With the imminent release of OpenAI's -o3 reasoning model and Deepseek's impressive R1 release, it's clear that reasoning models are improving rapidly. But, many aren't sure where they can be best applied or how to use them effectively. Here is a short primer, reviewing the scaling paradigm (RL on Chat-Of-Thought), practical prompting tricks, a set of emerging use-cases for reasoning models, and poractical usage of OpenAI's -o1.
Video notes:
https://mirror-feeling-d80.notion.site/Reasoning-Models-177808527b17809293ffd82abb3c05aa?pvs=4