The Toolkit for Advanced Discriminative Modeling (TADM) is a C++
implementation for estimating the parameters of discriminative models,
such as maximum entropy models. It uses the PETSc and TAO toolkits to
provide high performance and scalability. 

TADM is a rebranding and relicensing of a system written by Rob Malouf
and distributed previously as "estimate".  A paper describing the use
of the original system for evaluating various algorithms for training
maximum entropy models is available here:

http://bulba.sdsu.edu/~malouf/papers/conll02.pdf

TADM currently has very little documentation, so it can be rough for
first time users to work out how to use it. Look in the docs directory
for a start and try using the python scripts, which handle a lot of
low-level details and accept some well-known formats for stating
features.

See docs/install.txt for instructions on compiling and installing TADM
on your machine.

See the data directory for two example data sets formatted to work
with the Python front end and instructions for testing them.

If you need help, please post a message to the TADM sourceforge help forum:

http://sourceforge.net/forum/forum.php?forum_id=473054

Asking questions on the forum is much preferable to contacting the
developers directly so that everyone can benefit from the answers
which are provided.

A lot of useful information regarding maximum entropy modeling and
software is available on Zhang Le's maxent page:

http://homepages.inf.ed.ac.uk/s0450736/maxent.html

Finally, Jason Baldridge has used TADM in his Computational
Linguistics II class at the University of Texas at Austin. Homework 3
from the Fall 2006 version of the class is likely to be quite useful
for people who would like something like a tutorial for using
TADM. You can download it from the following location:

http://comp.ling.utexas.edu/jbaldrid/courses/2006/cl2/cl2-hw3.tgz