Commits

Author Commit Message Labels Comments Date
David McClosky
Add Java and Python SWIG wrappers for the second stage reranker. swig/wrapper.i: The wrapper itself simple-api.*: Since much of the reranker code lives in headers, made a simple API for reranker functionality for easier wrapping in SWIG. Makefile: add swig-java, swig-python, swig-python-test, swig-clean targets Remove tags files in real-clean.
David McClosky
.hgignore: Add wlle program binaries, *.orig (made by 'hg revert' commands)
David McClosky
Add Java and Python SWIG wrappers for the first stage parser. swig/wrapper.i: The SWIG wrapper itself swig/*/test: Some simple examples/tests for the wrapper Makefile: refactored out lots of common object lists, removed cruft new target real-clean for making distributions add swig-java, swig-java-test, swig-python, swig-python-test, and swig-clean targets ThreadManager: system to manage "thread slots" when SWIG is used in multithreading m…
David McClosky
TRAIN/Makefile: add real-clean target
David McClosky
TRAIN: Don't crash when model names don't end in '/' and some minor cleanups Removed some commented out code and unused variables, made programs accurately identify themselves.
David McClosky
PARSE: Don't crash when model names don't end in '/'
David McClosky
Makefile: switch "make" to "$(MAKE)" so parallel building will work
David McClosky
Tokenizer and utils: bug fixes and cleanups Tokenizer (ewDciTokStrm): change interface so this always takes an istream. This removes any special logic for cin. removed old code for handling <DOC> and <COMMENT> tags fix bug where the last token in a sentence wouldn't be split fix bug where abbreviations in quotes weren't split properly (e.g. "G.B.") parseIt: adjusted to use new ewDciTokStrm istream interface Utils: standardi…
Ben Swanson
Apply Apache 2.0 license.
David McClosky
Merge
David McClosky
Merge pull request #8 from vene/fix-build Fix build issues on recent MacOS X
Vlad Niculae
FIX: consistent use of CC and CXX envs
Vlad Niculae
FIX: failing build on MacOS X
David McClosky
More robust handling of some parse failures when using external POS tags. If external POS tags are provided and the parse fails, we try to parse again without any POS constraints.
David McClosky
Change how we handle the PU tag in Chinese. We now only use the word as a tag if the word is a known tag. Otherwise, we continue to use the PU tag. Better error message for the TRAIN/ version.
David McClosky
Minor cleanups: Get rid of comment cruft.
David McClosky
Make cvlm-owlqn the default estimator. As a result, README now says that Petsc/Tao are optional.
David McClosky
Remove old print statement.
David McClosky
Support for longer words (1024 characters) and sentences.
David McClosky
Better error messages when using external POS tags. If you specify a tag which isn't in terms.txt, you'll get a warning and the tag will be ignored.
David McClosky
Bugfix for printing head indices from Matt Gerber.
David McClosky
Merge in Matt Gerber's new options for best-parses. "...changes for writing out head indexes (best-parses.cc and tree.h). I added two additional printing modes ("-m" switch), one for printing syntactic heads and one for printing semantic heads."
David McClosky
Merge in Matthew Gerber's changes to @ symbol handling. @ symbols no longer get special treatment when bracketed in <s>. Previously they were ignored, but presently it has the effect of skipping many sentences from Twitter corpora.
David McClosky
Merge in Mark Johnson's latest changes to the reranker (version as of December 2011). Among other improvements, this should now compile on Ubuntu 11.10 (g++ 4.6.1). It also includes support for many more new feature families.
David McClosky
Small README updates to document pre-tagged input option (-E) and "Frequently confusing errors"
David McClosky
Two additional hgignores.
David McClosky
Fix my previous fix for issue #3 which wasn't properly accounting for the contents of savedWrd_.
David McClosky
Fix issue #3 (behave better in pipe mode: parse each sentence immediately rather than waiting for the next token or EOF)
David McClosky
Ignore vim swap files and ctags output.
David McClosky
Fix compilation issue #1. Compiles on gcc compilers for Ubuntu 10.10 and 11.04.
  1. Prev
  2. Next