I think you're missing why people find German compound nouns interesting and notable. Does English have compound nouns and noun phrases? Certainly. But take your example: "gay clergy book row priest." The noun phrase literally represents the sum of its parts, i.e. a priest involved in a row over a gay clergy book. The phrase is complicated, but its meaning is straightforward. In German, however, it seems as if two unrelated nouns come together to form something unexpected. For instance the word for tortoise is "Schildkröte", translated literally as "shield toad." Is a tortoise a toad? Most native English speakers would say no. Does it have a shield? Again, most native English speakers would say no. So you have a case where there are two easily understood German words that come together to form something unexpected (to an English speaker). Perhaps a German would say that yes, a tortoise is a toad and yes, it has a shield. This is probably my own bias, but I don't think compound nouns in English form nouns in the same way as some German compound nouns do, or at least it feels to happen more often in German than in English.
I like AWS, but K8s has reduced my cloud needs to down to a Docker registry and a K8s cluster, which I can easily get anywhere. The one thing I really like about AWS is DynamoDB, which has very cheap and straightforward pricing but can also scale up seamlessly.
One reason I like SSR is that you need some form of SSR for public-facing websites anyway. Website previews (like in iMessage, Twitter, etc.) rely on Open Graph tags in the HTML, and these services expect the OG tags to be available without executing any JavaScript. Since you already need this step, you can make loading pages much faster if you inline any data you might have fetched from the client at view time.
My impression when studying German is that the German language has more compound words that are singular words in English. e.g. Armbanduhr (watch), Aussehen (appearance), Hauptstadt (capital), Staubsauger (vacuum), anything ending with Zeug, etc. That doesn't necessarily mean German is longer on average.
German is famous for its compound nouns. My feeling is compound nouns are noticed by English speakers due to their explicit verbosity, but I don't have any data to back that up.
Very few consoles, let alone games, last 15 years. Requiring an online experience to work well for 15 years, while nice, is not a realistic requirement for a majority of games.
I play a lot of SSBU online and it works well. Rollback netcode would be a huge improvement, but the game is definitely playable. Online play clearly isn't a focus for Nintendo, but it's hard to argue with their success. Perhaps their focus on building self-contained experiences is part of their secret sauce.
I was responding to someone who asked "Who could strike a nuclear power plant who couldn't use a nuclear bomb." Canada is a very valid answer to that questions. No one is talking about the likelihood of such an event.
"Procedural generation" usually means content generated by a human-crafted procedure. Given that a human made the algorithm by hand, its scope is limited and easily becomes repetitive.
I'm not making an arbitrary point. The end of my comment mentions word embeddings, which are learned from analyzing language and its usage in an unsupervised manner. You can use a word embedding to measure similarity between words. For instance, I just downloaded a pre-trained model of Stanford's GloVe embedding that was trained on 840 billion (yes, billion with a b) tokens. The cosine similarity can be used to measure how close two words are in the embedding. The cosine similarity of million and billion is 0.89. The similarity of duck and fuck is 0.27. The data indicates million and billion are much more similar than duck and fuck.
This result is intuitively obvious to me, which this article illustrates. Even if the resulting sentence is not factually true, a true-sounding sentence can be constructed by taking a sentence with the word million and replacing it with the word billion (in a majority of cases). This isn't true with duck and fuck, and duck is a noun and a verb used in generally different contexts than the word fuck.
I'm only speaking from my own personal experience, but I've never seen bananas priced at less than 1c each in the grocery store. I've also seen people misplace decimal points frequently. So, based on the thousands of times I've been to dozens of grocery stores, I'm using pattern matching to deduce that almost assuredly it's more likely that a banana listed as costing "0.19c" refers to $0.19 not $0.0019. If you ever encounter a true sub-cent banana in a retail supermarket, feel free to reach out to me and I'll be the first to admit I was wrong.
Which interpretation is idiotic? Being able to take in complex signals from the environment and form an instructive opinion is the foundation of human ingenuity. Given that the bananas obviously aren't on sale for $0.0019, it seems idiotic to me ignore alternative explanations for the sign.
It's not only about the letters. "Million" and "billion" are both numeric concepts. Numeric concepts are only a small subset of things expressible by language. So million and billion are close because they are spelled similarly and they deal with the same conceptual "thing." Duck and Fuck don't occupy the same conceptual category, so the distance between the two words is large despite being spelled similarly.
One interesting application of neural networks is the creation word embeddings. ML models are trained to place words in a vector space, which is useful for measuring distance between words, performing arithmetic on words, or finding the closet word. Using an embedding allows your to formalize the "distance" between words, and perform fun tricks like King + Woman = Queen.
Language is an imperfect model for representing ideas. Million is close to billion, linguistically. Certainly not off by a power of 1000 in the space of language. I’m surprised the author is just coming to this realization.
I agree 100%, but I think viable businesses will begin to emerge especially as these large models move from text to images (and eventually to video and 3d models). If the examples shown of DALL-E 2 are indicative of its quality, then a large number of creative jobs could be replaced with a single "creative director" using the model. But the high entry cost just to attempt to train such a model will likely remain a hurdle until more business value is proven.
The amount of capital needed to train these high-quality models is eye watering (not to mention the costs needed to acquire the data). Does anyone know of any well capitalized startups exploring this space?
You always have the right of way in the U.S. unless there is a stop sign or a yield sign. If both directions have stop signs, then the person who arrived first has the right-of-way. To me, it's more straight forward to always have right-of-way, but that's probably because I'm used to it.
He was awarded money for loss of wages and emotional distress. I don't think he lost seven figures of wages, and how much emotional distress did he suffer? At least personally, it would hard for me to say he suffered millions of dollars worth of emotional distress (which is something millions of people experience every day for free).
I find the "right of way" signs in Europe way more strange, specifically when the signs aren't present. When you're on a main thoroughfare, it's pretty obvious you have the right of way. It's on the smaller side streets where you need to yield to the person to your right that things get a little weird.
I don't think you can call procedural generation "zero marginal content." When dealing with PG, the content is the procedure, not the output, and the procedure never changes. There is a distinction between PG and a generative model like DALL-E. PG involves procedures that are written and understood by humans. Something like DALL-E, while technically a deterministic procedure taking inputs and producing outputs, operates via a process that is not directly understandable to humans.
This just means the demand curve is very price sensitive, but it doesn't mean the supply curve doesn't exist. There is an element of survivorship bias to all of this. Products on the market exist because there is a point the supply and demand curves intersect. It's totally that the price people are willing to pay for Arizona ice tea will be less than the cost to produce it, at which point Arizona ice tea will cease to exist.
I don’t know if you’ve even seen a supply and demand graph, but there is a price a supplier is willing to sell at, and a price the buyer is willing to pay. The supplier’s price is based off of their costs. Saying the price is only determined by the buyer makes no sense.