EN / RU
← Back to essays
· Essay · 1 min

Day 2/12: Reinforcement Fine-Tuning

OpenAI is adding new capabilities to fine-tune models, but it's a complex and expensive process.

It seems OpenAI is adding new capabilities to fine-tune models. But let's admit, fine-tuning is generally an expensive and complex process that isn't necessary for most tasks.

In fact, the steps look like this:

1) Prompting
2) RAG
3) Fine-tuning - only if everything else hasn't helped.

Well, let's observe what the details are.

Live stream: https://www.youtube.com/watch?v=fMJMhBFa_Gc

#openai #ai #agi #gpt

—————————
Thoughts of Rvachev
—————————

Day 2/12: Reinforcement Fine-Tuning — illustration