Model-J: MAE Model (model_idx_0254)
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 | 8 |
| Max Train Steps | 2664 |
| Batch Size | 64 |
| Weight Decay | 0.007 |
| Seed | 254 |
| Random Crop | False |
| Random Flip | True |
Performance
| Metric | Value |
|---|
| Train Accuracy | 0.9552 |
| Val Accuracy | 0.8296 |
| Test Accuracy | 0.8306 |
Training Categories
The model was fine-tuned on the following 50 CIFAR100 classes:
crocodile, chair, orchid, sunflower, snail, shrew, butterfly, can, crab, seal, cattle, baby, orange, mountain, whale, palm_tree, worm, cockroach, tulip, pickup_truck, couch, dinosaur, train, girl, sea, lobster, house, kangaroo, pine_tree, bridge, lawn_mower, maple_tree, tiger, skyscraper, cup, television, bee, lion, keyboard, apple, plate, boy, man, chimpanzee, trout, tank, oak_tree, spider, lamp, possum
