Open Source ChatGPT for developers 👩‍💻 #programming #software #technology #chatgpt #ai

[CodingWithLewis]

Highly recommend you try out local models!

[amarjeetsingh]

Companies first make open source project so developers put their efforts then later on they change it to closed one

[DingleFlop]

Can't wait for this stuff to be more accessible. Remember, if we don't experiment ourselves, companies will develop and gatekeep it and sell it back to us at a premium.

[RandomMinecraftUser]

I do love Ollama! Their discord is quite good for help with it and chatting about AI as well!

[samarthverma]

Used it with the smallest model, cpu usage - 101% 😢

[shirakuyanai]

AAAND it's not as censored as Gemini or ChatGPT, already a big selling point.

For comparison, I asked eache of them how to pick a lock and here's their response:
-ChatGPT and Gemini: I don't want you to do it. And you can't do it if I won't tell you how. And here's why you should not do it.
-Ollama: Don't do it. But here's how anyway.

[sabeerbikba]

if thinking about install in local machine complete time

[markosth]

Are the weights of phi3 with functions available somewhere?

[ridabrahim]

Ma man. You need a performant computer to run the models.

[Nick_With_A_Stick]

Oobabooga isn’t as optimized as I would like, and its OpenAI api is very finicky when not loading models through the HF loader. I wish Oobabooga would support VLLM as a loader or VLLM could get a UI, or an easier back end like OLLAMA. Because vllm can get you 1, 200 tk/s on a single 3090 with a 7b.

[azazella]

You mentioned Docker, and I always hear the term. I recognize that it is a server related programming language, but what is it really?

Is it a database, just a way to connect backend to frontend, or something completely different? Of you were to use MERN stack as an analogy, where does Docker fit?

[combax_techx]

I wanted to try bigger llamas than 8B but my GPU is poop

[EricSykes-TheUmp]

Bro that'd not you talking that's ai

[averagememer]

Why is he talking like an AI? He keeps saying "Uhh"

[CommentGuard]

What? Its llama 405B thats better than GPT4-o its the ONLY one better.

And it 405B parameters so you better hope you have a $50, 000 h200 lying around

[Janemayank]

I don't know why I'm following you. Maybe because your videos are presented nicely. Other than that, your content makes no sense. What is your goal? Would you sometime explain?