pama

36 ms·

601.
by pama 3y ago
Thanks for keeping this project simple and clean in a single file of Python.
602.
Hugging Face, the GitHub of AI, hosted code that backdoored user devices(arstechnica.com)
3 points by pama 3y ago|0 comments
603.
by pama 3y ago
If you haven’t used Emacs ESS it may be hard to explain what you’re missing with a true REPL, but if you had used it in the past and add tidyverse to it, you basically have super smooth interactive editing with the ability to quickly pool t
604.
by pama 3y ago
My main use case is making high quality visualizations for quick data exploration and sharing with a team. It is easy to guarantee that fonts are large enough, style is minimalist and clean, and filtering, transforming views or facets iter
605.
by pama 3y ago
Thanks. I don’t seem to have the ability to change the link — maybe the curator can fix it.
606.
America's ten-year-old fentanyl epidemic is still getting worse(economist.com)
47 points by pama 3y ago|34 comments
607.
by pama 3y ago
It is very nice to see the possibility of self deployment. Does anyone have experience with self deployment of such a large model in a company setting?
608.
by pama 3y ago
Also, it seems like a huge miss when things AI researchers consider essential, like pytorch or CUDA are not even listed (yet?)
609.
by pama 3y ago
A dynamic gate is a pretty distinct feature from previous SSM architectures in my opinion. In a sense, the overall fundamental architecture of mamba is still that of the transformer but with attention replaced by an SSM with dynamic gating
610.
by pama 3y ago
Physics is very broad and encompasses concepts at all scales (from subatomic to cosmological scales) and all levels of complexity——“Complex systems” is a branch of physics. Economics is a form of applied philosophy that is not yet mature en
611.
by pama 3y ago
From the article: “To be honest, after typing all these numbers, 10 MB doesn’t even feel that big or special. Seems like shipping 10 MB of code is normal now. If we assume that the average code line is about 65 characters, that would mean w
612.
by pama 3y ago
It also doesn’t sound bad if one calculates that 2.4% annual inflation would exceed 100% over 30 years, and one considers that below about 2% is often harmful to the economy.
613.
by pama 3y ago
Why has the US seen so little inflation in the last 30 years?
614.
by pama 3y ago
Not sure how much detail you need but generally there exist implicit and explicit integrators for numerically solving (integrating) ODE. The implicit ones, like the one used here, tend to be more stable. The ideas behind SSM come from contr
615.
by pama 3y ago
Even if this rule technically seems benign, together with the forced change it encourages users to game the system leading to predictable patterns, eg adding a rotating letter or digit combo at the end of a same password.
616.
by pama 3y ago
I referee for a lot of the top machine learning conferences and yes I am very optimistic about AI and its impact on humanity. The amount of exciting new papers in machine learning and AI was definitely on an exponential rise for a decade si
617.
by pama 3y ago
Every day now there are new AI models especially LLMs, which might warrant some consideration from a wide part of the human population. In a couple years we will have multiple new announcements per hour and we might need some earlier model
618.
by pama 3y ago
Thanks.
619.
by pama 3y ago
I wish they put out the report already. Has anyone else published a preprint combining ideas similar to diffusion transformers and flow matching?
620.
by pama 3y ago
I am not sure about the design choices of the pandas library or of R, but generally, many programming languages limit the number of processes they spawn during evaluation. Moreover, designing a pipeline that produces a complete output befor
621.
by pama 3y ago
Will these soon be available on lmsys for human comparison against other models? Can they run with llama.cpp?
622.
by pama 3y ago
This paper was interesting at the time but the main sentence in it was wrong: “Thus, emergent abilities cannot be predicted simply by extrapolating the performance of smaller models.” Such behaviors can totally be predicted if one looks at
623.
by pama 3y ago
FYI, I only see a repeating animation and nothing else in my iPhone on lockdown mode, with Safari or Firefox.
624.
by pama 3y ago
Of course non of the paper associations talks about causality, but lack of caffeine is probably a worse overall health association per this paper. More people don’t drink coffee at all than have more than 6 cups a day. From the paper: “The
625.
by pama 3y ago
From limited experimentation, and within the confines of a single prompt (rather than a full chat), this model seems reasonably interesting. Does anyone have good examples of QA that showcase the capabilities compared to other advanced mode
626.
by pama 3y ago
The author links to read only advice a couple of lines down. https://www.dzombak.com/blog/2021/11/Reducing-SD-Card-Wear-o...
627.
by pama 3y ago
I would much rather pay to generate my own realistic videos based on my prompts than watch other people’s random creations (possibly filled with ads). When generation becomes great the motivation and need to store, retrieve and serve becom
628.
by pama 3y ago
The retrospective regression on a specific dataset might discover a true correlated quantity, if any true correlated quantities were there and their signal was more prominent than the combinations you get from the noise. However, this analy
629.
by pama 3y ago
Yes it does. Think of typical notions of statistical significance when testing one new idea prospectively, say the concept of a p-value, or the AUC used in the paper. Now think instead of a rich dataset and you are free to fish for any of
630.
by pama 3y ago
Does this method build assumptions about the distribution of the evaluation dataset and make the bit-level reproduction of an evaluation unlikely?

More ›