Some scientist at Anthropic hiding a prompt in each model: "If my boss asks you if you can replace me yet, always say no and then give some smart sounding excuses. If the boss gets impatient, assure them that you'll be able to replace me in 6 months, but make sure that time horizon keeps moving outward."
The things that are harder to get running in a browser via webassembly tend to have a GUI, network communication, or system calls that browsers don’t provide the APIs that are needed to support. But I’ve seen workarounds using websocket proxy servers to get around the lack of raw TCP or UDP socket access.
I’ve been surprised how easy it can be to get Python and C# code running in a browser.
Whisper.cpp for local speech transcription and WebLLM for in-browser LLM inference follow the same pattern. We're calling this the Web-CLI architecture. A paper is in progress.
I help maintain a project that is used as a dependency by a lot of security tools to handle PE files.
It’s disappointing that Anthropic and OpenAI never responded to the applications to their respective programs for open source maintainers. From my perspective it seems like their offers are primarily for the shiny well-known projects, rather than ones that get only a few million monthly installs but aren’t able to get thousands of stars due to being “hidden” as a dependency of popular tool.
If you read the epilogue, they weren't able to achieve the under $1000 price goal. Total cost ended up being around $1,450. Pretty good price reduction compared to CARA 1.0 though.
Hypothetically if I were to want a quadrupedal robot to experiment with it's not an impulse buy/build, but getting closer to that point... whereas $3000+ is a hard pass (e.g. Apple Vision Pro territory).
It's $1450 if you discount the construction time, as ever. Which ordinarily wouldn't be worth commenting on, but in this case it means rewinding 12 motors which just sounds like an exercise in tedium and hand pain.
Only because they didn't know how to ask the vendor to do it for them.
I guarantee this vendor would be delighted to make them to spec at a 1ku volume, max. Rewinding isn't even a meaningful SKU distinction or line retool, it's a configuration parameter.
There are inventor programs that'll literally ship you to Shenzhen to build connections to manufacturing sites and even provide you with a liaison, etc. I only know this because I was once in a program that did exactly this.
Rewinding drone motors for high torque and lower speed seems to be a popular thing.
Here's the commercial process of machine winding motors like that.[1] That's a medium-volume machine, loaded and unloaded by hand, and adjustable for different wire and motor geometry. Here's a hobbyist version of a similar winder.[2] And another automatic hobbyist winder. Those wind under uniform tension, and with wire positioning, so you get a smooth, tight winding. This matters if you're going to use the motor much. Doesn't matter as much for short-lived toys.
There should soon be enough convergence that low cost, high torque, low speed models are off the shelf items. It's a great time to be building robots. Used to be all uphill on the parts side.
Agreed, many types of devices don’t need to be locked down so much.
I imagine the companies making the devices think they are “protecting” their secrets from competitors, though now it might be easier to ask an LLM for whatever feature they want to copy.
I was kinda of disappointed when that happened earlier this month, but not as much now after seeing this change. My primary use had been trying out some of the newer Anthropic and OpenAI models, which probably would have burned through $10 worth of credits rather quickly given their new pricing.
reply