The mert optimization is unlikely to be a bottleneck unless you've
changed the size of the n-best list, which I believe defaults to 100.
At that setting it can usually finish in a few minutes on a single
processor. At a high setting of 1000, it can take over an hour in
later iterations.
Hi,
there is also a stage in the mert training (the actual mert optimization,
not running the decoder) that is currently not parallelized, so you
cannot expect any gains on that. It is my understanding that we will
have a better implementation of mert that also parallelizes the
optimization loop,
Thanks, increasing to 1G seem to "unstuck" the processes.
My training pairs is only about 250K of sentences, so I had thought 0.5G should
be sufficient; perhaps the issue is with the size of the LM.
Nevertheless, I'm doing mert on a dev-set which has only about 1000 source
sentences. For this set
Yee Seng Chan wrote:
> However, when I tried to parallelize it by submitting say.. 10
> jobs, I don’t get faster MERT iterations. In fact, it’s slower.
> Sometimes, a job can be stuck on one of the grid nodes and after
> hours, it’s still not completed. Its corresponding output-file e.g…
>
Hi,
First, I must say that I've used the mert script and it works well.
However, when I tried to parallelize it by submitting say.. 10 jobs, I don't
get faster MERT iterations. In fact, it's slower. Sometimes, a job can be
stuck on one of the grid nodes and after hours, it's still not compl
