Instructions to use naclbit/trinart_characters_19.2m_stable_diffusion_v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use naclbit/trinart_characters_19.2m_stable_diffusion_v1 with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("naclbit/trinart_characters_19.2m_stable_diffusion_v1", 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
- Draw Things
- DiffusionBee
Float32 Checkpoint and EMA
#2
by hakurei - opened
Hello! Really nice work on the model, I love it! Do you have any plans on releasing the float32 and/or ema+non-ema weights?
EMA might come in handy for those prefers low conditioning strength/CFG, but SD's stock ddim script breaks on ema+img2img combination so was not thinking about that.
Do you think ema+non-ema checkpoint (just optimizers removed) would do the job? Personally I don't really see much benefit from float32 inferencing other than for those trying to have it running on CPU (if any).
Yep, the model with just the optimizer states removed should do the job. I'd love to poke around at the model! Thanks in advance