Heeyyyy
I’m new to this website and a you can call me a noob or a newbie but I wanted to learn about all this so if anybody can help me please so. I want a basic intro to this platform
1 Like
Here are some resources. I hope they’re useful…
3 Likes
Thanks John for your help but let me give you a basic overview, I completed basic ml course from uni but it wasn’t good. So I know basics of ml concepts but not thorough and I started reading “How to build llm from scrath” by Sebastian Rachska but after seeing these many resources like I wanna also do course on agents by hf, there are so many things to learn I feel overwhelmed. So nay tip what should I do first?
1 Like
Well, there are really a lot of them.
It’s probably best to start with something that interests you.
How about trying out a few actual models in Spaces to find a field that interests you?
Most Spaces allow you to view the source code, which is mostly Python. You can also copy it. The model source code is also available on the library’s GitHub.
Speaking of libraries, the Diffusion course and NLP course cover the basics of Diffusers and Transformers, which will be useful when using Hugging Face models.
Also, there are reading groups and a section for sharing works (I Made This / Cool Finds) or so on HF Discord.
2 Likes
Yeah man, there’s a lot of stuff online here. I came across HF when I just thought of making a personalised gpt like model that can ask on my personal context and the fun fact is I was asking all these to gpt then I came across that there’s a finetuning thing and I first I tried with a large model from HF on Google Collab and it just crashed all the time not knowing free tier cant handle large models but after some research came across unsloth but still there was a problem of dataset and I thought I dont understand any of these, I m jus trying to vibe code a model and that’s why I thought to learn but there are so many aspects to it like math, dl, nueral networks and my gosh
1 Like
I see, so you have a pretty clear goal in mind.
In that case, I think there are some people on Hugging Face’s Discord who are knowledgeable about the relationship between LLM and mathematics. They also seem to know of some online reference materials. I’ve seen discussions about it on ask-for-help channel before.
For information on LLM training itself, I think unsloth’s resources and their Discord are also a good place to start.
There are a significant number of students, researchers, and engineers on both Hub and Discord(s), so if you can make good connections, you should be able to get valuable information on the best order to learn things.
As for environments to experiment with, besides Hugging Face, there’s Google Colab and lightning.ai. I’ve also heard of Kaggle and TPU Research Cloud. Well, buying a GPU would be more efficient, but it’s expensive…
Speaking of which, ChatGPT and Gemini are the most knowledgeable about AI training. It seems that they are heavily included in the learning data…
2 Likes
Yeahh sure. That’s what I’m doing, talking with knowledgeable people like you so that I can help myself to earn some wages 
. Nahhh, its too early for me to think to buy gpu for working. Let’s see let me get some undeerstanding and also I will also learn the agent course from HF learning platform. So, that’s the plan for now.
1 Like