PyTorch implementation for Score-Based Generative Modeling through Stochastic Differential Equations (ICLR 2021, Oral)

Overview

Score-Based Generative Modeling through Stochastic Differential Equations

PWC

This repo contains a PyTorch implementation for the paper Score-Based Generative Modeling through Stochastic Differential Equations

by Yang Song, Jascha Sohl-Dickstein, Diederik P. Kingma, Abhishek Kumar, Stefano Ermon, and Ben Poole


We propose a unified framework that generalizes and improves previous work on score-based generative models through the lens of stochastic differential equations (SDEs). In particular, we can transform data to a simple noise distribution with a continuous-time stochastic process described by an SDE. This SDE can be reversed for sample generation if we know the score of the marginal distributions at each intermediate time step, which can be estimated with score matching. The basic idea is captured in the figure below:

schematic

Our work enables a better understanding of existing approaches, new sampling algorithms, exact likelihood computation, uniquely identifiable encoding, latent code manipulation, and brings new conditional generation abilities (including but not limited to class-conditional generation, inpainting and colorization) to the family of score-based generative models.

All combined, we achieved an FID of 2.20 and an Inception score of 9.89 for unconditional generation on CIFAR-10, as well as high-fidelity generation of 1024px Celeba-HQ images (samples below). In addition, we obtained a likelihood value of 2.99 bits/dim on uniformly dequantized CIFAR-10 images.

FFHQ samples

What does this code do?

Aside from the NCSN++ and DDPM++ models in our paper, this codebase also re-implements many previous score-based models in one place, including NCSN from Generative Modeling by Estimating Gradients of the Data Distribution, NCSNv2 from Improved Techniques for Training Score-Based Generative Models, and DDPM from Denoising Diffusion Probabilistic Models.

It supports training new models, evaluating the sample quality and likelihoods of existing models. We carefully designed the code to be modular and easily extensible to new SDEs, predictors, or correctors.

JAX version

Please find a JAX implementation here, which additionally supports class-conditional generation with a pre-trained classifier, and resuming an evalution process after pre-emption.

JAX vs. PyTorch

In general, this PyTorch version consumes less memory but runs slower than JAX. Here is a benchmark on training an NCSN++ cont. model with VE SDE. Hardware is 4x Nvidia Tesla V100 GPUs (32GB)

Framework Time (second per step) Memory usage in total (GB)
PyTorch 0.56 20.6
JAX (n_jitted_steps=1) 0.30 29.7
JAX (n_jitted_steps=5) 0.20 74.8

How to run the code

Dependencies

Run the following to install a subset of necessary python packages for our code

pip install -r requirements.txt

Stats files for quantitative evaluation

We provide the stats file for CIFAR-10. You can download cifar10_stats.npz and save it to assets/stats/. Check out #5 on how to compute this stats file for new datasets.

Usage

Train and evaluate our models through main.py.

main.py:
  --config: Training configuration.
    (default: 'None')
  --eval_folder: The folder name for storing evaluation results
    (default: 'eval')
  --mode: <train|eval>: Running mode: train or eval
  --workdir: Working directory
  • config is the path to the config file. Our prescribed config files are provided in configs/. They are formatted according to ml_collections and should be quite self-explanatory.

    Naming conventions of config files: the path of a config file is a combination of the following dimensions:

    • dataset: One of cifar10, celeba, celebahq, celebahq_256, ffhq_256, celebahq, ffhq.
    • model: One of ncsn, ncsnv2, ncsnpp, ddpm, ddpmpp.
    • continuous: train the model with continuously sampled time steps.
  • workdir is the path that stores all artifacts of one experiment, like checkpoints, samples, and evaluation results.

  • eval_folder is the name of a subfolder in workdir that stores all artifacts of the evaluation process, like meta checkpoints for pre-emption prevention, image samples, and numpy dumps of quantitative results.

  • mode is either "train" or "eval". When set to "train", it starts the training of a new model, or resumes the training of an old model if its meta-checkpoints (for resuming running after pre-emption in a cloud environment) exist in workdir/checkpoints-meta . When set to "eval", it can do an arbitrary combination of the following

    • Evaluate the loss function on the test / validation dataset.

    • Generate a fixed number of samples and compute its Inception score, FID, or KID. Prior to evaluation, stats files must have already been downloaded/computed and stored in assets/stats.

    • Compute the log-likelihood on the training or test dataset.

    These functionalities can be configured through config files, or more conveniently, through the command-line support of the ml_collections package. For example, to generate samples and evaluate sample quality, supply the --config.eval.enable_sampling flag; to compute log-likelihoods, supply the --config.eval.enable_bpd flag, and specify --config.eval.dataset=train/test to indicate whether to compute the likelihoods on the training or test dataset.

How to extend the code

  • New SDEs: inherent the sde_lib.SDE abstract class and implement all abstract methods. The discretize() method is optional and the default is Euler-Maruyama discretization. Existing sampling methods and likelihood computation will automatically work for this new SDE.
  • New predictors: inherent the sampling.Predictor abstract class, implement the update_fn abstract method, and register its name with @register_predictor. The new predictor can be directly used in sampling.get_pc_sampler for Predictor-Corrector sampling, and all other controllable generation methods in controllable_generation.py.
  • New correctors: inherent the sampling.Corrector abstract class, implement the update_fn abstract method, and register its name with @register_corrector. The new corrector can be directly used in sampling.get_pc_sampler, and all other controllable generation methods in controllable_generation.py.

Pretrained checkpoints

All checkpoints are provided in this Google drive.

Instructions: You may find two checkpoints for some models. The first checkpoint (with a smaller number) is the one that we reported FID scores in our paper's Table 3 (also corresponding to the FID and IS columns in the table below). The second checkpoint (with a larger number) is the one that we reported likelihood values and FIDs of black-box ODE samplers in our paper's Table 2 (also FID(ODE) and NNL (bits/dim) columns in the table below). The former corresponds to the smallest FID during the course of training (every 50k iterations). The later is the last checkpoint during training.

Per Google's policy, we cannot release our original CelebA and CelebA-HQ checkpoints. That said, I have re-trained models on FFHQ 1024px, FFHQ 256px and CelebA-HQ 256px with personal resources, and they achieved similar performance to our internal checkpoints.

Here is a detailed list of checkpoints and their results reported in the paper. FID (ODE) corresponds to the sample quality of black-box ODE solver applied to the probability flow ODE.

Checkpoint path FID IS FID (ODE) NNL (bits/dim)
ve/cifar10_ncsnpp/ 2.45 9.73 - -
ve/cifar10_ncsnpp_continuous/ 2.38 9.83 - -
ve/cifar10_ncsnpp_deep_continuous/ 2.20 9.89 - -
vp/cifar10_ddpm/ 3.24 - 3.37 3.28
vp/cifar10_ddpm_continuous - - 3.69 3.21
vp/cifar10_ddpmpp 2.78 9.64 - -
vp/cifar10_ddpmpp_continuous 2.55 9.58 3.93 3.16
vp/cifar10_ddpmpp_deep_continuous 2.41 9.68 3.08 3.13
subvp/cifar10_ddpm_continuous - - 3.56 3.05
subvp/cifar10_ddpmpp_continuous 2.61 9.56 3.16 3.02
subvp/cifar10_ddpmpp_deep_continuous 2.41 9.57 2.92 2.99
Checkpoint path Samples
ve/bedroom_ncsnpp_continuous bedroom_samples
ve/church_ncsnpp_continuous church_samples
ve/ffhq_1024_ncsnpp_continuous ffhq_1024
ve/ffhq_256_ncsnpp_continuous ffhq_256_samples
ve/celebahq_256_ncsnpp_continuous celebahq_256_samples

Demonstrations and tutorials

Link Description
Open In Colab Load our pretrained checkpoints and play with sampling, likelihood computation, and controllable synthesis (JAX + FLAX)
Open In Colab Load our pretrained checkpoints and play with sampling, likelihood computation, and controllable synthesis (PyTorch)
Open In Colab Tutorial of score-based generative models in JAX + FLAX
Open In Colab Tutorial of score-based generative models in PyTorch

Tips

  • When using the JAX codebase, you can jit multiple training steps together to improve training speed at the cost of more memory usage. This can be set via config.training.n_jitted_steps. For CIFAR-10, we recommend using config.training.n_jitted_steps=5 when your GPU/TPU has sufficient memory; otherwise we recommend using config.training.n_jitted_steps=1. Our current implementation requires config.training.log_freq to be dividable by n_jitted_steps for logging and checkpointing to work normally.
  • The snr (signal-to-noise ratio) parameter of LangevinCorrector somewhat behaves like a temperature parameter. Larger snr typically results in smoother samples, while smaller snr gives more diverse but lower quality samples. Typical values of snr is 0.05 - 0.2, and it requires tuning to strike the sweet spot.
  • For VE SDEs, we recommend choosing config.model.sigma_max to be the maximum pairwise distance between data samples in the training dataset.

References

If you find the code useful for your research, please consider citing

@inproceedings{
  song2021scorebased,
  title={Score-Based Generative Modeling through Stochastic Differential Equations},
  author={Yang Song and Jascha Sohl-Dickstein and Diederik P Kingma and Abhishek Kumar and Stefano Ermon and Ben Poole},
  booktitle={International Conference on Learning Representations},
  year={2021},
  url={https://openreview.net/forum?id=PxTIG12RRHS}
}

This work is built upon some previous papers which might also interest you:

  • Song, Yang, and Stefano Ermon. "Generative Modeling by Estimating Gradients of the Data Distribution." Proceedings of the 33rd Annual Conference on Neural Information Processing Systems. 2019.
  • Song, Yang, and Stefano Ermon. "Improved techniques for training score-based generative models." Proceedings of the 34th Annual Conference on Neural Information Processing Systems. 2020.
  • Ho, Jonathan, Ajay Jain, and Pieter Abbeel. "Denoising diffusion probabilistic models." Proceedings of the 34th Annual Conference on Neural Information Processing Systems. 2020.
Owner
Yang Song
PhD Candidate in Stanford AI Lab
Yang Song
PSTR: End-to-End One-Step Person Search With Transformers (CVPR2022)

PSTR (CVPR2022) This code is an official implementation of "PSTR: End-to-End One-Step Person Search With Transformers (CVPR2022)". End-to-end one-step

Jiale Cao 28 Dec 13, 2022
Instant-nerf-pytorch - NeRF trained SUPER FAST in pytorch

instant-nerf-pytorch This is WORK IN PROGRESS, please feel free to contribute vi

94 Nov 22, 2022
Thermal Control of Laser Powder Bed Fusion using Deep Reinforcement Learning

This repository is the implementation of the paper "Thermal Control of Laser Powder Bed Fusion Using Deep Reinforcement Learning", linked here. The project makes use of the Deep Reinforcement Library

BaratiLab 11 Dec 27, 2022
On the model-based stochastic value gradient for continuous reinforcement learning

On the model-based stochastic value gradient for continuous reinforcement learning This repository is by Brandon Amos, Samuel Stanton, Denis Yarats, a

Facebook Research 46 Dec 15, 2022
A embed able annotation tool for end to end cross document co-reference

CoRefi CoRefi is an emebedable web component and stand alone suite for exaughstive Within Document and Cross Document Coreference Anntoation. For a de

PythicCoder 39 Dec 12, 2022
[CVPR 2021] VirTex: Learning Visual Representations from Textual Annotations

VirTex: Learning Visual Representations from Textual Annotations Karan Desai and Justin Johnson University of Michigan CVPR 2021 arxiv.org/abs/2006.06

Karan Desai 533 Dec 24, 2022
The first machine learning framework that encourages learning ML concepts instead of memorizing class functions.

SeaLion is designed to teach today's aspiring ml-engineers the popular machine learning concepts of today in a way that gives both intuition and ways of application. We do this through concise algori

Anish 324 Dec 27, 2022
Learning Correspondence from the Cycle-consistency of Time (CVPR 2019)

TimeCycle Code for Learning Correspondence from the Cycle-consistency of Time (CVPR 2019, Oral). The code is developed based on the PyTorch framework,

Xiaolong Wang 706 Nov 29, 2022
A sample pytorch Implementation of ACL 2021 research paper "Learning Span-Level Interactions for Aspect Sentiment Triplet Extraction".

Span-ASTE-Pytorch This repository is a pytorch version that implements Ali's ACL 2021 research paper Learning Span-Level Interactions for Aspect Senti

来自丹麦的天籁 10 Dec 06, 2022
[BMVC2021] The official implementation of "DomainMix: Learning Generalizable Person Re-Identification Without Human Annotations"

DomainMix [BMVC2021] The official implementation of "DomainMix: Learning Generalizable Person Re-Identification Without Human Annotations" [paper] [de

Wenhao Wang 17 Dec 20, 2022
code for Image Manipulation Detection by Multi-View Multi-Scale Supervision

MVSS-Net Code and models for ICCV 2021 paper: Image Manipulation Detection by Multi-View Multi-Scale Supervision Update 22.02.17, Pretrained model for

dong_chengbo 131 Dec 30, 2022
Solutions of Reinforcement Learning 2nd Edition

Solutions of Reinforcement Learning, An Introduction

YIFAN WANG 1.4k Dec 30, 2022
Code for layerwise detection of linguistic anomaly paper (ACL 2021)

Layerwise Anomaly This repository contains the source code and data for our ACL 2021 paper: "How is BERT surprised? Layerwise detection of linguistic

6 Dec 07, 2022
End-to-end Temporal Action Detection with Transformer. [Under review]

TadTR: End-to-end Temporal Action Detection with Transformer By Xiaolong Liu, Qimeng Wang, Yao Hu, Xu Tang, Song Bai, Xiang Bai. This repo holds the c

Xiaolong Liu 105 Dec 25, 2022
The toolkit to generate auto labeled datasets

Ozeu Ozeu is the toolkit to autolabal dataset for instance segmentation. You can generate datasets labaled with segmentation mask and bounding box fro

Xiong Jie 28 Mar 28, 2022
Mix3D: Out-of-Context Data Augmentation for 3D Scenes (3DV 2021)

Mix3D: Out-of-Context Data Augmentation for 3D Scenes (3DV 2021) Alexey Nekrasov*, Jonas Schult*, Or Litany, Bastian Leibe, Francis Engelmann Mix3D is

Alexey Nekrasov 189 Dec 26, 2022
This is the official pytorch implementation for our ICCV 2021 paper "TRAR: Routing the Attention Spans in Transformers for Visual Question Answering" on VQA Task

🌈 ERASOR (RA-L'21 with ICRA Option) Official page of "ERASOR: Egocentric Ratio of Pseudo Occupancy-based Dynamic Object Removal for Static 3D Point C

Hyungtae Lim 225 Dec 29, 2022
OOD Generalization and Detection (ACL 2020)

Pretrained Transformers Improve Out-of-Distribution Robustness How does pretraining affect out-of-distribution robustness? We create an OOD benchmark

littleRound 57 Jan 09, 2023
We provided a matlab implementation for an evolutionary multitasking AUC optimization framework (EMTAUC).

EMTAUC We provided a matlab implementation for an evolutionary multitasking AUC optimization framework (EMTAUC). In this code, SBGA is considered a ba

7 Nov 24, 2022
VISNOTATE: An Opensource tool for Gaze-based Annotation of WSI Data

VISNOTATE: An Opensource tool for Gaze-based Annotation of WSI Data Introduction Requirements Installation and Setup Supported Hardware and Software R

SigmaLab 1 Jun 14, 2022