Model-J: MAE Model (model_idx_0182)

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

🌐 Project | 📃 Paper | 💻 GitHub | 🤗 Dataset

Model Details

Attribute	Value
Subset	MAE
Split	train
Base Model	`facebook/vit-mae-base`
Dataset	CIFAR100 (50 classes)

Training Hyperparameters

Parameter	Value
Learning Rate	9e-05
LR Scheduler	linear
Epochs	5
Max Train Steps	1665
Batch Size	64
Weight Decay	0.009
Seed	182
Random Crop	False
Random Flip	False

Performance

Metric	Value
Train Accuracy	0.9961
Val Accuracy	0.8915
Test Accuracy	0.8968

Training Categories

The model was fine-tuned on the following 50 CIFAR100 classes:

tractor, woman, couch, road, mushroom, cattle, cloud, crocodile, tank, clock, rocket, cup, lawn_mower, lobster, house, bee, boy, snail, caterpillar, palm_tree, chimpanzee, worm, flatfish, otter, bridge, wardrobe, sea, chair, can, forest, butterfly, beetle, possum, bus, maple_tree, pine_tree, fox, keyboard, porcupine, telephone, leopard, baby, television, willow_tree, sweet_pepper, oak_tree, castle, lamp, raccoon, skunk