south niche
Hello, I'm a software engineer, but my experience with AI and ML is very limited, so I want to make some projects for the sake of learning to get deeper into it.
As I have a bit of background, a project I perhaps can start with is fine tuning a base LLM for the sake of summarization(I know its not new, but I'm doing it for the sake of studying myself). Create a dataset for example by going over some news articles and picking the title as the summarization (it would be an expert of finding titles later), or using an exiting AI like chatGPT to create the summarizations for my dataset, as far as I know even a few hundred like these will already make a difference.
I can find a lot of information, but its either the theory in a nut shell, or very specific for one model or another, without any details for why was the model chosen, or any of the other details involved.
Is there a source of information that covers the issue of fine tuning (pure, with lora or qlora) well and has the details as well, apart for the super practical parts?
quasi timber
