Is Bard better at writing Julia than ChatGPT?

Tried to work on a Tetris playing bot in Julia using ChatGPT, but ran into the problem where every package’s API had changed since 2021.

Has anyone had any better luck using Bard? Never fun to start the process and then realize nothing is going to work

I use ChatGPT to help me write code in other languages. Fancy SQL or python things. I already know how to write Julia. I don’t want to learn python.

1 Like

I’ve actually found Claude to be surprisingly good at Julia!

1 Like

I don’t think its about not knowing how to code Julia or not wanting to learn. It’s about asking it how to use packages to accomplish tasks. As in:

Hey, i built a tetris game in julia a few years back. Now im building a tetris bot that can play it. I need your help to write code to solve the following problem (see next paragraph). Remember that you have access to the underlying Julia tetris code, so you can technically change anything you need to. To fill out any details, you can assume that all relevant endpoints of the tetris game are ones that you would have chosen if you had written the game. All code you use should be written in Julia. Feel free to use the Flux.jl and ReinforcementLearning.jl packages …

Why Claude? Never heard of it before. Is it really better than ChatGPT or Bard or CoPilot?

And related to the original question, what is the cut-off date?
Look through the docs for Flux and ReinforcementLearning and you will quickly discover that they all change their APIs pretty frequently. Example scripts that work one year are broken the next very frequently