = 32 ,
vq_orthogonal_reg_weight = 10 ,
vq_orthogonal_reg_max_codes = 128 ,
). cuda ()
trainer = VQGanVAETrainer (
vae , # VAE defined above
folder = '/path/to/images' , # path to images
lr = 3e-4 , # learning rate
num_train_steps = 100000 , # number of training steps
batch_size = 8 , # batch size
grad_accum_every = 4 # gradient accumulation (effective batch size is (batch_size x grad_accum_every))
)
trainer . train ()
# results and model checkpoints will be saved periodically to ./results Um NUWA zu trainieren, müssen Sie zunächst einen Ordner mit .gif Dateien und entsprechenden .txt Dateien mit deren Beschriftung organisieren. Es sollte als solches organisiert werden.
ex.
video-and-text-data
┣ cat.gif
┣ cat.txt
┣ dog.gif
┣ dog.txt
┣ turtle.gif
┗ turtle.txt
Anschließend laden Sie Ihr zuvor trainiertes VQGan-VAE und trainieren NUWA mit den Klassen GifVideoDataset und NUWATrainer .
import torch
from nuwa_pytorch import NUWA , VQGanVAE
from nuwa_pytorch . train_nuwa import GifVideoDataset , NUWATrainer
# dataset
ds = GifVideoDataset (
folder = './path/to/videos/' ,
channels = 1
)
# autoencoder
vae = VQGanVAE (
dim = 64 ,
image_size = 256 ,
num_layers = 5 ,
num_resnet_blocks = 2 ,
vq_codebook_size = 512 ,
attn_dropout = 0.1
)
vae . load_state_dict ( torch . load ( './path/to/trained/vae.pt' ))
# NUWA transformer
nuwa = NUWA (
vae = vae ,
dim = 512 ,
text_enc_depth = 6 ,
text_max_seq_len = 256 ,
max_video_frames = 10 ,
dec_depth = 12 ,
dec_reversible = True ,
enc_reversible = True ,
attn_dropout = 0.05 ,
ff_dropout = 0.05 ,
sparse_3dna_kernel_size = ( 5 , 3 , 3 ),
sparse_3dna_dilation = ( 1 , 2 , 4 ),
shift_video_tokens = True
). cuda ()
# data
trainer = NUWATrainer (
nuwa = nuwa , # NUWA transformer
dataset = dataset , # video dataset class
num_train_steps = 1000000 , # number of training steps
lr = 3e-4 , # learning rate
wd = 0.01 , # weight decay
batch_size = 8 , # batch size
grad_accum_every = 4 , # gradient accumulation
max_grad_norm = 0.5 , # gradient clipping
num_sampled_frames = 10 , # number of frames to sample
results_folder = './results' # folder to store checkpoints and samples
)
trainer . train () Diese Bibliothek hängt von dieser Vektorquantisierungsbibliothek ab, die eine Reihe von Verbesserungen mit sich bringt (verbessertes VQGAN, orthogonale Codebuch-Regularisierung usw.). Um eine dieser Verbesserungen zu nutzen, können Sie die Schlüsselwortparameter des Vektorquantisierers konfigurieren, indem Sie vq_ bei VQGanVAE Initialisierung voranstellen.
ex. Kosinussimulation in verbessertem VQGAN vorgeschlagen
from nuwa_pytorch import VQGanVAE
vae = VQGanVAE (
dim = 64 ,
image_size = 256 ,
num_layers = 4 ,
vq_use_cosine_sim = True
# VectorQuantize will be initialized with use_cosine_sim = True
# https://github.com/lucidrains/vector-quantize-pytorch#cosine-similarity
). cuda () @misc { wu2021nuwa ,
title = { N"UWA: Visual Synthesis Pre-training for Neural visUal World creAtion } ,
author = { Chenfei Wu and Jian Liang and Lei Ji and Fan Yang and Yuejian Fang and Daxin Jiang and Nan Duan } ,
year = { 2021 } ,
eprint = { 2111.12417 } ,
archivePrefix = { arXiv } ,
primaryClass = { cs.CV }
} @misc { esser2021taming ,
title = { Taming Transformers for High-Resolution Image Synthesis } ,
author = { Patrick Esser and Robin Rombach and Björn Ommer } ,
year = { 2021 } ,
eprint = { 2012.09841 } ,
archivePrefix = { arXiv } ,
primaryClass = { cs.CV }
} @misc { iashin2021taming ,
title = { Taming Visually Guided Sound Generation } ,
author = { Vladimir Iashin and Esa Rahtu } ,
year = { 2021 } ,
eprint = { 2110.08791 } ,
archivePrefix = { arXiv } ,
primaryClass = { cs.CV }
} @misc { ding2021cogview ,
title = { CogView: Mastering Text-to-Image Generation via Transformers } ,
author = { Ming Ding and Zhuoyi Yang and Wenyi Hong and Wendi Zheng and Chang Zhou and Da Yin and Junyang Lin and Xu Zou and Zhou Shao and Hongxia Yang and Jie Tang } ,
year = { 2021 } ,
eprint = { 2105.13290 } ,
archivePrefix = { arXiv } ,
primaryClass = { cs.CV }
} @misc { kitaev2020reformer ,
title = { Reformer: The Efficient Transformer } ,
author = { Nikita Kitaev and Łukasz Kaiser and Anselm Levskaya } ,
year = { 2020 } ,
eprint = { 2001.04451 } ,
archivePrefix = { arXiv } ,
primaryClass = { cs.LG }
} @misc { shazeer2020talkingheads ,
title = { Talking-Heads Attention } ,
author = { Noam Shazeer and Zhenzhong Lan and Youlong Cheng and Nan Ding and Le Hou } ,
year = { 2020 } ,
eprint = { 2003.02436 } ,
archivePrefix = { arXiv } ,
primaryClass = { cs.LG }
} @misc { shazeer2020glu ,
title = { GLU Variants Improve Transformer } ,
author = { Noam Shazeer } ,
year = { 2020 } ,
url = { https://arxiv.org/abs/2002.05202 }
} @misc { su2021roformer ,
title = { RoFormer: Enhanced Transformer with Rotary Position Embedding } ,
author = { Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu } ,
year = { 2021 } ,
eprint = { 2104.09864 } ,
archivePrefix = { arXiv } ,
primaryClass = { cs.CL }
} @inproceedings { ho2021classifierfree ,
title = { Classifier-Free Diffusion Guidance } ,
author = { Jonathan Ho and Tim Salimans } ,
booktitle = { NeurIPS 2021 Workshop on Deep Generative Models and Downstream Applications } ,
year = { 2021 } ,
url = { https://openreview.net/forum?id=qw8AKxfYbI }
} @misc { liu2021swin ,
title = { Swin Transformer V2: Scaling Up Capacity and Resolution } ,
author = { Ze Liu and Han Hu and Yutong Lin and Zhuliang Yao and Zhenda Xie and Yixuan Wei and Jia Ning and Yue Cao and Zheng Zhang and Li Dong and Furu Wei and Baining Guo } ,
year = { 2021 } ,
eprint = { 2111.09883 } ,
archivePrefix = { arXiv } ,
primaryClass = { cs.CV }
} @misc { crowson2022 ,
author = { Katherine Crowson } ,
url = { https://twitter.com/RiversHaveWings/status/1478093658716966912 }
}Aufmerksamkeit ist die seltenste und reinste Form der Großzügigkeit. - Simone Weil
