This is a basic repository for running experiments related to the Lottery Ticket Hypothesis. More recently, this repository has been used for demonstrating a set of basic code patterns for machine learning experiments.
The main components include:
- A
model.py
file, which includes allModel
objects. - A
trainer.py
file, which includes allTrainer
objects - aTrainer
typically implements a Pytorch Dataloader (if using a native dataset) and the training, test, and validation loops. - A
train.py
file, which acts as the run script and manages most loading and saving actions anhps.yaml
file, through which experiment hyperparameters are specified.
Users will want to install the PyYAML (pip install PyYAML
) and tqdm (pip install tqdm
) packages.