axg11

44 ms·

481.
by axg11 5y ago
Your positivity and enthusiasm is infectious. Thanks for being an optimist.
482.
by axg11 5y ago
Accuracy is improving rapidly though. I agree that the current accuracy levels are not high enough to be relied upon. > Humans ... they will usually recognize total blunder I question this assumption. I don't believe this is true, e
483.
by axg11 5y ago
I don't see a difference. Large language models can also return their sources, as in the example on the Gopher blog post. This will lead to a quicker answer and equal transparency.
484.
by axg11 5y ago
The closer we get to artificial intelligence, the more we raise the bar for what qualifies as AI (as we should). Gopher/GPT-3 are already much more accurate than the average human at technical information retrieval (trivial to see from
485.
by axg11 5y ago
Apple Glasses, if they ever come to fruition, would be a textbook play for Apple. As with the iPod, iPhone, iPad, Apple Watch and even ARM laptops, they weren’t the first in the market in terms of technology but they were the first to make
486.
by axg11 5y ago
Related question: is it possible to run HN and serve HN-level traffic on any of the major cloud providers' free tier?
487.
by axg11 5y ago
In an efficient market public investors would demand to know the true/accurate DAU metrics with bots removed.
488.
by axg11 5y ago
I believe there are real estate ETFs that can easily be shorted. Many of the more complex contracts and instruments still exist, there are just more limitations on how banks can invest in them.
489.
by axg11 5y ago
While this sounds plausible, I think there are a couple of factors that work against this theory: 1) Why layoff your data science division if they are predicting with accuracy? 2a) If you have enough conviction to call the top of the market
490.
by axg11 5y ago
That’s an interesting idea. Not sure how I could automatically maintain that list. Out of interest do you have a list of your own?
491.
by axg11 5y ago
Did you read the article? It pretty clearly states that the internal comms team raised a concern.
492.
Show HN: Track tweets and floor price for NFTs(nft.arshamg.com)
2 points by axg11 5y ago|3 comments
493.
by axg11 5y ago
Wasn't he doing the very opposite of grovelling to China with the original comment? It's American culture that is forcing him to apologize.
494.
Ask HN: Why don’t startups share their cap table and/or shares outstanding?
4 points by axg11 5y ago|2 comments
495.
by axg11 5y ago
https://hyperfine.io/ This is as close as we have right now.
496.
by axg11 5y ago
Just let the market decide ;)
497.
by axg11 5y ago
As others have said, this is the way ML is headed and it has been clear for the last couple of years. Assuming a multi-task and multi-modal architecture is achievable, won't we reach a point in the future where preprocessing data is ju
498.
by axg11 5y ago
It would be interesting to know if blind or blind-folded dogs also exhibit this behaviour. My intuition is that head-tilting behaviour is more analogous to how we (humans) look up and the to left/right when concentrating. That feels le
499.
by axg11 5y ago
This is an interesting line of thinking. As a point of reference (rather than a serious forecast), total market cap for gold is roughly 10x the total market cap for BTC
500.
by axg11 5y ago
Everything we do as a species has an environmental impact, even breathing. Perhaps this is one of the group of activities where the benefits outweigh the costs.
501.
by axg11 5y ago
What are the KPIs for any store of value? I don’t think you can value Bitcoin in the same way you value a revenue generating business. I think the value makes more sense when you look at it as an abstract and digital store of value to repla
502.
by axg11 5y ago
Great feedback. I will improve the copy - as you said, it's currently aimed at technologists. It looks like three.js is able to load .glb files (a format I can easily generate with the current pipeline) so I will try to add a three.js
503.
by axg11 5y ago
I've made the service free for now - I'd like people to use it to get feedback on the models themselves. I'm also looking into adding a webviewer - that is a great point.
504.
by axg11 5y ago
Sorry that you feel it's bad value. What would an appropriate price point be for you? There is an example 3D model on the homepage - that should give you an idea of how the output will look like.
505.
by axg11 5y ago
Great question. The account only exists to manage the payment and subscription. I toyed with the idea of providing the service for free but it is very compute intensive. Each model takes approximately 20-30s on a modern GPU.
506.
Show HN: Facemodel, create 3D models of your head from a selfie(facemodel.me)
7 points by axg11 5y ago|12 comments
507.
Show HN: Recognize any triplet of objects live (but not real-time)(tingy.video)
1 points by axg11 5y ago|0 comments
508.
Search through hours of footage with machine learning (tingy)(tingy.video)
1 points by axg11 5y ago|0 comments
509.
by axg11 5y ago
In this case I'd argue this is more than just the usual iterative fodder. The architecture itself is not the interesting part, it's the evidence that many decent architectures exist and performant networks don't have to be an
510.
by axg11 5y ago
As mentioned in the article, pharma is high-risk high-reward long term investment. Many companies tried and failed to produce an effective COVID vaccine (CureVac, Merck). Those companies invested R&D dollars well before the pandemic and

More ›