14
votes
Looking for resources about AI development
Hello,
I'm looking for resources on how to develop AI, aimed at people who already have experience with programming.
They don't have to be free, I would just like to aggregate different type of resources to pick from.
Thanks!
Well, it depends on what kind of A.I. you want to do. I have to imagine that you are referring to generative AI specifically, since that is all the rage right now.
Level 0: Basics of Deep Learning
3Blue1Brown's "But what is a neural network?" Purely theoretical. Nice visuals. Good starting point.
Practical Deep Learning for Coders 2022 The basics of deep learning. Covers neural networks and techniques surrounding them. Jeremy Howard is the GOAT.
GPT From Scratch . Posted on here the other day, really good article.
Level 1: Transformers
Tensor Flow's explanation of transformers Maybe a bit more complex for beginners, but nice visuals.
Machine Learning Mastery Tutorials.
Hugging Face Repository of models that can be plugged into applications.
Level 2: Working with LLMs
LangChain A framework for developing applications powered by language models. Really gives you an idea of how applications for LLMs are made.
These look promising, thank you!
I'm probably gonna edit it and keep adding more, so keep an eye out.
Andrej Karpathy's Neural Networks: Zero To Hero series is definitively worth watching. It starts with the very fundamentals (automatic differentiation and backprop) and works its way up to writing an implementation of a Generative Pre-trained Transformer.
This video series is really great. This is where I'd start @axolotl if you're an intermediate programmer or beyond.
A little lighter resource: https://youtube.com/@AllAboutAI
Mainly focused on OpenAI / ChatGPT, simple Python scripts towards OpenAI etc. You can see some methods in practical use and purchase a membership that will give you access to a GitHub repo with basic scripts used in the videos.
Thank you, I will take a look at this channel!
AI development can have several meanings, ranging from the research and math part to building software that leverages off-the-shelf ML models.
A good resource I've found is AI Canon, which already aggregates many of the things that you'd want to read up on to get up to speed on what the current state of machine learning is. It has papers, blog posts, courses, and guides that had a significant impact on the field of Machine Learning in recent years.
I also really like the links that @JesusShuttlesworth provided, and can't ever not recommend 3Blue1Brown!
The way I learned was from reading papers and experimenting with what they had done in PyTorch. This probably won’t work unless you have a background in math, but I think the first thing I did was implement a neural styling algorithm, which was extremely valuable for understanding how neural nets work.