Model-J: DINO Model (model_idx_0921)
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 | test |
| Base Model | facebook/dino-vitb16 |
| Dataset | CIFAR100 (50 classes) |
Training Hyperparameters
| Parameter | Value |
|---|
| Learning Rate | 7e-05 |
| LR Scheduler | constant_with_warmup |
| Epochs | 8 |
| Max Train Steps | 2664 |
| Batch Size | 64 |
| Weight Decay | 0.03 |
| Seed | 921 |
| Random Crop | False |
| Random Flip | True |
Performance
| Metric | Value |
|---|
| Train Accuracy | 0.9677 |
| Val Accuracy | 0.8669 |
| Test Accuracy | 0.8540 |
Training Categories
The model was fine-tuned on the following 50 CIFAR100 classes:
train, tractor, telephone, kangaroo, squirrel, lamp, possum, orchid, sweet_pepper, porcupine, shrew, seal, rocket, palm_tree, mushroom, pear, worm, woman, castle, girl, bed, crocodile, mountain, bus, cattle, pine_tree, keyboard, mouse, pickup_truck, plate, lobster, snail, man, couch, sunflower, skunk, streetcar, wardrobe, motorcycle, clock, house, cup, skyscraper, wolf, lion, raccoon, forest, oak_tree, fox, cockroach
