ProbeX/Model-J__SD_1k

Model-J: SD_1k (Bundled LoRA Models)

This repository contains SD_1k LoRA models from the Model-J dataset, introduced in:

Learning on Model Weights using Tree Experts (CVPR 2025) by Eliahu Horwitz, Bar Cavia, Jonathan Kahana*, Yedid Hoshen

<p align="center"> 🌐 <a href="https://horwitz.ai/probex" target="blank">Project</a> | 📃 <a href="https://arxiv.org/abs/2410.13569" target="blank">Paper</a> | 💻 <a href="https://github.com/eliahuhorwitz/ProbeX" target="blank">GitHub</a> | 🤗 <a href="https://huggingface.co/ProbeX" target="blank">Dataset</a> </p>

Overview

This repository bundles all SD_1k LoRA models into a single repo. Each model is a LoRA adapter fine-tuned on images of a specific ImageNet class using Stable Diffusion. Each model directory also contains the training images used for fine-tuning.

Repository Structure

{split}/model_idx_{XXXX}/
    pytorch_lora_weights.safetensors
    training_images/
    README.md

Splits: train, val, test, val_holdout, test_holdout

For detailed hyperparameters of each model, see the README.md in each model's directory or the Model-J dataset.