.\" DO NOT MODIFY THIS FILE!  It was generated by help2man 1.40.10.
.TH ESTIMATE-NGRAM "1" "January 2013" "MITLM" "User Commands"
.SH NAME
estimate-ngram \- estimates n\-gram language model
.SH SYNOPSIS
.B estimate-ngram
[\fIOptions\fR]
.SH DESCRIPTION
Estimates an n\-gram language model by cumulating n\-gram count statistics,
smoothing observed counts, and building a backoff n\-gram model.  Parameters
can be optionally tuned to optimize development set performance.
.PP
Filename argument can be an ASCII file, a compressed file (ending in .Z or .gz),
or '\-' to indicate stdin/stdout.
.SH OPTIONS
.TP
\fB\-h\fR, \fB\-help\fR
Print this message.
.TP
\fB\-verbose\fR <int>
Set verbosity level.
.IP
Default: 1
.TP
\fB\-o\fR, \fB\-order\fR <int>
Set the n\-gram order of the estimated LM.
.IP
Default: 3
.TP
\fB\-v\fR, \fB\-vocab\fR <file>
Fix the vocab to only words from the specified file.
.TP
\fB\-u\fR, \fB\-unk\fR <boolean>
Replace all out of vocab words with <unk>.
.IP
Default: false
.TP
\fB\-t\fR, \fB\-text\fR <files>
Add counts from text files.
.TP
\fB\-c\fR, \fB\-counts\fR <files>
Add counts from counts files.
.TP
\fB\-s\fR, \fB\-smoothing\fR <ML, FixKN, FixModKN, FixKN#, KN, ModKN, KN#>
Specify smoothing algorithms.
.IP
Default: ModKN
.TP
\fB\-wf\fR, \fB\-weight\-features\fR <features\-template>
Specify n\-gram weighting features.
.TP
\fB\-p\fR, \fB\-params\fR <file>
Set initial model params.
.TP
\fB\-oa\fR, \fB\-opt\-alg\fR <Powell, LBFGS, LBFGSB>
Specify optimization algorithm.
.IP
Default: Powell
.TP
\fB\-op\fR, \fB\-opt\-perp\fR <file>
Tune params to minimize dev set perplexity.
.TP
\fB\-ow\fR, \fB\-opt\-wer\fR <file>
Tune params to minimize lattice word error rate.
.TP
\fB\-om\fR, \fB\-opt\-margin\fR <file>
Tune params to minimize lattice margin.
.TP
\fB\-wb\fR, \fB\-write\-binary\fR <boolean>
Write LM/counts files in binary format.
.IP
Default: false
.TP
\fB\-wp\fR, \fB\-write\-params\fR <file>
Write tuned model params to file.
.TP
\fB\-wv\fR, \fB\-write\-vocab\fR <file>
Write LM vocab to file.
.TP
\fB\-wc\fR, \fB\-write\-counts\fR <file>
Write n\-gram counts to file.
.TP
\fB\-wec\fR, \fB\-write\-eff\-counts\fR <file>
Write effective n\-gram counts to file.
.TP
\fB\-wlc\fR, \fB\-write\-left\-counts\fR <file>
Write left\-branching n\-gram counts to file.
.TP
\fB\-wrc\fR, \fB\-write\-right\-counts\fR <file>
Write right\-branching n\-gram counts to file.
.TP
\fB\-wl\fR, \fB\-write\-lm\fR <file>
Write ARPA backoff LM to file.
.TP
\fB\-ep\fR, \fB\-eval\-perp\fR <files>
Compute test set perplexity.
.TP
\fB\-ew\fR, \fB\-eval\-wer\fR <files>
Compute test set lattice word error rate.
.TP
\fB\-em\fR, \fB\-eval\-margin\fR <files>
Compute test set lattice margin.
.PP
.SH SEE ALSO
\fBevaluate-ngram\fR(1), \fBinterpolate-ngram\fR(1)