#!/bin/csh


########################################################################
#	don't change these
########################################################################

set sfchome = "/net/alf14/usr/sfc"
set arch = `$sfchome/exec/hosttype.x`


########################################################################
#	here is a list of things you probably want to set
#
#	defaults have been set to what Matt used in the TREC eval decode,
#	when possible
########################################################################

#	directory to write out nbest lists to
set outdir = "$sfchome/lattice/work/m007/m007c"

#	number of files to skip at beginning of control file; number of
#	files to rescore in control file (use these flags to split
#	up control file into sections, for parallelization)
set ctloff = 0
set ctlcnt = 3

#	directory where lattices are located; list of lattice files sans
#	extension
#
#	for training decode, use:
#	set latdir = "/net/db7/usr1/robust/spokes/TREC/corpora/h496/TREC6_train/Decode/Output/stories.s3new-51k.1164.latdir"
#	set ctlfile = "$sfchome/lattice/work/l007/l007c.ctl"
#
#	for eval decode, use:
#	set latdir = "/net/db7/usr1/robust/spokes/TREC/STORES/Lattices/space-2/TREC6-Evaltest-51k"
#	set ctlfile = "$sfchome/error/work/a021/a021d4.ctl"
#	(this contains all 11372 lattices, except for k960611.23-14 which
#	is corrupted)
#
set latdir = "/net/db7/usr1/robust/spokes/TREC/corpora/h496/TREC6_train/Decode/Output/stories.s3new-51k.1164.latdir"
set ctlfile = "$sfchome/lattice/work/l007/l007c.ctl"


########################################################################
#	here is a list of things you might want to set
#
#	defaults have been set to what Matt used in the TREC eval decode,
#	when possible
########################################################################

#	size of nbest lists to generate
set nbest = 150

#	language weight; insertion penalty; noise penalty; noise file,
#	containing penalties for particular tokens
set langwgt = 9.5
set inspen = 0.2
set noisepen = 0.05
set noisefile = "/net/db7/usr6/robust/TREC/corpora/Eval/Decode/extras/filler-dt96.pen"

#	this is the LM used by Matt, it is actually
#	/net/db7/usr6/robust/TREC/corpora/Eval/Decode/extras/bn92-96-51k+_phrases+xb-1-1.arpabo.Z.DMP
#	converted to my format
set arpabo = "/net/alf9/usr2/sfc/tmp/m007a.dmp"

#	MB of memory rescorer will use (when memory limit is reached for
#	a lattice, search beams are narrowed)
set mem = 100

#	sets log base with which scores in nbest file are expressed
#	to internal log base, which is 1.000333333333; this is set
#	for compatibility with version 0.1; by default, is set to -10.0,
#	which outputs scores log 10
set logbaseout = -1


########################################################################
#	here is the actual command line
########################################################################

$sfchome/pub/$arch/bin/Lattice \
	-hub4 -nbest.3 \
	-langwgt $langwgt \
	-inspen $inspen \
	-noisepen $noisepen \
	-noisefile $noisefile \
	-latdir $latdir \
	-ctlfile $ctlfile \
	-ctloff $ctloff \
	-ctlcnt $ctlcnt \
	-outdir $outdir \
	-latext .lat.Z \
	-arpabo $arpabo \
	-outext .nbest.gz \
	-nbest $nbest \
	-mem $mem \
	-bbrelax -flLogBaseOut $logbaseout


########################################################################
#	
########################################################################