muse maskgit pytorch
0.3.5
Implementación de Muse: generación de texto a imagen mediante transformadores generativos enmascarados, en Pytorch
Únase si está interesado en ayudar con la replicación con la comunidad LAION.
$ pip install muse-maskgit-pytorch Primero entrena tu VAE - VQGanVAE
import torch
from muse_maskgit_pytorch import VQGanVAE , VQGanVAETrainer
vae = VQGanVAE (
dim = 256 ,
codebook_size = 65536
)
# train on folder of images, as many images as possible
trainer = VQGanVAETrainer (
vae = vae ,
image_size = 128 , # you may want to start with small images, and then curriculum learn to larger ones, but because the vae is all convolution, it should generalize to 512 (as in paper) without training on it
folder = '/path/to/images' ,
batch_size = 4 ,
grad_accum_every = 8 ,
num_train_steps = 50000
). cuda ()
trainer . train () Luego pase el VQGanVAE entrenado y un Transformer a MaskGit
import torch
from muse_maskgit_pytorch import VQGanVAE , MaskGit , MaskGitTransformer
# first instantiate your vae
vae = VQGanVAE (
dim = 256 ,
codebook_size = 65536
). cuda ()
vae . load ( '/path/to/vae.pt' ) # you will want to load the exponentially moving averaged VAE
# then you plug the vae and transformer into your MaskGit as so
# (1) create your transformer / attention network
transformer = MaskGitTransformer (
num_tokens = 65536 , # must be same as codebook size above
seq_len = 256 , # must be equivalent to fmap_size ** 2 in vae
dim = 512 , # model dimension
depth = 8 , # depth
dim_head = 64 , # attention head dimension
heads = 8 , # attention heads,
ff_mult = 4 , # feedforward expansion factor
t5_name = 't5-small' , # name of your T5
)
# (2) pass your trained VAE and the base transformer to MaskGit
base_maskgit = MaskGit (
vae = vae , # vqgan vae
transformer = transformer , # transformer
image_size = 256 , # image size
cond_drop_prob = 0.25 , # conditional dropout, for classifier free guidance
). cuda ()
# ready your training text and images
texts = [
'a child screaming at finding a worm within a half-eaten apple' ,
'lizard running across the desert on two feet' ,
'waking up to a psychedelic landscape' ,
'seashells sparkling in the shallow waters'
]
images = torch . randn ( 4 , 3 , 256 , 256 ). cuda ()
# feed it into your maskgit instance, with return_loss set to True
loss = base_maskgit (
images ,
texts = texts
)
loss . backward ()
# do this for a long time on much data
# then...
images = base_maskgit . generate ( texts = [
'a whale breaching from afar' ,
'young girl blowing out candles on her birthday cake' ,
'fireworks with blue and green sparkles'
], cond_scale = 3. ) # conditioning scale for classifier free guidance
images . shape # (3, 3, 256, 256) Para entrenar la superresolución, maskgit requiere que cambies 1 campo en la creación de instancias MaskGit (ahora deberás pasar cond_image_size , ya que se está condicionando el tamaño de imagen anterior)
Opcionalmente, puede pasar un VAE diferente como cond_vae para la imagen condicionada de baja resolución. De forma predeterminada, utilizará vae para tokenizar las imágenes de resolución súper y baja.
import torch
import torch . nn . functional as F
from muse_maskgit_pytorch import VQGanVAE , MaskGit , MaskGitTransformer
# first instantiate your ViT VQGan VAE
# a VQGan VAE made of transformers
vae = VQGanVAE (
dim = 256 ,
codebook_size = 65536
). cuda ()
vae . load ( './path/to/vae.pt' ) # you will want to load the exponentially moving averaged VAE
# then you plug the VqGan VAE into your MaskGit as so
# (1) create your transformer / attention network
transformer = MaskGitTransformer (
num_tokens = 65536 , # must be same as codebook size above
seq_len = 1024 , # must be equivalent to fmap_size ** 2 in vae
dim = 512 , # model dimension
depth = 2 , # depth
dim_head = 64 , # attention head dimension
heads = 8 , # attention heads,
ff_mult = 4 , # feedforward expansion factor
t5_name = 't5-small' , # name of your T5
)
# (2) pass your trained VAE and the base transformer to MaskGit
superres_maskgit = MaskGit (
vae = vae ,
transformer = transformer ,
cond_drop_prob = 0.25 ,
image_size = 512 , # larger image size
cond_image_size = 256 , # conditioning image size <- this must be set
). cuda ()
# ready your training text and images
texts = [
'a child screaming at finding a worm within a half-eaten apple' ,
'lizard running across the desert on two feet' ,
'waking up to a psychedelic landscape' ,
'seashells sparkling in the shallow waters'
]
images = torch . randn ( 4 , 3 , 512 , 512 ). cuda ()
# feed it into your maskgit instance, with return_loss set to True
loss = superres_maskgit (
images ,
texts = texts
)
loss . backward ()
# do this for a long time on much data
# then...
images = superres_maskgit . generate (
texts = [
'a whale breaching from afar' ,
'young girl blowing out candles on her birthday cake' ,
'fireworks with blue and green sparkles' ,
'waking up to a psychedelic landscape'
],
cond_images = F . interpolate ( images , 256 ), # conditioning images must be passed in for generating from superres
cond_scale = 3.
)
images . shape # (4, 3, 512, 512)Todos juntos ahora
from muse_maskgit_pytorch import Muse
base_maskgit . load ( './path/to/base.pt' )
superres_maskgit . load ( './path/to/superres.pt' )
# pass in the trained base_maskgit and superres_maskgit from above
muse = Muse (
base = base_maskgit ,
superres = superres_maskgit
)
images = muse ([
'a whale breaching from afar' ,
'young girl blowing out candles on her birthday cake' ,
'fireworks with blue and green sparkles' ,
'waking up to a psychedelic landscape'
])
images # List[PIL.Image.Image] StabilityAI por el patrocinio, así como a mis otros patrocinadores, por brindarme la independencia para abrir la inteligencia artificial de código abierto.
? Huggingface para los transformadores y la biblioteca de aceleración, ambos maravillosos.
prueba de extremo a extremo
separar cond_images_or_ids, no se hace bien
agregar código de entrenamiento para vae
añadir autoacondicionamiento opcional en empotrar
combinar con el artículo crítico simbólico, ya implementado en Phenaki
conecte el código de entrenamiento acelerado para maskgit
@inproceedings { Chang2023MuseTG ,
title = { Muse: Text-To-Image Generation via Masked Generative Transformers } ,
author = { Huiwen Chang and Han Zhang and Jarred Barber and AJ Maschinot and Jos{'e} Lezama and Lu Jiang and Ming-Hsuan Yang and Kevin P. Murphy and William T. Freeman and Michael Rubinstein and Yuanzhen Li and Dilip Krishnan } ,
year = { 2023 }
} @article { Chen2022AnalogBG ,
title = { Analog Bits: Generating Discrete Data using Diffusion Models with Self-Conditioning } ,
author = { Ting Chen and Ruixiang Zhang and Geo rey E. Hinton } ,
journal = { ArXiv } ,
year = { 2022 } ,
volume = { abs/2208.04202 }
} @misc { jabri2022scalable ,
title = { Scalable Adaptive Computation for Iterative Generation } ,
author = { Allan Jabri and David Fleet and Ting Chen } ,
year = { 2022 } ,
eprint = { 2212.11972 } ,
archivePrefix = { arXiv } ,
primaryClass = { cs.LG }
} @article { Lezama2022ImprovedMI ,
title = { Improved Masked Image Generation with Token-Critic } ,
author = { Jos{'e} Lezama and Huiwen Chang and Lu Jiang and Irfan Essa } ,
journal = { ArXiv } ,
year = { 2022 } ,
volume = { abs/2209.04439 }
} @inproceedings { Nijkamp2021SCRIPTSP ,
title = { SCRIPT: Self-Critic PreTraining of Transformers } ,
author = { Erik Nijkamp and Bo Pang and Ying Nian Wu and Caiming Xiong } ,
booktitle = { North American Chapter of the Association for Computational Linguistics } ,
year = { 2021 }
} @inproceedings { dao2022flashattention ,
title = { Flash{A}ttention: Fast and Memory-Efficient Exact Attention with {IO}-Awareness } ,
author = { Dao, Tri and Fu, Daniel Y. and Ermon, Stefano and Rudra, Atri and R{'e}, Christopher } ,
booktitle = { Advances in Neural Information Processing Systems } ,
year = { 2022 }
} @misc { mentzer2023finite ,
title = { Finite Scalar Quantization: VQ-VAE Made Simple } ,
author = { Fabian Mentzer and David Minnen and Eirikur Agustsson and Michael Tschannen } ,
year = { 2023 } ,
eprint = { 2309.15505 } ,
archivePrefix = { arXiv } ,
primaryClass = { cs.CV }
} @misc { yu2023language ,
title = { Language Model Beats Diffusion -- Tokenizer is Key to Visual Generation } ,
author = { Lijun Yu and José Lezama and Nitesh B. Gundavarapu and Luca Versari and Kihyuk Sohn and David Minnen and Yong Cheng and Agrim Gupta and Xiuye Gu and Alexander G. Hauptmann and Boqing Gong and Ming-Hsuan Yang and Irfan Essa and David A. Ross and Lu Jiang } ,
year = { 2023 } ,
eprint = { 2310.05737 } ,
archivePrefix = { arXiv } ,
primaryClass = { cs.CV }
}
