Model-J: MAE Model (model_idx_0557)
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 | 0.0003 |
| LR Scheduler | constant_with_warmup |
| Epochs | 5 |
| Max Train Steps | 1665 |
| Batch Size | 64 |
| Weight Decay | 0.005 |
| Seed | 557 |
| Random Crop | False |
| Random Flip | False |
Performance
| Metric | Value |
|---|
| Train Accuracy | 0.9414 |
| Val Accuracy | 0.7803 |
| Test Accuracy | 0.7840 |
Training Categories
The model was fine-tuned on the following 50 CIFAR100 classes:
clock, plate, tank, bear, squirrel, pear, house, poppy, woman, chair, lion, cloud, chimpanzee, skyscraper, camel, shark, raccoon, lizard, otter, dinosaur, bowl, beaver, rabbit, bee, possum, castle, lobster, beetle, man, whale, mouse, pine_tree, rose, forest, snake, willow_tree, orange, sweet_pepper, keyboard, bus, girl, palm_tree, road, train, bridge, cup, wardrobe, lamp, skunk, caterpillar
