random32840

6 ms·

31.
by random32840 6y ago
Computing & rendering all on the GPU without having to query back to the CPU at all seems a natural fit to me, but I don't have a deep technical understanding of it. My main thought is just that DL reduces the amount of computate y
32.
by random32840 6y ago
Heh
33.
by random32840 6y ago
If you think the stock market is overvalued, buy. If you don't, sell. Money talks, articles in the New Yorker don't.
34.
by random32840 6y ago
Yes, but it's branded to sell you products and ignores differences in each item's volatility & leverage.
35.
by random32840 6y ago
I can see deep learning shaking up dynamic cloth animation dramatically in the not too distant future. It's already happened for fluids.
36.
by random32840 6y ago
You can also use SkookumScript, although I wouldn't. Blueprints can also be compiled down to C++.
37.
by random32840 6y ago
Probably not, it's just more effort to verify you as legitimate and your baseline behaviour is closer to a robot, meaning when your account gets hijacked, it's more difficult to detect deviance.
38.
by random32840 6y ago
JS is generally very fast because these monolithic VMs have had so much money poured into them that their optimisation process is unparalleled. V8 can sometimes get performance on par with C. "Overhead" is arguable, JS may be garb
39.
by random32840 6y ago
You're right actually. I was missing the point. Yes, this would be very useful.
40.
by random32840 6y ago
Quickly introducing battle-tested data science frameworks, snippets and patterns without having to replicate them in Rust.
41.
by random32840 6y ago
Someone might have jumped that hurdle but the point is it circumvents that class of hurdle. What about the next framework which doesn't have bindings? What about leveraging existing Pytorch code that's written in Python?
42.
by random32840 6y ago
Google will more reliably return what you're looking for than any hand-built documentation system.