Model-J: DINO Model (model_idx_0218)
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 | 0.0003 |
| LR Scheduler | constant |
| Epochs | 4 |
| Max Train Steps | 1332 |
| Batch Size | 64 |
| Weight Decay | 0.03 |
| Seed | 218 |
| Random Crop | True |
| Random Flip | True |
Performance
| Metric | Value |
|---|
| Train Accuracy | 0.3575 |
| Val Accuracy | 0.3115 |
| Test Accuracy | 0.3276 |
Training Categories
The model was fine-tuned on the following 50 CIFAR100 classes:
man, cattle, shark, cloud, otter, elephant, bed, ray, clock, rabbit, tulip, bear, bus, motorcycle, mushroom, fox, bottle, bee, orchid, crocodile, train, bridge, crab, pickup_truck, caterpillar, cup, worm, house, dolphin, aquarium_fish, apple, table, lobster, tiger, bicycle, snake, kangaroo, squirrel, sunflower, castle, plate, whale, mountain, tractor, forest, chair, rocket, flatfish, shrew, butterfly
