I came back to to this post several times today to see the comments on this paper. I was curious what is significant about this paper given that it stays on Hacker News front page for hours.
Another curiosity is, what is the typical cost and GPU hours to train the model with these algorithms?
This extraction of graph structure is the "holy grail" of NLP in that it can break documents down into facts so that, say, you can store them in a database and query them in a more accurate and efficient way.
Also these science papers frequently have a very low comments to vote ratio compared to, say, articles about cars or the housing supply in California.
This could work well for small scale farm lands owned by small families as these farm lands use more manual labors than tractors. The farm lands I am talking about are like the ones in the highlands in South East Asia.
Yes. The linked post says Japan is the 4th largest importer of wood. My geography professor in 1981 told us about it being cheaper to send raw wood logged in the Pacific Northwest to Japan for finishing than it was to ship it across the country. And, in Collapse I think, Jared Diamond talks about Japan "exporting" its deforestation to less-developed countries. Other countries do this too (and for other natural resources too), of course, but any of the "sustainability" of these old practices is greatly outweighed by the damage done by mass logging in the less-developed countries.
That is a positive assesment. On the other hand, imagine if the aliens think that the senders of these probes were deliberately sending pathogens to the aliens' worlds. This is a statement of war.
“Certainly, it would be a global disaster for humanity; a disaster for the entire world,” Putin said, in an interview for a Russian documentary “The World Order 2018,” adding that “as a citizen of Russia and the head of the Russian state I must ask myself: Why would we want a world without Russia?”
Them memory leak on Gnome shell is fixed on version 3.28.4 for me. It was memory-leak on Ubuntu 14.04 but they fixed in 14.04.2. The "a decade" claim needs some evidence.
The memory leak is the result of a poorly designed interface between compiled code and javascript. The result being that memory wasn't truly leaked in the traditional fashion but rather was consumed rapidly and freed on no sane timescale.
This mismatch between compiled code and javascript existed from inception to now. The fix if I understand it correctly is to really aggressively free memory in a fashion that appears grossly inefficient but in practice performs acceptably.
The release of gnome 3 was in spring of 2011 the "fix" was released in spring of 2018. It would be more accurate to say that it was terribly broken only for the first 7 years of its existence.
Because their "fix" is a hacky workaround for the fact that the only true fix is to throw everything away and start over the bug is actually back in 3.34
Another curiosity is, what is the typical cost and GPU hours to train the model with these algorithms?