Model-J DINO
Collection
1002 items
โข
Updated
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
๐ Project | ๐ Paper | ๐ป GitHub | ๐ค Dataset
| Attribute | Value |
|---|---|
| Subset | DINO |
| Split | train |
| Base Model | facebook/dino-vitb16 |
| Dataset | CIFAR100 (50 classes) |
| Parameter | Value |
|---|---|
| Learning Rate | 3e-05 |
| LR Scheduler | cosine_with_restarts |
| Epochs | 3 |
| Max Train Steps | 999 |
| Batch Size | 64 |
| Weight Decay | 0.005 |
| Seed | 330 |
| Random Crop | True |
| Random Flip | False |
| Metric | Value |
|---|---|
| Train Accuracy | 0.9967 |
| Val Accuracy | 0.9304 |
| Test Accuracy | 0.9286 |
The model was fine-tuned on the following 50 CIFAR100 classes:
spider, house, chair, camel, crab, apple, ray, beaver, palm_tree, sweet_pepper, cup, dinosaur, butterfly, bed, sea, mouse, plate, kangaroo, lobster, baby, motorcycle, skunk, rose, mountain, orange, clock, flatfish, dolphin, seal, otter, wolf, television, tank, snake, crocodile, leopard, shrew, bottle, willow_tree, skyscraper, road, girl, tractor, cattle, castle, whale, beetle, plain, bicycle, raccoon
Base model
facebook/dino-vitb16