ninja3925

The big unknown are the tax implication? The article should clarify what it means, at least vis a vis Korean tax agency.

For anybody close to the topic and knowing how shocked the Israeli society was on Oct 7th, this paper makes a very strong claim that needs more than just circumstantial evidence (higher short trades). Like any circumstantial evidence, many reasons could trigger unusual trading patterns

This a poor argument which basically criticize any cost sharing agreement with multiple payers.

“Look, GM is getting gov subsidies but we still have to pay for a car!” Not a strong argument indeed.

23andme has democratized DNA testing by making it affordable when the numbers didn’t work. This is remarkable in itself. Kudos to them.

CLIP is a such a nice paradigm shift. Historically, CV things were quite limited:

- You could predict a class (from a static list such as [dog,cat, ...]) or ...

- You could use image embeddings disconnected from text (you could tell image look-alikes but not what they actually represent). By "embedding" text and images in the same latent space, you can now query your images with text query (such a "a large dog") and find the relevant photos. CLIP understands semantics but also is not limited to a set list of classes (thanks to the ability to use of web data in training).

This is a list compiled by OpenCLIP of high performance models (some better than MetaCLIP) for those interested in using CLIP: https://github.com/mlfoundations/open_clip/blob/main/docs/op...

I work for a large tech company and we use CLIP internally for a variety of things:

- Image search (aka Google Image) to find relevant photos given a text prompt. This is the biggest use case

- Automated Labeling (to answer questions such as "How many entities have attribute Y"?

We didn't fine tune CLIP though. If folks have done it successfully, I would be love to hear them!

(Worked at Lyft) Our number of active micro services was small in comparison. 4,000 is likely a overblown number to highlight the accomplishment possibly counting inactive ones

I somewhat understand your point but the facts imperfectly support it. If information was already priced in, say 90% chance of success, a positive outcome would still provide a lift from the remaining 10%.

The decrease sounds like something else is going on.

I used to work for Lyft and I can confirm that both teams (driverless and ride-share) contribute. Ride-share scaled later which may give you that false impression.

And Yes, it's a big investment for Lyft for the open source community.

An embedding is a numerical representation of a complex object.

In an image, each pixel is a dimension but it does not have any meaning in itself - you need to look at the rest of the image to understand that the pixel is part of a cat. Embeddings is a way to represent this meaning. Think about it as a “summary” of a image / document.

The main problem is that the embeddings are getting larger and larger (1,000+ dimensions). The pressure is then on reducing memory use through techniques such as Product Quantization while not losing too much accuracy.

Once this is done, the search heuristics are not difficult (find the cells to explore and return nearest neighbors).

This is a good opportunity to thank Sergey for making his extensive content for free to the community. Such a nice gesture. Thanks!

I particularly enjoy the detail-oriented yet approachable content (to someone like me that does not have a CS background).

What are a terrible thing to say. I am deeply saddened by your experience and I hope you will find fulfillment somewhere else.

All the best.

FYI: I didn't read the racial taint in the sentence.

The author (Gael Varoquaux) is deeply involved in numpy/Sklearn and a talented researcher. Quite an impressive guy worth following.

Great comment! After reading this, you may wonder: why didn't the armies try to bomb railroads to prevent reinforcement from being rushed to the front? It turns out, they tried very hard but the technology was just too rudimentary. The French made a bombing units of Voisins bombers but they could only carry light payload (130lbs for the Voisin 3). The Brits got the Handley in 1916, which was bigger and quite successful (this doc has a ton of details on British bombing: https://www.airuniversity.af.edu/Portals/10/AUPress/Books/B_...). If you squint hard enough, you can start seeing the emergence of patterns of WW2 warfare by the end of WW1. Fascinating.

I have noticed that politicians always use extremes (many standard deviations away from the average) to illustrate a population.For example, Warren Buffet wealth is in the top 10. Yet, Warren's policies target the top 1% (top 10 is the top 0.00033% of the top 1%). He is not representative of the top 1% in any way!! Show me the median and let's talk.

I am surprised people bite at this manipulative narrative. It seems to me that it discredits their arguments right away. In that sense, voter education would go a long way.

While true and something that needs fixing immediately, this is uncommon and therefore, cannot be used to describe the relationship as "exploitative".

Of course! You can choose: - Not to work - Move to a city/state with better opportunities - Take classes to specialize - Work hard and get promoted - Start your business

Aren't these valid options? No "cigar smoking fat cat" force workers to sign the contract under duress.

> I'm just advocating that tax laws exist & are enforced, so as to redistribute wealth, which I don't view as oppressive

Laws are strongly enforced in this country (USA).

Where is the bar for "oppressive"? I come from a country that thought that 70%+ taxation was acceptable. In the USA, we are now at 60%+ (across federal, state, local). It sure sounds oppressive to me and one way out is to give the choice to people to opt out.

“Off the back” sounds like the wealthy took advantage of folks when, in fact, they entered voluntary contracts in a win-win scenario. Creating a company is not exploitative but rather collaborative and mutually beneficial (with workers).

It is good that we have such process for people to push back against their government greed. Usually, the citizenry is at the mercy of it and can’t do anything about it.

You don’t like it? Fine. You have a choice and can get out.

Good safety valve. It keeps the populist governments in check.

If this figure is true, do you see the symbolism of it as I do? It's not 49%, where the individual will still work mostly for themselves but 51%, where the state owns the majority of the labor for each individual. I find this number chilling.

Just googling for RCT, I found one convincing example: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5025014/

I am very hopeful about renewable but I am also aware of their intrinsic weakness: They can't be provide base power (unless we build battery farms able to cover our needs).

Renewables are exciting. Jumping too fast into it like Germany did (and is now polluting as much as 8X France with 450g CO2 / kWh) is much less exciting.

I wish there was a bit more expert involvement in the way we choose our energy policies (and much less tribalism and populism).

source: https://www.electricitymap.org/zone/DE

Honestly, I am not sure why this study is noteworthy. As you pointed you, the evidence is weak. Plus, we have the results of a good randomized controlled experiment. I don't think we need anything more to make up our mind (yes, vitamin D3 works very, very well): https://www.sciencedirect.com/science/article/pii/S096007602...

Note: The study link is for treatment, not prevention.

I don't understand why people say Uber "bought" a law. Is every prop sponsored by a corporation a "purchase"? This type of speech undermines our collective trust in democracy and should not be OK.

Good operation for the founders. They have 42M stocks + 35M options which should yield $3-4B. 3 newly minted billionaires.

Looks like this work led the way to Germ Theory (Pasteur). Pretty cool. Thanks Semmelweis and Pasteur. https://hardydiagnostics.com/wp-content/uploads/2016/05/Semm...