Hacker News Reader

Comment by vedtam | Hacker News Reader

"turn -> FINAL -> schedule wake-up", this is where my excitement has faded unfortunately. Many of us are probably wondering about the same idea: bridging the gap between a reactive agent and my daily workflow or existence. But, this still feels too close to how Claude (or any other agent) runs as a process in the background (always ON), where you can use a custom channel to feed the dialog with external signals like chat, CI/CD events, whatsapp, etc.

Humans aren't scheduling a wake-up to the next thought. Ideally, a sub second agentic loop with no FINAL / wake-up, always "spinning" would get closer. I'm conscious about the waste of resources this would drag with it (because of current architectures), but exciting still.

PS. I love the take on using bash instead of Python (one less abstraction layer!) and using UNIX fundamentals as stepping stone when composing tools as agents are naturally drawn to using it on a box anyways.

View on HN

AFAIU, that's basically impossible with a typical transformer architecture. The "recursive" transforms mentioned above are indeed a little closer, but still. You could just always start a new inference after the last one finishes, of course, but that feels more like a really fast, one-sided conversation than a real approximation of the modality/lifecycle of human consciousness.

I think the main answer to this concern is that humans are absolutely scheduling wake-ups on a cognitive level -- the best example may be, y'know, sleep! But also on a moment-to-moment basis, which is especially noticable during periods of boredom.

Think of the head LLM as you, and the workers as your subconscious faculties (e.g. the part of you that knows how to ride a bike in ways that you have never had to consciously articulate). The looping part is the unconscious substrate that makes all of that possible, arguably with some room for the faculty above you (metacognition) to control what gets presented to your conscious mind and when. The vast, vast majority of input never makes it that far tho, by design.

(ETA: ...so, that means that we don't need to fundamentally change the architecture of LLMs in order to get some really scary stuff going.)

fair point. the headlong agent running on my mac mini, cleo, doesn't pause between calls to "generate next thought". for a while the implementation didn't support pausing between thoughts. a `thinker`, which is where the thinking loop lives, only had a boolean variable `self_trigger`, which caused it to recursively fire itself to run again (ie immediate generate the next thought). We added the ability for it to slow down its thinking and prep for sharing it with the world since it makes it easier to leave it running and not have it spend a few hundred dollars a day and you can still observe the same mechanism, which is the heart of what we want to get at here:

I don't think the blog post fully conveys this, but for me it's not whether there's a pause between thought generations, but instead whether the agent is designed to stop after some human-set goal is achieved to wait for more instructions

In contrast the idea here was to force the agent to pick its own next goal (even when it receives no explicit tasks from a human) -- even if that goal is implicit -- and then watch and see what it picks (or at least what it does with its time). and then figure out what sort of mechanisms we can introduce to the architecture at that outermost life/goal picking loop

Reply on news.ycombinator.com