Iteration 1, loss = 1.63217924
Iteration 2, loss = 1.29897925
Iteration 3, loss = 1.27730382
Iteration 4, loss = 1.25522668
Iteration 5, loss = 1.24093841
Iteration 6, loss = 1.22688958
Iteration 7, loss = 1.21007260
Iteration 8, loss = 1.19890125
Iteration 9, loss = 1.19354226
Iteration 10, loss = 1.17991083
Iteration 11, loss = 1.17395744
Iteration 12, loss = 1.16664669
Iteration 13, loss = 1.16354717
Iteration 14, loss = 1.14823746
Iteration 15, loss = 1.14443052
Iteration 16, loss = 1.13545109
Iteration 17, loss = 1.13031813
Iteration 18, loss = 1.11902953
Iteration 19, loss = 1.11372865
Iteration 20, loss = 1.10703823
Iteration 21, loss = 1.09622826
Iteration 22, loss = 1.08154347
Iteration 23, loss = 1.07831238
Iteration 24, loss = 1.07372885
Iteration 25, loss = 1.06327184
Iteration 26, loss = 1.06619736
Iteration 27, loss = 1.04959434
Iteration 28, loss = 1.04284265
Iteration 29, loss = 1.02761738
Iteration 30, loss = 1.02272489
Iteration 31, loss = 1.01919685
Iteration 32, loss = 1.00956499
Iteration 33, loss = 0.99425610
Iteration 34, loss = 0.99532218
Iteration 35, loss = 0.99069944
Iteration 36, loss = 0.98671093
Iteration 37, loss = 0.96517603
Iteration 38, loss = 0.95891830
Iteration 39, loss = 0.95367584
Iteration 40, loss = 0.94397054
Iteration 41, loss = 0.94235047
Iteration 42, loss = 0.93516203
Iteration 43, loss = 0.92985783
Iteration 44, loss = 0.92109383
Iteration 45, loss = 0.90783774
Iteration 46, loss = 0.92519276
Iteration 47, loss = 0.90078778
Iteration 48, loss = 0.90864331
Iteration 49, loss = 0.89139658
Iteration 50, loss = 0.89450404
Iteration 51, loss = 0.88815468
Iteration 52, loss = 0.86398860
Iteration 53, loss = 0.87716786
Iteration 54, loss = 0.87152422
Iteration 55, loss = 0.85379970
Iteration 56, loss = 0.85295028
Iteration 57, loss = 0.84762872
Iteration 58, loss = 0.85117170
Iteration 59, loss = 0.84828352
Iteration 60, loss = 0.83197196
Iteration 61, loss = 0.81883705
Iteration 62, loss = 0.82985756
Iteration 63, loss = 0.82917493
Iteration 64, loss = 0.82029196
Training loss did not improve more than tol=0.000100 for two consecutive epochs. Stopping.
Confusion matrix
668	102	32	109	21
158	1499	200	517	93
86	187	2155	1126	218
52	111	291	3784	428
11	66	120	1401	2187
************************************************************
	Recall		Precision	F1
A1	0.716738197425 0.685128205128 0.700576822234
A2	0.607620591812 0.762849872774 0.676444043321
B1	0.57131495228 0.770192994996 0.65601217656
B2	0.810972996142 0.54548075537 0.652245109024
C1	0.577807133421 0.74211062097 0.649732620321
------------------------------------------------------------
Average F1: 0.667002154292
Best classifier: siwoco-svalex-d-topic-bigram-compact-nh.csv-mlp-1000_500.pickle, F1: 0.667002154292
Iteration 1, loss = 1.41151680
Iteration 2, loss = 1.27024759
Iteration 3, loss = 1.23111660
Iteration 4, loss = 1.20230034
Iteration 5, loss = 1.18394528
Iteration 6, loss = 1.15792778
Iteration 7, loss = 1.14594665
Iteration 8, loss = 1.13226360
Iteration 9, loss = 1.11452620
Iteration 10, loss = 1.09820225
Iteration 11, loss = 1.07923586
Iteration 12, loss = 1.06433618
Iteration 13, loss = 1.04888692
Iteration 14, loss = 1.03050483
Iteration 15, loss = 1.01663433
Iteration 16, loss = 1.00652480
Iteration 17, loss = 0.98714291
Iteration 18, loss = 0.97500442
Iteration 19, loss = 0.95951131
Iteration 20, loss = 0.94523237
Iteration 21, loss = 0.93611174
Iteration 22, loss = 0.91550521
Iteration 23, loss = 0.91389205
Iteration 24, loss = 0.90308356
Iteration 25, loss = 0.89241600
Iteration 26, loss = 0.87864259
Iteration 27, loss = 0.87569415
Iteration 28, loss = 0.86544661
Iteration 29, loss = 0.85119932
Iteration 30, loss = 0.84301555
Iteration 31, loss = 0.83845531
Iteration 32, loss = 0.83966548
Iteration 33, loss = 0.82359338
Iteration 34, loss = 0.81910791
Iteration 35, loss = 0.82000863
Iteration 36, loss = 0.81091616
Iteration 37, loss = 0.80667950
Iteration 38, loss = 0.80104551
Iteration 39, loss = 0.79467179
Iteration 40, loss = 0.79057311
Iteration 41, loss = 0.78430801
Iteration 42, loss = 0.79344981
Iteration 43, loss = 0.78106335
Iteration 44, loss = 0.77546895
Iteration 45, loss = 0.76744378
Iteration 46, loss = 0.77006133
Iteration 47, loss = 0.76358232
Iteration 48, loss = 0.77591392
Iteration 49, loss = 0.75622101
Iteration 50, loss = 0.76075541
Iteration 51, loss = 0.75058510
Iteration 52, loss = 0.74988546
Iteration 53, loss = 0.74199794
Iteration 54, loss = 0.74396934
Iteration 55, loss = 0.74448147
Iteration 56, loss = 0.74018260
Iteration 57, loss = 0.73281807
Iteration 58, loss = 0.74013454
Iteration 59, loss = 0.73182494
Iteration 60, loss = 0.73216567
Iteration 61, loss = 0.72403816
Iteration 62, loss = 0.73906084
Iteration 63, loss = 0.73005007
Iteration 64, loss = 0.72139841
Iteration 65, loss = 0.71720542
Iteration 66, loss = 0.72225095
Iteration 67, loss = 0.71639589
Iteration 68, loss = 0.72093706
Iteration 69, loss = 0.71324224
Iteration 70, loss = 0.73219052
Iteration 71, loss = 0.71278867
Iteration 72, loss = 0.70307615
Iteration 73, loss = 0.69047978
Iteration 74, loss = 0.70566755
Iteration 75, loss = 0.71029788
Iteration 76, loss = 0.70625470
Training loss did not improve more than tol=0.000100 for two consecutive epochs. Stopping.
Confusion matrix
703	82	28	49	70
129	1712	90	191	345
77	176	2226	487	806
34	99	135	3205	1193
10	34	50	481	3210
************************************************************
	Recall		Precision	F1
A1	0.754291845494 0.737670514166 0.745888594164
A2	0.693960275638 0.814075130766 0.749234135667
B1	0.5901378579 0.880189798339 0.706554515156
B2	0.686883840549 0.726263312939 0.706024892609
C1	0.848084544254 0.570768136558 0.682325433096
------------------------------------------------------------
Average F1: 0.718005514139
Best classifier: siwoco-svalex-d-topic-bigram-compact-nh.csv-mlp-1000_200.pickle, F1: 0.718005514139
Iteration 1, loss = 1.39669513
Iteration 2, loss = 1.23092260
Iteration 3, loss = 1.18372219
Iteration 4, loss = 1.15007254
Iteration 5, loss = 1.11795176
Iteration 6, loss = 1.09203509
Iteration 7, loss = 1.06191715
Iteration 8, loss = 1.03713142
Iteration 9, loss = 1.00780452
Iteration 10, loss = 0.97911537
Iteration 11, loss = 0.95793701
Iteration 12, loss = 0.92457695
Iteration 13, loss = 0.89971460
Iteration 14, loss = 0.87730020
Iteration 15, loss = 0.86278500
Iteration 16, loss = 0.83398762
Iteration 17, loss = 0.81249432
Iteration 18, loss = 0.79202377
Iteration 19, loss = 0.77175226
Iteration 20, loss = 0.76830272
Iteration 21, loss = 0.75769455
Iteration 22, loss = 0.75178979
Iteration 23, loss = 0.73023983
Iteration 24, loss = 0.71083299
Iteration 25, loss = 0.71041000
Iteration 26, loss = 0.70697462
Iteration 27, loss = 0.68547457
Iteration 28, loss = 0.69224941
Iteration 29, loss = 0.68648225
Iteration 30, loss = 0.67430251
Iteration 31, loss = 0.66194001
Iteration 32, loss = 0.65417227
Iteration 33, loss = 0.66224307
Iteration 34, loss = 0.65107110
Iteration 35, loss = 0.67200574
Iteration 36, loss = 0.63604518
Iteration 37, loss = 0.65056890
Iteration 38, loss = 0.64771564
Iteration 39, loss = 0.63726401
Training loss did not improve more than tol=0.000100 for two consecutive epochs. Stopping.
Confusion matrix
722	59	32	81	38
141	1722	94	352	158
58	95	2468	792	359
25	65	122	3869	585
6	31	47	919	2782
************************************************************
	Recall		Precision	F1
A1	0.774678111588 0.758403361345 0.766454352442
A2	0.698013781921 0.87322515213 0.775850416761
B1	0.654294803818 0.893231994209 0.755317521041
B2	0.829189884269 0.643439215034 0.724599681618
C1	0.73500660502 0.709331973483 0.721941092513
------------------------------------------------------------
Average F1: 0.748832612875
Best classifier: siwoco-svalex-d-topic-bigram-compact-nh.csv-mlp-1000_500_200.pickle, F1: 0.748832612875
Iteration 1, loss = 1.33104228
Iteration 2, loss = 1.20346920
Iteration 3, loss = 1.16401840
Iteration 4, loss = 1.13328039
Iteration 5, loss = 1.10234616
Iteration 6, loss = 1.07127512
Iteration 7, loss = 1.04500742
Iteration 8, loss = 1.00936258
Iteration 9, loss = 0.97727403
Iteration 10, loss = 0.94370197
Iteration 11, loss = 0.91181290
Iteration 12, loss = 0.88612274
Iteration 13, loss = 0.86524172
Iteration 14, loss = 0.82825051
Iteration 15, loss = 0.81169525
Iteration 16, loss = 0.78714301
Iteration 17, loss = 0.77279375
Iteration 18, loss = 0.74496337
Iteration 19, loss = 0.74414057
Iteration 20, loss = 0.72335508
Iteration 21, loss = 0.70204829
Iteration 22, loss = 0.69990612
Iteration 23, loss = 0.68343503
Iteration 24, loss = 0.67307930
Iteration 25, loss = 0.67086144
Iteration 26, loss = 0.66069142
Iteration 27, loss = 0.65106786
Iteration 28, loss = 0.65464785
Iteration 29, loss = 0.64499308
Iteration 30, loss = 0.65077066
Iteration 31, loss = 0.61696463
Iteration 32, loss = 0.64215493
Iteration 33, loss = 0.61659225
Iteration 34, loss = 0.61198555
Iteration 35, loss = 0.61709048
Iteration 36, loss = 0.60790091
Iteration 37, loss = 0.61469324
Iteration 38, loss = 0.60138699
Iteration 39, loss = 0.59863811
Iteration 40, loss = 0.59568754
Iteration 41, loss = 0.59865195
Iteration 42, loss = 0.60172210
Iteration 43, loss = 0.59474056
Iteration 44, loss = 0.58407398
Iteration 45, loss = 0.57945125
Iteration 46, loss = 0.59990012
Iteration 47, loss = 0.59651762
Iteration 48, loss = 0.57310489
Iteration 49, loss = 0.57561555
Iteration 50, loss = 0.59178559
Iteration 51, loss = 0.56555400
Iteration 52, loss = 0.57246449
Iteration 53, loss = 0.58492291
Iteration 54, loss = 0.56626347
Training loss did not improve more than tol=0.000100 for two consecutive epochs. Stopping.
Confusion matrix
782	20	16	61	53
146	1789	63	241	228
57	70	2592	550	503
33	54	102	3674	803
5	6	55	604	3115
************************************************************
	Recall		Precision	F1
A1	0.839055793991 0.764418377322 0.8
A2	0.725172274017 0.922640536359 0.81207444394
B1	0.687168610817 0.916548797737 0.785454545455
B2	0.787398199743 0.716179337232 0.750102082483
C1	0.822985468956 0.662484049341 0.734063862378
------------------------------------------------------------
Average F1: 0.776338986851
Best classifier: siwoco-svalex-d-topic-bigram-compact-nh.csv-mlp-1000_500_200_50.pickle, F1: 0.776338986851
Iteration 1, loss = 1.32148322
Iteration 2, loss = 1.19000170
Iteration 3, loss = 1.14051782
Iteration 4, loss = 1.10326718
Iteration 5, loss = 1.06418372
Iteration 6, loss = 1.02701082
Iteration 7, loss = 0.98490259
Iteration 8, loss = 0.94788270
Iteration 9, loss = 0.91161392
Iteration 10, loss = 0.87645161
Iteration 11, loss = 0.84318358
Iteration 12, loss = 0.81503214
Iteration 13, loss = 0.79622325
Iteration 14, loss = 0.75960748
Iteration 15, loss = 0.74092739
Iteration 16, loss = 0.72395460
Iteration 17, loss = 0.70488073
Iteration 18, loss = 0.69911362
Iteration 19, loss = 0.69003583
Iteration 20, loss = 0.68276400
Iteration 21, loss = 0.66432495
Iteration 22, loss = 0.66111947
Iteration 23, loss = 0.65595776
Iteration 24, loss = 0.64046946
Iteration 25, loss = 0.63676308
Iteration 26, loss = 0.62556736
Iteration 27, loss = 0.64609855
Iteration 28, loss = 0.62989166
Iteration 29, loss = 0.62109403
Iteration 30, loss = 0.61390071
Iteration 31, loss = 0.61081286
Iteration 32, loss = 0.61573452
Iteration 33, loss = 0.59100513
Iteration 34, loss = 0.59330247
Iteration 35, loss = 0.61128970
Iteration 36, loss = 0.60331216
Training loss did not improve more than tol=0.000100 for two consecutive epochs. Stopping.