Luca Massaron

kaggledays-2019-gbdt. Kaggle Days Paris - Competitive GBDT Specification and Optimization Workshop

Machine-Learning-on-Tabular-Data. Code for the new Manning book on machine learning on tabular datasets

deep-learning-for-tabular-data. An updated (2025) guide to Deep Learning for tabular data, comparing a fine-tuned Keras 3 (PyTorch backend) DNN and an Optuna-optimized XGBoost model on the Kaggle Amazon Employee Access Challenge

gemma_from_scratch. A clean, minimal, and educational implementation of the Gemma 3 language model architecture using pure PyTorch and JAX

dl4dummies. Deep Learning For Dummies - code repository maintained by one of the authors

datasets.

Gemma-2-2B-IT-GRPO. Fine-tuning the Google/gemma-2-2b-it model using Generative Reward Post-Optimization (GRPO)

algo4d_2ed. Source code for Algorithms For Dummies revisited by removing reference to external packages

ml4dummies_2ed. Code and data repository for Machine Learning For Dummies 2nd edition

fine-tuning-workshop. Jupyter Notebook

petfinder_7th_place_solution. Kaggle PetFinder Competition: (Deep Learning) Machine Learning For Puppies

keras_tuner. Bayesian Optimization for Neural Architecture Search

Python-Data-Science-Essentials-Third-Edition. Python Data Science Essentials - Third Edition, Published by Packt - Code maintained by one of the authors

ml4dummies_3ed. Code and data repository for Machine Learning For Dummies 3rd edition

tunix-medical-finetune. Hands-on tutorial for adapting Google Gemma 3 to the medical domains using Tunix, a high-performance JAX-based fine-tuning library. While optimized for TPUs, this project demonstrates the hardware-agnostic power of JAX/XLA.

Gemma-3-Function-Calling. Making google/gemma-3-1b-it model ready for an agentic role with function calling tasks

gemini_hallucination_detection. Demonstrating how to use Gemini to detect hallucinations in LLMs

retinanet_demo. Demonstration of how to easily use Keras-RetinaNet, a project sponsored by the Dutch robotic company Fitz and made possible by many contributors (the top contributors are Hans Gaiser and Maarten de Vries)

gemma-4-benchmarks. Jupyter Notebook