And a group has published an independent working implementation today, nice to see:

https://github.com/tonbistudio/turboquant-pytorch

It has a lot clearer explanation of the method than Google's own post.

Well, yeah. Claude simplified it. That doesn't mean it's a better explanation.

Did it lose important detail?

Good post but link at the end is broken.

“”” For the full technical explanation with equations, proofs, and PyTorch pseudocode, see the companion post: TurboQuant: Near-Optimal Vector Quantization Without Looking at Your Data.“

Author here. Sorry still working on refining the post. Will share once the post is ready.

And the lower precision float variants.

TLDR the recorded media isn’t end-to-end encrypted and they aren’t selling it but instead using it to train their own systems. What is new here?

Ditto - completely appeals to me. I’m good starting with mail.

I’m the opposite - I much prefer a structured query language (ahem) for this type of thing. If I’m looking at someone’s (ie my own 6 months later) script I much prefer to see the explicit structure being queried vs “why are we feeling for foo or grabbing the 5th field based on squashed spaces as the separater”.

Nice use of CEL too. Neat all around.

Yes only on that page, not the rest of his blog. Guessing he ansible’d it to redirect ;)

Neat idea à la regex golf.

But doesn’t seem to do enough shell escaping or correctly. Also seems underspecified, ie “find 5 lines starting with ‘the” doesn’t require a pipe to head -5.

> find 5 lines

Especially since the previous two questions used head/tail. IMO, the wording would be better as "find all the lines" since that's what the command does.

I get what your saying but I think it’s not the best way to describe it - “GPL is property”? Hardly - it’s a societal common good that can be used by anyone interested in helping that common good.

Are parks “proprietary”? I can’t run my car dealership from one, so it’s …proprietary? No. So using the terminology of “proprietary” doesn’t do justice to what it actually is.

The phrasing is a little awkward but I like the sentiment: gpl software is owned by the public/humanity/the commons/etc in the same way something like the grand canyon should be.

Public property is different from private property, which is different from personal property.

Lol I was hoping for that too