Model-J: MAE Model (model_idx_0840)
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 |
| Epochs | 6 |
| Max Train Steps | 1998 |
| Batch Size | 64 |
| Weight Decay | 0.007 |
| Seed | 840 |
| Random Crop | True |
| Random Flip | False |
Performance
| Metric | Value |
|---|
| Train Accuracy | 0.9147 |
| Val Accuracy | 0.7245 |
| Test Accuracy | 0.7174 |
Training Categories
The model was fine-tuned on the following 50 CIFAR100 classes:
bottle, leopard, shrew, bed, kangaroo, sweet_pepper, orchid, flatfish, otter, camel, bowl, bear, streetcar, plate, cup, dinosaur, orange, porcupine, rabbit, butterfly, hamster, lizard, pine_tree, boy, rose, trout, bee, turtle, girl, woman, television, aquarium_fish, beetle, possum, lobster, shark, palm_tree, can, willow_tree, rocket, raccoon, dolphin, mouse, clock, poppy, mountain, elephant, bridge, skunk, telephone
