dbish

18 ms·

241.
by dbish 3y ago
It’s very model and hardware dependent
242.
by dbish 3y ago
Don’t you think if this were true there would be more breakout technical (and by extension startup) successes in Europe which doesn’t seem to be the case.
243.
by dbish 3y ago
Fair point, just saying they definitely won’t get people who can help with or understand SOTA at these prices unless they’re basically volunteering.
244.
by dbish 3y ago
That’s not going to get good people unless they’re doing it to help.
245.
by dbish 3y ago
Sorry but as someone who works in ML, no one is paying hundreds of people to design a machine to replace a pizza influencer, that may be one task that is possible from a far larger purpose, but it's one of quite a few. Pizza making rob
246.
by dbish 3y ago
I don't think it's egotistical, I would love to ask some questions to my great great grandparents, learn more about our family, what things were like for them back then, and more. I assume (and hope) my descendants would be intere
247.
by dbish 3y ago
"We don't need AI's impersonating humans in creative fields" <- Why? What's the harm (assuming they're not being used to copy and steal music/art/etc., but instead to creatively use and learn from
248.
by dbish 3y ago
You can post and talk about whatever you’d like and see if others agree. Others disagreeing doesn’t mean society is toxic imho
249.
by dbish 3y ago
Automation helps every day. Farms are becoming more and more automated for one, truly reducing human workload. This is so pessimistic. You can look to a million places to see how tech and automation improves every day. I don't follow w
250.
by dbish 3y ago
Anyone can have a vision, talk to people around the world in an instant, and find others who might align or be inspired by what you're doing. We live in a world where you can create a website, ship a product, post to a public forum wha
251.
by dbish 3y ago
Grand new futures have to take baby steps and actually be used along the way to get to something incredible. I don't think it's right to completely write off things like this since they are testing and proving grounds. For example
252.
by dbish 3y ago
Airlines have gotten particularly bad with staffing, delays, hidden fees, service, and more. Wish they were a better business so people would come compete properly. Best wishes for this guy for sure
253.
by dbish 3y ago
We're a new startup that has recording -> docs, not slides yet, but it's easy enough to go that direction. We just started a beta of a product that is a drop in loom replacement and “smarter” in that we both generate doc artifa
254.
by dbish 3y ago
We just started a beta of a product that is a drop in loom replacement and “smarter” in that we completely index not just what you said but what you showed on screen (OCR) and what you did (captured actions) to make everything easier to fin
255.
by dbish 3y ago
We just started a beta of a product that is a drop in loom replacement and “smarter” in that we completely index not just what you said but what you showed on screen (OCR) and what you did (captured actions) to make everything easier to fin
256.
by dbish 3y ago
Classic Microsoft naming issues
257.
by dbish 3y ago
That's still an example of better quality results that should be quantifiable, that's ranking. We have things like precision@n/NDCG@n/etc. where it should be straight forward to show a metric for some smaller n where Kag
258.
by dbish 3y ago
Agreed, and I think there is a bit of a shift to where open source is being used more frequently as a marketing technique/getting the first traction for funding and eyeballs then land and expand to nearby non-open source areas which mu
259.
by dbish 3y ago
This already happened with Lex Friedman. Longer than most podcasts, far fewer original or deep questions per episode, pretty monotone voice that would be easy to get any current TTS engine to emulate, I give it like a 30% chance.
260.
by dbish 3y ago
Natural language generation as well
261.
by dbish 3y ago
But the real question is how does it compare to the real one, do people listen to it at the same level and for the same amount of time (no). It’s neat, and I work in ML/AI, but it’s missing depth, coherency, and in some way an actual w
262.
by dbish 3y ago
Given that some actors effectively play the same character every movie or show they’re in (ex: the rock) and people still seem to mostly like them, why couldn’t you capture an actor’s “spark” (mannerisms) once and replay in different settin
263.
by dbish 3y ago
Eh, I think anything that we have a lot of data on (ex: recorded acting) will and can be replicated in the very near future. Sure, you will have the occasional Oscar worthy actor that is some level of creativity that doesn’t work, but vast
264.
by dbish 3y ago
Why not?
265.
by dbish 3y ago
Shouldn’t they start with the actors in their union then if that’s what they really care about? Some of the people protesting make crazy amounts of money themselves, if they want to show equality matters to them, should they start by redist
266.
by dbish 3y ago
They have a VC arm that just looks for and does investments which doesn’t necessarily means salesforce proper is interested in the tech (though I’m sure there is some of that), kind of like Google Ventures.
267.
by dbish 3y ago
That is the issue for sure
268.
by dbish 3y ago
Do you put artists in quotes for anyone who uses a computer for their art at all? Are photographers artists in your opinion?
269.
by dbish 3y ago
There is expectation you return value though. I think that’s what most people mean when saying “pay it back” regarding investments like this.
270.
by dbish 3y ago
If you’re hosting for your own use, unclear why you wouldn’t just do that directly on cloud instances. If you’re trying to share models with others publicly, then HF is the best option right now.

More ›