Joshua Technical Support

Joshua has become an official Apache Incubator project. Its new website is: http://joshua.incubator.

Thanks very much, Gideon! I just made a note on your PR that you should submit the patch to KenLM

Hi, I have a few questions about Joshua. How does the translation compare with the Moses system or

All right. Feel free to post problems (and solutions!) I've just noticed myself that a number of

Okay, great to hear! matt On Feb 25, 2016, at 11:02 PM, theresa...@gmail.com wrote: Hi Matt, I

Great! Let us know how it goes. On Feb 9, 2016, at 9:56 PM, theresa...@gmail.com wrote: I tried

Ah; these were recently moved to $JOSHUA/src. Updating the paths, or creating a symlink, should fix

The pipeline command is executed now successfully. Thank you Matt. -- Mohamed On 22 December 2015 at

I figured out that and it was hard coded in $JOSHUA/bin/joshua-decoder ## memory usage; default is 4

Thanks Matt, for the script to generate glue grammar. I have emailed Juri (cc co-authors) about

Whoops! New directory. Added it to the packager and re-uploaded, please try grabbing it again, thanks

Hi, Joshua 4 is quite old. You must mean Joshua 6.0.4. I'll need a lot more information to be

Thanks a lot Matt. I'll try to build using the master branch on github and see if it works.

Hi i have downloaded the joshua decoder file but unable to find .exe file to install i am using

One thing I've seen with lower-resource languages is that the Berkeley aligner generates much

Hello, The run bundler script can be found in $JOSHUA/scripts/support/run_bundler.py, and is

That is a *very* large tuning set. I suggest using no more than 5k, perhaps sample from it randomly.

Hi, Try this: echo "This is a sample sentence" | ./ngram [arguments -ppl - matt On Aug 15,

Hi, Probably something went wrong with tuning. What was your pipeline.pl invocation? Can you paste in

I would try rebuilding kenlm; it looks like you have an incompatible binary. You can do this with cd

Yes, this shouldn't be too hard. I don't have an existing script to do it, though. Here is

It doesn't include one, but Moses' script is easy enough to use. matt (from my phone) On Jul

To clarify, you want to continue tuning, initializing parameters from a previous run? You could use

Fixed! On Jun 2, 2015, at 9:44 AM, gabriell...@gmail.com wrote: Good morning, Is the link to

> Back on the actual Quick Start guide, it seems to stop short of > actually using the trained

We haven't produced any, do you have a particular request? We do have some parallel data for some

You could also use MERT or PRO for tuning, which don't depend on Moses. matt On Apr 1, 2015, at 3

It looks like creation of the grammar failed for some reason. You'll have to investigate that

It looks like your training data is too small for Kneser-Ney smoothing. You can try adding the --

(Apologies for multiple copies.) This is a join announcement and call for participation and papers