Top
Best
New

Posted by pella 18 hours ago

GLM-5.3: Frontier coding with emergent cyber capabilities(z.ai)
1010 points | 499 commentspage 6
himata4113 7 hours ago|
There goes the last argument that anthropic had. I think beyond this point we're entering the 'dark scary world' that dario predicted which in fact result in things going on as usual. Really, the amount of fear mongering is astonishing.

Hopefully they will drop it all together and focus on making models that are useful for everyone like their original mission was instead of playing games with politics.

Jacopos311 13 hours ago||
This looks very interesting indeed!
aizk 17 hours ago||
The model releases just don't stop!
cubefox 16 hours ago||
> Open Source: We will release the weights in two weeks after launch, once safety evaluation and hardening are complete.

What safety evaluation? What safety hardening? They already evaluated it and found it to be highly capable at exploiting security vulnerabilities. So we know it is not "safe", and they don't seem to plan to do anything against it. What could be more dangerous than hacking? Biological weapons research? I don't think Chinese labs are doing anything against this either.

h8hawk 1 hour ago||
Are you against open-source models?

Data and content related to "biological weapons" already exist on the internet, in books, etc. The real issue is access to facilities and tools. There are models that help researchers, but they are not LLMs, rather they are models trained specifically on biological data (like AlphaFold).

Cybersecurity is basically used like a dog whistle pioneered by Anthropic to achieve regulatory capture. Otherwise, the widespread availability of good tooling for security analysis would eliminate more of these cyber threats, rather than gatekeeping them for a few private companies.

thepasch 8 hours ago|||
I wouldn't be surprised if more resources were put into abliteration resistance the more capable open weight models become. It's something you don't need at all to start hosting the model on your own, but something you need to take care of before you release the weights (if you do care about it at all).
gpm 12 hours ago|||
I'm curious what they mean by that too... They might be trying to weaken the cyber capabilities... Or I guess they might mean safety evaluation and hardening of the open source (and perhaps closed source Chinese) software ecosystem...
alightsoul 16 hours ago||
They need to make money. Let them do it. They deserve it. Also, this is what inference engines like vLLM want to have "zero day" supporr
tw1984 17 hours ago||
just imagine the world without these open weight models - we'd probably have to reverse mortgage our homes to pay for tokens to those trillion $ companies to have access to their models.
yogthos 10 hours ago||
I'm so glad I managed to get their subscription when it was on sale for 250 bucks a year back when it was 5.1. Back then it was just ok, but after 5.2, it's become my main workhorse. And 5.3 is looking fantastic.
ofjcihen 12 hours ago||
The capabilities of open models approaching or meeting that of SOTAs is good in every way except for our short-sighted economic reliance on their success (in the US at least).
smurf9852 12 hours ago|
" a judge agent then attempts each task to verify that it is actually solvable "

I understand you need to verify the goal is achievable. But if the judge agent has the same goal as the training agent (solve), and both are of the same model, then aren't the judge and the training agent doing the exact same thing? What is the point then? Can someone explain this to me.

More comments...