Excellent, thanks! Just took a look at "TCP/IP Illustrated: Volume 1" and was exactly what I was looking for. Any book along those lines that is compressed/water down (just to get started over a weekend)?
thanks. it'd be great to have a quick tutorial on doing so.
this is close to my dream of creating Frankenstein apps with the web platform instead of graal :)
You can stay on 3.x. The license on 3.x shows a watermark and a license key will hide it. New commercial licenses will still work for 3.x too, in case you’re unable to upgrade, though 4.x has only a few small breaking changes.
Maybe they aren't, they could just be hosting these interviews as a fresh source of data for Claude to slurp up! ... sorry, I shouldn't be stirring up conspiracy theories this late on a Friday.
I feel building your own is probably not the best way to benefit from a CAS, unless you are doing it for fun! But I had this question too in the past and found some resources, including the fact that SymPy is a good project to learn from.
Not directly, but COPRA which they compare to (and show about 3% more theorems proved than) is based on GPT-4 using an agent framework. And in the COPRA paper, they compare to using GPT-4 directly as a one-shot, and find it can only get 10% of theorems in miniF2F, as opposed to 23% with their agent based approach. So if the eval line up correctly, we would see GPT-4 one-shot proving 10.6% of theorems in miniF2F, as opposed to Llemma-7b proving 26.23%, a pretty significant improvement. Still not as good as other specialized tools though, especially when you look at tools in other theorem proving languages (see my other comment for more detail about a cross-language comparison).