Kaze vs ChatGPT o1 full video

[@saymehname]

Based Hallucinating Robot

[@mirabilis]

It's just a little tipsy.

[@chosencode5881]

You're a legend bro! Would be nice if you got someone to edit this into the best bits! Edit: LOL NVM saw your other video!

[@chosencode5881]

LOL nvm just saw the other video on your channel XD

[@ImYourCat2]

that's a cool thumbnail

[@Catdevzsh01]

cousin? lol im porting b3313 to mac we're both cats !

[@ogbaxstar]

@45:30 pretty epic that it reasoned to try that scaling indicies solution. This iteration of the coding AI is waaaaaayyy improved. It understands quite a lot. Just in time for you to dis it saying it hadn't helped you out yet LOL. Great video. Enjoyed watching this one a lot more than your previous attempts at using AI to improve and optimize code. Excited to watch the rest. Edit: just got to 1:03:35, rip my comment :/ LOL

[@Catdevzsh01]

wah wah wah Just use codellama 70b if u have sick gpu cards i use 34b or codestral

[@brokendoop]

ChatGPT can be a great learning tool if you ask it very simple questions for things, or to explain certain concepts. It's fun to read code it generates and come to the conclusion that what it just made is complete nonsense, reaffirms that you are making progress. Sometimes the poor code it generates can actually make you learn something new, in most cases what NOT to do lmao.

[@Fuzzycuffsqt]

Watching the proposed optimization completely lose track of mario's rotation data was funny. Thank you for the window into your process! TBH it's a miracle I understood anything in this coding video at all, so I'm gonna quit while I'm ahead

[@Catdevzsh01]

also theres personlization memory

[@Catdevzsh01]

this is what i watch toput me to bed lmao

[@realdomdom]

krasset ding

[@Catdevzsh01]

u need to use o1 preview ;p

[@Catdevzsh01]

Kaze wanna here a cringe joke why did the mips was faster than the VRAM Bus because he couldnt the VRAM Bus woke up late

[@b00marrows]

It seems improved in some aspects, understanding more of the question and code and stays more on point... But still fails to follow all instructions... I do wonder if it uses whats said to it to learn. Like, if you tell it "this increases instruction count by 2" it will take that issue into account harder. I also wonder if it defaults to positive, as in if you dont respond to something it did it considers it a success... Man AI stuff both intrigues me and pisses me off at the same time.

[@SuperXzm]

Wrestling with hallucinations really pisses me off. Can we have a service of a middleman? Like you talk to a person and then he screws with all the gpts and bings and alexas for an hour and gives you the result?

[@psionick5106]

22:44 do (1) { ... } while (1); idk if u checked it after but i'm curious