NanoGPT using Simpsons Data: Get Started with Large Language Models

preview_player
Показать описание
NanoGPT is a simple, fast repository for training/finetuning medium-sized GPTs. I recommend it to get a better handle on large language models. This video walks through using it on a Simpsons dataset. It covers why I chose nanoGPT, how I munged the Simpson dataset, how I trained my first model, and ways to keep learning.

Chapters:
00:00:00 intro
00:00:17 Why NanoGPT
00:00:52 Simpons dataset
00:01:47 Using the Google Colab notebooks
00:02:24 pull into nanogpt_simpsons repo
00:04:18 using the config files
00:05:36 training the model
00:06:12 getting predictions
00:07:16 using weights and biases for experiment management

━━━━━━━━━━━━━━━━━━━━━━━━━
★ Rajistics Social Media »
━━━━━━━━━━━━━━━━━━━━━━━━━
Рекомендации по теме
Комментарии
Автор

This is so cool! Thank you for sharing this!

dongnguyenanh
Автор

Have you gerbil language model github that's supposed to be useless it would it small enough to be a good learning too if you can get it to work

sadface