cdec-dtrain-legacy/compound-split/de, branch json_serial Mirror of https://github.com/pks/cdec-dtrain-legacy.git https://git.simianer.de/mirrored/cdec-dtrain-legacy/atom?h=json_serial 2014-01-19T02:57:32Z new tuning of crf compound splitter for wmt14 2014-01-19T02:57:32Z Chris Dyer cdyer@allegro.clab.cs.cmu.edu 2014-01-19T02:57:32Z urn:sha1:3c1e7365d8708846dfffd598158571d4c4abd097 new de split 2014-01-17T09:07:16Z Chris Dyer redpony@gmail.com 2014-01-17T09:07:16Z urn:sha1:8bce53d2a076ef76f3fea34b2bddb0a713ec4ea8 fix 2012-11-06T23:47:04Z Chris Dyer cdyer@allegro.clab.cs.cmu.edu 2012-11-06T23:47:04Z urn:sha1:eb03a413713c070edf07e466575e0f58085aa258 fix up some frequencies to give better results on a few common words 2012-11-06T23:43:45Z Chris Dyer cdyer@allegro.clab.cs.cmu.edu 2012-11-06T23:43:45Z urn:sha1:d6db35c6d718f48dec30c041c4a0e469c00a6739 larger training data for semi-crf word segmenter 2012-11-06T04:41:55Z Chris Dyer cdyer@allegro.clab.cs.cmu.edu 2012-11-06T04:41:55Z urn:sha1:9e5107a05bfabb76ce547d2849173c5a11aeba60 noun-segmenting weights for german word segmenter 2011-10-24T17:32:13Z Chris Dyer cdyer@cs.cmu.edu 2011-10-24T17:32:13Z urn:sha1:62f98f92a3ccf9a614eafe115dbfd04c79c55ea0 Updated kenlm. Includes left state support but not the cdec-side use of it. Updated binary format. 2011-09-21T22:23:50Z Kenneth Heafield kenlm@kheafield.com 2011-09-21T22:23:50Z urn:sha1:f111672dd611f78656fceb3df3729a290453ef56 Fix broken klm file for de compounding 2011-03-10T21:10:34Z Kenneth Heafield kheafiel@cluster10.lti.ece.cmu.local 2011-03-10T21:10:34Z urn:sha1:faf9e73708e745e43405531c6fc1a33bd192260c remove dependency on SRILM 2011-03-10T06:58:30Z Chris Dyer cdyer@cs.cmu.edu 2011-03-10T06:58:30Z urn:sha1:4f9933d668d247ea5831c3f2af0b996a94da28f7 more german fixes 2011-01-17T17:49:06Z Chris Dyer cdyer@cs.cmu.edu 2011-01-17T17:49:06Z urn:sha1:b49944ee0e5f347a936df244a7c354a867c79c93