Model-J: MAE Model (model_idx_0040)
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 | train |
| Base Model | facebook/vit-mae-base |
| Dataset | CIFAR100 (50 classes) |
Training Hyperparameters
| Parameter | Value |
|---|
| Learning Rate | 3e-05 |
| LR Scheduler | cosine |
| Epochs | 7 |
| Max Train Steps | 2331 |
| Batch Size | 64 |
| Weight Decay | 0.005 |
| Seed | 40 |
| Random Crop | True |
| Random Flip | False |
Performance
| Metric | Value |
|---|
| Train Accuracy | 0.9837 |
| Val Accuracy | 0.8835 |
| Test Accuracy | 0.8904 |
Training Categories
The model was fine-tuned on the following 50 CIFAR100 classes:
shark, cattle, dinosaur, castle, oak_tree, chimpanzee, caterpillar, tulip, road, clock, pine_tree, seal, telephone, wardrobe, leopard, butterfly, snake, lion, bridge, turtle, elephant, cockroach, mountain, rocket, table, bowl, tiger, rabbit, man, lawn_mower, fox, bicycle, keyboard, maple_tree, crab, pear, raccoon, tank, trout, tractor, shrew, whale, lamp, mouse, mushroom, orange, can, spider, bottle, apple
