scaling_test

#!/bin/bash

dataset="fisher_test"
epochs=30
alpha=1
weight=1

WC_OUT="$(wc -l data/fisher-callhome/corpus/ldc/${dataset}.es.nopunc.lower)"
for token in $WC_OUT; do
	SIZE=$token
	break
done
CUTS=10
DIFF=$((SIZE/CUTS))

# The number of sentences per minute
callhome_rate=$((20875/(20*60)))

for ((i=0;i<CUTS;i++)); do
	num_sents=$((SIZE-i*DIFF))
	printf "%b\t" $((num_sents/callhome_rate))
	#./run $dataset $epochs $alpha $weight $num_sents
	#./wer $dataset $epochs $alpha $weight $num_sents
	./run_giza $dataset $num_sents
	./decode_giza $dataset $num_sents > data/out/giza-pp/${dataset}.n${num_sents}.es
	python ../nlp/per.py \
			--ref data/fisher-callhome/corpus/ldc/${dataset}.es.nopunc.lower \
			--hypo data/out/giza-pp/${dataset}.n${num_sents}.es
done
#./run $dataset $epochs $alpha $weight 0
#./wer $dataset $epochs $alpha $weight 0