komuher

9 ms·

31.
by komuher 5y ago
Man nim is such a cool language it just need some more community and better docs to get some hype.
32.
by komuher 6y ago
I love ur transformer repos, its always joy to re-implement stuff from ur code its easy to read and clean :D
33.
by komuher 6y ago
If u cant even read code dont lie xD n = length(ARGS) > 0 ? parse(Int, ARGS[1]) : 100 left = calc(101) # <------- THIS IS COLD START JITTING CALL right = -18.67 if abs(left - right) > 0.1 println(stderr,
34.
by komuher 6y ago
Wait wat, jax and also pytorch is used in a lot more areas then NN's. Jax is even consider to do better in that department in terms on performance then all of julia so wat are u talking about
35.
by komuher 6y ago
nim-lang
36.
by komuher 6y ago
Sounds like poland and hungary today :)
37.
by komuher 6y ago
Correcting false statement?
38.
by komuher 6y ago
This is the same stuff like julia community is doing. Spam language in every Python/R related topic (D guys target Nim/Rust)
39.
by komuher 6y ago
Flax/jax can :)
40.
by komuher 6y ago
This is my favourite comment about julia for like last few years +1 on that. Ecosystem is extremly poor outside very few niches and most of the Deep Learning stuff isn't even faster than python api (+C ofc.) so swaping is just usless i
41.
by komuher 6y ago
Can u elaborate on that?
42.
by komuher 6y ago
Still same as it was 2 years ago Flux + some other smaller repos.
43.
by komuher 6y ago
+1 I can fully agree its so salty i cant read it
44.
by komuher 6y ago
Not even cloese, jax.jit allow you to compute almost anything using lax.for_loops, lax.cond and other lax and jax contsturts pytorch jit does not allow that its just extra optimization for static pytorch functions.
45.
by komuher 6y ago
No GPU so ye its very bad :P
46.
by komuher 6y ago
Can they just start investing in software so folks from ml/deep learning field can just get out from nvidia monopoly -.-
47.
by komuher 6y ago
If u want to do anything related to more then very basic machine learning then NVIDIA gpu is must have
48.
by komuher 6y ago
Its without XLA and without cudnn and cuda 11.1 for ga102 lets wait for proper drivers to see full results :P
49.
by komuher 6y ago
It was hyped and probably are used in even more places (in my previous work i even wrote a full Julia based parsers etc.) But still i didn't see much advantage of writing in Julia now 2 years and even maybe 1 year ago there were few (b
50.
by komuher 6y ago
What purpose did you use Julia for? Julia was hyped like 2 years ago but right now i didn't see any new cool lib or even a blog posts and DL libs are lagging hard compared to python DL libs (torch, tf, jax etc.).
51.
by komuher 6y ago
Meh another shit-post for some clout.
52.
by komuher 6y ago
Yes they all did
53.
by komuher 6y ago
arch
54.
by komuher 6y ago
Dunno why (probably dataset) but open source Speech Recognition models are performing very poorly on real world data compared to google speech to text or azure cognitive.
55.
by komuher 7y ago
Nope
56.
by komuher 7y ago
Looking forward to S4TF 1.0 lets see how much progress they'll manage to make in next few months without Chris Lattner
57.
by komuher 7y ago
Sure (dont have twitter yet) but will post it here on hacker news in next week probably :)
58.
by komuher 7y ago
It is compiled to XLA so should be a lot faster then pure PyTorch but probably will be slower then TVM ( https://tvm.apache.org/ ) i can prepare some benchmarks in next few days if u are interested :)
59.
by komuher 7y ago
JAX is focusing more on performance and probably will be next big think for google machine learning teams like https://github.com/google/flax (most of the people hate tensorflow) Dunno about zygote future
60.
by komuher 7y ago
Sorry im not native :( I tried my best

More ›