HN user
Software developer in machine learning and natural language processing. Physics PhD.
Since gravity is the curvature of spacetime, quantizing gravity would mean quantizing spacetime (or quantizing geometry) also, which would lead to their being smallest units of space and time, perhaps somewhere around the Planck length and Planck time.
Lying to local (and state?) police is legal (I think). Lying to federal LEOs is a crime.
The cost of Rust's abstractions are largely at compile time, not at runtime. For example, see https://idursun.com/posts/rust_zero_cost_abstractions_in_act... or https://carette.xyz/posts/zero_cost_abstraction/.
This reminds me a bit of a Clojure library called Plumbing (formerly Graph): https://github.com/plumatic/plumbing. It also let you create a DAG for structured computation. It was used for a web service, at that time.
It wasn't a foreign agent being used for spying. It was software the government paid for.
Checkout Datomic. It’s a temporal database that uses datalog as it’s query language. There’s also Datascript, which does the same thing.
Research into belief perseverance shows that even when people are shown overwhelming evidence contradicting one of their beliefs that they dig in their heels and stick to the belief anyway. [1]
There's also Frank Lunz's observation that many people form opinions based solely on the emotive content of the words (Russell conjugation [2]) they're presented with, regardless of the facts. [3]
[1] https://en.wikipedia.org/wiki/Belief_perseverance [2]: https://en.wikipedia.org/wiki/Emotive_conjugation [3]: https://en.wikipedia.org/wiki/Frank_Luntz#Use_of_language
Libraries like this enable differentiable programming, which lets you backprop through more than just neural networks. For instance, people have built a differentiable raytracer and plugged a physics engine into reinforcement learning to accelerate training.
It's not quite what you're asking for, but there's a comparison to the 1050Ti: https://www.macrumors.com/2020/11/16/m1-beats-geforce-gtx-10...
Rust can use reference counting though, in addition to borrow checking.
This is true for me also. It's much easier for me to make connections between things when I have information about them in my head already.
I don’t think anyone is entirely writing off x86.
Apple seems to be.
The C extensions are partly why Python reigns in the machine learning and data science world.
About a month ago, Open AI demoed a language model that does python code generation, using a natural language prompt. It was trained on Github repos.
There were only two professors I was on a first name basis with in college. One was my advisor, I also worked in his lab for two years. The other was another professor in the department. It was different in grad school. It was first names for professors you worked with. At that point, they don't know more than you do about your research.
I'm hoping the SwiftUI updates will include a TextView that can handle attributed strings.
The current way to get a lot of efficient computation on a modern computer is to engage the GPU of said machine. This means that any language hosted on the CPU is going to be at a disadvantage.
That's only true if your calculations are mostly the kind of things that take advantage of GPUs, like working with deep neural nets or graphics. For most other things using multiple CPU cores will be better.
The author of the linked post (Karpathy) recommends learning from Stanford's CS231n class (free) instead: http://cs231n.stanford.edu/.
This seems like something that could be useful for data scientists.
It seems like the author is wrong about the weak Sapir-Whorf hypothesis not applying to programming languages.
This is anecdotal, but once I learned Clojure, some things about the way I thought about problems changed. I didn't think about things in a functional way before that. The same has happened since I started learning Rust. Now I'm thinking about ownership and lifetimes. And in C I had to manually allocate and free memory.
Perhaps the hypothesis should be amended to refer to programming language paradigms, or something along those lines.
Since TFHE does the encryption at the binary level, it looks like it would need to also work on the float level to be useful for a lot of data science/machine learning use cases. I could be wrong though.
This was shared in 2017. https://news.ycombinator.com/item?id=15249289
Meh. The new logo looks like it was made to go on toys.
I'd be surprised if it's dead. At least two of the people on the Swift for Tensorflow team were building a deep learning library in Swift before Swift for Tensorflow was a thing. Much of their work is a part of Swift for Tensorflow.
I've not heard of Pointing and Calling before, but this perhaps explains what I do to keep from forgetting my wallet or phone when leaving the house. I pat my pockets before I exit and say "wallet" or "phone" as I pat the appropriate pocket.
It looks like the main differences are the bidirectional links and the automated creation of those links.
Non-paywalled version: https://outline.com/WvZKTv
if he took his papers, put them in a stack, and put a staple in the upper left corner, then they would accept that stack as his dissertation
When I was finishing my PhD, my advisor pulled out his dissertation to show me what it looked like. It was literally his papers sandwiched between the covers.
(Edit: formatting)
This work is part of the spintronics field. There are some good youtube videos about it:
TEDxCaltech - David Awschalom - Spintronics: Abandoning Perfection for the Quantum Age https://www.youtube.com/watch?v=g9NAUPbqAvE
Some of the readme reminds me of Borges' Library of Babel [1].