welcome

If you are like most people these days, LLMs are already part of your everyday toolkit. Maybe you have asked an LLM to proofread an email, debug a tricky piece of code, or explain a concept that sent you down an unexpected rabbit hole. Since 2022, and the launch of ChatGPT, these models have moved from experimental novelties to essential tools in our daily work and learning.

It’s been quite a journey to get here. The earliest GPT models, introduced in 2018, could generate text that was more or less human-like, but they were primarily text-completion models, largely unable to answer even simple queries, and the response quality was nowhere near that of the LLMs we use today.

Then came instruction fine-tuning and alignment with human preferences, which ChatGPT popularized in 2022. The techniques behind ChatGPT transformed LLMs into the everyday problem solvers we use today. Currently, we are in the latest phase: developing reasoning models. Reasoning is the ability for an LLM to tackle more complex problems step-by-step.