Comment by qrsjutsu
1 month ago
If you are not on that hype-train, yet, then
don't waste time skimming over, reading and understanding any LLM and AI papers.
Read about ELIZA. Build your own.
Get Tensors, Vectors, Fields, Linguistics, Computer Architectures, Networks.
Focus on the subjects themselves, not them in the context of Neural Networks, "Deep Learning" et al.
The list seems deliberately focussed on learning practical topics and not the foundations. I think that's what makes it interesting - there are any number of places recommending you start by learning linear algebra, stats, probability...
Personally I like to learn the foundations but there's genuinely room for useful knowledge of SOTA techniques even without the foundations. To be honest I feel that any amount of learning about computer architecture and vector fields is unhelpful if you are trying to understand good eval benchmarks or prompt engineering techniques.
> unhelpful if you are trying to understand good eval benchmarks or prompt engineering techniques.
You are absolutely correct. I jumped to conclusions when I saw the list and read "AI Engineer". The reading list isn't addressing people who want to build AIs, but those who want to maximize and optimize their results with the existing ones.
My bad.
what is ELIZA? github link?