---------------------------------------------------

Some examples to learn how to *program* in Torch3.
Of course you can use them without any changes,
but you will not use 1/10 of the library if you
don't want to code something!

kmeans.cc  : simple K-means models
gmm.cc     : simple Gaussians Mixtures Models
hmm.cc     : simple Hidden Markov Models (see speech and 
             decoder example for speech recognition)

---------------------------------------------------

You will find two examples of datasets in data/.
Torch3 is able to deal directly with gziped data,
but you can uncompress them if you want to have a
look. (Or do a zless, zcat, zmore).

---------------------------------------------------

Examples of command line:


echo "GMM"
./Linux_OPT_FLOAT/gmm -save gmm_model -one_file data/train.amat.gz

or
echo data/train.amat.gz>list;./Linux_OPT_FLOAT/gmm -save gmm_model list

./Linux_OPT_FLOAT/gmm --retrain gmm_model -one_file data/train.amat.gz
./Linux_OPT_FLOAT/gmm --test gmm_model -one_file data/test.amat.gz

echo "HMM"
./Linux_OPT_FLOAT/hmm -save hmm_model -one_file data/train.amat.gz
./Linux_OPT_FLOAT/hmm --retrain hmm_model -one_file data/train.amat.gz
./Linux_OPT_FLOAT/hmm --test hmm_model -one_file data/test.amat.gz

echo "KMeans"
./Linux_OPT_FLOAT/kmeans -save kmeans_model -one_file data/train.amat.gz
./Linux_OPT_FLOAT/kmeans --test kmeans_model -one_file data/test.amat.gz

---------------------------------------------------
