min-dalle-test/README.md

35 lines
1.3 KiB
Markdown
Raw Normal View History

2022-06-27 19:46:04 +00:00
# min(DALL·E)
2022-06-27 17:38:35 +00:00
2022-06-27 21:26:05 +00:00
[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/kuprel/min-dalle/blob/main/min(DALL·E).ipynb)
2022-06-27 17:51:48 +00:00
This is a minimal implementation of [DALL·E Mini](https://github.com/borisdayma/dalle-mini) in both Flax and PyTorch
2022-06-27 18:34:10 +00:00
### Setup
2022-06-27 17:51:48 +00:00
2022-06-27 21:26:05 +00:00
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
2022-06-27 17:51:48 +00:00
2022-06-27 18:34:10 +00:00
### Run
2022-06-27 17:51:48 +00:00
Here are some examples
2022-06-27 17:38:35 +00:00
```
2022-06-27 20:49:42 +00:00
python image_from_text.py --seed=7 --text='alien life'
2022-06-27 19:46:04 +00:00
```
![Alien](examples/alien.png)
2022-06-27 20:49:42 +00:00
2022-06-27 19:46:04 +00:00
```
2022-06-27 20:49:42 +00:00
python image_from_text.py --mega --seed=4 --text='a comfy chair that looks like an avocado'
2022-06-27 17:38:35 +00:00
```
![Avocado Armchair](examples/avocado_armchair.png)
```
2022-06-27 20:49:42 +00:00
python image_from_text.py --mega --seed=100 --text='court sketch of godzilla on trial'
2022-06-27 17:38:35 +00:00
```
![Godzilla Trial](examples/godzilla_trial.png)