ninjin

58 ms·

601.
by ninjin 8y ago
As a Type 1 diabetic in the UK I feel inclined to agree and disagree. Yes, the drug price negotiation is great and keeps my insulin affordable. However, the NHS comes with its own ailments that we should admit. For Type 1 diabetics in parti
602.
by ninjin 8y ago
I would have been inclined to give the same charitable interpretation, but then I stumbled upon this Arstechnica article [1] from the time of the sale with a quoted statement from the Stardock CEO Brad Wardell: “We won't be making any
603.
by ninjin 8y ago
This looks really really nice, do you have any plan to add relational annotation support as well? Asking since a student is about to embark on a project that would require it. In hindsight, I regret that we did not get around to implementin
604.
by ninjin 8y ago
This certainly is what I would call a “neat hack”. Out of curiosity I had to check what it rotates Japanese into. Turns out, mostly Korean: “日本語はどうかな？” becomes “ື걅갿개갡걀等”.
605.
by ninjin 8y ago
The code in question: https://cvsweb.openbsd.org/src/usr.bin/nc/ Unless I am mistaken, it was not linked in the article.
606.
by ninjin 8y ago
Bangs are truly amazing, my only disagreement would be in that I use !sp instead of !g as a fallback since they have a better privacy record. Other favourites include !m and !osm, but I should probably read up on even more of them.
607.
by ninjin 8y ago
For those unfamiliar with this reference to the idea of “diffusion of responsibility” [1], this is fascinating stuff. [1]: https://en.wikipedia.org/wiki/Execution_by_firing_squad#Blan...
608.
by ninjin 8y ago
Okay, with this comment your comment on preferring C++(11?) and pybind11 [1] I think I am finally getting your angle. Let’s see if I can bridge the perspectives. If I am reading you correctly, you are a library builder, an infrastructure cr
609.
by ninjin 8y ago
> I think that the ML community's decision to write core operations in C or C++ and provide Python wrappers is the way to go about it if you want the flexibility of scripting. Isn’t Google’s “Swift for TensorFlow” move in direct opp
610.
by ninjin 8y ago
I wholeheartedly agree, one could also note how this is Java-esque. My hope is that someone far more well-versed than me in this area will find the time and push for it. On that note, one wonderful thing about Julia is how easy it is to con
611.
by ninjin 8y ago
Compilation on the target machine, in my case. Performance trumps portability for me, but I can totally see why this is not the case for everyone.
612.
by ninjin 8y ago
Indeed, which comes with its own advantages and disadvantages. As someone that has to always care about performance, Machine Learning, I am very happy not to have to worry about dependencies not being compiled to take advantage of the lates
613.
by ninjin 8y ago
It certainly matters because the whole Machine Learning field was pretty much 100% open access before DeepMind “went big” and started pursuing Nature publications from the Atari paper and onwards. I completely understand their move as Natur
614.
by ninjin 8y ago
I hate to misrepresent facts, but after looking over what I wrote and reading what I linked for the first time in many years I don’t see any misrepresentation of the issue. Yes, what you write adds additional detail, but nowhere do I or wha
615.
by ninjin 9y ago
Regarding 1., not just ugly workarounds, remember how Debian broke their OpenSSL PRNG for several years to silence warnings for analysis tools [1] – read the impact section and shiver if you had not heard of this one before. There are real
616.
by ninjin 9y ago
First of all, I am deeply involved with Julia, so be wary of any biases as I try to stay objective. Lattner et al. outright says that "[We] picked Swift over Julia because Swift has a much larger community, is syntactically closer to P
617.
by ninjin 9y ago
Indeed, “Grave of the Fireflies” was the first ever, and so far only, movie that has brought me to tears. If anyone ever questions if an animated movie can ever convey emotions like “real” movies, think of this masterpiece.
618.
by ninjin 9y ago
I never watched Louis CK, but I managed to dig up a reference for that quote [1]. Regarding Foxconn, @evanpw is right though in that the suicide rates were not significantly different from those in the rest of the country. There is no denyi
619.
by ninjin 9y ago
Is there a good illustration of this method? I have tried to interpret it several times over the years and never experienced any benefit.
620.
by ninjin 9y ago
I would recommend David P. Sanders talks at JuliaCon 2015 and 2016. https://www.youtube.com/watch?v=gQ1y5NUD_RI&index=3&list=PLP... http://juliacon.org/2016/workshops.html#invitation Personally
621.
by ninjin 9y ago
One of my students tried something along these lines for Natural Language Inference (NLI) last year. [1] The results where not conclusive, but perhaps Machine Translation is a better target? My reason for believing this is that the specif
622.
by ninjin 10y ago
Haven't read the work, but it appears that Project Gutenberg has a freely available version. http://www.gutenberg.org/ebooks/71
623.
by ninjin 10y ago
While Python does the trick just fine, I find how Julia handles "shelling out" and IO with the shell to be very pleasing. I am probably still a bit ignorant in terms of what is out there, so I am reading the comments with great i
624.
by ninjin 10y ago
Todai has a strong traditional science profile and traditionally science deals mainly with calculus and statistics. This leads to less of an emphasis on algebra for these exams since they are based on the undergraduate curriculum. It is r
625.
by ninjin 10y ago
Japanese and English are the only languages for the exams. Unless the curriculum has changed radically I would not recommend doing a MSc without already being proficient in Japanese. It would be far better to enter as an undergraduate on
626.
by ninjin 10y ago
> Wait, so you didn't have to get them all right? I am not sure when this would even be desirable. If a student fails a question regarding a definition (simple, memorisation) and succeeds in proving the complexity of a novel algori
627.
by ninjin 10y ago
The general structure of the problems is that they start out with easy questions and get progressively more difficult. If you don't get a single "last question" right, the committee will most likely tear you to pieces. At l
628.
by ninjin 10y ago
Goodness me, this brings back memories having taken the entrance exam back in 2009. I am glad that they have removed the "We do not guarantee the accuracy of the English translation of the original Japanese version of the exam" c
629.
by ninjin 10y ago
Thank you for the link and I agree with your scepticism towards the analogy tasks. I have had a beef with intrinsic evaluation for years, but never found the time to pursue this line of research. However, there is a consensus at this poin
630.
by ninjin 10y ago
I agree, you don't deserve downvotes for what you said. Your concerns are legitimate, it is just that quantifying the impact of changes in data domain is difficult. My experience is that my own vectors trained only on Wikipedia has p

More ›