Commits

Show all
Author Commit Message Labels Comments Date
David Mimno
updates to topic evaluation
Tags
2.0.5
David Mimno
support for polylingual topics
David Mimno
null pointer checks
David Mimno
adding polylingual topic model
David Mimno
bug fix: training instances that have been rewritten with expanded testing vocabularies before topic training
David Mimno
initializing betaSum
David Mimno
initial version
Kedar Bellare
Maxent trainer uses L-BFGS instead of orthant L-BFGS. Naive Bayes EM trainer now takes doc length normalization into account.
Gregory Druck
Fixed bug in LabelDistributionEvaluator.
David Mimno
better handling of punctuation and marks (useful for Indic scripts)
David Mimno
tools for held-out likelihood estimation
David Mimno
deprecation messages
David Mimno
deprecation messages
David Mimno
helper method for beta optimization
David Mimno
concentration parameter optimization for topic-word distributions, bug fixes, log likelihood reported per word
David Mimno
bug in topic inference
David Mimno
new method to estimate concentration parameter only (not base multinomial) from histograms
David Mimno
utilities for multivariate normals
David Mimno
added --config default option to read values from files, more meaningful variable names
David Mimno
constant time sampling of gamma variates with alpha greater than 1.0
David Mimno
trigamma function
Andrew McCallum
Implement XML reporting of topics for both individual words and phrases.
David Mimno
set random seed before initialization
Gregory Druck
A few minor fixes and better error messages for GE training of MaxEnt models.
David Mimno
Added tag 2.0RC4 for changeset b35c729b13e8
Gregory Druck
GE for MaxEnt code updated to use ParallelTopicModel.
Tags
2.0RC4
David Mimno
bug fix
David Mimno
added accessor methods for sorted words
Gregory Druck
Fixed NaN bug when target expectation is 0 in MaxEntOptimizableByGE.
David Mimno
Significant changes to PagedInstanceList courtesy of Mike Bond
  1. Prev
  2. Next