It’s not lazy anymore! I’m pleasantly surprised that it was able to generate 400 lines of python code with comments etc over 4 responses. Neat. No fighting back and forth. Three days ago I was just trying to refactor the same python file I tried today and I decided to give up before trying one more time today. I’m glad I did.
What changed?
Glad it’s working better! We switched back to a prompt from before gpt-4 0125 (even though it’s much better than 1106, it is still sometimes pretty lazy)
Ohh ! You guys finally adopted turbo ! Interesting.
I can’t get it, sorry for a stupid question: you switched back to 1106 or 0125 but with an old prompt?
Sorry, wasn’t very clear. We are on 0125!
I mean in another topic one dev said “cmd-k uses gpt-4-turbo, but we may be rolling back due to the laziness problem. Chat will stay unaltered though” so I wonder what you decide? No need to roll back for cmd-k or it’s still a question?
Haven’t decided yet! Will update on the forum if/when we do!
Thanks for the update.
Because of the different cut-off dates, It’s pretty important for me… If laziness can be surpassed by 1-2 additional tries i’d better go with a new model…
… And as others said the option to choose would be just perfect. Maybe you could show the “recommended” model and let us choose smth else if it doesn’t work.