thx67

Google owns the backbone. They definitely have free egress.

Prediction markets can solve this.

And free bandwdith. Free bandwidth is nice.

from the xcancel, https://xcancel.com/zai_org/status/2072349453361557898

Introducing ZCode, the official development environment for GLM-5.2

GLM Coding Plan subscribers: now 1.5x usage quota in ZCode

BYOK supported: works with your existing subscriptions and APIs

Available on macOS, Windows, and Linux

Download now: zcode.z.ai/en

It was a joke, nothing could save this "paper". I don't think the author wrote anything. They pointed claude at a directory and said "write a paper"

You obviously haven't read it, because it is clunky garbage.

> 19.4 Pacing compiles after a failure

> A failed compile is not free of side effects on the shared compile service. A compile that fails restarts the service, which takes a few seconds to come back, and failures that keep arriving faster than the service can restart between them keep it from making progress, so unrelated compiles slow down until the failures stop. The effect is a function of how fast failures arrive, not how many occur: failures spaced out past the restart interval cause no degradation at all. On detecting a failed compile, wait at least one restart interval, roughly 15 seconds, before the next compile, so a burst of failures cannot accumulate. No hard failure-count cap is needed.

The whole document is less nutritious than a wonderbread miracle whip sandwich.

This is obvious Claude slop writing, the author would be advised to use vale [1] with samples of their own writing as a guide.

> Performance begins with the roofline. On the M1 the engine holds about 12 fp16 TFLOP/s of compute against a DRAM-bandwidth ceiling. The roofline has a ridge point near 141 FLOP per byte, a 2 MB working-set threshold, a 0.23 ms floor under any single dispatch, and efficiency near 0.37 picojoules per FLOP at the compute optimum. On a 256-channel 3x3 convolution it runs about 3.8 times faster than the same chip’s GPU and 9 times more energy-efficient. The roofline pairs the engine’s throughput ceilings with its measured power.

> Reaching the engine is not the same as running an arbitrary graph on it. The operations the engine executes are distinct from the ones a capability bit only advertises. A feature attested in the hardware tables or accepted by the compiler frontend counts only once a compile-and-run confirms it, and several advertised operations, three-dimensional convolution among them, never lower to the engine at all. Weight compression on the direct path cuts bandwidth, not only stored size. On the unentitled engine, int4 lookup-table weights run about 2.37 times faster than fp16, and structured sparsity 1.55 to 1.64 times faster at 0.43 times the bytes.

https://vale.sh/

The whole world benefits when our infrastructure can stay on spec and those specs are freely available for everyone. Specs are the vaccines holding civilization together.

I can't believe the screenshot isn't running xeyes. At least it is TWM.

In the world they operated in when this started was in a big corporate environment, gatekeeping was a feature. Anyone who needed a standard could already get it for free through their companies records department.

At my first corporate job the first thing I did was checkout and read all the MPEG standards.

But I agree, the whale we need to go after is IEEE.

His page links to https://en.wikipedia.org/wiki/Nuclear_close_calls which is a harrowing thing to read. We keep rolling the dice with no changes to the game.

When you are at capacity and in a degraded state, you have no additional headroom to get out of that state. Why wounds won't heal, or the poor stay poor.

This is where the industry made and continues to make mistakes wrt autonomous driving.

They should be able to quantitatively say how many crashes were reduced, avoided and spotted. The autonomous safety system should be running all the time and it should detect not only issues with primary vehicle but it should also catalog issues it sees in other vehicles in its vicinity.

We shouldn't have gotten AD before we got automated crash avoidance.

The problem is that management witnesses the pain, but the response isn't to adjust behavior, it is then to punish the limb where the pain originated from. The reason that people pull heroics is also because the organization isn't healthy, and cannot reflect on its actions. Papering over organizational flaws is a symptom of a larger, often unseen problem. If it was healthy, someone would have already said, "hey, I think we need to work on this networking component" and it would have been looked at.

Pain propagation, to use the corpus metaphor isn't enough.