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
Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: other
|
| 3 |
+
license_name: ideogram-4-non-commercial
|
| 4 |
+
license_link: https://huggingface.co/ideogram-ai/ideogram-4-fp8/blob/main/LICENSE.md
|
| 5 |
+
tags:
|
| 6 |
+
- text-to-image
|
| 7 |
+
- lora
|
| 8 |
+
- diffusers
|
| 9 |
+
base_model: ideogram-ai/ideogram-4-fp8
|
| 10 |
+
|
| 11 |
+
---
|
| 12 |
+
# Ideogram 4 Unconditional LoRA
|
| 13 |
+
|
| 14 |
+
This is a LoRA that was initialized by extracting the difference of the Ideogram 4 conditional and unconditional model weights.
|
| 15 |
+
It was further tuned using student teacher training on real data and a loss was performed on a per layer basis to more closely
|
| 16 |
+
match the unconditional model. This can be used on the conditional Ideogram 4 model during the unconditional pass
|
| 17 |
+
as a replacement to the full 9B paramiter unconditional model.
|
| 18 |
+
|
| 19 |
+
Using the full unconditional model will likely yield better results, but this will work as a light weight alternative. It was
|
| 20 |
+
originally trained to be used in [Ostris AI Toolkit](https://github.com/ostris/ai-toolkit) so samples would be more in line
|
| 21 |
+
with what the full pipeline would produce without needing to load the entire unconditional model.
|