GPT-3/LLMs' Achilles heel is short context length - how many "in-context" examples they can consume to learn a new task.

Enter "Structured Prompting": scale your examples from dozens => 1,000+

Here's how:

=> Get 1000s of in-context samples

=> split them into M groups, each small enough to fit in regular context length

=> encode each of M groups using LLM encoder

=> combine these encoded groups and attend over a scaled version of the combination simultaneously

Paper: https://arxiv.org/pdf/2212.06713.pdf

Code: https://github.com/microsoft/LMOps

AI has a lying problem

AI Alignment

AI Alignment

What will finally enable this?

The student

Honor Roll student of the future.

the moment when teachers figure out they can grade 300 essays in 15 min with chatGPT

AI has a wokeness problem,
and a lying problem.

what would steve jobs think of the woke mob?

ChatGPT has been a game changer for school essays

I learn through online school in a small city without about 300 students, classes cycle every 3 weeks, English for the first semester opening up just about when ChatGPT launched. So, I decided to see how it, along with Quillbot would fare. The teacher fr called to congratulate me for being one of the best writers he's taught and my mom was so proud. Was just sitting there trying so hard not to laugh. AI really is to essays what Calculators were to math