tbalsam

12 ms·

91.
by tbalsam 3y ago
They aren't the government, silly billy. Just because it's written down doesn't mean that it has value, it's just an (effectively unfortunate) deterrent, since oftentimes a court has to decide that it's illegal. Ho
92.
by tbalsam 3y ago
I keep misreading all of these comments.
93.
by tbalsam 3y ago
Yes, this is partially due, among other reasons, to the fact that inference induces a domain shift compared to training due to the use of teacher forcing.
94.
by tbalsam 3y ago
HF is okay for off the shelf prototypes and very quick, multi-hour hacks, and I greatly appreciate their contributions to the community, but unfortunately in my (and sadly, many other people's experiences), their code is a continuum of
95.
by tbalsam 3y ago
I think you're missing a crucial difference here: > ... write 20k LOC of complex code ... Vs the all important: > ... write _and maintain_ 20k LOC of complex code ... The answer to the latter, for me, is a hard no.
96.
by tbalsam 3y ago
A lot of attention window stuff is fluff IMPE, just gotta look at the benchmarks regardless of what the raw numbers sat.
97.
by tbalsam 3y ago
Thank you confused_boner, I agree that this will be a very impactful update for our future.
98.
by tbalsam 3y ago
It's not tattling if it's well against the intended effects of the system.
99.
by tbalsam 3y ago
Apparently designed for mobile inference too, I've heard the weights on the nano model were quantized down to uint4. Will be exciting to see how all of that plays out in terms of 'LLMs on phones', going forward. People who kn
100.
by tbalsam 3y ago
Once again, it's not quite as simple as that. This debate did not start just last year, lol, it has been going on for years. There's a reason you can't just simply slap a discriminator on things and get away scot free. It&#x2
101.
by tbalsam 3y ago
> in place ._. Pain.
102.
by tbalsam 3y ago
I you're really maybe missing the simulation business w.r.t. computing, etc. It's like cells in a dish debating the world outside them without any context of it. It's pretty trivial to have this reality be a 'container&#
103.
by tbalsam 3y ago
springer_what_look.gif
104.
by tbalsam 3y ago
Honestly, he has some really good takes mixed with....really bad ones. I ended up unfollowing him as the curmudgeony approach combined with some of the far out stuff was just a bit too much for me. Doesn't really improve discourse in t
105.
by tbalsam 3y ago
Usually this is due to autism, please be kind.
106.
by tbalsam 3y ago
In hlb-CIFAR10 the MaxPooling ops are the slowest kernels now and take longer than all 7 of the convolution operations combined, if I understand correctly. Memory-bound operations seem to rather consistently be the limiting factor in my per
107.
by tbalsam 3y ago
*shot for shot, apologies
108.
by tbalsam 3y ago
I am sure it adds another layer of ironic, wry, situational comedy to that movie as many other layers of it are. When I learned about how it was basically a shot for shit remake of Zero Hour, AND the story of how and why they basically got
109.
by tbalsam 3y ago
Okay, I found the MonkeyBall adventure reference, at about 2:17:00 or so (it's so low in the mix usually you just here it when falling out, and there's this sorta terrible whining noise above it), but I remember it sticking out to
110.
by tbalsam 3y ago
I have a sinking feeling under some further investigation that it may be! I am pretty sure that I've heard it in a few places, one of which being the Disney movie Up. I'm not sure how to describe it, it is a sensorially very uniqu
111.
by tbalsam 3y ago
I'm on the spectrum, and I hate loud noises with a burning passion (due to the fact that the all of the information usually hurts like all get-out), but something ever since being a child that has gotten me extremely excited whenever I
112.
by tbalsam 3y ago
yes i do.
113.
by tbalsam 3y ago
You will get error regardless of whether it is analog or digital. Digital trades bandwidth for error correction, but this does not mean it is better as many ML algorithms are noise-tolerant and would benefit greatly from the increased throu
114.
by tbalsam 3y ago
Every "innocent until proven guilty" still has a p value.
115.
by tbalsam 3y ago
Pregnancy seems to be _the_ primary critical period for autism. It is one of the unifying threads in all of the research. I am autistic. I have/experience the fever effects, so my symptoms significantly improve when I have a fever (yes
116.
by tbalsam 3y ago
Oh, it's a vast field of corruption. Don't look around or into if you want to keep your blood pressure low. Cause, y'know. It'll cost an arm and a leg to take care of that. With your insurance.
117.
by tbalsam 3y ago
This seems more political than scientific, despite originating from scientists. I came to similar beliefs from some very personal experiences outside of the field, and it's just that -- a belief. Trying to turn something that's no
118.
by tbalsam 3y ago
I'm curious how this relates to MDL-tied regularization. I don't see any explicit reference to the L2 norm here, and I'm not too smart about graph neural networks. But it seems like overfitting is almost necessary unless some
119.
by tbalsam 3y ago
https://www.gutenberg.org/browse/categories/1
120.
by tbalsam 3y ago
I've certainly seen game engines trying to advertise much more-involved particle systems over the last few years, that is for sure. ;PPPP Which in some ways I guess makes sense, as the shift towards AI has heavily pushed GPUs almost to

More ›