zwaps's comments

On: “Math 2.0” will need to value mathematical progress more holistically

[zwaps]

Gary Marcus has has a long career now of shifting goalposts

[vidarh → zwaps]

Does he have time to do anything but shift goalposts, given how fast his are moving?

[Chance-Device → vidarh]

I believe he now just leaves them in his truck and drives it a bit further each day.

On: Kolibri: A Sovereign Open-Weight Model

[zwaps]

Such a crazy change from the times of Luminous, when they published a three pager with a claim that the model is similar good as „gpt 3“ (which??) with some graphs without y axis.

Bravo team!

On: Clef: Open-weight decision models, and new RL fine-tuning platform

[zwaps]

No mention of calibration. Is it just another llm finetune?

[kflansburg → zwaps]

> Our post-training utilizes label-smoothed cross-entropy for valid schema outputs paired with a Brier loss to refine probability calibration.

On: Burnham must be clear: no slavery reparations will be paid

[zwaps]

There is museum in London called the British Museum which has zero artifacts on display from Britain and is entirely filled with stolen goods from other countries No one there finds this strange, bad, or anything that should be adressed

Brits are entirely deranged when it comes to their former empire.

[mna_ → zwaps]

>There is museum in London called the British Museum which has zero artifacts on display from Britain

That's not true at all. What's your agenda?

On: If AI coding is lowering your code quality, you're not managing quality right

[zwaps]

You are holding it wrong!

On: I built non-autoregressive decision models with RL a year ago

[zwaps]

I am sympathetic but this buries the lede, hard.

You are competitive with Jev only if you fine tune on the train dataset and calibrate per question.

As much as I dislike literally everything about typesafes behavior, they have an API model that works on any problem without fine tuning, and that is the key.

To be honest everyone who can finetune can likely finetune a BERT for a specific task and get similar results to yours. And that has been true for years

The key to Jevs success is that it works without fine tuning

On: Alan's Random Insult Generator (1999)

[zwaps]

Hey, don’t copy my daily affirmations

On: Hyper-Markdown – Markdown for knowledge graphs

[zwaps]

Thanks Claude for writing that

On: Show HN: Shared memory graph for Claude and ChatGPT, over MCP

[zwaps]

I literally struggle to read it

On: Handbook.md shows that long policy documents do not reliably govern agents

[zwaps]

If the authors are reading: Why did you let an AI model author parts like "Design Principles"? It's not good writing and it is obvious.