Alexander Kozlov

Published in Towards Data Science

·Feb 24, 2021

PyTorch Lightning and Optuna: Multi-GPU hyperparameter optimisation

How to quickly set up multi-GPU training for hyperparameter optimisation with PyTorch Lightning — Probably most people reading this post have at least once trained an ML model that took a considerable amount of time. This was certainly the case for me when I was fine-tuning RoBERTa for high accuracy text classification. Utilising GPU for training can shave off an order of magnitude from…

Machine Learning

5 min read

PyTorch Lightning and Optuna: Multi-GPU hyperparameter optimisation

Published in Towards Data Science

·Oct 19, 2020

Electricity demand in Victoria during coronavirus pandemic

Has electricity demand pattern in Australia’s second-largest state changed during COVID-19 pandemic? — Foreign media widely regard Australia as an example of a developed country that successfully contained COVID-19 outbreak. In July 2020, after a significant rise in new daily cases, Victoria, the second-largest state in Australia by population, introduced some of the strictest movement control measures in the world. Travel restrictions, widely…

Time Series Forecasting

5 min read

Electricity demand in Victoria during coronavirus pandemic

Published in Towards Data Science

·Sep 4, 2020

High accuracy text classification with Python

How to fine-tune Huggingface models to get a 99% accuracy text classifier. — As of the time of writing this piece, state-of-the-art results on NLP and NLU tasks are obtained with Transformer models. There is a trend of performance improvement as models become deeper and larger, GPT 3 comes to mind. Training small versions of such models from scratch takes a significant amount…

NLP

6 min read

Fine-tuning BERT and RoBERTa for high accuracy text classification in PyTorch