2 Comments
User's avatar
Rainbow Roxy's avatar

Hey, great read as always, I would appreciate a clarification on the 'o desetinku lepší' improvement for ChatGPT; could you elaborate on which specific performance metric or capability saw this noted percentage gain in its latest iteracion?

Martin Kopta's avatar

I don't want to come across as a jack of all trades, so I'll recommend articles by my colleagues at Substack for evaluation:

https://msukhareva.substack.com/p/gpt-52-and-meaningless-benchmarks

https://thezvi.substack.com/p/gpt-52-is-frontier-only-for-the-frontier

https://natesnewsletter.substack.com/p/new-chatgpt-52-complete-teardowni

https://simonw.substack.com/p/gpt-52-and-useful-patterns-for-building

BTW: I appreciate your courage in reading the original Czech text. ❤️