Model-J: SupViT Model (model_idx_0519)
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 | SupViT |
| Split | train |
| Base Model | google/vit-base-patch16-224 |
| Dataset | CIFAR100 (50 classes) |
Training Hyperparameters
| Parameter | Value |
|---|
| Learning Rate | 7e-05 |
| LR Scheduler | cosine |
| Epochs | 3 |
| Max Train Steps | 999 |
| Batch Size | 64 |
| Weight Decay | 0.005 |
| Seed | 519 |
| Random Crop | True |
| Random Flip | False |
Performance
| Metric | Value |
|---|
| Train Accuracy | 0.9954 |
| Val Accuracy | 0.9557 |
| Test Accuracy | 0.9546 |
Training Categories
The model was fine-tuned on the following 50 CIFAR100 classes:
rocket, motorcycle, poppy, cattle, sweet_pepper, bed, pear, shark, castle, mountain, sea, cup, sunflower, lion, tractor, wolf, otter, aquarium_fish, orchid, road, can, table, cockroach, train, bridge, maple_tree, caterpillar, palm_tree, willow_tree, man, clock, porcupine, mushroom, flatfish, seal, lawn_mower, house, tiger, crab, wardrobe, lizard, baby, skyscraper, bear, snail, kangaroo, plain, bicycle, fox, lamp
