README.md

---
base_model: microsoft/resnet-101
library_name: transformers
pipeline_tag: image-classification
tags:
  - probex
  - model-j
  - weight-space-learning
---

# Model-J: ResNet Model (model_idx_0311)

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** | ResNet |
| **Split** | train |
| **Base Model** | `microsoft/resnet-101` |
| **Dataset** | CIFAR100 (50 classes) |

## Training Hyperparameters

| Parameter | Value |
|---|---|
| Learning Rate | 0.0005 |
| LR Scheduler | constant_with_warmup |
| Epochs | 6 |
| Max Train Steps | 1998 |
| Batch Size | 64 |
| Weight Decay | 0.03 |
| Seed | 311 |
| Random Crop | False |
| Random Flip | False |

## Performance

| Metric | Value |
|---|---|
| Train Accuracy | 0.9892 |
| Val Accuracy | 0.8675 |
| Test Accuracy | 0.8722 |

## Training Categories

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

`kangaroo`, `camel`, `dolphin`, `tractor`, `cockroach`, `ray`, `orange`, `mountain`, `man`, `wolf`, `telephone`, `beetle`, `bridge`, `bee`, `turtle`, `elephant`, `porcupine`, `table`, `couch`, `flatfish`, `rose`, `rabbit`, `plate`, `maple_tree`, `tiger`, `girl`, `squirrel`, `lamp`, `raccoon`, `train`, `hamster`, `lizard`, `lobster`, `streetcar`, `television`, `whale`, `beaver`, `seal`, `skyscraper`, `dinosaur`, `butterfly`, `caterpillar`, `baby`, `otter`, `cup`, `crab`, `chimpanzee`, `road`, `crocodile`, `bear`