gigagan pytorch Unduh - gigagan pytorch Unduh kode sumber

Saya juga akan menambahkan beberapa temuan dari ringan gan, untuk konvergensi yang lebih cepat (lewati eksitasi lapisan) dan stabilitas yang lebih baik (rekonstruksi kerugian tambahan dalam diskriminator)

Ini juga akan berisi kode untuk upsampler 1k - 4k, yang menurut saya menjadi sorotan makalah ini.

$ pip install gigagan-pytorch

 import torch

from gigagan_pytorch import (
    GigaGAN ,
    ImageDataset
)

gan = GigaGAN (
    generator = dict (
        dim_capacity = 8 ,
        style_network = dict (
            dim = 64 ,
            depth = 4
        ),
        image_size = 256 ,
        dim_max = 512 ,
        num_skip_layers_excite = 4 ,
        unconditional = True
    ),
    discriminator = dict (
        dim_capacity = 16 ,
        dim_max = 512 ,
        image_size = 256 ,
        num_skip_layers_excite = 4 ,
        unconditional = True
    ),
    amp = True
). cuda ()

# dataset

dataset = ImageDataset (
    folder = '/path/to/your/data' ,
    image_size = 256
)

dataloader = dataset . get_dataloader ( batch_size = 1 )

# you must then set the dataloader for the GAN before training

gan . set_dataloader ( dataloader )

# training the discriminator and generator alternating
# for 100 steps in this example, batch size 1, gradient accumulated 8 times

gan (
    steps = 100 ,
    grad_accum_every = 8
)

# after much training

images = gan . generate ( batch_size = 4 ) # (4, 3, 256, 256)

 import torch
from gigagan_pytorch import (
    GigaGAN ,
    ImageDataset
)

gan = GigaGAN (
    train_upsampler = True ,     # set this to True
    generator = dict (
        style_network = dict (
            dim = 64 ,
            depth = 4
        ),
        dim = 32 ,
        image_size = 256 ,
        input_image_size = 64 ,
        unconditional = True
    ),
    discriminator = dict (
        dim_capacity = 16 ,
        dim_max = 512 ,
        image_size = 256 ,
        num_skip_layers_excite = 4 ,
        multiscale_input_resolutions = ( 128 ,),
        unconditional = True
    ),
    amp = True
). cuda ()

dataset = ImageDataset (
    folder = '/path/to/your/data' ,
    image_size = 256
)

dataloader = dataset . get_dataloader ( batch_size = 1 )

gan . set_dataloader ( dataloader )

# training the discriminator and generator alternating
# for 100 steps in this example, batch size 1, gradient accumulated 8 times

gan (
    steps = 100 ,
    grad_accum_every = 8
)

# after much training

lowres = torch . randn ( 1 , 3 , 64 , 64 ). cuda ()

images = gan . generate ( lowres ) # (1, 3, 256, 256)

Lari yang sehat akan memiliki G , MSG , D , MSD dengan nilai berkisar antara 0 hingga 10 , dan biasanya tetap konstan. Jika suatu saat setelah 1.000 langkah pelatihan, nilai-nilai ini tetap pada tiga digit, itu berarti ada sesuatu yang salah. Tidak masalah jika nilai generator dan diskriminator kadang-kadang turun menjadi negatif, tetapi nilai tersebut akan berayun kembali ke kisaran di atas.

GP dan SSL harus didorong menuju 0 . GP terkadang bisa melonjak; Saya suka membayangkannya sebagai jaringan yang mengalami pencerahan

Kelas GigaGAN kini dilengkapi dengan ? Akselerator. Anda dapat dengan mudah melakukan pelatihan multi-gpu dalam dua langkah menggunakan CLI accelerate

$ accelerate launch train . py

 @misc { https://doi.org/10.48550/arxiv.2303.05511 ,
    url     = { https://arxiv.org/abs/2303.05511 } ,
    author  = { Kang, Minguk and Zhu, Jun-Yan and Zhang, Richard and Park, Jaesik and Shechtman, Eli and Paris, Sylvain and Park, Taesung } ,  
    title   = { Scaling up GANs for Text-to-Image Synthesis } ,
    publisher = { arXiv } ,
    year    = { 2023 } ,
    copyright = { arXiv.org perpetual, non-exclusive license }
}

 @article { Liu2021TowardsFA ,
    title   = { Towards Faster and Stabilized GAN Training for High-fidelity Few-shot Image Synthesis } ,
    author  = { Bingchen Liu and Yizhe Zhu and Kunpeng Song and A. Elgammal } ,
    journal = { ArXiv } ,
    year    = { 2021 } ,
    volume  = { abs/2101.04775 }
}

 @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 }
}

 @inproceedings { Karras2020ada ,
    title     = { Training Generative Adversarial Networks with Limited Data } ,
    author    = { Tero Karras and Miika Aittala and Janne Hellsten and Samuli Laine and Jaakko Lehtinen and Timo Aila } ,
    booktitle = { Proc. NeurIPS } ,
    year      = { 2020 }
}

 @article { Xu2024VideoGigaGANTD ,
    title   = { VideoGigaGAN: Towards Detail-rich Video Super-Resolution } ,
    author  = { Yiran Xu and Taesung Park and Richard Zhang and Yang Zhou and Eli Shechtman and Feng Liu and Jia-Bin Huang and Difan Liu } ,
    journal = { ArXiv } ,
    year    = { 2024 } ,
    volume  = { abs/2404.12388 } ,
    url     = { https://api.semanticscholar.org/CorpusID:269214195 }
}