Anonymous ID: d7ad6e March 27, 2023, 7:48 a.m. No.18589114   🗄️.is 🔗kun   >>9124 >>9148

>>18589023 pb

It is only expensive if you run it on someone else's server. Github is full of full-fledged AI / NLP systems that can be run locally.

 

The trick comes with training models and fine-tuning. Shit isn't easy, nor is the documentation easy to get through for someone not trained in ML.

Anonymous ID: d7ad6e March 27, 2023, 8:01 a.m. No.18589175   🗄️.is 🔗kun   >>9190 >>9233

>>18589148

first of all, SCREW YOU BUDDY (sorry, just trying to match the current tone)

 

python man. all the academics love it, which is why all the ML stuff is written in it. but its slow af. well written python reads like a narrative.

Anonymous ID: d7ad6e March 27, 2023, 8:23 a.m. No.18589249   🗄️.is 🔗kun   >>9260 >>9272

>>18589233

NLP can draw upon not just the body of q posts, but a wealth of human knowledge. Done correctly, it can put together things mere humans can not. It can tie things together across all q posts and bring understanding to things we have missed.

Think of it like a million part jigsaw puzzle.

Anonymous ID: d7ad6e March 27, 2023, 8:34 a.m. No.18589298   🗄️.is 🔗kun   >>9433

>>18589272

That is exactly what I want.

I've got all of the text of all posts from the link at the top of the breads. I've got several different types of NLP frameworks downloaded, but training models is foreign to me. There are plenty of tutorials, and I'm working through several of them, some work with text, pics, and even video.

 

I'm hoping someone in here has experience with stuff like this.>>18589272

Anonymous ID: d7ad6e March 27, 2023, 9:26 a.m. No.18589501   🗄️.is 🔗kun   >>9506

>>18589493

curiosity-ai looks like a lot of work that has already been done in python. there is a lot to be said for working in a comfortable language, but good coders code. great coders steal.