Given one of several implementations in PyTorch / TensorFlow.
- collecting all relevant information : code, datasets, pre-trained models
- running experiments on Jupyter notebooks to test the working implementation
- collect, plot, comment metrics, improve data collection/annotation, use relevant data augmentation / regularisation, tune hyperparameters using hyperparameter-tuning/overview/
- apply best practice improvements of implementations for speed and/or portability.
- structuring the code around our productionisation template
- understand the optimisation process such as layers fusion and quantisation, and evaluate before/after metrics
(removed by Toogit admin)
Your code needs to be so good that you are confident to publish it publicly in the relevant repos, to contribute to the community effort.
About the recuiterMember since May 20, 2018 Prashant Sethia
from Jamalo-Nenets, Russia