RestlessAPI's comments

On: Muse Spark: Scaling towards personal superintelligence

[RestlessAPI]

Token cost really matters here. I want too know what API pricing is. As we see, this model is like 85% as good as the frontier models? What if its priced at $0.2 in / $0.5 out Mtok? All of a sudden, this model is A LOT more appealing to me.

On: OpenAI’s Windsurf deal is off, and Windsurf’s CEO is going to Google

[RestlessAPI]

I use Windsurf so I remain in the driver's seat. Using AI coding tools too much feels like brain rot where I can't think sharply anymore. Having auto complete guess my next edit as I'm typing is great because I still retain all the control over the code base. There's never any blocks of code that I can't be bothered to look at, because I wrote everything still.

On: Do You Even Need Antivirus Software in 2024?

[RestlessAPI]

Antivirus is useful for those who are tech illiterate, like those who will click on any link and install whatever it is that comes out of it.

Antivirus is likely not useful for the vast majority of users that regularly browse HN.

On: Ask HN: Alternative to Mint.com?

[RestlessAPI]

+1 for Quicken Simplifi. YNAB is more of a methodology, and you pay for the UI. Simplifi is more of tool that lets you look at your money as you see fit.

On: Spotting LLMs with Binoculars: Zero-Shot Detection of Machine-Generated Text

[RestlessAPI]

In the same way that giving a graphing calculator to students wont make them ace math classes, giving a student an English calculator (GPT4) wont make them ace anything with writing. Laziness will always yield poor results.

On: Signs that it's time to leave a company

[RestlessAPI]

So privately held companies? """Shareholders"""" are the worst...

[marcusb → RestlessAPI]

Privately held companies have shareholders, too, and some of the inter-shareholder drama at a small company can impact your day-to-day life as an engineer more than at a BigCorp.

On: Things are about to get worse for generative AI

[RestlessAPI]

Such a thing happened with DALLE, Midjourney, and Stable Diffusion.

Stable Diffusion, when used to its fullest with thing like Control Net and LoRAs, blows the pants off of other proprietary models.

On: DRµGS: Deep Random Micro-Glitch Sampling

[RestlessAPI]

I wonder if this works for Stable Diffusion as well...

[acheong08 → RestlessAPI]

From what I understand, Stable Diffusion starts out with random noise that is then step by step made into the final image. Randomness is there at the beginning.

Current LLMs first find the n most likely next token and only then gets randomness on the choice among the top-n. This injects randomness to the initial search for the top-n.

So I think, from my ignorance, that Stable Diffusion is already doing this in some sense

[samus → RestlessAPI]

There are transformer-based approaches to vision models. It should work for those.

On: Ask HN: Is there a low cost way to learn real K8s, after exhausting minikube?

[RestlessAPI]

Is a "lot" literally a unit of 6 computers (or however many)? Is this saying I can buy six computers for $170?

On: Do you avoid the news? You’re in growing company

[RestlessAPI]

Cable TV and Internet are tools. They are technologies. They are not people.

News companies, specifically news company executives looking to sell as much ad revenue as possible, and are willing to produce any content that gets viewership, are to blame.