I think the raw tokens embeddings shouldn't have been included in the PCA (past the first or second graph). Since they're disjoint from the conversation embeddings, you're essentially losing a dimension and compressing all the conversations into a plane. There may well have been much more obvious structure in the language embeddings that you'd be able to see.

You wrote that other apps can't access it because of TCC Full Disk Access. Presumably the OS itself can read it, so do you have any information on whether parts of the graph (or fingerprints, or inferences made with it) are ever sent to the cloud/apple servers?

Not really relevant in the case of embedding models. We might note that compressing a lossless PNG to 99% quality JPEG will also give you a different embedding. You don't care about bit for bit equality, just the distance in embedding space. And the same text embedded on CPU vs GPU will give very very similar embeddings.

You seem to misunderstand what full screen mode is. It doesn't just make the terminal fullscreen, it causes pi to maintain its own scrollback buffer and a UI around it, rather than letting the terminal emulator handle scrolling (ie, letting the session context actually live in the terminal history)

The "why" is probably that they're getting lots of bugs about flickering and scroll jumping in different set-ups.

Structured decoding limits next-token probabilities to ensure valid JSON. The main issue is that if the model puts a substantial probability on an invalid token, then it was already confused, and in that case, you don't actually want whatever the next-most-likely valid token is: even if it's syntactically valid, it's likely semantically erroneous.

Not sure why you think everyone has moved on from this. Valid JSON is only one aspect. It almost certainly has far more uptake today than ca. Sonnet 3.7. Grammar engines can enforce a lot of other things too. Models are actually really good at emitting valid JSON, but complex schemas effectively require it.

So whats the alternative? Letting it codegen a file and the piping that? What a worse workflow

Alternative is to fail the validation and regenerate the request, but I have no idea what this person is talking about. It is absolutely not the case that everyone has moved on from structured generation. Absurd claim.

sounds like an argument in favor of structured output, is that right?