Model-J: MAE Model (model_idx_0252)
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 | 0.0003 |
| LR Scheduler | cosine |
| Epochs | 4 |
| Max Train Steps | 1332 |
| Batch Size | 64 |
| Weight Decay | 0.007 |
| Seed | 252 |
| Random Crop | True |
| Random Flip | False |
Performance
| Metric | Value |
|---|
| Train Accuracy | 0.9810 |
| Val Accuracy | 0.8581 |
| Test Accuracy | 0.8534 |
Training Categories
The model was fine-tuned on the following 50 CIFAR100 classes:
cockroach, spider, rabbit, telephone, bee, shark, dinosaur, caterpillar, baby, raccoon, sweet_pepper, oak_tree, plain, porcupine, orange, skyscraper, house, fox, seal, lobster, skunk, palm_tree, bus, table, camel, man, kangaroo, woman, hamster, snake, keyboard, bowl, cattle, cup, flatfish, bottle, streetcar, bicycle, apple, squirrel, road, beaver, shrew, couch, beetle, butterfly, rocket, television, maple_tree, mushroom
