README.md

---
base_model: google/vit-base-patch16-224
library_name: transformers
pipeline_tag: image-classification
tags:
  - probex
  - model-j
  - weight-space-learning
---

# Model-J: SupViT Model (model_idx_0349)

This model is part of 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>

![ProbeX](https://raw.githubusercontent.com/eliahuhorwitz/ProbeX/main/imgs/poster.png)

## Model Details

| Attribute | Value |
|---|---|
| **Subset** | SupViT |
| **Split** | train |
| **Base Model** | `google/vit-base-patch16-224` |
| **Dataset** | CIFAR100 (50 classes) |

## Training Hyperparameters

| Parameter | Value |
|---|---|
| Learning Rate | 7e-05 |
| LR Scheduler | cosine_with_restarts |
| Epochs | 9 |
| Max Train Steps | 2997 |
| Batch Size | 64 |
| Weight Decay | 0.05 |
| Seed | 349 |
| Random Crop | False |
| Random Flip | False |

## Performance

| Metric | Value |
|---|---|
| Train Accuracy | 1.0000 |
| Val Accuracy | 0.9645 |
| Test Accuracy | 0.9626 |

## Training Categories

The model was fine-tuned on the following 50 CIFAR100 classes:

`keyboard`, `poppy`, `chair`, `kangaroo`, `bottle`, `train`, `mushroom`, `otter`, `lion`, `snail`, `tiger`, `ray`, `girl`, `seal`, `snake`, `palm_tree`, `bed`, `turtle`, `beetle`, `rabbit`, `raccoon`, `flatfish`, `elephant`, `table`, `skyscraper`, `lobster`, `rocket`, `bus`, `lawn_mower`, `fox`, `mountain`, `pickup_truck`, `apple`, `plate`, `cup`, `crab`, `sunflower`, `tulip`, `sweet_pepper`, `butterfly`, `shark`, `dolphin`, `baby`, `cattle`, `telephone`, `oak_tree`, `mouse`, `bridge`, `squirrel`, `orange`