How good are LLMs at mathematical reasoning? A new paper by researchers at Apple modifies the standard database of Grade School Math (GSM8k) problems in three ways to test how ChatGPT etc perform. 🧵 1/
In 2013, I wrote a book about educational myths.
I *didn't* include learning styles. Why?
Because even in 2013, so much research had already been published showing they were bunk. Didn't feel worth me going over it all again.
It's now 2024. What's in @TheEconomist ?
@cripincantation It was for sure unexpected, but I think it fits really well. For a minute I thought "wow, now Cynic?", but it serves to make a great contrast between the Stargate and The Message. I mean, they literally crossed the Stargate, the other side has to be different.