Simon Willison’s Weblog

Tuesday, 6th September 2022

dolthub/jsplit (via) Neat Go CLI tool for working with truly gigantic JSON files. This assumes files will be an object with one or more keys that are themselves huge lists of objects—it than extracts those lists out into one or more newline-delimited JSON files (capping their size at 4GB) which are much easier to work with as streams of data. # 8:27 pm

karpathy/minGPT (via) A “minimal PyTorch re-implementation” of the OpenAI GPT training and inference model, by Andrej Karpathy. It’s only a few hundred lines of code and includes extensive comments, plus notebook demos. # 2:52 pm

2022 » September