Auto-translation used

Experiments with Gemini and GPT-4o

I got into an OpenAI experiment to work with the latest GPT-4o model. I decided to test and compare the results of writing ML code using Google Gemini Advanced and OpenAI GPT-4o.

The results reminded us of the sacred truth, which, one day, was told by a friend from Rambler:

— when I drink a glass of cognac in the evening, I immediately start writing code faster. And if the second or third, then in general it flows like from a stream. But for now there is one problem

— Which one? Does your head hurt in the morning?

— in the morning, when I read, I need to delete almost everything 😂

That's the same with the current versions of neural navigators. They write the code quickly, but according to quality metrics, the resulting modules:

— for AdaBoost, they work as a constant classifier

— for Gradient Boosting - compare the quality of one trained tree

Current development status (as our C++ teacher said in the SHAD):

#define private public

# happy debugging

(c) Public Morozov

😉 So it 's still relevant:

Application testing course 

ML course 

#work #study

Comments 0

Login to leave a comment