Update regression tests

4 files changed, 156 insertions, 0 deletions

diff --git a/tests/training/restart/.gitignore b/tests/training/restart/.gitignore
index 60d3c29..504d7ac 100644
--- a/tests/training/restart/.gitignore
+++ b/tests/training/restart/.gitignore
@@ -12,3 +12,9 @@ corpus_sqlite
 corpus_sqlite.expected
 corpus_mini
 corpus_mini.expected
+corpus_one
+corpus_one.expected
+corpus_maxi
+corpus_maxi.expected
+corpus_fit
+corpus_fit.expected
diff --git a/tests/training/restart/test_corpus_restoration_maxi_batch.sh b/tests/training/restart/test_corpus_restoration_maxi_batch.sh
new file mode 100644
index 0000000..3d339e0
--- /dev/null
+++ b/tests/training/restart/test_corpus_restoration_maxi_batch.sh
@@ -0,0 +1,50 @@
+#!/bin/bash -x
+
+# Exit on error
+set -e
+
+# Test code goes here
+rm -rf corpus_maxi corpus_maxi*.log
+mkdir -p corpus_maxi
+
+test -e vocab.de.yml
+test -e vocab.en.yml
+
+extra_opts="--seed 1111 --maxi-batch 20 --mini-batch 32 -o sgd --dim-emb 128 --dim-rnn 256 --disp-freq 4 --restore-corpus"
+
+$MRT_MARIAN/build/marian \
+    -m corpus_maxi/model_full.npz -t train.max50.{en,de} -v vocab.{en,de}.yml \
+    --after-batches 70 $extra_opts \
+    --log corpus_maxi.log
+
+test -e corpus_maxi/model_full.npz
+test -e corpus_maxi.log
+
+cat corpus_maxi.log | $MRT_TOOLS/strip-timestamps.sh | grep "Ep\. " | sed 's/ : Time.*//' > corpus_maxi.expected
+
+$MRT_MARIAN/build/marian \
+    -m corpus_maxi/model.npz -t train.max50.{en,de} -v vocab.{en,de}.yml \
+    --after-batches 40 $extra_opts \
+    --log corpus_maxi_1.log
+
+test -e corpus_maxi/model.npz
+test -e corpus_maxi_1.log
+
+cat corpus_maxi_1.log | $MRT_TOOLS/strip-timestamps.sh | grep "Ep\. " | sed 's/ : Time.*//' > corpus_maxi_1.out
+cp corpus_maxi/model.npz.yml corpus_maxi/model.npz.1.yml
+
+$MRT_MARIAN/build/marian \
+    -m corpus_maxi/model.npz -t train.max50.{en,de} -v vocab.{en,de}.yml \
+    --after-batches 70 $extra_opts \
+    --log corpus_maxi_2.log
+
+test -e corpus_maxi/model.npz
+test -e corpus_maxi_2.log
+
+cat corpus_maxi_2.log | $MRT_TOOLS/strip-timestamps.sh | grep "Ep\. " | sed 's/ : Time.*//' > corpus_maxi_2.out
+cat corpus_maxi_1.out corpus_maxi_2.out > corpus_maxi.out
+
+$MRT_TOOLS/diff-floats.py corpus_maxi.out corpus_maxi.expected -p 0.1 > corpus_maxi.diff
+
+# Exit with success code
+exit 0
diff --git a/tests/training/restart/test_corpus_restoration_mini_batch_fit.sh b/tests/training/restart/test_corpus_restoration_mini_batch_fit.sh
new file mode 100644
index 0000000..5cbc4a0
--- /dev/null
+++ b/tests/training/restart/test_corpus_restoration_mini_batch_fit.sh
@@ -0,0 +1,50 @@
+#!/bin/bash -x
+
+# Exit on error
+set -e
+
+# Test code goes here
+rm -rf corpus_fit corpus_fit*.log
+mkdir -p corpus_fit
+
+test -e vocab.de.yml
+test -e vocab.en.yml
+
+extra_opts="--seed 5555 --maxi-batch 8 --maxi-batch-sort src --mini-batch 32 --mini-batch-fit -w 100 -o sgd --dim-emb 128 --dim-rnn 256 --disp-freq 4 --restore-corpus"
+
+$MRT_MARIAN/build/marian \
+    -m corpus_fit/model_full.npz -t train.max50.{en,de} -v vocab.{en,de}.yml \
+    --after-batches 80 $extra_opts \
+    --log corpus_fit.log
+
+test -e corpus_fit/model_full.npz
+test -e corpus_fit.log
+
+cat corpus_fit.log | $MRT_TOOLS/strip-timestamps.sh | grep "Ep\. " | sed 's/ : Time.*//' > corpus_fit.expected
+
+$MRT_MARIAN/build/marian \
+    -m corpus_fit/model.npz -t train.max50.{en,de} -v vocab.{en,de}.yml \
+    --after-batches 60 $extra_opts \
+    --log corpus_fit_1.log
+
+test -e corpus_fit/model.npz
+test -e corpus_fit_1.log
+
+cat corpus_fit_1.log | $MRT_TOOLS/strip-timestamps.sh | grep "Ep\. " | sed 's/ : Time.*//' > corpus_fit_1.out
+cp corpus_fit/model.npz.yml corpus_fit/model.npz.1.yml
+
+$MRT_MARIAN/build/marian \
+    -m corpus_fit/model.npz -t train.max50.{en,de} -v vocab.{en,de}.yml \
+    --after-batches 80 $extra_opts \
+    --log corpus_fit_2.log
+
+test -e corpus_fit/model.npz
+test -e corpus_fit_2.log
+
+cat corpus_fit_2.log | $MRT_TOOLS/strip-timestamps.sh | grep "Ep\. " | sed 's/ : Time.*//' > corpus_fit_2.out
+cat corpus_fit_1.out corpus_fit_2.out > corpus_fit.out
+
+$MRT_TOOLS/diff-floats.py corpus_fit.out corpus_fit.expected -p 0.1 > corpus_fit.diff
+
+# Exit with success code
+exit 0
diff --git a/tests/training/restart/test_corpus_restoration_one_epoch.sh b/tests/training/restart/test_corpus_restoration_one_epoch.sh
new file mode 100644
index 0000000..dd15815
--- /dev/null
+++ b/tests/training/restart/test_corpus_restoration_one_epoch.sh
@@ -0,0 +1,50 @@
+#!/bin/bash -x
+
+# Exit on error
+set -e
+
+# Test code goes here
+rm -rf corpus_one corpus_one*.log
+mkdir -p corpus_one
+
+test -e vocab.de.yml
+test -e vocab.en.yml
+
+extra_opts="--seed 9999 --maxi-batch 1 --maxi-batch-sort none --mini-batch 32 -o sgd --dim-emb 128 --dim-rnn 256 --disp-freq 4 --restore-corpus"
+
+$MRT_MARIAN/build/marian \
+    -m corpus_one/model_full.npz -t train.max50.{en,de} -v vocab.{en,de}.yml \
+    --after-batches 70 $extra_opts \
+    --log corpus_one.log
+
+test -e corpus_one/model_full.npz
+test -e corpus_one.log
+
+cat corpus_one.log | $MRT_TOOLS/strip-timestamps.sh | grep "Ep\. " | sed 's/ : Time.*//' > corpus_one.expected
+
+$MRT_MARIAN/build/marian \
+    -m corpus_one/model.npz -t train.max50.{en,de} -v vocab.{en,de}.yml \
+    --after-batches 25 $extra_opts \
+    --log corpus_one_1.log
+
+test -e corpus_one/model.npz
+test -e corpus_one_1.log
+
+cat corpus_one_1.log | $MRT_TOOLS/strip-timestamps.sh | grep "Ep\. " | sed 's/ : Time.*//' > corpus_one_1.out
+cp corpus_one/model.npz.yml corpus_one/model.npz.1.yml
+
+$MRT_MARIAN/build/marian \
+    -m corpus_one/model.npz -t train.max50.{en,de} -v vocab.{en,de}.yml \
+    --after-batches 70 $extra_opts \
+    --log corpus_one_2.log
+
+test -e corpus_one/model.npz
+test -e corpus_one_2.log
+
+cat corpus_one_2.log | $MRT_TOOLS/strip-timestamps.sh | grep "Ep\. " | sed 's/ : Time.*//' > corpus_one_2.out
+cat corpus_one_1.out corpus_one_2.out > corpus_one.out
+
+$MRT_TOOLS/diff-floats.py corpus_one.out corpus_one.expected -p 0.1 -n 1 > corpus_one.diff
+
+# Exit with success code
+exit 0