Model-J: MAE Model (model_idx_0098)
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 | 3e-05 |
| LR Scheduler | cosine_with_restarts |
| Epochs | 4 |
| Max Train Steps | 1332 |
| Batch Size | 64 |
| Weight Decay | 0.01 |
| Seed | 98 |
| Random Crop | True |
| Random Flip | False |
Performance
| Metric | Value |
|---|
| Train Accuracy | 0.9117 |
| Val Accuracy | 0.8443 |
| Test Accuracy | 0.8310 |
Training Categories
The model was fine-tuned on the following 50 CIFAR100 classes:
baby, mushroom, crab, cattle, possum, tiger, wardrobe, rabbit, mouse, boy, porcupine, skyscraper, snake, couch, seal, wolf, television, shrew, rose, elephant, butterfly, orange, pine_tree, chair, squirrel, man, lion, trout, lizard, bed, camel, train, castle, lobster, ray, palm_tree, orchid, lawn_mower, oak_tree, bottle, girl, plate, keyboard, bicycle, worm, dolphin, hamster, shark, house, clock
