gessha

It’s the biggest present cryptobros left us with. Before they started dabbling with diffusion models, they talked about the release of NFTs and crypto coins using the term. It’s an instant tell for me if a person is an ML insider or outsider.

Maybe they're talking about parameter count?

Second paragraph

They can enshittify their product as much as they want, it just makes the case for removing every piece of Google garbage out of my life and never using a single piece of software from them.

I’ve been going back and forth between opus and local Qwen 27B and a good number of times opus makes absolutely stupid things compared to Qwen.

For a large number of tasks you definitely don’t need a frontier model but a good harness around it.

If you don’t can’t/want to experiment with local model, have Claude implement the exact same feature with haiku/sonnet/opus and see what the differences are. You might be surprised.

What does Nvidia do with the parts they buy back? Resell to others or grind to dust?

I totally forgot about that. I guess my memory wanted them to be good around that time.

Tech at least in my eyes peaked around 2018-19. That also coincided with the first more mainstream public-aware introduction of vision and NLP systems into tech products. A few years later it all went downhill for the tech companies.

Once the bubble bursts we can finally sweep up the datacenter GPUs and jerry rig them to our home servers!

I’m on the same opinion. Why do I need to limit myself to something proprietary and in somebody else’s control when I can build and morph a tool to my needs.

The only reason I even look at frontier lab apps is for inspiration or shortcuts to my own implementation.

Using their apps feels like QA testing a trainwreck anyway with all of their UI/UX being an afterthought.

Big Tech found the wrong slider

Is it patronizing if that’s all the geopolitical forces have left the workers as a choice?

Because of scale, potential for automating harm and lack of regulation around ML models. Even Texas probably has regulation around firearms and limiting the blast radius but there aren't any in the US against ML models.

My argument is more that things haven’t changed. Some people end up posting as if this is Reddit and others would point that out. I don’t think this type of interaction will go away because there’s nothing currently that explicitly guides you about how to post here other than community feedback and if you post something egregious - mod actions. And that’s okay. Downvote it, comment on it, etc. Make the community yours.

Similar things have happened in comedy where comedians would complain: “I miss the old days when you could joke about this or that.” If you check newspapers, you will see this trend going all the way back to 1900s.

Jaded part of me thinks companies are going frontier-model-vibe-code their security into security chips everywhere, encrypted firmware everywhere, locked down APIs everywhere and it’s going to take away the fun out of it.

He bought a whole bunch of TVs and he’s testing them right now. Check out his YT channel.

It’s quite something when western companies act more like authoritarian governments than actual authoritarian governments. I just wish RAM was less expensive ;(

People complaining about orange site degradation are probably as old as the site itself.

Re 2: There’s so many gov agencies with overlapping responsibilities that I’m actually surprised you ask this. You can totally close a number of them and still have a functioning government on paper. Reality might turn out different.

As per your second question, I have no answer. Dems are so dysfunctional it’s beyond parody.

That’s because diffusion models struggle with memorizing sequences while it’s essentially what autoregressive models excel at.

I’ve tried to fine tune a diffusion model to memorize an MD5 hash and after days of hacking on it, it failed to completely memorize it. Autoregressive models of any size learn it over <1000 iterations.

Do you keep a lot of Tesla and SpaceX stocks in your portfolio?

I’m failing to see what’s so advanced about LLM usage in the big labs. If it was so advanced, they’d show it so it drives demand/pricing/IPO but I don’t see anything that local models can’t do. Businesses haven’t adopted the tech because it takes time/thought/economic pressure to innovate.

“Got another blackout in Santo Domingo. Netrunners are at it again, poking holes in the power grid!”

Does a hacker need to explain why they wired up 10 thingamajigs and made them go beep boop?

What is an American if not a temporarily foreign worker ;)

What does it mean to “genuinely” contribute to mathematics? Because the definition you use is load-bearing ;) and it might be different from that of others.

This seems like a step towards their plan for building a platform in competition of Google/Apple. Once the generative UIs are polished and more useful than individual apps, their hardware can now ship a device which circumvents the app store moat.

\s

I totally agree. I propose the opposite of what this lawsuit is fighting: let's merge Visa and Mastercard. Throw in AmEx while you're at it. Since the concepts of competition, interoperability and standards are too foreign to people and having too many types of payment processors is confusing, we need to have a single payment processor. That way we can grumble at only a single party when they jack up processing fees for editing database rows or straight up debank us. Efficiency!

\s

I believe every individual and organization should understand the tradeoffs between the two and have a backup plan. The backup is typically unnecessary assuming stable times but when incentives change, everyone should be vigilant for rug pulls. As an individual, I can calmly build a self-hosted inference stack based on EOL crypto mining accelerators but when building for clients, support and replacements becomes more important than budget and novelty.

You can contribute to the consumer rights wiki: https://consumerrights.wiki/w/Main_Page