Let's reproduce GPT-2 (124M)

We reproduce the GPT-2 (124M) from scratch. This video covers the whole process: First we build the GPT-2 network, then we optimize its training to be really fast, then we set up the training run following the GPT-2 and GPT-3 paper and their hyperparameters, then we hit run, and come back the next...

Intermediate - 241 min - AI & Data - Andrej Karpathy

The full course: Learn Deep Learning free

What you'll need

• GPU-ready laptop: Train small models locally.
• Laptop stand: Screen at eye level for long study sessions.
• Noise-cancelling headphones: Focus through long video lessons.

Books

1. Deep Learning with Python by Francois Chollet - View on Amazon
2. Deep Learning by Ian Goodfellow - View on Amazon

Next lessons

• Let's build GPT: from scratch, in code, spelled out. (116 min, Andrej Karpathy, Beginner)
• Transformers, the tech behind LLMs | Deep Learning Chapter 5 (27 min, 3Blue1Brown, Intermediate)
• But what is a neural network? | Deep learning chapter 1 (19 min, 3Blue1Brown, Beginner)
• Gradient descent, how neural networks learn | Deep Learning Chapter 2 (21 min, 3Blue1Brown, Intermediate)
• The spelled-out intro to neural networks and backpropagation: building micrograd (146 min, Andrej Karpathy, Beginner)
• Attention in transformers, step-by-step | Deep Learning Chapter 6 (26 min, 3Blue1Brown, Beginner)
• How might LLMs store facts | Deep Learning Chapter 7 (23 min, 3Blue1Brown, Intermediate)
• Let's build the GPT Tokenizer (134 min, Andrej Karpathy, Intermediate)