| Age | Commit message (Collapse) | Author | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
and the parser segfaults (observed with grammars full of unary rules)
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
Add IgnoredStateSize() to FeatureFunction
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
1. alignment.h, argument_reorder_model.h, src_sentence.h, tree.h,
   tsuruoka_maxent.h -> decoder/ff_const_reorder_common.h.
2. Trainers source files (argument_reorder_model.cc and
   constituent_reorder_model.cc) are moved to training/const_reorder.
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
span limits so it is not enabled, but it seems to be functional.
 | 
 | 
Mr.MIRA compatibility
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
Add --show_derivations_mask to control what to print when --show_derivations
 | 
 | 
 | 
 | 
 | 
 | 
 |