Code for paper "Adversarial score matching and improved sampling for image generation"

Overview

Adversarial score matching and improved sampling for image generation

This repo contains the official implementation for the ICLR 2021 paper Adversarial score matching and improved sampling for image generation. It it a highly extended version of the original repo on score matching.

Discussion and more samples at https://ajolicoeur.wordpress.com/adversarial-score-matching-and-consistent-sampling.


Denoising score matching with Annealed Langevin Sampling (DSM-ALS) is a recent approach to generative modeling. Despite the convincing visual quality of samples, this method appears to perform worse than Generative Adversarial Networks (GANs) under the Frechet Inception Distance, a popular metric forgenerative models. We show that this apparent gap vanishes when denoising thefinal Langevin samples using the score network. In addition, we propose two improvements to DSM-ALS: 1) Consistent Annealed Sampling as a more stable alternative to Annealed Langevin Sampling, and 2) a hybrid training formulation, composed of both denoising score matching and adversarial objectives. By combining both of these techniques and exploring different network architectures, we elevate score matching methods and obtain results competitive with state-of-the-art image generation on CIFAR-10

Adversarial LSUN-Churches


Citation

If you find this code useful please cite us in your work:

@inproceedings{
  jolicoeur-martineau2021adversarial,
  title={Adversarial score matching and improved sampling for image generation},
  author={Alexia Jolicoeur-Martineau and R{\'e}mi Pich{\'e}-Taillefer and Ioannis Mitliagkas and Remi Tachet des Combes},
  booktitle={International Conference on Learning Representations},
  year={2021},
  url={https://openreview.net/forum?id=eLfqMl3z3lq}
}

Setup

Needed

Hyperparameter choice

  • Use "main.py --compute_approximate_sigma_max" to choose model.sigma_begin based on the current dataset (based on Technique 1 from https://arxiv.org/abs/2006.09011)
  • Use calculate_number_of_steps.R to choose model.num_classes (based on Technique 2 from https://arxiv.org/abs/2006.09011)
  • tune sampling.step_lr manually for consistent or non-consistent with n_sigma=1 (see Appendix B for how to extrapolate to n_sigma > 1 from the step_lr at n_sigma = 1)
  • Everything else can be left to default

To Replicate paper

To train a non-adversarial score network

python main.py --config cifar10_9999ema.yml --doc cifar10_bs128_L2_9999ema --ni

Log files will be saved in <exp>/logs/cifar10_bs128_L2_9999ema.

To train an adversarial score network

python main.py --config cifar10_9999ema.yml --doc cifar10_bs128_L2_9999ema_adam0_9_adamD-5_9_LSGAN_ --ni  --adam --adam_beta 0 .9 --D_adam --D_adam_beta -.5 .9 --adversarial

Log files will be saved in <exp>/logs/cifar10_bs128_L2_9999ema_adam0_9_adamD-5_9_LSGAN_.

To sample from a pre-trained score network (ex: cifar10_bs128_L2_9999ema, consistent, nsigma=1)

python main.py --sample --config cifar10_9999ema.yml -i cifar10_bs128_L2_9999ema --ni --consistent --nsigma 1 --step_lr 5.6e-6 --batch_size 100 --begin_ckpt 250000

Samples will be saved in <exp>/image_samples/cifar10_bs128_L2_9999ema.

To compute the FID for a range of checkpoints from a pre-trained score network (ex: cifar10_bs128_L2_9999ema, at 100k to 300k iterations)

python main.py --fast_fid --config cifar10_9999ema.yml -i cifar10_bs128_L2_9999ema --ni --consistent --nsigma 1 --step_lr 5.6e-6 --batch_size 4000 --fid_num_samples 10000 --begin_ckpt 100000 --end_ckpt 300000

FIDs will be saved in {args.fid_folder}/log_FID.txt.

Pretrained Score Network Checkpoints

Link: https://www.dropbox.com/s/dltiobdlsb2vhyo/DSM_ScoreNetwork_Pretrained.zip?dl=0

Download and unzip it to the exp folder.

FID statistics (for FID evaluation)

Link: https://www.dropbox.com/s/nhvp2tf1unxj08g/fid_stats.zip?dl=0

Download and unzip it to the exp/datasets folder.

Owner
Alexia Jolicoeur-Martineau
I'm a researcher in statistics and machine learning. I am particularly interested in GANs, denoising score-matching, and statistical divergences/metrics.
Alexia Jolicoeur-Martineau
A Discord Server Cloner With Lot Of New Features.

Technologies Screenshots Table of contents About Installation Links Deployed Features Website Score Contribution Need Help? Instagram Discord About A

NotSakshyam 25 Dec 31, 2022
Free TradingView webhook alert for basic plan users.

TradingView-Free-Webhook-Alerts Project start on 01-02-2022 Providing the free webhook service to the basic plan users in TradingView. Portal ↠ Instal

Freeman 31 Dec 25, 2022
Desafio de projeto sobre Git/Github

Maçã ou Laranja? 🤔 Desafio Projeto Dio para Git/Github 🔶 Para esse primeiro repositório, decidir adicionar o primeiro algoritmo de inteligência arti

José Filipe 2 Oct 23, 2022
Lumi-Bot - Discord bot that fetches cryptocurrency prices utilizing CoinGeko API

Lumi-Bot Discord bot that fetches and monitors cryptocurrency prices utilizing C

Diego Castro 2 Oct 08, 2022
Some 3Commas helper bots, AltRank, GalaxyScore, Watchlist, Auto-Compound

3Commas Cyber Bot Helpers A collection of 3Commas bot helpers I wrote. (collection will grow over time) Disclaimer THE SOFTWARE IS PROVIDED "AS IS", W

Ron Klinkien 176 Jan 02, 2023
摩尔庄园手游脚本

摩尔庄园 BlueStacks 脚本 手游上线,情怀再起,但面对游戏中枯燥无味的每日任务和资源采集,你是否觉得肝疼呢? 本项目通过生成 BlueStacks 模拟器的宏脚本,帮助玩家护肝。 使用脚本请阅读 使用方式 和对应的 功能及说明 联系 Telegram 频道 @mole61 Telegram

WH-2099 43 Dec 16, 2022
A Discord Self-Bot in Python

👨‍💻 Discord Self Bot 👨‍💻 A Discord Self-Bot in Python by natrix Installation Run: selfbot.bat Python: version : 3.8 Modules

natrix_dev 3 Oct 02, 2022
Source code of BobuxAdmin bot from Bobux Bot Development server.

BobuxAdmin Source code of BobuxAdmin bot from Bobux Bot Development server. The bot is written with usage of disnake and SQLite database. Functionalit

Bobux Bot Developers 3 Dec 29, 2022
Weee - Advanced project's versions bumper

Weee - Advanced project's versions bumper

Yan Kurbatov 2 Jun 06, 2022
Bringing Ethereum Virtual Machine to StarkNet at warp speed!

Warp Warp brings EVM compatible languages to StarkNet, making it possible to transpile Ethereum smart contracts to Cairo, and use them on StarkNet. Ta

Nethermind 700 Dec 26, 2022
Bot para automatizacao de registros no Vacivida para o COVID19

VACIBOT v.06 - Bot para automatizacao de registros no Vacivida para o COVID19 by Victor Fragoso - Prefeitura Municipal de Santo André Email:

Prefeitura de Santo André 22 Sep 19, 2022
A Python implementation of a discord bot that acts as a server scanner similar to Copenheimer.

Bad Copenheimer A Python impelentation of a discord bot that acts as a server scanner. This is a discord bot that will scan ip adresses to see if they

pilot1782 69 Dec 16, 2022
This is a tutorial on how to make a Discord Bot using the discord.py library

HowToMakeADiscordBot This Github repository is here to help you code a Discord Bot using the discord.py library! 1 - Setup: Download the code inside t

Baz 1 Oct 31, 2021
AWS Blog post code for running feature-extraction on images using AWS Batch and Cloud Development Kit (CDK).

Batch processing with AWS Batch and CDK Welcome This repository demostrates provisioning the necessary infrastructure for running a job on AWS Batch u

AWS Samples 7 Oct 18, 2022
Anchor Protocol Script that can save you from being liquidated!

Why My day job requires a fairly good amount of automation from time to time. Besides, I do like computers to work on what I cannot while I'm sleeping

126 Oct 16, 2022
SI_EXPLAINER_tg_bot: This bot is an assistant for medical professionals in interpreting the results of patient clustering.

SI_EXPLAINER_tg_bot This bot is an assistant for medical professionals in interpreting the results of patient clustering. ABOUT This chatbot was devel

Alexander Kanonirov 1 Jan 21, 2022
The first open-source PyTgCalls-based project.

SU Music Player — The first open-source PyTgCalls based Pyrogram bot to play music in voice chats Requirements FFmpeg NodeJS 15+ Python 3.7+ Deploymen

Calls Music 74 Nov 19, 2022
Tools for Twitter

Tools for Twitter Data This is a start of a collection of tools to use for collecting data via the Twitter API. If you do not have a Twitter Developer

DiscoverText 36 Oct 13, 2022
A Python AWS Lambda Webhook listener that generates a permanent URL when an asset is created in Contentstack.

Webhook Listener A Python Lambda Webhook Listener - Generates a permanent URL on created assets. See doc on Generating a Permanent URL: https://www.co

Contentstack Solutions 1 Nov 04, 2021
C Y B Ξ R UserBot is a project that simplifies the use of Telegram. All rights reserved.

C Y B Ξ R USΞRBOT 🇦🇿 C Y B Ξ R UserBot is a project that simplifies the use of Telegram. All rights reserved. Automatic Setup Android: open Termux p

C Y B Ξ R 0 Sep 20, 2022