Go to file
2022-06-27 17:28:40 -04:00
examples simplified 2022-06-27 15:46:04 -04:00
min_dalle fixed an issue with argument parser 2022-06-27 16:49:42 -04:00
.gitignore examples 2022-06-27 13:39:00 -04:00
image_from_text.py fixed an issue with argument parser 2022-06-27 16:49:42 -04:00
LICENSE license and cleanup 2022-06-27 14:34:10 -04:00
min(DALL·E).ipynb Created using Colaboratory 2022-06-27 17:23:30 -04:00
README.md readme formatting 2022-06-27 17:28:40 -04:00
requirements.txt first commit 2022-06-27 11:57:56 -04:00
setup.sh first commit 2022-06-27 11:57:56 -04:00

min(DALL·E)

Open In Colab

This is a minimal implementation of DALL·E Mini in both Flax and PyTorch

Setup

Run sh setup.sh to install dependencies and download pretrained models. The only required dependencies are flax and torch. In the bash script, GitHub LFS is used to download the VQGan detokenizer and the Weight & Biases python package is used to download the DALL·E Mini and DALL·E Mega transformer models. These models can also be downloaded manually:

VQGan: https://huggingface.co/dalle-mini/vqgan_imagenet_f16_16384

DALL·E Mini: https://wandb.ai/dalle-mini/dalle-mini/artifacts/DalleBart_model/mini-1/v0/files

DALL·E Mega: https://wandb.ai/dalle-mini/dalle-mini/artifacts/DalleBart_model/mega-1-fp16/v14/files

Run

Here are some examples

python image_from_text.py --seed=7 --text='alien life'

Alien

python image_from_text.py --mega --seed=4 --text='a comfy chair that looks like an avocado'

Avocado Armchair

python image_from_text.py --mega --seed=100 --text='court sketch of godzilla on trial'

Godzilla Trial