Instructions to use terminusresearch/pixart-900m-1024-untrained with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use terminusresearch/pixart-900m-1024-untrained with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("terminusresearch/pixart-900m-1024-untrained", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
PixArt-900M Template Model
This is a template model, and it is untrained.
Making use of these weights will result in next to zero outputs, as it requires a large amount of compute (8x A100 or better) and training samples (3 million or more, ideally).
To use an under-construction version of PixArt-900M, see the following repositories:
- ptx0/pixart-900m-1024-ft-large
- Actively training on a mix of real photos, Midjourney, DALLE-3, and Nijijourney data. It is considered "large" because of the data scale of its sets.
- ptx0/pixart-900m-1024-ft
- Unlikely to (currently) function well, training on a mix of SFW anime data, Midjourney, Nijijourney, real photos, and more. It has been caching VAE/text embeds. It's recommended currently to make use of the
pixart-900m-ft-largemodel instead.
- Unlikely to (currently) function well, training on a mix of SFW anime data, Midjourney, Nijijourney, real photos, and more. It has been caching VAE/text embeds. It's recommended currently to make use of the
- Downloads last month
- -