Tag
A 45-minute video tutorial on creating synthetic datasets and training tiny (100M parameter) local language models for narrow tasks, with code and resources provided.