import torch
from diffusers import DiffusionPipeline
# switch to "mps" for apple devices
pipe = DiffusionPipeline.from_pretrained("FreedomIntelligence/Janus-4o-7B", dtype=torch.bfloat16, device_map="cuda")
pipe.load_lora_weights("peterface/ThinkClearly")
prompt = "Make the sea shell red"
image = pipe(prompt).images[0]ThinkClearly
https://github.com/peterN908/ThinkClearly/tree/main
Model description
A conceptual exploration of tool use in AR image models
We fine-tune Janus 4o models to produce `<CLEAR>` instruction tokens when making text+image to image edits. This indicate that this token patch should be an exact pixel match to the input image
Download model
Download them in the Files & versions tab.
- Downloads last month
- -
Model tree for peterface/ThinkClearly
Base model
deepseek-ai/Janus-Pro-7B Finetuned
FreedomIntelligence/Janus-4o-7B