Recently I tried to train mt-ar tagger and I got the same error:
segmentation fault during "Kupiec's initialization of transition
and emission probabilities..." stage.


It seems that the problem is somehow related to the number of
def-mults: with more than 1200 def-mults I get segmentation fault,
with about 450 def-mults I don't; and it seems that it really has
nothing to do with the actual contents of def-mults.


With limited number of def-mults the tagger is training and
I get a .prob file; unfortunately it does not work properly,
I get additional tags in the tagger output:

echo "ma jifhmux" | lt-proc mt-ar.automorf.bin 
^ma/ma<adv>$  ^jifhmux/fehem<vblex><tv><pres><p3><mf><pl>+x<neg> \
/fehem<vblex><tv><pres><p3><m><sg>+u<prn><p3><m><sg>+x<neg>$

$ echo "ma jifhmux" | lt-proc mt-ar.automorf.bin | apertium-tagger -g mt-ar.prob
^ma<adv>$ ^fehem<vblex><tv><pres><p3><mf><pl>+x<neg>+u<prn><p3><m><sg>+x<neg>$


Does anyone know what could be the reason?


Best regards,
Maria Fronczak


------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Apertium-stuff mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/apertium-stuff

Reply via email to