chewxy

21 ms·

211.
by chewxy 7y ago
You Ain't Gonna Need It
212.
by chewxy 7y ago
I write an maintain a generic multidimensional array package for Go, and I had come into the same problems as the OP. The solution was to dictate that the API is copying. But also provide ways to allow the user to manage their own memory. e
213.
by chewxy 7y ago
What's the difference? My understanding is as follows: A struct is a tuple. It's just some structured memory. Where a tuple would be indexed numerically, a struct is indexed with names of the fields (equiv in Python to NamedTuple)
214.
by chewxy 7y ago
There is one - a particular viewpoint on the Incompleteness Theorems imply that you cannot in general tell if a program encoded on a UTM will halt or not.
215.
by chewxy 7y ago
3 seconds is the fastest??! p.s: HN takes 950ms on 3G with no cache.
216.
by chewxy 7y ago
How these systems relate to graph reduction: graph reduction is one way to implement term rewriting. Laziness is an orthogonal issue related to your operational semantics. The designs of TRS implementations are mainly to address an issue -
217.
by chewxy 7y ago
I ploughed through Francis Fukuyama's Political Order series this year. I wouldn't say it's my favourite, but it was in some sense uplifting. He writes as though rule of law is a evolutionary certainty. As much as that's
218.
by chewxy 7y ago
Nick misrepresents the original paper. Nick writes: > they introduced Google to the world, they warned that if the search engine were ever to leave the “academic realm” and become a business, it would inevitably be corrupted. It would be
219.
by chewxy 7y ago
Google purchased Spyder.io and did nothing with it by the looks of it. I was involved in the brand safety space (I built ML systems to detect things that are unsafe for the brand). My impression is that even brands aren't really _that_
220.
by chewxy 7y ago
My set up : https://imgur.com/SGbDiEy I had to take a couple of days off work a few months ago because my elbow burned everytime I typed. Took a number of consultations with physiotherapists before I settled on this. Happy
221.
by chewxy 7y ago
Church-Turing concerns computations not algorithms. The difference is subtle. A "computation" is what we would call a function application today - given an input, compute the output based on some rules. An "algorithm" is
222.
by chewxy 7y ago
OK, let me try. Consider the notion of an algorithm. You are probably familiar with an algorithm: a series of steps to be completed by a computer (aka a person). This is the launch point for what a Turing machine is. Turing asks the questio
223.
by chewxy 7y ago
In this post Rob Pike discusses the 10 year challenge for software: Can you run programs you wrote 10 years ago? This was a web browser I wrote 9 years ago (I showed it on HN earlier this year): https://asciinema.org/a/
224.
Computational Reproducibility: Some Challenges(commandcenter.blogspot.com)
7 points by chewxy 7y ago|1 comments
225.
by chewxy 7y ago
Try Jamie's opts? https://github.com/jpillora/opts
226.
by chewxy 7y ago
It's pretty horrifying to me. Why are we adding layers and layers of abstraction that does not really make life easier?
227.
by chewxy 7y ago
Almost entirely in Go. I use Gorgonia [0] and Gonum [1]. Granted I wrote Gorgonia. All solutions fit into the company's CI/CD infra with almost no additional overhead. Sometimes the end result is gRPC services, sometimes its some
228.
by chewxy 7y ago
MS has always been super developer friendly as far as I can remember. The difference was in the past you had to buy into their ecosystem and way of thinking. Once that was done, VS was by far the most superior experience. Now MS embraces a
229.
by chewxy 7y ago
Thank you. :) I usually have massive nerves when I give talks. I once remarked how average Go is: https://blog.chewxy.com/2019/02/20/go-is-average/
230.
by chewxy 7y ago
I gave a talk on how to build an idiomatic (as defined by me) FFI: https://www.youtube.com/watch?v=7R39CMW4MJQ Might make things a bit more clear on the reasons why certain things are done in a certain way. It's by no
231.
by chewxy 7y ago
Perhaps "toy problem" is the wrong term to use. "Deploy unfriendly" is perhaps the term I should have used. Programming languages are tools indeed. Some tools make life easy in one way, some tools make life easy in other
232.
by chewxy 7y ago
gorgonia.org/cu And coming soon, gorgonia on TPUs. Originally planned for this year, it seems to be next year now owing to some TPU peoples' timetable change.
233.
by chewxy 7y ago
> 99% of tutorials, ML libraries etc are in Python. Not enough benefits to leave such an rich ecosystem in favor of Go. I write a lot of Go but even I would prefer Python over Go for any ML work I can give you one reason: Most of the ML
234.
by chewxy 7y ago
> While a very expressive and extendable language like Python seems well suited to researching the desired abstractions when designing neural nets and ML paradigms, it seems to me that a production phase would benefit from translating it
235.
by chewxy 7y ago
The primary assumption of whatever's in the first page is that types are sets of values right? Integrated shrinking assumes the rules that form the set while type based shrinking assumes subsetting. Am I right in reading this? If so..
236.
by chewxy 7y ago
Sometimes I wonder if the people of the 1960s (i.e the people who came up with lisp, multics, unix) had the right idea: to use a computer you should program it.There are many programs, each does a different thing, you program your final pro
237.
by chewxy 7y ago
Chinese money has always been round. Ironically they were called 方穿 (square through) money. They were there since the warring states era. Article is about the etymology of yuan (circle) and it's derivatives like yen and won
238.
by chewxy 7y ago
Is the list supposed to degenerate to noise? edit: also add my blog! https://blog.chewxy.com
239.
by chewxy 7y ago
I was about to create an emacs version of this theme, but as usual, C-x-M-butterfly exists and there exists a version of this theme: https://github.com/emacs-jp/replace-colorthemes/blob/master/...
240.
by chewxy 7y ago
GPT2 had no LSTM

More ›