This video provides a wide overview of the current (as of May 2024) proprietary and open LLM situations, explaining the major differences along the way. After that there’s a technical deep dive into concepts that make up large language models.

This is the first in a series of videos designed to provide a wide overview of the large langage model landscape.

The primary metaphor is that the large language model is a symphony.

Note: Running models locally, referred to as inference, can be complicated. The breadth of options and the massive research occurring makes it difficult to keep track of. There are lots of resources around running local models, here's one.

Models Referenced:

• GPT4
• Google’s Gemini
• Anthropic’s Claude
• Meta’s Llama 3
• Bloom
• Bert
• HuggingFace Models (639,008 as of the moment I wrote this)

Glossary:

• Backpropogation
• Training (covers data preparation requirements)
• Attention