Love these. I’ve also made some visual explanations for ml for Amazon, available at https://mlu-explain.github.io/ Big fan of your early work, Jay, a big inspiration for me!
That's beautiful! Hope you're getting to do more of these!
Show HN: Visual intuitive explanations of LLM concepts (LLM University)
31–37 of 37 posts
Re: Show HN: Visual intuitive explanations of LLM concepts (LLM University)
#32Hi Jay,
I really loved your [explainer on AI Art](https://www.youtube.com/watch?v=MXmacOUJUaw), and I've already added more of your videos and articles on my watch-later read-later lists! Can't wait to spend more time with them this weekend.
Thank you for creating such wonderful resources!
Re: Show HN: Visual intuitive explanations of LLM concepts (LLM University)
#33Looks really helpful. Joined the community as well.
Re: Show HN: Visual intuitive explanations of LLM concepts (LLM University)
#34> We've just published a lot of original, visual, and intuitive explanations of concepts to introduce people to large language models. Kinda frustrating that the main link dumps me onto what reads like a university syllabus, and nothing original, visual, or intuitive. If I click through the sections in order, there are 5 "preamble" sections describing logistical and other meta-information about the course. All text.…
I read and watched almost all the modules and for me it as it is, perfectly accomplishes the intention of the course as stated by the op. Your suggestion may work for other intents (like having a Schaum's Outline of LLM's) and I would also love to have that additional material (maybe yourself could provide it as it seems you have a clear idea)
I'm not proposing anything WRT the depth or scope of the material.
If the premise of the material is that phrases like "dot product" can be used freely or with minimal explanation then images like the "place the apple quiz" make even less sense. For that person, not much more needs to be said than "We want to represent words as vectors so we can do linear algebra with them. If the representations preserve structure we care about then here are some cool things that happen: (examples of good and bad embeddings)."
Then go deep, having given them an adequate advanced organizer.
Re: Show HN: Visual intuitive explanations of LLM concepts (LLM University)
#35Earlier quoted context omitted.
Thanks, Jay! I wasn't aware of that BERT explainability paper - will be reading it, and watching your video. Are there any more recent Transformer Explainability papers that you would recommend - maybe ones that build on this and look at what's going on in later layers?
Additional ones that come to mind now are: Transformer Feed-Forward Layers Are Key-Value Memories https://arxiv.org/abs/2012.14913 The Dual Form of Neural Networks Revisited: Connecting Test Time Predictions to Training Patterns via Spotlights of Attention https://arxiv.org/abs/2202.05798 https://github.com/neelnanda-io/TransformerLens
Another piece of the puzzle seems to be transformer "induction heads" where attention heads in consecutive layers work together to provide a mechanism that is believed to be responsible for much of in-context learning. The idea is that earlier instances of a token pattern/sequence in the context are used to predict the continuation of a similar pattern later on.
In the most simple case this is a copying operation such that an early occurrence of AB predicts that a later A should be followed by B. In the more general case this becomes A'B' => AB which seems to be more of an analogy type relationship.
https://arxiv.org/abs/2209.11895
This is still only a low level mechanistic type of operation, but at least a glimpse into how transformers are operating at inference time.
Re: Show HN: Visual intuitive explanations of LLM concepts (LLM University)
#36Love these. I’ve also made some visual explanations for ml for Amazon, available at https://mlu-explain.github.io/ Big fan of your early work, Jay, a big inspiration for me!
That's beautiful! Hope you're getting to do more of these!
Thanks, am definitely trying to squeeze them out as I’m able!
Re: Show HN: Visual intuitive explanations of LLM concepts (LLM University)
#37I'm not sure how much is actually known to write about, but what I'd like to see explained is how transformer-based LLMs/AI really work - not at the mechanistic level of the architecture, but in terms of what they learn (some type of world model ? details, not hand waving!) and how do they utilize this when processing various types of input ? What type of representations are being used internally in these models ? We…
No one can tell you much about that. Interpretability is still very poor.
You don't know what they learn beforehand (else deep learning wouldn't be necessary) so you have to try and figure it out afterwards.
But artificial parameters aren't beholden to any sort of "explainabilty rule". No guarantee anything is wired in a way for humans to comprehend. And even if it was, you're looking at hundreds of billions of parameters potentially.