nextos

63 ms·

931.
by nextos 3y ago
I think it gets to a sufficient depth if you also consider the Supplementary Material and Jupyter Notebooks hosted on GitHub. But for those with no ML background, the place to start is: https://mml-book.github.io/
932.
by nextos 3y ago
SafetyNet is a good example. It makes it pretty hard to run your own OS and, at the same time, useful applications such as banking ones. We take the freedom associated with personal computers as granted, but these devices might be the excep
933.
Simple Haskell(simplehaskell.org)
3 points by nextos 3y ago|0 comments
934.
by nextos 3y ago
It can get you two major things. Docker-like reproducibility for any development project in a much more convenient way. And the ability to try lots of software without installing it. For me, the last one is a gamechanger, as I am no longe
935.
by nextos 3y ago
I meant this kind of issue: https://discourse.nixos.org/t/can-i-use-flakes-within-a-git-...
936.
by nextos 3y ago
Because Flakes are simultaneously discouraged and labeled as unfinished, yet used by many. Some aspects of Flakes are a bit contentious. For example, official adoption of Flakes will mean Nix will no longer be a mono-repository. Plus, Flake
937.
by nextos 3y ago
Less extreme, but Simon Peyton Jones never did a PhD. Yet he eventually became a Lecturer and a Professor at UCL and Glasgow, respectively: https://en.wikipedia.org/wiki/Simon_Peyton_Jones#Education
938.
CausalPy – causal inference for quasi-experiments(causalpy.readthedocs.io)
4 points by nextos 3y ago|0 comments
939.
by nextos 3y ago
Maybe, but this is something I experienced on Firefox ESR.
940.
by nextos 3y ago
> Finally there is some movement on making Google Docs better. Recently, they have also improved the UI on Google Docs and Gmail making it much less laggy. On my old NUC computer, it makes a big difference. It seems that middle-managemen
941.
by nextos 3y ago
I think the main outcome from Anglican was a great theoretical textbook: https://arxiv.org/abs/1809.10756 But it is probably too expressive, and therefore too slow, for most practical inference problems.
942.
by nextos 3y ago
Yes, that is confusing as the intro in the README says "SynJax is a neural network library for JAX structured probability distributions". However, contrary to my expectations after reading that, they do not have lots primitives to
943.
by nextos 3y ago
Also relevant, Rich Hickey listed the books that inspired him to create Clojure in his own Amazon account: https://www.amazon.com/ideas/amzn1.account.AFAABBRGIVOWVKTHP... Easier to browse list: https://www.g
944.
Rich Hickey's Clojure Bookshelf(amazon.com)
5 points by nextos 3y ago|1 comments
945.
by nextos 3y ago
Easier to browse version: https://www.goodreads.com/list/show/137472.Rich_Hickey_s_Clo...
946.
by nextos 3y ago
My understanding from benchmarks is that M1 and M2 have an excellent performance / energy consumption ratio. But in terms of raw performance, many Intel and AMD options can still deliver more. Another very interesting aspect of Apple&#
947.
Key-Locked Rank One Editing for Text-to-Image Personalization(research.nvidia.com)
1 points by nextos 3y ago|0 comments
948.
by nextos 3y ago
And patchy support for M2, which is understandable. A shame, as the basic Mac Mini M2 is really competitive in terms of pricing and energy usage for a home desktop or server. It is below $600 with an education discount.
949.
by nextos 3y ago
You can start with a review [1], and then BRML [2]. I think applications that are very focused on control tend to come from the military and thus not so well documented. [1] An introduction to factor graphs. https://ieeexplore.ie
950.
by nextos 3y ago
Message passing can lead to extremely fast maximum likelihood or variational inference. I have used it for realtime control problems. It is a very interesting domain!
951.
by nextos 3y ago
The options I suggested above are not necessarily MCMC, they are mostly message passing algorithms. ForneyLab and Gen in particular are designed for online inference.
952.
by nextos 3y ago
A probabilistic programming language can make it much easier to implement a non-linear filter. Essentially, you implement the generative model, and you get inference compiled for you. Interesting options to look at include ForneyLab.jl, Inf
953.
by nextos 3y ago
I don't see it this way. Juniors will be rarely appointed as reviewers by Nature . Double-blinding gives unknown groups and juniors the chance to publish at top venues, which are incredibly biased towards big-name universities and big
954.
by nextos 3y ago
The problem with LLMs, which are really impressive, is that they have too many parameters. This makes them fragile, as they lack regularization. IMHO, a very interesting research route is to combine small mechanistic models with big network
955.
by nextos 3y ago
GTK used to support a similar thing up to version 2? One needed to enable it with: gsettings set org.gnome.desktop.interface gtk-key-theme "Emacs" Sadly, I think support was removed. Part of the problem was that this and
956.
by nextos 3y ago
I agree with what you say but the problem is that journals such as Nature do not have blinded peer review. Reviewers know who you are. This is quite shocking to discover if you come from Math or CS.
957.
by nextos 3y ago
Sure. It's hard to make this justice on a single comment, as there are lots of applications. Basically, any scenario with small or medium size datasets where you would use generative models. PPLs are just a way to encode generative mod
958.
by nextos 3y ago
It's much more expensive to train models. Besides, compilers are not that smart yet. E.g. a HMM implemented in a PPL is far from the efficiency of hand-rolled code. For many use cases, they are still a leaky abstraction. However, in
959.
by nextos 3y ago
Very interesting, seems to support fairly new devices (e.g. Jasper Lake). Although newest ones tend to have unsupported boot or audio (e.g. Lenovo 100e Chromebook Gen 4, codename rusty): https://eupnea-linux.github.io/docs&#
960.
by nextos 3y ago
By having a math degree you could possibly leverage your knowledge to get into some niche ML areas which still hire well and might turn out to be more interesting. I'd consider physics-informed ML, and time-series modeling. Julia has a

More ›