Model-J: MAE Model (model_idx_0902)
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 | MAE |
| Split | test |
| Base Model | facebook/vit-mae-base |
| Dataset | CIFAR100 (50 classes) |
Training Hyperparameters
| Parameter | Value |
|---|
| Learning Rate | 0.0005 |
| LR Scheduler | cosine |
| Epochs | 5 |
| Max Train Steps | 1665 |
| Batch Size | 64 |
| Weight Decay | 0.05 |
| Seed | 902 |
| Random Crop | True |
| Random Flip | False |
Performance
| Metric | Value |
|---|
| Train Accuracy | 0.7884 |
| Val Accuracy | 0.5624 |
| Test Accuracy | 0.5828 |
Training Categories
The model was fine-tuned on the following 50 CIFAR100 classes:
castle, seal, orange, shark, streetcar, bicycle, girl, spider, road, wolf, otter, cattle, snail, raccoon, palm_tree, snake, bowl, tank, plate, tulip, lawn_mower, pine_tree, baby, lizard, caterpillar, can, keyboard, squirrel, chimpanzee, rocket, sunflower, orchid, beetle, cockroach, tractor, apple, kangaroo, forest, flatfish, crab, boy, table, pickup_truck, worm, maple_tree, mushroom, leopard, shrew, cloud, train
