charlescurt123

6 ms·

1.
by charlescurt123 2y ago
Companies that have no interest in growth and are already heavily entrenched have no purpose for increasing output though. They will fire everyone and behave the same? On the other hand that means they are weaker if competition comes along
2.
by charlescurt123 2y ago
comparing a laptop to a A100 (312 teraFLOPS) or H100 (~1P FLOPS) server is a stretch to say the least. An M2 is according to a reddit post around 27 tflops So < 1/10 the performance of just computation. let alone the memory. What wo
3.
by charlescurt123 2y ago
I'm not proposing a change in model size; rather, I'm suggesting a higher dimensionality within the current structure. There’s an interesting paper on LLM explainability, which found that individual neurons often represent a super
4.
by charlescurt123 2y ago
I find the entire field lacking when it comes to long-horizon problems. Our current, widely used solution is to scale, but we're nowhere near achieving the horizon scales even small mammal brains can handle. Our models can have trillio
5.
by charlescurt123 2y ago
I believe they do know this. However the real challenge would be: 1. bring this mapping into a AI framework for inferencing 2. We don't know the "OS" on how it runs. Just randomly triggering a neuron probably wouldn't wo
6.
by charlescurt123 2y ago
Possibly, but as of now it's a completely unsolved problem and to my knowledge nobody has shown even a tiny model being able to perform it. Based on the top page today I may even be able to make the argument we can't even simulate
7.
by charlescurt123 2y ago
continually learn with a fixed computational budget over massively extended task horizons.
8.
by charlescurt123 2y ago
with these methods the issue is the log scale of compute. Let's say you ask it to solve fusion. It may be able to solve it but the issue is it's unverifiable WHICH was correct. So it may generate 10 Billion answers to fusion and o
9.
by charlescurt123 2y ago
It's RL so that means it's going to be great on tasks they created for training but not so much on others. Impressive but the problem with RL is that it requires knowledge of the future.
10.
by charlescurt123 2y ago
So to preface, I'm not looking for a job (trying to build my own company) When I do interviews (probably limited compared to you but some) I do it like I wish someone would interview me. I focused purely on curiosity. how many things d
11.
by charlescurt123 2y ago
Self taught, come from EE background. Was originally building firmware and hardware for previous company while testing DS on their systems. They liked my work and I switched to DS and ended up a team lead. Honestly have never had a take hom
12.
by charlescurt123 2y ago
I'm not bad at getting jobs, just feel when I am looking I believe people think I am lying about my project experience. I do these things because I see a place in the company to grow value and do it. I can write almost all basic coding
13.
by charlescurt123 2y ago
So I feel I strongly fall in a poor performer interview category any time any code problems come up. How would I convince you I do not have a fraudulent resume? I study hours every day for many years now. I know many complex systems however
14.
by charlescurt123 2y ago
I imagine we could do this now but not the way you think. have a human created story and text as a guideline. With that have genAI make the text per stage, you would get different statements every time and would stay on track. Would be inte
15.
by charlescurt123 2y ago
I think hallucinations are actually the sign that LLMs are far closer to a real brain than we realize. I think hallucinations are a major unsearched gateway to AGI.
16.
by charlescurt123 2y ago
Good question, I'm working on exactly this, I suppose you could call it the replacement of RAG. It's actually not very easy to achieve this. I could give a very long winded answer (don't tempt me) but suffice to say it's
17.
by charlescurt123 2y ago
Doing any job for more than an hour without completely forgetting it's goals and tasks
18.
The Forever AI Agent(medium.com)
2 points by charlescurt123 2y ago|0 comments