Remplementation of paper https://arxiv.org/abs/1705.08665. This repo utilizes code from Bayesian Optimization and Bayesian Tutorial
| Network | Dataset | Epochs | Accuracy(before) | Accuracy(after) | Compression Rate |
|---|---|---|---|---|---|
| 3-Layer MLP | MNIST | 50 | 98.33% | 98.33% | 1.3 |
| 3-Layer MLP | CIFAR 10 | 50 | 56.26% | 54.47% | 3.5 |
| LeNet | MNIST | 50 | 99.24% | 99.26% | 1.4 |
python example.py --dataset mnist --nettype mlp --epochs 50- Clean the train-prune code
- Modularize Bayesian Layer and Module
- Fix bug in Convolution