File size: 2,189 Bytes
79001f7 81d3c6b 79001f7 81d3c6b 79001f7 81d3c6b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 |
---
license: creativeml-openrail-m
base_model: runwayml/stable-diffusion-v1-5
tags:
- stable-diffusion
- stable-diffusion-diffusers
- text-to-image
- diffusers
- jax-diffusers-event
inference: true
datasets:
- ChristophSchuhmann/improved_aesthetics_6plus
---
# Stable Diffusion Nano
Stable Diffusion Nano was built during the [JAX/Diffusers community sprint 🧨](https://github.com/huggingface/community-events/tree/main/jax-controlnet-sprint#jaxdiffusers-community-sprint-).
We spent most of the time in the sprint working on [Stable Diffusion Nano 2.1](https://huggingface.co/bguisard/stable-diffusion-nano-2-1), which we'd recommend over this version.
prompt: A watercolor painting of an otter
![images_0)](./images_0.png)
prompt: Marvel MCU deadpool, red mask, red shirt, red gloves, black shoulders, black elbow pads, black legs, gold buckle, black belt, black mask, white eyes, black boots, fuji low light color 35mm film, downtown Osaka alley at night out of focus in background, neon lights
![images_1)](./images_1.png)
## Training details
All parameters were initialized from the runwayml/stable-diffusion-v1-5 model. The unet was fine tuned as follows:
- 100,000 steps training the full unet, learning rate = 1e-5, batch size = 512 (128 per TPU).
- Trained on [LAION Improved Aesthetics 6plus](https://huggingface.co/datasets/ChristophSchuhmann/improved_aesthetics_6plus).
## License
This model is open access and available to all, with a CreativeML OpenRAIL-M license further specifying rights and usage. The CreativeML OpenRAIL License specifies:
- You can't use the model to deliberately produce nor share illegal or harmful outputs or content.
- The authors claims no rights on the outputs you generate, you are free to use them and are accountable for their use which must not go against the provisions set in the license.
- You may re-distribute the weights and use the model commercially and/or as a service. If you do, please be aware you have to include the same use restrictions as the ones in the license and share a copy of the CreativeML OpenRAIL-M to all your users (please read the license entirely and carefully) Please read the full license here. |