4 ms·
Resources to deepen LLMs understanding for software engineers
I am a FE developer and am overall interested in LLMs (although who isn't atm?).
I have a broad understanding of how LLMs work and a decent knowledge of software, but I clearly don't have the same knowledge an ML engineer would have to understand such a topic.
Therefore I find myself wanting to learn more, but I either find basic resources that already cover what I know, or too advanced stuff that is meant for another audience.
I was wondering if there is somebody producing content for my segment instead. Ideally somebody is very technical on the topic, but that explains it for a broader audience.
To be clear, I don't want to become an ML engineer, but I'm interested in learning about updates and breakthroughs (and why they are such), as well as deeply understanding the challenges, limits, applications, and potential of LLMs.
- duckkg5 3y agosentdex on YouTube has a good library of videos on the subject.
- tikkun 3y agohttps://news.ycombinator.com/item?id=36195527 https://news.ycombinator.com/item?id=36195527
- coolThingsFirst 3y agoThe problem with LLMs is that even if you learn them you can't train them. Not enough power in consumer GPUs
- a2code 3y agoIsnt training as a service a possible temporary solution here?
- coolThingsFirst 3y agoIt's really expensive, I'm talking about new people learning and experimenting with LLMs
- aborsy 3y agoAren’t there smaller ones, like 7B llama2, that could be trained in a powerful computer?
- jyu 3y agothis has been a great resource. approachable and great for practitioners. it's frequently updated with new papers and techniques https://www.promptingguide.ai/ https://www.promptingguide.ai/
- davidajackson 3y agoLlamaIndex docs, LangChain docs, Youtube vids on intro to neural networks, tutorials on RAG, cross encoders, etc. should keep you busy for 20-40 hours at least