Instructions to use ostris/ideogram_4_unconditional_lora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use ostris/ideogram_4_unconditional_lora with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("ideogram-ai/ideogram-4-fp8", dtype=torch.bfloat16, device_map="cuda") pipe.load_lora_weights("ostris/ideogram_4_unconditional_lora") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Inference
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
File size: 1,054 Bytes
f56068d | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 | ---
license: other
license_name: ideogram-4-non-commercial
license_link: https://huggingface.co/ideogram-ai/ideogram-4-fp8/blob/main/LICENSE.md
tags:
- text-to-image
- lora
- diffusers
base_model: ideogram-ai/ideogram-4-fp8
---
# Ideogram 4 Unconditional LoRA
This is a LoRA that was initialized by extracting the difference of the Ideogram 4 conditional and unconditional model weights.
It was further tuned using student teacher training on real data and a loss was performed on a per layer basis to more closely
match the unconditional model. This can be used on the conditional Ideogram 4 model during the unconditional pass
as a replacement to the full 9B paramiter unconditional model.
Using the full unconditional model will likely yield better results, but this will work as a light weight alternative. It was
originally trained to be used in [Ostris AI Toolkit](https://github.com/ostris/ai-toolkit) so samples would be more in line
with what the full pipeline would produce without needing to load the entire unconditional model. |