Implementation of Denoising Diffusion Probabilistic Models in PyTorch
First please install tensorfn
It is simple convenience library for machine learning experiments. Sorry for the inconvenience.
First prepare lmdb dataset:
python prepare_data.py --size [SIZES, e.g. 128,256] --out [LMDB NAME] [DATASET PATH]
Then run training looop!
python train.py --n_gpu [NUMBER OF GPUS FOR TRAINING] --conf config/diffusion.conf
Also, you can tried to use improved configurations from https://openreview.net/forum?id=-NEXDKk8gZ (Cosine beta schedule, more multi-headed self attention, affine modulation for time conditions, new training hyperparameters.)
python train.py --n_gpu [NUMBER OF GPUS FOR TRAINING] --conf config/improved.conf
Samples from FFHQ
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4
