Martin DittgeninTowards Data ScienceImplementing LoRA from ScratchHow to implement LoRA from scratch and some practical tipsDec 12, 20235Dec 12, 20235
Martin DittgenTraining a BERT Model from Scratch on a Single Nvidia RTX 3060In this blogpost I will show you how I trained my own BERT-like model, called DitBERT, one a single Nvidia RTX 3060.Jul 21, 20231Jul 21, 20231
Martin DittgenReproducing the “Attention is all you need” Paper from ScratchEver since ChatGPT’s release in December 2022, the excitement surrounding transformer models has been on a steady incline.Apr 11, 20231Apr 11, 20231
Martin DittgeninTowards Data ScienceBenchmarking simple models with feature extraction against modern black-box methodsCan interpretable models, like linear or logistic regression compete with black-box algorithms like boosted trees or neural networks?Oct 13, 20192Oct 13, 20192