#For anyone looking for a highly
minor monolith
imma read this whole thing
tawny nexus
Nice, I'll check it out later, and give my thoughts 🍵
tawny nexus
Good compilation of terms related to language models. I was under the impression that this would be targeted at prompt engineering. However, It was good to look over some of the jargon.
The tone was generally academic, with bits of pieces of creative writing and opinion interjected which felt a bit jarring to me, but could be interesting for others. Things such as the AI being omnipotent.
I think that you could figure out exactly what you'd like to focus on, as some parts are getting a lot of attention, while other parts get little.
I wouldn't say this is a guide. As it has very few instructions on implementing things, except the small part on prompt engineering. It mainly comes off as a reference for ML and language model definitions and terms.
I think that this is a good read to skim through, but to understand the ideas will require further research that breaks things down into simple parts. The way I think about teaching things is to make everything obvious in isolation.
Overall, great job. It takes a lot of work to craft up a document like that, and it's great that you are sharing your work freely with others. I hope you enjoyed the process of putting it together and continue to help out the community 🍵
pallid nebula
Good compilation of terms related to language models. I was under the impression...
Thank you so much for taking the time to lay this review out for me. The Idea is to take a ground-up approach for anyone looking for any level of information about this - eventually when complete it will contain all necessary information related to XLLMs down to how to build one from scratch. I kind of excitedly released my 2 weeks in version to the public because it does contain a lot of very good general information to sort of work off of.
I did not claim AI was omnipotent, AGI would be the Omnipotent version or "goal" ofAI. It was intended to make the readers think about AI and its use today. As while it is really cool, it is also just recycled human communication right now, and not at all "intelligent" in any form or capacity. This was in hopes to inspire people to stop trying to use it as a pet and feel very disappointed when it cannot do their impossibly complex requests with 100% perfection, as I see so many immediately do. Maybe even question if such an "intelligent" AGI is even possible? Perhaps that section needs more work.
Also, I threw in the Prompt Engineering section in like a day... so while it contains all of the info you really need to write prompts, I still intend to grow it by at least 5 pages with more detailed and specific strategy. Right now it at least has all of the stuff needed to grow that strategy if you are clever enough 😁 That is a section I wish to make much much better if you have the time to critique it specifically.