micromind checkpoints for MNIST
This repository contains checkpoints for the MNIST dataset for the following networks:
| Model | Top 1 Accuracy | Top 5 Accuracy |
|---|---|---|
PhiNet(alpha=0.5, beta=1, t_zero=6, num_layers=4, resolution=28) |
98.96% | 100% |
PhiNet(alpha=0.75, beta=1, t_zero=6, num_layers=5, resolution=28) |
99.03% | 99.98% |
PhiNet(alpha=0.35, beta=1, t_zero=6, num_layers=7, resolution=28) |
98.72% | 99.99% |
PhiNet(alpha=0.25, beta=1, t_zero=6, num_layers=7, resolution=28) |
98.84% | 99.99% |
PhiNet(alpha=0.25, beta=1, t_zero=5, num_layers=7, resolution=28) |
98.76% | 99.97% |
To download and use this repo:
from micromind import PhiNet
model = PhiNet.from_pretrained("MNIST", alpha=0.5, beta=1.0, t_zero=6, num_layers=4, num_classes=10, resolution=28)