nextos

34 ms·

601.
by nextos 2y ago
The name is a bit unfortunate given that Julia's most popular ML library is called Flux. See: https://fluxml.ai . This library is quite well known, 3rd most starred project in Julia: https://juliapackages.com&#x2F
602.
by nextos 2y ago
Thanks, I meant other Lean resources. This is also interesting, though.
603.
by nextos 2y ago
Lean is quite nice, but last time I checked this was the only software verification-oriented book. Anything else worth knowing? Most work in Lean is geared towards mathematics, in contrast to Isabelle and Coq. Actually, The Hitchhiker&#x27
604.
by nextos 2y ago
Yes, the article linked to ashr is quite famous.
605.
by nextos 2y ago
If you add a multilevel structure to shrink your (generalized) linear predictors, this framework becomes incredibly powerful. There are entire statistics textbooks devoted to multilevel linear models, you can get really far with these. Shri
606.
The Hitchhiker's Guide to Logical Verification [pdf] (2023)(browncs1951x.github.io)
100 points by nextos 2y ago|14 comments
607.
Applied Regression and Causal Inference(statmodeling.stat.columbia.edu)
2 points by nextos 2y ago|0 comments
608.
by nextos 2y ago
I'd say Kevin Murphy's two-volume PML book is a wonderful outline of almost everything that concerns ML these days: https://probml.github.io/pml-book . Note there is a ton of companion code to recreate figures and
609.
by nextos 2y ago
Me too, and I'd like it to become mainstream. The major problem right now is that it doesn't have anything that is close to Torch or JAX in performance and robustness. Flux et al. are 90% there, but the last 10% requires a massi
610.
by nextos 2y ago
I suggest a simple Linux machine running a Gnome DE and learning Racket with Dr Scheme or, perhaps, Smalltalk. Alternatively, if the kid is interested in hardware, go the Arduino route. I learned BASIC and then Logo when I was 4-8. It made
611.
by nextos 2y ago
The MINIX Z100 is a popular N100 fanless desktop.
612.
by nextos 2y ago
I think you have hit the nail on the head, focusing on the wrong things. The problem is that software engineering is still, in many ways, in a pre-scientific era [1]. For example, civil engineers know how to build bridges that are good enou
613.
by nextos 2y ago
Also, x86 is much better supported than ARM by OS and software. A N100 inside a fanless case is a great setup. I hope some of these low-power Intel CPUs start popping up in laptops, where competition for Apple M is truly welcome.
614.
by nextos 2y ago
I agree, that's essentially what I said. Entry-level faculty positions and funding are much more common in the US.
615.
by nextos 2y ago
My experience involves really well funded research areas in e.g. Oxbridge, sometimes better funded than their US counterparts I have interacted with. So I don't think its only about funding, but how funding is allocated. EU funding has
616.
by nextos 2y ago
After working at several universities, and interacting with lots of institutions on both sides of the pond, I think this freedom issue is also what makes American Academia more successful than European Academia these days. In Europe, academ
617.
by nextos 2y ago
AFAIK, 0.5% or even 1% is normal in those cases.
618.
by nextos 2y ago
> the model should be consistent with knowledge about the problem [...] which I don't consider the norm in ML at all. I don't think that is so niche. Murphy's vol II, a mainstream book, starts with this quote: "Intel
619.
by nextos 2y ago
Sure, but Lean has very little support for software problems compared to Isabelle, Coq or Dafny right now. Those 3 also have a lot of training data as well. Hoping Lean gets more support as it is very friendly. As a basic learning resource
620.
by nextos 2y ago
These kind of LLMs are also very interesting for software engineering. It's just a matter of replacing Lean with something that is more oriented towards proving software properties. For example, write a formal specification of a functi
621.
by nextos 2y ago
It's a fantastic and atypical program. It's not surprising faculty there wrote and teach from this masterpiece: https://www.routledge.com/Physical-Biology-of-the-Cell/Phill...
622.
by nextos 2y ago
Check the work of David Unwin from NHS, who has reversed T2D in many patients using dietary interventions: https://www.diabetes.co.uk/blog/2015/08/dr-david-unwin-publi... This publication is a good starting p
623.
by nextos 2y ago
That is absolutely not true. Even genetic variants from your father that are not inherited by you still influence your chances of survival and how likely you are to pass your genes to an offspring. Imagine your dad has a rare genetic varian
624.
by nextos 2y ago
Simplifying a lot, once you age, your cells have accumulated lots of genetic and epigenetic defects, so they are more likely to become cancerous if you let them divide quickly. Besides, you have less lymphocytes looking for carcinogenic cel
625.
by nextos 2y ago
Yes, the trick will be to prevent too much growth and therefore the appearance of rogue mutants that may lead to e.g. osteosarcoma. Lots of organs in the body lack regeneration capabilities once you age as an evolutionary mechanism to avoid
626.
by nextos 2y ago
Yes it was a reply to https://news.ycombinator.com/item?id=41027805
627.
by nextos 2y ago
Quite plausible. Dell still sells lots of Linux machines which come with a slightly customized Ubuntu, but support is not very good. My employer bought me a $20k workstation a few years back. Hardware was 100% mainstream x86_64. It was expe
628.
by nextos 2y ago
It's also my favorite beginners resource. And free. Straightforward and short. Takes you to the interesting bits very quickly.
629.
by nextos 2y ago
Garrity's All the Math You Missed book, mentioned elsewhere in the comments, draws a nice map of subjects, along with little introductions and book recommendations. The map is good for continuous mathematics, but IMHO fails to consid
630.
by nextos 2y ago
In NixOS, you have a bootloader to load your OS. Unless you botch your bootloader, you can't paint yourself into an unbootable state. If one system configuration doesn't work, you reboot and choose the prior one before the OS begi

More ›