guided-diffusion

This is the codebase for Diffusion Models Beat GANS on Image Synthesis.

This repository is based on openai/improved-diffusion, with modifications for classifier conditioning and architecture improvements.

Usage

Training diffusion models is described in the parent repository. Training a classifier is similar. We assume you have put training hyperparameters into a TRAIN_FLAGS variable, and classifier hyperparameters into a CLASSIFIER_FLAGS variable. Then you can run:

mpiexec -n N python scripts/classifier_train.py --data_dir path/to/imagenet $TRAIN_FLAGS $CLASSIFIER_FLAGS

Make sure to divide the batch size in TRAIN_FLAGS by the number of MPI processes you are using.

Here are flags for training the 128x128 classifier. You can modify these for training classifiers at other resolutions:

TRAIN_FLAGS="--iterations 300000 --anneal_lr True --batch_size 256 --lr 3e-4 --save_interval 10000 --weight_decay 0.05"
CLASSIFIER_FLAGS="--image_size 128 --classifier_attention_resolutions 32,16,8 --classifier_depth 2 --classifier_width 128 --classifier_pool attention --classifier_resblock_updown True --classifier_use_scale_shift_norm True"

For sampling from a 128x128 classifier-guided model, 25 step DDIM:

MODEL_FLAGS="--attention_resolutions 32,16,8 --class_cond True --image_size 128 --learn_sigma True --num_channels 256 --num_heads 4 --num_res_blocks 2 --resblock_updown True --use_fp16 True --use_scale_shift_norm True"
CLASSIFIER_FLAGS="--image_size 128 --classifier_attention_resolutions 32,16,8 --classifier_depth 2 --classifier_width 128 --classifier_pool attention --classifier_resblock_updown True --classifier_use_scale_shift_norm True --classifier_scale 1.0 --classifier_use_fp16 True"
SAMPLE_FLAGS="--batch_size 4 --num_samples 50000 --timestep_respacing ddim25 --use_ddim True"
mpiexec -n N python scripts/classifier_sample.py \
    --model_path /path/to/model.pt \
    --classifier_path path/to/classifier.pt \
    $MODEL_FLAGS $CLASSIFIER_FLAGS $SAMPLE_FLAGS

To sample for 250 timesteps without DDIM, replace --timestep_respacing ddim25 to --timestep_respacing 250, and replace --use_ddim True with --use_ddim False.

This is the codebase for Diffusion Models Beat GANS on Image Synthesis.

Related tags

Overview

guided-diffusion

Usage

Owner

OpenAI

Using pretrained GROVER to extract the atomic fingerprints from molecule

Inflated i3d network with inception backbone, weights transfered from tensorflow

LeafSnap replicated using deep neural networks to test accuracy compared to traditional computer vision methods.

The original implementation of TNDM used in the NeurIPS 2021 paper (no longer being updated)

CONditionals for Ordinal Regression and classification in PyTorch

Credit fraud detection in Python using a Jupyter Notebook

Doing the asl sign language classification on static images using graph neural networks.

Corruption Invariant Learning for Re-identification

PyTorch wrappers for using your model in audacity!

Solution to the first stage Quiz of Hamoye internship: Introduction to Python for Machine Learning

ML models and internal tensors 3D visualizer

FedMM: Saddle Point Optimization for Federated Adversarial Domain Adaptation

Home for cuQuantum Python & NVIDIA cuQuantum SDK C++ samples

PyTorch implementation of HDN(Homography Decomposition Networks) for planar object tracking

We have implemented shaDow-GNN as a general and powerful pipeline for graph representation learning. For more details, please find our paper titled Deep Graph Neural Networks with Shallow Subgraph Samplers, available on arXiv (https//arxiv.org/abs/2012.01380).

Rule Extraction Methods for Interactive eXplainability

REGTR: End-to-end Point Cloud Correspondences with Transformers

Multiple Object Extraction from Aerial Imagery with Convolutional Neural Networks

Unit-Convertor - Unit Convertor Built With Python

Neuralnetwork - Basic Multilayer Perceptron Neural Network for deep learning