codelord

As a machine learning researcher for over a decade I clicked to understand what "brain-based AI" is. I learned it's a made-up term by the author to maximize click-through rate. Pass.

I understand what you are saying, and I'm saying it's a pointless argument which doesn't add any value to the discussion.

Why is that an important difference? To capture wind and solar you have to build turbines and solar panels which aren't free. You also have to have batteries to store them because of seasonal nature of these sources. You also need to transmit the energy thousands of miles through electrical grid because these sources of energy are limited to certain areas. Building the electrical grid itself just needs cutting out millions of more trees. So overall, the environmental impact of going full solar/wind would be worse than a combination of solar, wind, nuclear, and even maybe natural gas.

The proof is in the pudding.

Re: negotiating. You negotiate your comp once, it takes 1 hour maybe. Compare that to the effort it takes to maintain a fragile career based on lies.

Encouraging people to take two full-time jobs by lying and cheating their way around is such a terrible advice even if you only look at it purely based on self-interests with no moral considerations.

Negotiate well when starting a new job (If you haven't find a new job and negotiate well). Do great work. Use the extra time to invest on yourself: read, exercise, learn new skills, etc. You'll be happier and more successful in the long run.

You can see from comments here on hackernews that at least a few people interpreted this as a change to their ranking algorithm. Writing ambiguously to support a narrative is also misreporting.

I know anti-FB sentiment is strong on hackernews. But seriously look at that screenshot and tell me what's wrong with that? How's that different from promotions by Apple on ios/app store, Microsoft on windows, twitter, google, youtube, etc. etc. They are just promoting their company on their own platform. When did your Facebook newsfeed become this untouchable sacred place where nothing beside chronological posts from your friends is allowed?

I want to hate facebook as much as anyone else but I really don't see news here.

A screenshot from the actual experiment: https://twitter.com/joeosborne/status/1440456340673949700 Looks like the article is overblown.

How much experience do you have with Unreal engine and game dev? If you think art-style is dependent on game engine you probably don't know what you are talking about. Modern game engines are pretty much like an operating system, there's little constraint in terms of what you can achieve artistically.

This is cool, but I find it hard to find any use case for it. I did plenty of work as an indie game dev in the past, shipped and made money out of games based on my own C++/OpenGL game engine and also shipped a few indie games with some external indie game engines. Then I moved away from gamedev, but recently started playing around with Unreal engine again and was blown away by how much Unreal does for you. It felt like with Unreal engine, game dev is mostly about art and game design and not so much about programming anymore. Given game dev has become so much more competitive and harder to do than in the old days, and the fact that you can get a AAA game engine like Unreal for free (and only 5% royalty after 1-5M in sales), it doesn't make any sense to use anything else other than Unreal/Unity IMO.

This is great. Thanks for sharing.

> The surrounding blue box is the parent element.

(R: 93, G: 200, B: 195) That's more green than blue. When you consider that humans have stronger green color preceptors it's definitely green.

To understand if Copilot is infringing the licenses of codes used in its training data we have to get into the details of what it does and how it works. We can't make a general statement for any code generation software that was trained on open source code.

It is a possible that at some point maybe even in not so distant future we will have ML models so good they can understand abstract concepts, learn and invent new algorithms and implementations by reading code. Such a ML model can be argued is learning similar to a human and hence it's not infringing any copyrights because it's not copying implementations it's learning concepts and ideas.

But we are not there yet. When we get there we will know. Because at that point Siri would be able to have seamless conversations with you. At least half the jobs would disappear in favor of robots in a short time. The world would be a different place.

Let's talk about what Copilot actually can do. It can copy snippets of code from Github while changing variable names. It can autocomplete trivial boilerplate code. If it's automagically generating a function for you that actually does something useful like sorting an array, you can be absolutely sure that it's just copy pasting it from an existing repo with some cosmetic changes.

It's not really comparable to a pen. Because a pen by itself doesn't copy someone else's code/written words. It's more like copying code from Github or if you wrote a script that did that automatically. You have to be actively cautious that the material that you are copying is not violating any copyrights. The problem is Copilot has enough sophistication to for example change variable names and make it very hard to do content matching. What I can guarantee it won't be able to do is to be able to generate novel code from scratch that does a particular function (source: I have a PhD in ML). This brute-force way of modeling computer programs (using a language model) is just not sophisticated enough to be able to reason and generate high level concepts at least today.

Similar taste can be achieved with non-Persian tea. I think brewing method maybe more important. For the brand I buy loose leaf english breakfast and use a transparent teapot to see when the tea is ready. Usually takes ~5 minutes for the perfect taste.

Tea: https://teapigs.com/products/english-breakfast?_sid=3aed062b... (choose the loose leaf version)

Teapot: https://www.amazon.com/gp/product/B00KIW0T9C

Yes. We call it "Zereshk Polo". I didn't know it's also popular in eastern Europe. Interesting!

Actually color can affect the perceived taste. Red/vibrant colors make the drink taste more sweet.

Saucer is optional. I drink directly from the cup. Make sure the cup is made of clear glass though to see the color of the tea.

I feel like people in the west need to become more familiar with Persian cooking. Every year I learn that one more ingredient that we commonly use in Persian cooking is good for your health. So based on this definitive evidence I'm gonna conclude everything we eat is good for you. To get you ahead here are some other things to try:

1. Sabzi Khordan: Elevate your breakfast with Sabzi Khordan, a mixed of herbs that are commonly eaten as breakfast with cheese and walnuts.

2. Drink properly brewed tea: If your idea of tea is what you get from Starbucks, you don't know what tea is. Try Persian brewed black tea.

3. Persian barberry rice: Try Persian style rice cooked with rice, salt, butter, barberries, and saffron.

4. Add turmeric to pretty much every stew.

5. Try Fesenjan, Ghormeh Sabzi, and Khoresh Gheymeh.

I would just read the linked post:

https://bugs.python.org/issue38980?fbclid=IwAR0cyfahpBywNzbq...

As it contains almost the same info without the rant and with better explanation.

Oops. You are right.

I haven't used TF32 personally, but I think the sales pitch is not too far off. Most of the time I use mixed-precision training which should be similar to FP16/TF32 in terms of performance. It does tremendously speed up training.

It's not really fair to compare a discrete GPU to a mobile GPU, I only provided this as a comparison for someone who maybe has one of these at home. And btw, you are talking about TF32 performance not FP32. TF32 actually uses 16 bits. A100's FP32 performance is actually lower than 3090, it's 19.5 TFLOPS: https://www.nvidia.com/content/dam/en-zz/Solutions/Data-Cent...

FP16 performance is also relevant as a lot of people now train in FP16. The default for pytorch/TF is still FP32.

Seems like AMD has been using Vega 20 to refer to two different things.

I was talking about the mobile GPU on MacBook Pros which is based on a 14nm chip. The full name is Radeon Pro Vega 20:

https://www.amd.com/en/graphics/radeon-pro-vega-20-pro-vega-...

https://www.techpowerup.com/gpu-specs/radeon-pro-vega-20.c32...

Vega 20 seems to also refer to a discrete GPU. This has been later rebranded to Radeon VII (maybe because of this confusion). The number you are quoting is for the discrete GPU.

On the one hand it seems Google is unstoppable. Google can do everything. But I think iPhone showed an alternative path. Now most people are actually using apps (not Google) on mobile for different functionalities. I guess this is the argument that Google has been making but I buy that. My guess is 20 years from now Android, YouTube, Google Maps, etc. would be more valuable for Google than search.

M1 and AMD GPU support. I'm personally more interested in the latter as I haven't yet upgraded my MacBook Pro and I expect that my Vega 20 to be faster than M1 at ML training.

The raw compute power of M1's GPU seems to be 2.6 TFLOPS (single precision) vs 3.2 TFLOPS for Vega 20. This can give you an estimate of how fast it would be for training.

Just for reference Nvidia's flagship desktop GPU(3090)'s FP32 performance is 35.5 TFLOPS.

I don't believe mining Bitcoin quite falls into the category of "most productive things we can use energy for". Drug discovery, carbon capture, agriculture, construction, etc. etc. come to mind as slightly more useful. Financial transactions in traditional banking systems which Bitcoin mining is more comparable with consume far far far less energy. Also Bitcoin is designed to adjust PoW difficulty based on network's average mining performance. In other words, the more compute power you provide the less efficient it becomes.

On your website that you linked says "We conduct our business in a way that protects ... our planet's climate against warming." Can you elaborate how providing infra to mine Bitcoin is helping the planet against global warming? Aren't you doing the exact opposite? If you are using 100% green energy, that energy can be used for something productive not providing PoW. Note Bitcoin alone currently consumes 0.55% of world's TOTAL electricity production and this is going to get even worse with more adoption.

Reference: https://hbr.org/2021/05/how-much-energy-does-bitcoin-actuall...

Fixed mostly to relatively. That wasn't my main point really. I thought the data points were interesting.

This is from earlier: https://www.cebm.net/covid-19/excess-mortality-across-countr... Interestingly Sweden that got famous for defying lockdowns saw an uptick of 1.5% excess mortality per 100K people compared to 12.9% for the US that was relatively more locked down. I'm not sure what to make of it, but it sure is an interesting data point.

Not sure why would you go and make a statement about me without any evidence (And btw yes to all your questions). Your point would be stronger if you don't target a person and maybe talk about the average consumer. My point exactly that a large chuck of the population don't follow CDC guidelines, if they did we collectively would be a healthier nation.