GPT is trained to predict the input (estimating p(x)), versus predicting a label given an input (p(y|x)). So in the case of GPT you can use the input dimensionality as a "label", as another responder has mentioned. ImageNet classification is different (excepting recent semi-supervised or unsupervised approaches to image recognition).

The ability to generalize in the typical imagenet setting is, as the article says, a byproduct of SGD with early stopping, which in practice limits the number of functions a deep neural network can express (something not considered in an analysis which only considers parameter count).

It's not about sticking it to them. It's to increase the cost of their appeals and pressure them to drop the case.

The first stay went without incident. Upon getting ready to check in to the second place, however, I logged on to the website and discovered that the second booking was cancelled. I was aghast, and tried to get in contact with customer service. After many hours I got someone on the line, and was told that the second payment had not go through, and so they had cancelled the reservation.

I was shocked because I was not notified by this at all and would have been happy to put another card down. After digging, I found that 5 emails from automated@airbnb.com had landed in my spam inbox. I would have never thought to check the spam folder, because in the meanwhile I had received many emails from Airbnb about our other reservations, and gotten like 20 texts with details of the other reservations. Why hadn't they just used the other email or texted?

In the meantime, they had relisted the unit and it was immediately booked by someone else. They also took the $3000 we put down on the unit because it was cancelled after the date of a refund.

This had severe consequences for us. We couldn't find a comparable place to rebook and so our friends cancelled their whole trip (they were just flying in for the latter half). We are now spending our remaining time just trying to cut our losses and figure out how to end the vacation early, which has made everything extremely stressful. Overall I think we'll lose over $5k. I was also going to propose on this trip, but will not anymore because we're both so sad. Airbnb's customer service is completely unresponsive and unempathetic.

Anyway, to the question. Who's at fault here? Am I dumb and deserve this for not checking my spam folder all the time? Is it GMail's fault? Is Airbnb at fault?

Comments URL: https://news.ycombinator.com/item?id=27102110

Points: 1

# Comments: 2

If you don't believe me, consider that many ML researchers track their commits (or exact code versions) extremely carefully, because oftentimes they will make some change (or changes) they think are inconsequential and later find that actually, their model broke. If they made too many changes, whoops, guess you have to binary search over the diff to see what happened since your last "good run".

If the people who spent months (if not years) tuning a model can't tell whether it will work from the code, how could anyone else? Most ML researchers will not bother with most code that doesn't give proof of results (in terms of a model that can actually be evaluated) because it is just so unlikely that it will actually work well. Now, it might "work" in the sense that it converges and does something when you prompt it with examples. But will this GPT-3 reimplementation actually outperform say, the 10x smaller T5 checkpoint that was released by Google, or the other smaller language models others have released? If it doesn't, it's hard to argue that its very useful at all.

I think that's the spirit of why the original commenter said what they did, but I still do applaud the efforts of this team (and hope that their implementation is, in fact, highly performant!)

Comments URL: https://news.ycombinator.com/item?id=26532543

Points: 2

# Comments: 1

Alternative take: the hard part is that the US is heterogeneous and people just don't trust each other to not abuse benefits. (You could also say it's racist). How could we give equity to every person when we can't even seem to agree that they deserve basic healthcare?