This is at a high level how one of our Google internal black box optimizers behaves. The term used is "infeasible region" but it's the same idea as using a nan.
I would caution against using nan to always mean infeasible. Instead users should catch experiments outside the feasibility region and return a special infeasible value. This will increase visibility into the behavior of the optimizer, because it leaves nan to be used for values inside the region of constraint that are still problematic (due to bugs, numerical instability, etc)
Computation existed for more than 40 years before apps started making a ton of money on it. If BD develops robotic platform-"workers" that can be trained in 5 seconds to do any manual labor, that's the new "app" that one would be making money off of in the future.
"On the joyous and satisfying occasion of this, our first successful journey to Mars, we think back not only on all those who died along the way, but also on the Slack clone that allowed us avert disaster on this mission..."
I do wonder why we want humans on Mars. Given the huge amount of fossil fuels that must be burned on this planet to get us there.
We may find Mars is dry and arid when we get there, and so is Earth.
Was looking at buying the quest 2 today and decided to wait until the weekend to see how the launch would go and then I saw this. I won't buy a quest 2 unless the account requirement is removed
But most people are capped to how much they'll spend on leisure by their vacation time. So there probably isn't much headroom to be gained for people who are regular travellers post this pandemic.
Sure, but like vacation money, vacation time has also been curtailed. I'm currently not on a vacation I had previously planned and working instead, saving both time and money I would have otherwise spent.
I'm in the same boat. Dearly wish I could take a vacation but the state has banned short term rentals for some reason (hotels still open though). So when they let us out of house arrest I plan to get out of town asap.
Does the use of positional embeddings mess with the gnn for formulation? I'm not familiar with the requirements for something to be a gnn, but positional embeddings mean the graphs have to capture order of occurrence, and the graph in the shared page doesn't seem to do that.
In graph terms, Positional encodings are useful for adding sequential/temporal properties to each node in the graph. Indeed, there are works on position-aware GNNs.
That isn't true. Downtown Toronto has multiple Google engineers. I know one in brain and one in cloud. One has been in TO for ~2.5 years, the other for 1 year. As far as I know the only way to get into the office is by special permission. One of them only got their role because they said they were quitting due to the waterloo commute.
Fwiw I'm a Canadian that just signed a Google offer to work out of downtown SF because I hate the winter too. I think people care more or less about the weather based on their hobbies / interests.
It absolutely is true in actual practice; unless you're some unicorn ML researcher all our engineering talent is in Waterloo (or Montreal). If it was possible to work in Toronto I wouldn't have had to have sold my house and moved my family years back.
We do have a rather comfortable bus which runs from Toronto, if you don't mind the 2 hour (each way) commute.
(The company I worked for was bought by Google back in 2011. We had people in NY and Toronto; all our Toronto people were told we had to move to Waterloo and that was that. Most of us have stayed @ Google but for some the practice of having to rip up their entire lives in Toronto to move wasn't ideal and they left.)
The number of engineers at Google Toronto is so small that they're the exception that proves the rule, though. For all intents and purposes beyond some very special cases (and the Hinton stuff) you can treat the site as having no engineering.
That is not what I understand it to mean -- rather, the opposite. That is, finding an engineer at the Toronto office is so improbable that it is remarkable, hence you are reminded that it is normally an engineer free zone.
There are some idioms that do mean it is unexceptional. You might be familiar with the Pratchettism "Million-to-one chances happen nine times out of ten."
Google can preview results for clicks by exposing a site's content in the search results.
E.g if you google Obama you will see a wiki blurb about him on the right
I took this quote to mean the following:
Engineers are well equipped and positioned to understand the set of solutions that are available to them and the material negative impacts of those solutions (e.g if we build a dam at spot x, the region y might have an increase in floods). But engineers should not be given the ability to make decisions about which of set of negative impacts a society should bear, and that should instead be a decision made collectively.
So the interface between the engineers and the people is not to find solutions together, or to find flaws together, but for engineers to find solutions and flaws, and the people to pick the one that they are happiest with.
(I'm not certain my understanding of the quote is right, just sharing my interpretation / an alternative to the ones you shared)
The costs(in various dimensions) of an Engineering decision in projects of this magnitude is borne by the whole population itself. Therefore the populace must have a seat at the decision-making table to choose amongst alternatives.
Original software for it is a bit hard to work with, improvements in https://github.com/ongdexter/ar3_core