futurisold

6 ms·

31.
by futurisold 1y ago
Btw, besides the prompt, the other input to the technical report (the gdrive link) was this repo: https://github.com/elder-plinius/CL4R1T4S/tree/main
32.
by futurisold 1y ago
Yes, pretty much. We wanted it be functional from the start. Even low level, everything's functional (it's even called functional.py/core.py). We're using decorators everywhere. This helped a lot with refactoring, extend
33.
by futurisold 1y ago
Wanted to do just that, thank you
34.
by futurisold 1y ago
I didn't expect this -- I was supposed to be sleeping now, but I guess I'll chat with whoever jumps in! Good thing I've got some white nights experience.
35.
by futurisold 1y ago
That's gonna be a very, very, long answer. What's funny is that not much changed since 2022 (eoy) when the projected started; the models just got better, but we had a good chunk of the primitives since gpt-3. What's more rece
36.
SymbolicAI: A neuro-symbolic perspective on LLMs(github.com)
224 points by futurisold 1y ago|62 comments
37.
"Why SymbolicAI" Is the Wrong Question(futurisold.github.io)
4 points by futurisold 1y ago|0 comments
38.
We just compressed months of math research into 24 hours(twitter.com)
1 points by futurisold 1y ago|0 comments
39.
Large Language Models Can Self-Improve at Web Agent Tasks(arxiv.org)
1 points by futurisold 2y ago|0 comments
40.
by futurisold 3y ago
You might want to give a read to "Scaling Data-Constrained Language Models" [1]. They basically generalized the Chinchilla scaling law by investigating behavior on multi-epoch runs. [1] https://arxiv.org/abs/2
41.
by futurisold 3y ago
Thank you! We have integration with graphistry, their tool is amazing for dealing with graphs.
42.
by futurisold 3y ago
Right. Even though we provide support for both OS models and custom ones, this part of the framework lacks proper maintenance. In this case less was definitely NOT more, and it's something we're aware of. To briefly present our cu
43.
Show HN: SymbolicAI(github.com)
26 points by futurisold 3y ago|5 comments
44.
SymbolicAI – Retrieval Augmented Generation, Multimodal Inputs, User Packages [video](youtube.com)
3 points by futurisold 3y ago|0 comments
45.
by futurisold 3y ago
What a champ.
46.
by futurisold 3y ago
Great article!
47.
by futurisold 3y ago
> There is enough misinformation and woo surrounding QM already. There is absolutely no reason to entertain the notion that there is some deep connection between QM and any macroscopic phenomenon simply because both involve a wave equati
48.
by futurisold 3y ago
There's nautil.us too.
49.
by futurisold 3y ago
Very insightful piece. Excellent foresight.
50.
by futurisold 3y ago
I also don't believe that "Willpower needed to make decisions is a limited resource.", (and in the context) is one reason why todo lists are weighing one down. The only person I currently found discussing another way of looki
51.
by futurisold 3y ago
It's actually the only thing that worked for me. I have a natural skill at deconstructing complex things, and my hunch is this the main reason I never enjoyed todo lists, even though if a task is extremely complex I have to break it do