Does that not prove this test should be well burried by now?

Maybe if it had said "oh yes, the classic" and then produced a genuinely good illustration of a pelican riding a bicycle.

For what it’s worth, I enjoy the pelican test. I don’t care if it’s scientific. It’s a fun thing I look forward to when a new model drops.

The cost there is multiple rounds of review tokens making it both slow and expensive.

There are many models that are neither slow nor expensive that are suitable for targeted review tasks.

These things are still very far from human-level intelligence. Maybe a touch beyond a golden retriever in processing power. Ratcheting the intelligence up a notch from there is expensive. It’s a lot cheaper to simplify the problem it’s solving instead.

Man-computer symbiosis is the void: mechanically extended man or AI.

Distilling LLMs are a reversal of that.

But isn't it running basically 1 request at a time? This would make agentic coding difficult right? Compared to running as many sequential tests as you want via api?

Yea I don’t use sub-agent style workflows. I often use planning patterns and generate markdown for really complex tasks, but never got into the sub-agent thing. I am likely closer to AI-assisted, though I am heavily using pi coding agent and my editor is basically just for viewing files and changes now.

wait until you hear why, and that you can optionally use e2e but lose other functionality.

And end-to-end encryption not supported on web client and doesn't sync properly between devices. Useless.

>> doesn't sync properly between devices

which is good, actually. We don't want a way through which a server could tell a new device all the information it needs to decrypt messags

I thought this was a long solved problem: the server syncs encrypted data, and the user provides the decryption key from another device (via QR codes, BLE, …)

Make it device-to-device then

yeah, it is good! I hate usability features too!

You either have E2EE or you don't. If you have usability, you don't have E2EE.

I can't have E2EE and also have cross device data sync?

counter point https://github.com/GrayHatter/toxcore/tree/multi-device (I guess technically you'd need https://github.com/uTox/uTox/tree/mdev too)

Because the two sets of skills aren't related. Library and design work are not the same thing. That's why

It's really that weak?

worked for me.

and yet 5 more popped up.

Deleting them as they appear.

roughly ~50–56GB, although this is somewhat configurable with iogpu.wired_limit_mb. By default, macOS reserves ~25% of memory for the system.

There are none. People forget the other side wanted to shut down research entirely, not just release. No idea why people think the other side would have been any better, it would have been even worse. On top of that, anthropic got exactly what they wanted.