We unfortunately do live in a post-post privacy world now: so LLMs (and personal assistants based on them) with higher-than GPT-4 levels of quality that are so efficient to train and use that we can target the next generation of state-of-the-art personal computing devices. All-in on on-the-edge and power efficiency (data center energy consumption will increasingly become ridiculous; as will Nvidia’s power…).
Why does a child need so much less data for “training” to reach “adult brain”?
Apple can’t be our only hope in this regard, can it?
Do they even employ anyone who could substantially push the envelope on the software side there? I would guess they would need to headhunt people from Google and OpenAI and at completely outlandish terms (less of an issue for them; rather than convincing to leave respective AAA “AI” teams). Academia probably the only hope, I’d straight go there if I’d be Apple. It will probably need a complete overhaul of the concepts / architecture of current ways to go about LLMs. Or maybe not, maybe just needs a “small fix”? I for sure don’t know.
I’ve got two downvotes here, really curious on the why? It’s clearly not only the hardware that is “still too slow” at this point? Again just for training then, why does a LLM basically need the whole internet and then some to arrive at “uncanny valley intelligence” at best?
I didn't downvote but I imagine they came from either (a) asking for what amounts to a personalized AGI on a phone, which doesn't seem to fit the spirit of the question or (b) talking about Apple like they're significantly related to (a).
> Why does a child need so much less data for “training” to reach “adult brain”?
If we count all sensory input for years I do not think a child needs less data, but more.
If, in addition, we consider that the child gets new data (decides new interactions) based on past data, there is an optimization in the child that the LLM lacks: the LLM cannot just decide to get some new data based on the data it just saw. It is stuck with the data fed to it for training.
Out of interest, what exactly do people use Grafana for? Is it always monitoring infrastructure/ systems or is it more general purpose than that? What is special about it?
I come previously ran an analytics team and I work on an open source BI tool (https://github.com/evidence-dev/evidence) but I have never actually used grafana or come across it when talking to other "business analytics" folks. Everyone in my world is just using tableau or looker or jupyter notebooks.
Easy to configure, good looking dashboards with a lot of different integrations.
Meaning pretty much any team with basic know-how can get a monitoring dashboard going, or several for different resources and cases. It's main focus is monitoring.
Fast and featureful collection/file/library management tools. If I never have to write nested loops to rename a bunch of files ever again I would be very happy.
I take voice notes. I would really like a way for voice notes to be transcribed with timestamps. There are SaaS services that let you do that but they are for professionals and too expensive for stream of conscious personal notes.
A while ago, I did find a tutorial that would do this in Python but never got around to it.
I use https://goodsnooze.gumroad.com/l/macwhisper for this (no affiliation, just saw on HN before). IIRC there are also some open source solutions in the space compatible with Linux/win but I haven't tried them.
26 comments
[ 0.17 ms ] story [ 104 ms ] thread+ real-time collaborative version
Why does a child need so much less data for “training” to reach “adult brain”?
Apple can’t be our only hope in this regard, can it?
Do they even employ anyone who could substantially push the envelope on the software side there? I would guess they would need to headhunt people from Google and OpenAI and at completely outlandish terms (less of an issue for them; rather than convincing to leave respective AAA “AI” teams). Academia probably the only hope, I’d straight go there if I’d be Apple. It will probably need a complete overhaul of the concepts / architecture of current ways to go about LLMs. Or maybe not, maybe just needs a “small fix”? I for sure don’t know.
Your reply contains a lot of things, except an answer.
Lots of people are probably already working on that but it doesn’t exist yet.
If we count all sensory input for years I do not think a child needs less data, but more.
If, in addition, we consider that the child gets new data (decides new interactions) based on past data, there is an optimization in the child that the LLM lacks: the LLM cannot just decide to get some new data based on the data it just saw. It is stuck with the data fed to it for training.
Genode and Hurd aren't ready yet.
I'm shocked that Grafana is really the only OSS game in town for user-customizable realtime visualization.
p.s. I would love nothing more than to be proven wrong.
I come previously ran an analytics team and I work on an open source BI tool (https://github.com/evidence-dev/evidence) but I have never actually used grafana or come across it when talking to other "business analytics" folks. Everyone in my world is just using tableau or looker or jupyter notebooks.
Meaning pretty much any team with basic know-how can get a monitoring dashboard going, or several for different resources and cases. It's main focus is monitoring.
Once you have your data sources connected, it's easy to create all kind of graphs and alerts.
"Hey, don't forget me!" (displays sad_puppy_eyes.jpg)
- my wife.
A while ago, I did find a tutorial that would do this in Python but never got around to it.