kadushka

20 ms·

331.
by kadushka 2y ago
I clearly remember 12 year olds who were active on LiveJournal and a couple other less known social networking sites back in 2002-2003. The only difference it was done via a desktop computer rather than an iPhone.
332.
by kadushka 2y ago
These researchers are being recruited regardless of what license they put on their academic code. In fact, I really doubt anyone in the industry cares about the license for this work. It's not a patent.
333.
by kadushka 2y ago
If you can make it bigger for use as a regular monitor, I'd pay good money for it.
334.
by kadushka 2y ago
Imagine what Hinton or Lecun feel!
335.
by kadushka 2y ago
Yes, it's part AI, part human. How it should be, in my opinion.
336.
by kadushka 2y ago
If you're looking for control, check out https://www.aiva.ai It's MIDI based, and provides a piano roll interface for editing.
337.
by kadushka 2y ago
That might be because he didn't like what he discovered, or the results didn't make sense to him. But it was the intuition that got him there in the first place. But you have a good point. Ilya got us so much closer to AGI, but he
338.
by kadushka 2y ago
You have used an excellent term: AI intuition. This quality is extremely rare. Einstein probably had a similar kind of intuition in physics, and maybe that's why he was so successful. The ability to see what direction to pursue. Ilya h
339.
by kadushka 2y ago
Interesting. One other potential benefit is an easier quantization of the activations.
340.
by kadushka 2y ago
In case it's not clear - I'm willing to pay extra for this.
341.
by kadushka 2y ago
Because I'm willing to pay for it?
342.
by kadushka 2y ago
I'd prefer to pay for no-ad version.
343.
by kadushka 2y ago
No worries, we're all friends here! it's just you happened to accidentally describe the idea behind RWP, which is a gradient-free optimization method, so I thought I should point it out.
344.
by kadushka 2y ago
Change of scenery.
345.
by kadushka 2y ago
it comes down to "does increasing this parameter improve the performance on this task? Yes? Move the dial up a bit. No? Down a bit..." x 1e9 This is not how gradient based NN optimization works. What you described is called &quo
346.
by kadushka 2y ago
Top academic conferences [1] will have plenty of high level discussions. Top minds in the field and people developing next gen tech will be there. Other conferences will have a higher ratio of BS artists and people pitching their AI startup
347.
by kadushka 2y ago
There are two other problems with FPGAs: 1. They are hard to use (program). If you're a regular ML engineer, there will be a steep learning curve with Verilog/VHDL and the specifics of the chip you choose, especially if you want t
348.
by kadushka 3y ago
This explanation is better, because it puts things into perspective, but you don't seem to realize that your 1 and 2 are almost trivial compared to 3 and 4. At the heart of it are "methods of creating distributed representation of
349.
by kadushka 3y ago
I agree that the visualization could be better, but it actually seems the differences between the three groups are not that large.
350.
by kadushka 3y ago
It's not evident. I've had a fairly long career in tech (both as an IC and as a manager) and many execs I worked with didn't do much of really impactful work. Sure, they attended a lot of meetings, created spreadsheets and sl
351.
by kadushka 3y ago
Is Korea Advanced Institute of Science and Technology a commercial entity?
352.
by kadushka 3y ago
I've done this 12 years ago. Network Engineer --> Sales Engineer --> Product Manager --> ML Researcher. Though I ended up getting a PhD in ML after I quit my PM job to pursue AI. It was a different era, DL field was a lot simp
353.
by kadushka 3y ago
OpenAI is not a threat to Google OpenAI + Microsoft is a threat to Google.
354.
by kadushka 3y ago
I've read the second book (by Tunstall et al). It's excellent.
355.
by kadushka 3y ago
Before any "singularity" can happen, we first need to build AGI (human level intelligence). As we get closer to building AGI, more and more jobs will get automated. At some point, this might result in you losing your job. That sho
356.
by kadushka 3y ago
It will learn. It's called "in-context learning". This learning will be limited to the current context window, because it won't get encoded in the model parameters. In order to make it permanent, you would need to retrai