You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

94 lines
3.4 KiB

2 years ago
# min(DALL·E)
2 years ago
2 years ago
[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/kuprel/min-dalle/blob/main/min_dalle.ipynb)
 
2 years ago
[![Replicate](https://replicate.com/kuprel/min-dalle/badge)](https://replicate.com/kuprel/min-dalle)
2 years ago
 
[![Join us on Discord](https://img.shields.io/discord/823813159592001537?color=5865F2&logo=discord&logoColor=white)](https://discord.gg/xBPBXfcFHd)
2 years ago
2 years ago
This is a fast, minimal implementation of Boris Dayma's [DALL·E Mini](https://github.com/borisdayma/dalle-mini). It has been stripped for inference and converted to PyTorch. The only third party dependencies are numpy, requests, pillow and torch.
2 years ago
To generate a 3x3 grid of DALL·E Mega images it takes
2 years ago
- **35 seconds** with a P100 in Colab
2 years ago
- **15 seconds** with an A100 on Replicate
2 years ago
- **TBD** with an H100 (@NVIDIA?)
2 years ago
2 years ago
The flax model and code for converting it to torch can be found [here](https://github.com/kuprel/min-dalle-flax).
2 years ago
## Install
2 years ago
2 years ago
```bash
2 years ago
$ pip install min-dalle
```
2 years ago
2 years ago
## Usage
2 years ago
### Python
2 years ago
2 years ago
Load the model parameters once and reuse the model to generate multiple images.
2 years ago
2 years ago
```python
from min_dalle import MinDalle
2 years ago
2 years ago
model = MinDalle(is_mega=True, models_root='./pretrained')
2 years ago
```
2 years ago
The required models will be downloaded to `models_root` if they are not already there. Once everything has finished initializing, call `generate_image` with some text and a seed as many times as you want.
2 years ago
```python
2 years ago
text = 'a comfy chair that looks like an avocado'
2 years ago
image = model.generate_image(text)
display(image)
2 years ago
```
2 years ago
<img src="https://github.com/kuprel/min-dalle/raw/main/examples/avocado_armchair.png" alt="drawing" width="200"/>
2 years ago
```python
2 years ago
text = 'court sketch of godzilla on trial'
image = model.generate_image(text, seed=6, grid_size=3)
display(image)
2 years ago
```
2 years ago
<img src="https://github.com/kuprel/min-dalle/raw/main/examples/godzilla_trial.png" alt="drawing" width="300"/>
2 years ago
```python
text = 'Rusty Iron Man suit found abandoned in the woods being reclaimed by nature'
image = model.generate_image(text, seed=0, grid_size=3)
display(image)
```
2 years ago
<img src="https://github.com/kuprel/min-dalle/raw/main/examples/ironman.png" alt="drawing" width="300"/>
2 years ago
2 years ago
```python
text = 'a funeral at Whole Foods'
image = model.generate_image(text, seed=10, grid_size=3)
display(image)
```
2 years ago
<img src="https://github.com/kuprel/min-dalle/raw/main/examples/funeral.png" alt="drawing" width="300"/>
2 years ago
```python
text = 'Jesus turning water into wine on Americas Got Talent'
image = model.generate_image(text, seed=2, grid_size=3)
display(image)
```
2 years ago
<img src="https://github.com/kuprel/min-dalle/raw/main/examples/jesus.png" alt="drawing" width="300"/>
2 years ago
```python
text = 'cctv footage of Yoda robbing a liquor store'
image = model.generate_image(text, seed=0, grid_size=3)
display(image)
```
2 years ago
<img src="https://github.com/kuprel/min-dalle/raw/main/examples/yoda.png" alt="drawing" width="300"/>
2 years ago
2 years ago
### Command Line
2 years ago
Use `image_from_text.py` to generate images from the command line.
2 years ago
```bash
$ python image_from_text.py --text='artificial intelligence' --seed=7
```
2 years ago
<img src="https://github.com/kuprel/min-dalle/raw/main/examples/artificial_intelligence.png" alt="drawing" width="200"/>
2 years ago
```bash
2 years ago
$ python image_from_text.py --text='trail cam footage of gollum eating watermelon' --mega --seed=1 --grid-size=3
2 years ago
```
2 years ago
<img src="https://github.com/kuprel/min-dalle/raw/main/examples/gollum_trailcam.png" alt="drawing" width="300"/>