ML Reproducibility Challenge 2020

Description

This repository contains all of the code used to replicate the results of the ICLR 2020 conference paper

Dataset	Network	Optimizer	Learning Rate
CIFAR 10	ResNet-56	Nesterov SGD $\beta = 0.9$ Batch Size: 128 Weight Decay: 0.0002 Epochs: 182	$\alpha = 0.1 \quad \forall t \in [0,91)$ $\alpha = 0.01 \quad \forall t \in [91, 136)$ $\alpha = 0.001 \quad \forall t \in [136, 182]$
ImageNet	ResNet-34	Nesterov SGD $\beta = 0.9$ Batch Size: 1024 Weight Decay: 0.0001 Epochs: 90
ImageNet	Resnet-50	Nesterov SGD $\beta = 0.9$ Batch Size: 1024 Weight Decay: 0.0001 Epochs: 90
WMT16 EN-DE	GNMT	Adam $\beta_1 = 0.9$ $\beta_2 = 0.999$ Batch Size: 2048 Epochs: 5

Name		Name	Last commit message	Last commit date
Latest commit History 57 Commits
result_plots		result_plots
trained_models		trained_models
.gitignore		.gitignore
README.md		README.md
Training Environment.ipynb		Training Environment.ipynb
accuracy_logs.json		accuracy_logs.json
main.py		main.py
plot_utils.py		plot_utils.py
requirements.txt		requirements.txt
resnet.py		resnet.py
snip.py		snip.py
test.py		test.py