easytorch 0.3

Easy Neural Network Experiments with pytorch

EasyTorch setup

1. Install pytorch and torchvision from Pytorch official website
2. pip install easytorch

Higlights

• A convenient framework to easily setup neural network experiments.
• Minimal configuration to setup a newu experimenton new dataset:
  • Only need to initialize neural network architecture, if needed.
  • Create a python dictionary pointing to data ,ground truth, and mask directory(dataspecs.py).
  • Automatic k-fold cross validation.
  • Automatic logging and model checkpointing.
  • Works an all sort of classification and regression task.
  • GPU enabled metrics like precision, recall, f1, overlap, and confusion matrix with maximum GPU utilization.
  • Ability to combine all dataset with correct dataspecs. Combining dataset and running experiments is hassle free.

Link to a full working example

Sample usecase as follows:

import argparse

import dataspecs as dspec
from easytorch.utils.defaultargs import ap
from easytorch.runs import run, pooled_run
from classification import MyTrainer, MyDataset

ap = argparse.ArgumentParser(parents=[ap], add_help=False)

dataspecs = [dspec.DRIVE, dspec.STARE]
if __name__ == "__main__":
    run(ap, dataspecs, MyTrainer, MyDataset)
    pooled_run(ap, dataspecs, MyTrainer, MyDataset)

Training+Validation+Test

* $python main.py -p train -nch 3 -e 3 -b 2 -sp True

Only Test

* $python main.py -p test -nch 3 -e 3 -b 2 -sp True

References

Please cite us if you use this framework(easytorch) as follows: @misc{easytorch, author = {Khanal, Aashis}, title = {Quick Neural Network Experimentation}, year = {2020}, publisher = {GitHub}, journal = {GitHub repository}, url = {https://github.com/sraashis/easytorch} }