Model-J: MAE Model (model_idx_0966)
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 | linear |
| Epochs | 3 |
| Max Train Steps | 999 |
| Batch Size | 64 |
| Weight Decay | 0.009 |
| Seed | 966 |
| Random Crop | False |
| Random Flip | True |
Performance
| Metric | Value |
|---|
| Train Accuracy | 0.9177 |
| Val Accuracy | 0.8488 |
| Test Accuracy | 0.8530 |
Training Categories
The model was fine-tuned on the following 50 CIFAR100 classes:
orchid, bear, lobster, couch, lamp, hamster, mountain, bus, rabbit, skunk, table, leopard, ray, seal, television, man, sweet_pepper, cockroach, tulip, clock, butterfly, caterpillar, plain, apple, rocket, road, kangaroo, sunflower, streetcar, crocodile, snake, house, castle, wolf, fox, shark, porcupine, girl, orange, worm, palm_tree, beetle, poppy, chimpanzee, shrew, squirrel, lizard, bottle, willow_tree, keyboard
