Notices by Dash 🇸🇬 (dashrandom@kopiti.am)

@mustardsauce yesterday 😂😂😂

Whoever said robots are going to replace manual workers have obviously not researched the course of these things yet

Started using Higgsfield and I can say that video editors, VFX/SFX staff, and videographers are NOT going to lose their jobs. Just like programmers, have AI makes you work faster but if you have no grasp of the basics, you're not going to benefit much.

Actors and actresses however...

@christaingoay hi Christian, sure. Tell me a bit more about yourself? Your profile doesn't say much.

The more I understand how Agentic AI works, the more I think it's going to be a flop. LLMs don't reason and creating workflows and validation loops for them sounds like a plan OpenAI or Nvidia concocted to get you to burn more tokens.

It's better to have well trained models, fine tuned or trained to specific usecases, sufficient RAG and context or well crafted system prompts. It's a lot more evident in text-to-image and image-to-video than it is for say, coding or text based work.

@eviltofu podcast on the sneakernet

What is it with this weird rumour that Singapore has a US military base on Chinese social media?

If history has taught Singapore anything, it's that we can't rely on outsiders to defend our country. Chinese social media retards are a truly special breed.

WH40K commissar, except irl

https://www.bbc.com/news/articles/cz7gw3l395ro

TIL Hyper-V is pretty crap. If you reassign a different Virtual Network Switch to a VM, restart the VM, it sometimes doesn't unassign correctly and stays on both network switches at the same time. You need to delete (??) the original Virtual Network Switch to unassign it.

I had an internal network switch assigned to a VM, reassigned it a bridged network switch and it could connect to the internet fine and all but all other services continued to run on the internal switch.

I couldn't RDP, couldn't connect via SSMS, yet the new IP address on the new virtual switch responded to pings. Spent like 4 hours troubleshooting this until I decided on a whim to delete the old Virtual Network Switch and then voila, everything magically works.

tl;Dr: microshit sucks, use Linux.

Decided to see what other things are associated with thoughtcrime.org, ran it through dnsdumpster.

And I get, mx records: alt1.aspmx.l.google.com

Moxie uses Gmail. What a giant ass fucking privacy hypocrite. No wonder there are so many claims Signal is just a honeypot and the TextSecure protocol has backdoors.

Fake ass hacker name and aura Matthew Rosenfeld.

Just learnt that the Signal package name is org.thoughtcrime.securesms. Was very interested with this contrasting package name and it's supposed privacy objectives, so I went to check on who owns the thoughtcrime.org domain name (in case you're not familiar with Android package names, they tend to use the reverse the FQDN order of the domain name providing the signing key to sign the package).

So thoughtcrime.org now redirecte to moxie.org, Moxie Marlinspike's personal page. No surprises there.

Maybe the Fallout timeline will come true after all.

@wynlim Did you do any upgrades of WordPress versions recently? WordPress uses a wp-cron.php job to run certain things but it's a fake cron scheduler, it only runs when someone visits the page. And if I'm not mistaken, one of the things it schedules is RSS feed refreshing.

If most of your visitors are using rss to get content from your page, then it would make sense that wp-cron doesn't run as often anymore and the feed isn't updated as often.

@echoz I'd say that gif is a perfect reason why projects need to be separate from operations 😂

@cyd Deepseek only censored before the API call on the official frontend, the model itself is completely uncensored. If you make a direct API call on the official API, you'll get uncensored replies.

@eviltofu Disclaimer: This is the simplified versiont that removes a lot of nuance.

Usually there are prebuilt prompts to ask it to confirm its own work. Additionally it develops its own prompts to prompt itself to investigate other aspects of the original line of reasoning. There are a few approaches to this "chain of reasoning" approach such as Mixture of Experts, Chain of Thoughts, Tree of Thoughts, etc.

But at the base, it's still an LLM that predicts the next most likely word.

@skribe yeeeeaaaaaa that long simmer is sign of a good sauce

@houfu are you using anything that would allow you to store context? I would highly recommend taking a look at OpenWebUI

The point of Batman is that anyone can be the Batman. The point of LKY is that anyone (well you know what I mean) can be the Prime Minister of Singapore.