cdec
cdec copied to clipboard
redpony
Decoder, aligner, and model optimizer for statistical machine translation and other structured prediction models based on (mostly) context-free formalisms
Problematic port binding happens in the sentence server module: it first tests if a port is available, and it tells that the port is available and immediately try to bind...
I have newer version of boost installed in my home directory and have set LIBRARY_PATH and LD_LIBRARY_PATH. However, python/setup.py calls g++ to link like this: g++ -pthread -shared -O3 build/temp.linux-x86_64-2.6/cdec/_cdec.o...
``` bash python -m cdec.sa.extract -c ini -g grammars -j 2 < corpus > sgm ``` If the ini file contains invalid paths, this will launch a bunch of child...
The so-called Python Lattice Format (PLF) syntax supported by cdec is more constrained than I realized: unlike the equivalent Python data structure, - commas are required after all nodes and...
Would be pretty useful! Maybe not that hard to do? Cython works with Python 3 these days...
I've observed that corpus/tokenize-anything.sh does not necessarily preserve the number of lines on the files it is fed. I'll try and isolate the problem and give more details.
We should probably fix this when moving to Paul's new interface, but it can be annoying
When i try do this: python -m cdec.sa.extract -c extract.ini -g dev.grammars -j 1 < dev.lc-tok.ru-en > dev.lc-tok.ru-en.sgm I've got this error: File "/usr/lib/python2.7/runpy.py", line 162, in _run_module_as_main "__main__", fname,...
Currently both functions call ScoreCCandidate in SentenceScorer to get a ScoreP smart pointer (which, when non-null, gives access to non-static versions of Score::GetZero or Score::GetOne). However, most actual implementations of...
Tokenizing bitext with `` tags fails: ``` echo "x ||| x" | ~/tools/cdec/corpus/tokenize-anything.sh x ||| x ``` Why these tags are in my corpus is another problem :tired_face:
