README.md

---
base_model: facebook/dino-vitb16
library_name: transformers
pipeline_tag: image-classification
tags:
  - probex
  - model-j
  - weight-space-learning
---

# Model-J: DINO Model (model_idx_0925)

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** | DINO |
| **Split** | val |
| **Base Model** | `facebook/dino-vitb16` |
| **Dataset** | CIFAR100 (50 classes) |

## Training Hyperparameters

| Parameter | Value |
|---|---|
| Learning Rate | 9e-05 |
| LR Scheduler | linear |
| Epochs | 9 |
| Max Train Steps | 2997 |
| Batch Size | 64 |
| Weight Decay | 0.01 |
| Seed | 925 |
| Random Crop | False |
| Random Flip | False |

## Performance

| Metric | Value |
|---|---|
| Train Accuracy | 0.9999 |
| Val Accuracy | 0.8973 |
| Test Accuracy | 0.9116 |

## Training Categories

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

`tulip`, `flatfish`, `fox`, `sweet_pepper`, `palm_tree`, `aquarium_fish`, `crocodile`, `willow_tree`, `mountain`, `keyboard`, `castle`, `raccoon`, `lizard`, `couch`, `woman`, `otter`, `streetcar`, `sea`, `boy`, `tiger`, `cattle`, `plate`, `skyscraper`, `rocket`, `snake`, `snail`, `whale`, `maple_tree`, `beaver`, `bear`, `orange`, `television`, `lobster`, `possum`, `shrew`, `kangaroo`, `squirrel`, `plain`, `mushroom`, `bowl`, `cockroach`, `clock`, `cup`, `baby`, `beetle`, `lion`, `lawn_mower`, `worm`, `ray`, `girl`