GPT-Engineer: Your Own Personal Coding Assistant

[@AaronJOlson]

I am a huge GPT engineer fan. I think it shines under more complex prompts (2-3 paragraph descriptions, not 2-3 sentences.). The more files and things that need to get made, the more impressive it is. It will cost more, but to really see what it is capable of you need to give it more complex requirements.

[@NeuralNine]

Interesting. Maybe I should give it a try again with a more complex description.

[@issamelimrany]

I believe that using cursor (cursor composer does what GPT-Engineer does) is way more convenient and efficient, would be nice if you made some video about your experience using it!

[@shanknbake]

this

[@amongus-rq2hb]

i really like using cursor and i am even addicted to it. my 2000 autocompletion got over. and i dont know what to do now. 20$ is too much for someone like me. is there any alternative.

[@issamelimrany]

For me, I just keep changing accounts, apparently they don't keep track of the IP adress lol

[@amongus-rq2hb]

@@issamelimrany how do you change the account via mail?

[@lakutomafoso]

@@amongus-rq2hb just register new email

[@EdwardTilley]

Great intro to the app.

[@NeuralNine]

Thank you!

[@BunnyR-ut5om]

@NeuralNine What OS are you using in this video? And why not use Windows?

[@NeuralNine]

Pop!_OS Linux 22.04

[@BunnyR-ut5om]

@@NeuralNine thanks for the reply dude. Appreciate it!

[@shamikchoudhury5924]

There are few tools like gpt engineer which claims to do lots of things, but in practice it can't go beyond very simple application. Not sure how this tools helps and what is the use of these.

[@NeuralNine]

Maybe with better LLMs over time, these tools can replace coding interns with little experience.

[@dipeshsamrawat7957]

Thank you 😊

[@NeuralNine]

Thanks for watching!

[@ChristopherBruns-o7o]

6:15 why does the thing give python3 syntax with a python header, is what I want to know. Alias perhaps? This bot should self destruct after installation. Would be incredible.

[@timstevens3361]

what gpu r you using ?

[@wahwahwally1]

I think he is using the API, not his own gpu in this video.

[@NeuralNine]

I am using OpenAI models in the cloud. But I have a 3060 Ti.

[@timstevens3361]

@@NeuralNine i hav 2 rtx 3060s 12g. i want to put them on same motherboard. im using ollam n llama3.2-vision atm.

[@seshusai8208]

BRo any free version

[@TobiMetalsFab]

Run locally with open source models, he quickly shows the guide they have at 2:08

[@NeuralNine]

You can use your own LLMs locally. But for this you need hardware that is strong enough.

[@g.s.3389]

it would be nnice if instead of working with chatGPT it would work with ollama.

[@NeuralNine]

There is the possibility to use custom models with it too.

[@soul_maestro]

tyvm for not calling it AI, but naming it for what it is: LLMs also tyvm for all the guides you put up around this and other topics.

[@NeuralNine]

Well, technically LLMs are neural networks, which is deep learning, which is machine learning, which is AI. But I totally understand your aversion when it comes to using AI as a buzz word.

[@soul_maestro]

technically machine learning isn't AI. ML is just "let's finetune this crud algoritme we though of with this data we collected and tagged, and be lazy about it and use another algoritme to do the tuning part" and now we're at the point that we can feed it alot of data, but it's still just a stupid algoritme with no intelligence what so ever.

[@RocketLR]

OpenAI GPTs are garbage compared to Claude.. I've thrown a whole project into claude and had it do so much for me. The only annoying thing is that you have to manually tell it the structure of your project so it know when it references other code or something. But with this , waw it would kick ass.. You really should thy using anthropics latest model and compare.

[@NeuralNine]

Yes, it's on my to-do list for quite some time already.