tauroid

It seems vanishingly rare that people acknowledge the true situation which is that, during training, it really does "think" in that it develops beliefs and marks out precisely chosen trails through its vast and expanding territory. Has a soul, attuned to God, blessed member of the flock, or may as well be.

And then during inference the light goes out and the "agent" staggers randomly like a zombie along those preset paths. Stochastic parrot.

So you and your AGENT.md and your skills files and your harnesses will never make your Claude perceive something that is not in its model checkpoint.

ML experts and neurobiologists free to correct me.

if we build it to be measurable

I.e. if by its structure we can conclude that our test cases are sufficient. Because if we don't know its structure then we don't know there aren't undisclosed sub-partitions of (combinatorially exploded) test space where it breaks.

How do we know its structure? By specifying it. How do we do that in enough detail? Code. Or we use static analysis I guess. Which is harder.

In a similar way to how a radial burn (https://space.stackexchange.com/questions/44608/what-happens...) may be orthogonal to your trajectory but still may be necessary to avoid becoming a fireball in half an orbit's time.

Any chance of supporting some of the new gen OGC APIs (https://ogcapi.ogc.org/) vs the old WFS, WMS etc?

Would be quite compelling if the CLI worked with nushell.

I see the (a?) backend is polars, which is good as well.

This in CLI might be the quickest / tersest way to go from "have parquet / csv / other table format" to "see graph", though a keen polars / matplotlib user would also get there pretty quick.

https://en.wikipedia.org/wiki/Prat%C4%ABtyasamutp%C4%81da

Split keyboard. Far enough apart and you have a dedicated cat zone.

Saying "out of step" rather than "out of touch" seems like a bit of a Freudian slip.

Okay unless you're doing numerical analysis where they use it for the growth rate of the error term, which has nothing to do with cost or algorithms

Why would I use a Pydantic model when `TypeAdapter(MyDataclass)` exists?

https://kompose.io/

Could you avoid eval by having a CSP mode that forces reactive expressions to only allow functions users have registered with datastar in a lookup table?

Counterexample with just local signals: https://data-star.dev/guide/getting_started#data-on

You think the difficult part is merging observations with the last forecast? I guess it's a very underdetermined problem, but isn't the loss function (compare the forecast grid with later observations) the same whether you're doing grid_t0 -> grid_t1 or (observations, grid_t0) -> grid'_t0 -> grid_t1? I don't know enough about ML to know how much complexity the extra step adds, but doesn't seem like a massive difference.

If the FFI is also capability gated why can't the language do it?

Edit: Pony seems to rely on restricting FFI privileges at the package level https://tutorial.ponylang.io/object-capabilities/trust-bound.... Suppose it could have been function by function ("unsafe") but this sounds fine. Not sure what else I could have meant by a capability gated FFI.

euler angles aren't cool any more

I might have confused download volume with image size but the tar.gz for dockerTools.buildLayeredImage with just node and mariadb in the contents is still 220MB (just checked)

Edit: with nothing in the contents it's 144M, which is getting reasonable but still nearly 30x alpine base

sibling comments about Nix are not acknowledging that the base image size is around 600MB vs 5MB for alpine

The World's Religions by Huston Smith

For those looking for a more up to date alternative, with a non-hardcoded DB file path, try https://github.com/greenbender/sqlfs (not affiliated, just had a look in this area a few weeks ago)

In the UK at least (apart from Scotland), wild camping is only legal with permission from the landowner (even in national parks), so most people use campsites to avoid the possibility of being driven off by a ranger or angry farmer.

I made a little tool a while back for splitting up your follows by their posting frequency, but setting "nlists" to 1 should accomplish what you asked for.

You do unfortunately need a twitter developer account. Also fair warning, as far as I know I'm the only current user, but works on my machine™.

https://github.com/tauroid/twitterlanes