mapo-t2i/mapo

Official codebase for Margin-aware Preference Optimization for Aligning Diffusion Models without Reference (MaPO).

37
/ 100
Emerging

This project helps machine learning engineers fine-tune existing text-to-image diffusion models to generate images that align better with human preferences. You input a pre-trained diffusion model and a dataset of image pairs with human preference labels. The output is a refined diffusion model capable of producing higher-quality, more preferred images.

No commits in the last 6 months.

Use this if you are a machine learning engineer working with large diffusion models and want to improve their image generation quality based on explicit human feedback, without needing a reference image for comparison.

Not ideal if you don't have access to substantial GPU resources (like an H100 with 40GB VRAM) or are looking for a pre-packaged, low-code solution for image generation.

Diffusion Models Image Generation Model Alignment Fine-tuning AI Art
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 9 / 25
Maturity 16 / 25
Community 12 / 25

How are scores calculated?

Stars

82

Forks

9

Language

Python

License

Apache-2.0

Last pushed

Jun 11, 2024

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/mapo-t2i/mapo"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.