One user suggested using a model already trained on Wikipedia using llamafile, which can be downloaded along with several models, put on a USB drive or hard drive, and sent to the boy via DHL. The uncle appreciated the suggestion but mentioned that the boy is keen on tinkering and learning more about how these models work.
Key takeaways:
- A 16-year-old in East Africa is seeking advice on how to train an open source LLM using an on disk Wikipedia.
- The nephew has limited internet access, making the task more challenging.
- One suggestion is to use a model already trained on Wikipedia using llamafile, which can be downloaded and sent to him via a physical drive.
- The nephew is already proficient in programming, including developing Android apps and Python, and is keen to learn more about how these models work.