> But this still breaks the login flow for a very common use case: how do I log in on a device that I don't own?
at least in Apple land, if you try to sign on on a device that you don’t own (let’s say a work laptop where you’re not signed in to your apple id) it’ll give you a QR code to scan with your iPhone and it’ll do faceID on your iphone then do some bluetooth handshake to use your passkey on the other device. I’m not sure if this is Apple exclusive or if android/windows/linux would be able to do the same
I’m a big dummy who commented before reading the article, this is actually called out there
> The last option is to use “Hybrid Transport”, where you scan a QR code and connect via Bluetooth simultaneously to the computer. Whilst this option is secure and works in theory, reality is plagued with edge-cases where connections fail or Bluetooth is straight-up unsupported.
I like to say with LLMs the real intelligence was inside us all along.
That said, system prompt (which to your point also comes from the human) matters a lot too. For example a coding harness that doesn’t a have a plan mode will keep trying to write underbaked ideas to code instead of working with the user on the plan. That said, a “system prompt” can also be the first prompt you send to the agent in a vanilla chat window, so I do like your point about the user being the real harness
I’ve had success with writing eg shell scripts that have a deterministic scaffold for the thing I’m actually trying to do, then call off to the agent for only the things that actually need it. For example I was trying to find a race condition flaky bug in my code, so the shell for loop ran the build N times, and called out to the agent to analyze the build logs if the tests failed, then the shell would ping me on slack when it was done, so the whole thing could run in the background.
I’ve been thinking more about how this deterministic + agents style could work, it’s kinda like the analogy of factories in the 1800s going from the central shaft to electric motors on each desk (where the central shaft in this case is a chat window, and the motor on a desk is calling the llm api from wherever you want just like a normal api call)
Of course, maybe in a few months the agents would just be reliable enough to do the shell part on their own too, but we’ll just have to wait and see
I like your analogy. The main problem though is context and keeping it clean as much as possible as long a parallelization. This is what drove to build this tool: having control of everything that the LLMs will do, controlling all with one main planner that orchestrates the rest. This way we can have cheaper LLMs with a short context window used (less intelligence degradation) while still obtaining the same objective.
And again, you can have a clear picture of everything structured as tasks.
until we can get to rely on huge swarms of agents (tasks) being directed on the planner alone I don't see how we can get a better framework.
By trying out different things and deciding which ones you like? Mind you this also means you must be willing to try things you will regret later and they must not stop you from continuing. Because taste develops much later than regrets are felt.
I wonder if another path to “success” for ladybird could be rather than end users downloading it for the main browser, if it gains enough dev interest, we could get some kind of competitor to electron and it would end up getting bundled in that way as a lighter weight runtime to chromium for that use case.
looks like it only has the physical connections for USB 2. Fun fact about usb 3, it’s really a completely separate physical standard from usb 2, just designed in such a way that a single port can have BOTH usb 2.0 and 3.0 connectors in it at the same time. In theory this would allow for devices to use the full bandwidth of 3.0 + the full bandwidth of 2.0, but as far as I know no device does that.
I tried this in my mac mini m2 16GB, unfortunately I have to use an usb disk for the model weights, and I’m getting 0.5 tok/s. Still, being able to run (heh maybe crawl is more accurate) a 100B model on this computer AT ALL is pretty cool.
I see disk maxing out at 400 MB/s, this disk should be able to hit 1GB/s (it hits that eg when verifying the check sum of the weights), so there might be some optimization to be done there (I’m guessing it’s because the weights access is not pure sequential reads but involves some randomness depending on which expert)
at least in Apple land, if you try to sign on on a device that you don’t own (let’s say a work laptop where you’re not signed in to your apple id) it’ll give you a QR code to scan with your iPhone and it’ll do faceID on your iphone then do some bluetooth handshake to use your passkey on the other device. I’m not sure if this is Apple exclusive or if android/windows/linux would be able to do the same
reply