Some of the techniques improve over linear scaling of the baseline models. For example, from the article:

> Conditional computation avoids applying all model parameters to all tokens from the input sequence. [CoLT5] applies heavy computations only to the most important tokens and processes the rest of the tokens with a lighter version of layers. It will speed up both training and inference.

[CoLT5]: https://arxiv.org/abs/2303.094752

Magic Leap 2 has a segmented dimmer that works fairly well. It's a small probably-LCD panel that sits in front of your eye. It lets the headset black out part of your view, leaving a kinda blurry shadow around objects.

I ran into this the other day, and it took me a moment to realise that I could just use `matches!`:

    // want to do this
    if not let Some("pattern") = val {
        doSomething();
    }

    // can instead do
    if !matches!(val, Some("pattern")) {
        doSomething();
    }

That won’t allow extracting a value from the optional though, will it? i.e. I can’t do this something along the lines of:

    if !matches!(val, Some(let x))

Indeed not, so it's not a full replacement. I'm not entirely sure that I like the idea of an "if let" variant creating bindings that escape the visual block, but I'd probably get over it.

I ran into this the other day, and it took me a moment to realise that I could just use `matches!`:

    // want to do this
    if not let Some("pattern") = val {
        doSomething();
    }

    // can instead do
    if !matches!(val, Some("pattern")) {
        doSomething();
    }

edit: formatting

Here is one (recently created) that operates on the whole program rather than a single function: https://crates.io/crates/no-panics-whatsoever