You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@joshua.apache.org by Tommaso Teofili <to...@gmail.com> on 2016/09/29 13:01:03 UTC

NN for MT

https://research.googleblog.com/2016/09/a-neural-network-for-machine.html

Although probably everyone has already read this by now, I thought it was
worth sharing.

Regards,
Tommaso

Re: NN for MT

Posted by Matt Post <po...@cs.jhu.edu>.

Yes, interesting, particularly the human evaluations, though I would have liked to see something focused on comprehension of the output, instead of just pairwise comparisons.

This is also an interesting calculation (suggesting neural MT is more energy-efficient than humans), though the comparison isn't actually relevant:

	https://twitter.com/stephenroller/status/780842250083311616

matt

> On Sep 29, 2016, at 9:01 AM, Tommaso Teofili <to...@gmail.com> wrote:
> 
> https://research.googleblog.com/2016/09/a-neural-network-for-machine.html
> 
> Although probably everyone has already read this by now, I thought it was
> worth sharing.
> 
> Regards,
> Tommaso