Given how tech has gone from nerdy underdog to Orwellian, Machiavellian, dopamine-peddling overlord, I’m actually a bit surprised people aren’t more angry / upset.

Over the last 40 years there has been two career paths with “disproportionate leverage”: tech and finance.

I believe in capitalism, contracts and the rule of law. A founder starting a company and attracting capital, employees and customers and generating tremendous wealth I see as an opportunity, not a bug.

But if I were in a normal career the wealth generated by tech & finance would certainly look & feel like some form of cheat code.

The last forty years have been a huge transient. Massive. AI will probably push it even further.

I hope we as a society and democracy can survive the strain.

(That can of course change very quickly, yes)

For very simple geometries it works great, but it very quickly becomes apparent that there’s a bit of a disconnect between “LLM views image” and “LLM emits scad that looks like that image” when it comes to anything non-trivial.

Still gives me a starting point I can mess with, which is great since I have zero CAD training or experience.

(I’m not the commenter you replied to)

Did end up just adding them to our WAF blocklist, which is weirdly ironic - hosting on their infra & using their services to block their AI scraper...

I love the promise of nuclear energy, and I understand that every single engineering decision has tradeoffs, but these tradeoffs just seem so bad? Are there really no better options?

Tensor Spy lets you quickly inspect the contents of numpy & pytorch tensors locally (your tensors are not uploaded to any servers).

This is useful to validate your deep learning data pipelines, to check which layers in your diverging model are actually going haywire, and just because it's kind of cool & a lot more convenient for one-off inspections than loading things up in python.

If you work with diffusion models, inspecting the latent space can be quite informative: you want some "noise" in there but it should probably be fairly smooth for your LDM (Latent Diffusion Model) to be able to target it well.

Also, if you haven't looked at your data, it's probably not what you think it is ;)

Basic stats are auto-computed, and any inf/nan values are both counted and rendered with contrasting colors, to help you quickly identify issue hotspots.

The site is free, and our broad intention is to keep it that way (we run a bunch of pro-bono little utility sites in addition to our commercial ones, they're all linked on the about page).

Would love to hear your thoughts, I'm sure there are some stats or utility features we missed, so please give it a spin and let us know!

---

Agentic coding is a brave new world. Three years ago, after the initial rush of ChatGPT's launch, I commented to some friends that "we're standing on the beach and the water just receded". The tsunami is really hitting now. As in: this project took about 2 weeks, and not only would we not have done it without agentic coding, it would have taken months using "traditional methods". With agentic coding, adding .pt/.pth support was basically a single request. And it just worked. Time to adapt yet again.

Comments URL: https://news.ycombinator.com/item?id=47221645

Points: 22

# Comments: 1

I guess they get such a large input of queries that they can only realistically check and therefore use a small fraction? Though maybe they've come up with some clever trick to make use of it anyway?

The boiling / pressure water reactors all have requirements on active cooling being maintained in emergencies - I’m not familiar with this design nor to what extent the salt is intended to fulfill such a function, but it’s plausible that it could buffer things for idk 1h-3d maybe?

The holy grail is the “walk away safe” reactor, I would hope / presume all the novel / modern ones fulfill that?

Rather than using prompts (slow, and how do you even describe a font?) we've pre-generated a large catalog, and organized it into "vibe" folders for easy exploration.

For those not familiar, typical font sites show you 10-20 crippled specimens in seemingly random order, have limited & inconsistent catalogs, are slow to load, and use complicated licensing terms that make it hard to know what you're allowed to do.

Font Hero solves these problems: hundreds of instantly updating custom specimens organized by "vibe", more than a million fonts with consistent & broad character sets, blazingly fast load times, and simple licensing:

  All fonts are free for commercial use while in beta.

About the AI: we have trained a custom generative AI model from scratch using a dataset of non-copyrighted raster specimens. As with most first stabs at Gen AI, the results are on the one hand amazing (there are a lot of good-to-great fonts in the collection), but of course the error rate is non-zero, so if you dig around you'll soon find some "six fingered" glyphs too ;)

Please check it out, and we'd love to hear your thoughts!

Comments URL: https://news.ycombinator.com/item?id=45609080

Points: 3

# Comments: 0

Great for conferences and networking events.

Free, works on any modern phone, no app installation, and no frustrating typos.

---

I'm experimenting with vibe coding, and literal SPAs are a really good fit: the entire page, CSS, HTML & plain vanilla javascript all fit into the AI's context, so even with "just ChatGPT" I can paste the entire thing in, ask it to make some changes, and it spits out an entire result (or sometimes diffs).

(Yes, I should try Cursor et al, all in due time...)

It's quite neat, and while this particular app is super-simple, it is still a meaningful nugget of utility (I hope) that the AI can do all the (not particularly) heavy lifting on.

Basically no more excuses when it comes to shipping micro-SPAs ;)

Thanks for checking it out, would love to hear your thoughts & comments!

Comments URL: https://news.ycombinator.com/item?id=45071537

Points: 4

# Comments: 1

(And then it can of course get derailed, but that's a separate story)

Free, works on any modern phone, no app installation, and no more messages going to the wrong number / email.

https://qr.clv.ws/

Comments URL: https://news.ycombinator.com/item?id=45015594

Points: 1

# Comments: 0

A lot of employees at successful startups & FAANG make most of their money from the stock, no? And they need to buy houses and send their kids to fancy schools too, no? So sure, we can reduce it to stock holders, but I’d bet dollars to donuts the 90% of employees who aren’t posting on hn are at least passively ok with “improving metrics”, and some ambitious ones are driving the enshittification initiatives hard.

> "What we've uncovered is that the turtles are actually swimming,"

> the trick was developing flexible solar-powered tags that could hang onto shells long enough to send back data.

> Eventually the GPS tags slough off because "the outside of a young turtle's shell sheds as they grow very quickly,"