gushogg-blake

> bullshit detection

This doesn't sound like anything that's not accepted by mainstream medicine to me. If they'd started bringing pseudoscientific ideas in then sure, but breathing exercises are a well-known way to regulate your nervous system aren't they?

It's an amazing project but I always found the docs lacking. Maybe you'd have to sit down with the code for a while or ask AI to explain it, but there were concepts, like "facets", that seemed crucial but were never really explained. I also found the way the docs sites were structured to be really frustrating and confusing - it always felt like there were pages missing and I was never sure if I was on the right page for the thing I was trying to read about (or even on the right site, as it's split into multiple projects).

Having written a canvas-based editor with some non-trivial features like Tree-sitter based multi-language syntax highlighting, performance is actually what impresses me the most about VS Code. Maybe my standards are different but I don't think I've ever noticed it lag even on large files.

Currently doing myo, and Buteyko! Glad to see someone else here. I've also done a MARPE.

Multiple opinions won't necessarily help unless you research first. I'm not sure about other disciplines but in orthodontics for example there is a very definite split between whether they are "airway focused" or mainstream. I saw 6 or 7 different doctors, dentists and orthos in the UK over a period of about 10 years and none of them saw anything wrong with my jaws. Saw an airway focused ortho and a surgeon specialising in maxillo-mandibular advancement surgery and they recommended aggressive surgery. I agree with them, mainstream medicine is ignoring an epidemic of jaw underdevelopment because it challenges current practices in orthodontics.

Exactly what I was looking for, thanks!

This is exactly what I was looking for, thanks!

I haven't found an explanation yet that answers a couple of seemingly basic questions about LLMs:

What does the input side of the neutral network look like? Is it enough bits to represent N tokens where N is the context size? How does it handle inputs that are shorter than the context size?

I think embedding is one of the more interesting concepts behind LLMs but most pages treat it as a side note. How does embedding treat tokens that can have vastly different meanings in different contexts - if the word "bank" were a single token, for example, how does embedding account for the fact that it can mean river bank or money bank? Do the elements of the vector point in both directions? And how exactly does embedding interact with the training and inference processes - does inference generate updated embeddings at any point or are they fixed at training time?

(Training vs inference time is another thing explanations are usually frustrating vague on)

Too many popups