Model-J: DINO Model (model_idx_0986)
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 | 6 |
| Max Train Steps | 1998 |
| Batch Size | 64 |
| Weight Decay | 0.05 |
| Seed | 986 |
| Random Crop | False |
| Random Flip | True |
Performance
| Metric | Value |
|---|
| Train Accuracy | 0.4426 |
| Val Accuracy | 0.3837 |
| Test Accuracy | 0.3878 |
Training Categories
The model was fine-tuned on the following 50 CIFAR100 classes:
table, house, wolf, cattle, mountain, beaver, can, fox, crocodile, castle, possum, couch, bed, tractor, camel, road, poppy, caterpillar, sweet_pepper, palm_tree, sunflower, flatfish, elephant, kangaroo, pine_tree, bridge, worm, chimpanzee, wardrobe, bee, otter, skyscraper, lizard, rabbit, shrew, motorcycle, ray, lawn_mower, bus, apple, bicycle, cockroach, tiger, cloud, spider, crab, porcupine, forest, lamp, aquarium_fish
