Model-J DINO
Collection
1002 items
โข
Updated
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
๐ Project | ๐ Paper | ๐ป GitHub | ๐ค Dataset
| Attribute | Value |
|---|---|
| Subset | DINO |
| Split | train |
| Base Model | facebook/dino-vitb16 |
| Dataset | CIFAR100 (50 classes) |
| Parameter | Value |
|---|---|
| Learning Rate | 0.0003 |
| LR Scheduler | constant_with_warmup |
| Epochs | 8 |
| Max Train Steps | 2664 |
| Batch Size | 64 |
| Weight Decay | 0.009 |
| Seed | 810 |
| Random Crop | True |
| Random Flip | True |
| Metric | Value |
|---|---|
| Train Accuracy | 0.4940 |
| Val Accuracy | 0.4053 |
| Test Accuracy | 0.4160 |
The model was fine-tuned on the following 50 CIFAR100 classes:
dinosaur, bowl, caterpillar, bus, plain, cup, girl, wardrobe, bottle, crab, aquarium_fish, flatfish, crocodile, rose, chair, plate, palm_tree, castle, bridge, poppy, trout, table, bee, oak_tree, bed, boy, road, tractor, snail, raccoon, wolf, lion, tulip, spider, skunk, lawn_mower, keyboard, bear, tiger, ray, cattle, lizard, fox, clock, bicycle, mouse, sweet_pepper, couch, baby, possum
Base model
facebook/dino-vitb16