sbszllr

6 ms·

31.
by sbszllr 3y ago
“Think of the children” is, as usual, just to get the foot in the door. They use it as a justification, because it works. Of course CSAM is bad, shouldn’t we do everything in our power to prevent it? If you implement client-side scanning, y
32.
by sbszllr 3y ago
I don’t know CCPA laws that well, but I do the EU ones. As it stands “the right to erasure / to be forgotten” is extremely vague on this, and there doesn’t seem to be a wide precedent. In general the law is applicable to raw data recor
33.
by sbszllr 3y ago
Twitter’s real value was its “global RSS” nature. You could get short, quick updates about local news, published papers, sports events, political happenings, and more. It made it easy to source info from your followers (or their followers)
34.
by sbszllr 3y ago
The family of sigmoid functions has nice gradient properties with theoretical backing. Good starting read: https://stats.stackexchange.com/questions/162988/why-sigmoid...
35.
by sbszllr 3y ago
OP is right in that his change would make the softmax in the attention output zero if it "has nothing to add" (QuietAttention, as he said). Buuut, it's missing the forest for the trees. The goal of the last step of attention
36.
Ask HN: Why is the printer industry so scammy?
132 points by sbszllr 4y ago|208 comments
37.
by sbszllr 4y ago
Source: I work in the field. This is a current limitation, and an artifact of the data+method but not something that should be relied upon. If we do some adversary modelling, we can find two ways to work around this: 1) actively generate an
38.
by sbszllr 5y ago
Didn't read bc of the paywall so maybe someone can enlighten me. My educated guess is that <at best> UN is using homomorphic encryption for basic analysis... or they have achieved a breakthrough and have FHE /s.