Model-J: DINO Model (model_idx_0645)
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
ProbeXModel Details
| Attribute | Value |
|---|
| Subset | DINO |
| Split | train |
| Base Model | facebook/dino-vitb16 |
| Dataset | CIFAR100 (50 classes) |
Training Hyperparameters
| Parameter | Value |
|---|
| Learning Rate | 0.0001 |
| LR Scheduler | cosine_with_restarts |
| Epochs | 8 |
| Max Train Steps | 2664 |
| Batch Size | 64 |
| Weight Decay | 0.009 |
| Seed | 645 |
| Random Crop | True |
| Random Flip | False |
Performance
| Metric | Value |
|---|
| Train Accuracy | 0.9995 |
| Val Accuracy | 0.9109 |
| Test Accuracy | 0.9026 |
Training Categories
The model was fine-tuned on the following 50 CIFAR100 classes:
can, palm_tree, squirrel, rocket, house, forest, cup, pear, whale, tractor, raccoon, wardrobe, train, hamster, kangaroo, rose, spider, caterpillar, man, turtle, sunflower, worm, possum, chair, lamp, shrew, keyboard, maple_tree, tiger, bicycle, poppy, boy, sweet_pepper, bed, dolphin, wolf, snake, mountain, butterfly, plate, lion, ray, motorcycle, lobster, cockroach, shark, girl, oak_tree, mushroom, mouse
