Fix guided alignment regressions tests (#84)

* fix guided alignment regressions tests
* update regression tests

16 files changed, 118 insertions, 101 deletions

diff --git a/tests/interface/input-tsv/test_tsv_train_with_align.sh b/tests/interface/input-tsv/test_tsv_train_with_align.sh
index a95ad9d..c2d79a7 100644
--- a/tests/interface/input-tsv/test_tsv_train_with_align.sh
+++ b/tests/interface/input-tsv/test_tsv_train_with_align.sh
@@ -14,10 +14,11 @@ mkdir -p train_align
 
 # Run marian command
 $MRT_MARIAN/marian \
-    --cost-type ce-mean --no-shuffle --clip-norm 0 --seed 5555 --dim-emb 32 --dim-rnn 64 --maxi-batch 1 --maxi-batch-sort none --optimizer sgd --learn-rate 0.001 \
+    --cost-type ce-mean --no-shuffle --clip-norm 0 --seed 5555 --maxi-batch 1 --maxi-batch-sort none --optimizer adam --learn-rate 0.001 \
+    --dim-emb 32 --transformer-dim-ffn 64 --type transformer --enc-depth 3 --dec-depth 3 \
     -m train_align/model.npz --tsv -t train2.de-en-aln.tsv -v $MRT_MODELS/rnn-spm/vocab.deen.{spm,spm} \
     --after-batches 100 --disp-freq 4 \
-    --guided-alignment 2 --guided-alignment-weight 1.0 \
+    --guided-alignment 2 --guided-alignment-weight 1.0 --guided-alignment-cost ce \
     --log train_align.log
 
 
diff --git a/tests/interface/input-tsv/test_tsv_train_with_align_and_weights.sh b/tests/interface/input-tsv/test_tsv_train_with_align_and_weights.sh
index efa1793..ee0832d 100644
--- a/tests/interface/input-tsv/test_tsv_train_with_align_and_weights.sh
+++ b/tests/interface/input-tsv/test_tsv_train_with_align_and_weights.sh
@@ -14,10 +14,11 @@ mkdir -p train_align_weights
 
 # Run marian command
 $MRT_MARIAN/marian \
-    --cost-type ce-mean --no-shuffle --clip-norm 1 --seed 7777 --dim-emb 32 --dim-rnn 64 --maxi-batch 1 --maxi-batch-sort none --optimizer sgd --learn-rate 0.1 \
+    --cost-type ce-mean --no-shuffle --clip-norm 1 --seed 7777 --maxi-batch 1 --maxi-batch-sort none --optimizer adam --learn-rate 0.1 \
+    --dim-emb 32 --transformer-dim-ffn 64 --type transformer --enc-depth 3 --dec-depth 3 \
     -m train_align_weights/model.npz --tsv -t train2.de-w-aln-en.tsv -v $MRT_MODELS/rnn-spm/vocab.deen.{spm,spm} \
     --after-batches 60 --disp-freq 4 \
-    --guided-alignment 2 --guided-alignment-weight 1.0 --data-weighting 1 \
+    --guided-alignment 2 --guided-alignment-weight 1.0 --data-weighting 1 --guided-alignment-cost ce \
     --log train_align_weights.log
 
 
diff --git a/tests/interface/input-tsv/test_tsv_train_with_align_and_weights_inputtypes.sh b/tests/interface/input-tsv/test_tsv_train_with_align_and_weights_inputtypes.sh
index c0e175e..a049bc1 100644
--- a/tests/interface/input-tsv/test_tsv_train_with_align_and_weights_inputtypes.sh
+++ b/tests/interface/input-tsv/test_tsv_train_with_align_and_weights_inputtypes.sh
@@ -14,10 +14,11 @@ mkdir -p train_align_weights_intypes
 
 # Run marian command
 $MRT_MARIAN/marian \
-    --cost-type ce-mean --no-shuffle --clip-norm 1 --seed 7777 --dim-emb 32 --dim-rnn 64 --maxi-batch 1 --maxi-batch-sort none --optimizer sgd --learn-rate 0.1 \
+    --cost-type ce-mean --no-shuffle --clip-norm 1 --seed 7777 --maxi-batch 1 --maxi-batch-sort none --optimizer adam --learn-rate 0.1 \
+    --dim-emb 32 --transformer-dim-ffn 64 --type transformer --enc-depth 3 --dec-depth 3 \
     -m train_align_weights_intypes/model.npz --tsv -t train2.de-w-aln-en.tsv -v $MRT_MODELS/rnn-spm/vocab.deen.{spm,spm} \
     --after-batches 60 --disp-freq 4 \
-    --input-types sequence weight alignment sequence --guided-alignment-weight 1.0 \
+    --input-types sequence weight alignment sequence --guided-alignment-weight 1.0 --guided-alignment-cost ce \
     --log train_align_weights_intypes.log
 
 # Check if files exist
diff --git a/tests/interface/input-tsv/test_tsv_train_with_align_pos0.sh b/tests/interface/input-tsv/test_tsv_train_with_align_pos0.sh
index 34c829f..c7e9929 100644
--- a/tests/interface/input-tsv/test_tsv_train_with_align_pos0.sh
+++ b/tests/interface/input-tsv/test_tsv_train_with_align_pos0.sh
@@ -14,10 +14,11 @@ mkdir -p train_align0
 
 # Run marian command
 $MRT_MARIAN/marian \
-    --cost-type ce-mean --no-shuffle --clip-norm 0 --seed 5555 --dim-emb 32 --dim-rnn 64 --maxi-batch 1 --maxi-batch-sort none --optimizer sgd --learn-rate 0.001 \
+    --cost-type ce-mean --no-shuffle --clip-norm 0 --seed 5555 --maxi-batch 1 --maxi-batch-sort none --optimizer adam --learn-rate 0.001 \
+    --dim-emb 32 --transformer-dim-ffn 64 --type transformer --enc-depth 3 --dec-depth 3 \
     -m train_align0/model.npz --tsv -t train2.aln-de-en.tsv -v $MRT_MODELS/rnn-spm/vocab.deen.{spm,spm} \
     --after-batches 100 --disp-freq 4 \
-    --guided-alignment 0 --guided-alignment-weight 1.0 \
+    --guided-alignment 0 --guided-alignment-weight 1.0 --guided-alignment-cost ce \
     --log train_align0.log
 
 
diff --git a/tests/interface/input-tsv/test_tsv_train_with_align_shuffle.sh b/tests/interface/input-tsv/test_tsv_train_with_align_shuffle.sh
index 7a0503c..925b3ec 100644
--- a/tests/interface/input-tsv/test_tsv_train_with_align_shuffle.sh
+++ b/tests/interface/input-tsv/test_tsv_train_with_align_shuffle.sh
@@ -14,10 +14,11 @@ mkdir -p train_align_shuffle
 
 # Run marian command
 $MRT_MARIAN/marian \
-    --cost-type ce-mean --clip-norm 1 --seed 4444 --dim-emb 32 --dim-rnn 64 --maxi-batch 1 --maxi-batch-sort none --optimizer sgd --learn-rate 0.1 --sync-sgd \
+    --cost-type ce-mean --clip-norm 1 --seed 4444 --maxi-batch 1 --maxi-batch-sort none --optimizer adam --learn-rate 0.1 --sync-sgd \
+    --dim-emb 32 --transformer-dim-ffn 64 --type transformer --enc-depth 3 --dec-depth 3 \
     -m train_align_shuffle/model.npz --tsv -t train2.aln-de-en.tsv -v $MRT_MODELS/rnn-spm/vocab.deen.{spm,spm} \
     --after-batches 100 --disp-freq 4 \
-    --guided-alignment 0 --guided-alignment-weight 1.0 \
+    --guided-alignment 0 --guided-alignment-weight 1.0 --guided-alignment-cost ce \
     --log train_align_shuffle.log
 
 # Check if files exist
diff --git a/tests/interface/input-tsv/test_tsv_train_with_align_shuffle_in_ram.sh b/tests/interface/input-tsv/test_tsv_train_with_align_shuffle_in_ram.sh
index 49e74de..a5db769 100644
--- a/tests/interface/input-tsv/test_tsv_train_with_align_shuffle_in_ram.sh
+++ b/tests/interface/input-tsv/test_tsv_train_with_align_shuffle_in_ram.sh
@@ -14,10 +14,11 @@ mkdir -p train_align_shuffle_ram
 
 # Run marian command
 $MRT_MARIAN/marian \
-    --cost-type ce-mean --shuffle-in-ram --clip-norm 1 --seed 4444 --dim-emb 32 --dim-rnn 64 --maxi-batch 1 --maxi-batch-sort none --optimizer sgd --learn-rate 0.1 \
+    --cost-type ce-mean --shuffle-in-ram --clip-norm 1 --seed 4444 --maxi-batch 1 --maxi-batch-sort none --optimizer adam --learn-rate 0.1 \
+    --dim-emb 32 --transformer-dim-ffn 64 --type transformer --enc-depth 3 --dec-depth 3 \
     -m train_align_shuffle_ram/model.npz --tsv -t train2.aln-de-en.tsv -v $MRT_MODELS/rnn-spm/vocab.deen.{spm,spm} \
     --after-batches 100 --disp-freq 4 \
-    --guided-alignment 0 --guided-alignment-weight 1.0 \
+    --guided-alignment 0 --guided-alignment-weight 1.0 --guided-alignment-cost ce \
     --log train_align_shuffle_ram.log
 
 # Check if files exist
diff --git a/tests/interface/input-tsv/test_tsv_train_with_align_stdin.sh b/tests/interface/input-tsv/test_tsv_train_with_align_stdin.sh
index 6eb5386..4e0bd32 100644
--- a/tests/interface/input-tsv/test_tsv_train_with_align_stdin.sh
+++ b/tests/interface/input-tsv/test_tsv_train_with_align_stdin.sh
@@ -14,10 +14,11 @@ mkdir -p train_align_stdin
 
 # Run marian command
 cat train2.aln-de-en.tsv | $MRT_MARIAN/marian \
-    --cost-type ce-mean --no-shuffle --clip-norm 0 --seed 5555 --dim-emb 32 --dim-rnn 64 --maxi-batch 1 --maxi-batch-sort none --optimizer sgd --learn-rate 0.001 \
+    --cost-type ce-mean --no-shuffle --clip-norm 0 --seed 5555 --maxi-batch 1 --maxi-batch-sort none --optimizer adam --learn-rate 0.001 \
+    --dim-emb 32 --transformer-dim-ffn 64 --type transformer --enc-depth 3 --dec-depth 3 \
     -m train_align_stdin/model.npz -t stdin -v $MRT_MODELS/rnn-spm/vocab.deen.{spm,spm} \
     --disp-freq 4 \
-    --guided-alignment 0 --guided-alignment-weight 1.0 \
+    --guided-alignment 0 --guided-alignment-weight 1.0 --guided-alignment-cost ce \
     --log train_align_stdin.log
 
 
diff --git a/tests/interface/input-tsv/train_align.expected b/tests/interface/input-tsv/train_align.expected
index 6d1c8b3..8114a79 100644
--- a/tests/interface/input-tsv/train_align.expected
+++ b/tests/interface/input-tsv/train_align.expected
@@ -1,25 +1,25 @@
-267.55108643
-238.90954590
-203.89498901
-194.40493774
-201.26391602
-196.71656799
-205.61053467
-190.75955200
-190.06002808
-175.82437134
-171.81805420
-183.65437317
-183.88174438
-195.11131287
-181.43255615
-182.07211304
-168.76817322
-167.46075439
-175.70928955
-179.71203613
-187.29899597
-175.05770874
-175.59832764
-165.63943481
-163.86834717
+349.90066528
+330.57034302
+301.72637939
+294.93923950
+311.47195435
+307.98052979
+318.67773438
+297.93566895
+289.55316162
+265.69094849
+258.18420410
+271.55532837
+267.22616577
+277.04589844
+259.60607910
+252.39709473
+232.80690002
+225.52084351
+237.79919434
+234.49322510
+244.92973328
+230.92803955
+226.21130371
+210.91824341
+204.61116028
diff --git a/tests/interface/input-tsv/train_align_shuffle.expected b/tests/interface/input-tsv/train_align_shuffle.expected
index 4c7909c..2db8857 100644
--- a/tests/interface/input-tsv/train_align_shuffle.expected
+++ b/tests/interface/input-tsv/train_align_shuffle.expected
@@ -1,25 +1,25 @@
-267.62625122
-260.25723267
-256.07604980
-251.13800049
-248.25172424
-242.42607117
-243.15475464
-226.86355591
-237.34564209
-228.19276428
-212.93722534
-214.65142822
-218.42727661
-213.79447937
-213.66851807
-201.51606750
-210.07385254
-198.46676636
-197.96029663
-205.45782471
-209.31356812
-201.94723511
-187.90913391
-197.95065308
-196.05712891
+305.95520020
+252.33764648
+241.97651672
+237.92004395
+241.01110840
+235.38877869
+244.05845642
+228.61547852
+244.48324585
+240.67021179
+225.93069458
+230.75164795
+239.92446899
+234.91127014
+236.48362732
+225.34948730
+236.05422974
+226.67541504
+227.21850586
+240.31086731
+247.07595825
+240.54412842
+221.21253967
+337.10803223
+236.50607300
diff --git a/tests/interface/input-tsv/train_align_stdin.expected b/tests/interface/input-tsv/train_align_stdin.expected
index e7a4610..6d8095f 100644
--- a/tests/interface/input-tsv/train_align_stdin.expected
+++ b/tests/interface/input-tsv/train_align_stdin.expected
@@ -1,7 +1,7 @@
-267.55108643
-238.90954590
-203.89497375
-194.40493774
-201.26391602
-196.71656799
-205.61053467
+349.90066528
+330.57034302
+301.72637939
+294.93923950
+311.47195435
+307.98052979
+318.67773438
diff --git a/tests/interface/input-tsv/train_align_weights.expected b/tests/interface/input-tsv/train_align_weights.expected
index 4092789..77d155e 100644
--- a/tests/interface/input-tsv/train_align_weights.expected
+++ b/tests/interface/input-tsv/train_align_weights.expected
@@ -1,15 +1,15 @@
-341.56268311
-328.85687256
-300.68945312
-295.04937744
-322.07330322
-306.52780151
-332.50936890
-305.94641113
-290.83953857
-265.52841187
-256.46743774
-278.98114014
-265.27020264
-292.67654419
-273.39343262
+366.47402954
+311.10235596
+278.69085693
+266.78692627
+289.20837402
+277.70391846
+308.79412842
+286.08660889
+281.75717163
+269.53497314
+256.59310913
+292.10559082
+316.60882568
+305.43435669
+286.01943970
diff --git a/tests/training/features/guided-alignment/rnn.expected b/tests/training/features/guided-alignment/rnn.expected
index 0cefbcf..3b07346 100644
--- a/tests/training/features/guided-alignment/rnn.expected
+++ b/tests/training/features/guided-alignment/rnn.expected
@@ -1,10 +1,10 @@
-236.62251282
-256.09661865
-241.58700562
-236.70104980
-224.99372864
-204.95254517
-235.38345337
-205.25239563
-215.87429810
-202.22142029
+297.51348877
+323.35086060
+308.66668701
+304.00738525
+290.16757202
+264.55703735
+307.27072144
+267.86627197
+281.67095947
+264.18920898
diff --git a/tests/training/features/guided-alignment/test_guided_alignment_rnn.sh b/tests/training/features/guided-alignment/test_guided_alignment_rnn.sh
index 925f36f..ca9f030 100644
--- a/tests/training/features/guided-alignment/test_guided_alignment_rnn.sh
+++ b/tests/training/features/guided-alignment/test_guided_alignment_rnn.sh
@@ -18,7 +18,7 @@ $MRT_MARIAN/marian \
     --no-shuffle --clip-norm 1 --seed 1111 --dim-emb 32 --dim-rnn 64 --mini-batch 32 --maxi-batch 1 --maxi-batch-sort none --optimizer sgd --cost-type ce-mean \
     -m rnn/model.npz -t corpus.bpe.{en,de} -v vocab.en.yml vocab.de.yml \
     --after-batches 100 --disp-freq 10 \
-    --guided-alignment corpus.bpe.align --guided-alignment-weight 1.0 --learn-rate 0.1 \
+    --guided-alignment corpus.bpe.align --guided-alignment-weight 1.0 --guided-alignment-cost ce --learn-rate 0.1 \
     --log rnn.log
 
 # Check if files exist
diff --git a/tests/training/features/guided-alignment/test_guided_alignment_transformer.sh b/tests/training/features/guided-alignment/test_guided_alignment_transformer.sh
index cd28f1c..cdca3c4 100644
--- a/tests/training/features/guided-alignment/test_guided_alignment_transformer.sh
+++ b/tests/training/features/guided-alignment/test_guided_alignment_transformer.sh
@@ -15,10 +15,10 @@ mkdir -p transformer
 
 # Run marian command
 $MRT_MARIAN/marian --type transformer \
-    --no-shuffle --clip-norm 0 --seed 2222 --dim-emb 32 --dim-rnn 64 --mini-batch 32 --maxi-batch 1 --maxi-batch-sort none --optimizer sgd --cost-type ce-mean \
+    --no-shuffle --clip-norm 0 --seed 2222 --dim-emb 32 --dim-rnn 64 --mini-batch 32 --maxi-batch 1 --maxi-batch-sort none --optimizer adam --cost-type ce-mean \
     -m transformer/model.npz -t corpus.bpe.{en,de} -v vocab.en.yml vocab.de.yml \
     --after-batches 100 --disp-freq 10 \
-    --guided-alignment corpus.bpe.align --guided-alignment-weight 1.0 --learn-rate 0.1 \
+    --guided-alignment corpus.bpe.align  --guided-alignment-weight 1 --guided-alignment-cost ce --learn-rate 0.01 \
     --log transformer.log
 
 # Check if files exist
diff --git a/tests/training/features/guided-alignment/test_guided_alignment_transformer_sync.sh b/tests/training/features/guided-alignment/test_guided_alignment_transformer_sync.sh
index 963052d..eb1bb70 100644
--- a/tests/training/features/guided-alignment/test_guided_alignment_transformer_sync.sh
+++ b/tests/training/features/guided-alignment/test_guided_alignment_transformer_sync.sh
@@ -15,10 +15,10 @@ mkdir -p transformer_sync
 
 # Run marian command
 $MRT_MARIAN/marian --type transformer \
-    --no-shuffle --clip-norm 0 --seed 2222 --dim-emb 32 --dim-rnn 64 --mini-batch 32 --maxi-batch 1 --maxi-batch-sort none --optimizer sgd --cost-type ce-mean --sync-sgd \
+    --no-shuffle --clip-norm 0 --seed 2222 --dim-emb 32 --dim-rnn 64 --mini-batch 32 --maxi-batch 1 --maxi-batch-sort none --optimizer adam --cost-type ce-mean --sync-sgd \
     -m transformer_sync/model.npz -t corpus.bpe.{en,de} -v vocab.en.yml vocab.de.yml \
     --after-batches 100 --disp-freq 10 \
-    --guided-alignment corpus.bpe.align --guided-alignment-weight 1.0 --learn-rate 0.1 \
+    --guided-alignment corpus.bpe.align  --guided-alignment-weight 1 --guided-alignment-cost ce --learn-rate 0.01 \
     --log transformer_sync.log
 
 # Check if files exist
diff --git a/tests/training/features/guided-alignment/transformer.expected b/tests/training/features/guided-alignment/transformer.expected
index e69de29..56a6ff2 100644
--- a/tests/training/features/guided-alignment/transformer.expected
+++ b/tests/training/features/guided-alignment/transformer.expected
@@ -0,0 +1,10 @@
+276.90930176
+258.53686523
+244.95483398
+240.12124634
+234.31253052
+208.04850769
+234.38320923
+205.47990417
+212.96972656
+199.58921814