PinnedPublished inTowards Data ScienceCan Transformers Learn to Strategize?TicTacGPT for playing simple board gamesSep 8, 2023Sep 8, 2023
Some quick thoughts on learning compressed representations and GPT-5 speculationLanguage models are impressive, but I think they are cheating. We give them an abstract vocabulary as a way to compress the external world…Mar 11Mar 11
Published inTowards Data ScienceImage Registration for Medical DatasetsFrom SimpleElastix to Spatial Transformer NetworksFeb 22, 20233Feb 22, 20233
Published inTowards Data ScienceData analytics principles at an AI startupRules for a scientific approach to data analysis and machine learningSep 21, 2022Sep 21, 2022
Published inTowards Data ScienceAchieving state-of-the-art for offensive tweet prediction using transformersUtilising DistilBERT and fine-tuning for text classificationMay 2, 2022May 2, 2022
Published inTowards Data ScienceAn introduction to transformers and Hugging FaceLeveraging the power of NLP from a top-down approachApr 26, 20225Apr 26, 20225
Published inTowards Data ScienceEnd-to-end Pytorch model in five minutesStandardising dataloaders, model, training and validation — so you don’t have toFeb 16, 2022Feb 16, 2022
Published inTowards Data ScienceUsing Deep Learning for Text-Generation of Wikipedia ArticlesUsing AWD-LSTMs for natural language models in fastai and PytorchJan 27, 20222Jan 27, 20222
Predicting the price of Bitcoin with multivariate Pytorch LSTMsUsing multivariate, multi-output forecasting models for financial dataJan 25, 20229Jan 25, 20229
Published inTowards Data SciencePytorch LSTMs for time-series dataUsing the Pytorch functional API to build temporal models for univariate time-seriesJan 12, 20223Jan 12, 20223