Uploaded the model

final-model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:540573dc73c98f5b1fa42c789f4465e7f1b2c7f326d7461dfdeada7d2522644b
+size 444998061

loss.tsv

@@ -0,0 +1,11 @@
+EPOCH	TIMESTAMP	BAD_EPOCHS	LEARNING_RATE	TRAIN_LOSS	DEV_LOSS	DEV_PRECISION	DEV_RECALL	DEV_F1	DEV_ACCURACY
+1	02:16:47	4	0.0001	0.2182895945059601	0.0355144739151001	0.7661	0.8947	0.8254	0.722
+2	03:26:13	4	0.0000	0.13729935181138425	0.015243684872984886	0.9007	0.926	0.9132	0.8548
+3	04:35:09	4	0.0000	0.11197359439314927	0.016585879027843475	0.9119	0.9342	0.9229	0.8697
+4	05:44:00	4	0.0000	0.09147635538963178	0.016923826187849045	0.9132	0.9296	0.9213	0.8708
+5	06:52:15	4	0.0000	0.07495889990317275	0.017464155331254005	0.9377	0.9246	0.9311	0.8831
+6	08:00:45	4	0.0000	0.061747689342078395	0.01982131227850914	0.9348	0.9369	0.9358	0.8909
+7	09:09:20	4	0.0000	0.0519030773124998	0.02467426098883152	0.9395	0.9315	0.9355	0.892
+8	10:18:00	4	0.0000	0.04503195115695853	0.02364770695567131	0.9306	0.9438	0.9371	0.8939
+9	11:26:37	4	0.0000	0.040509963133028556	0.026182951405644417	0.9328	0.9394	0.9361	0.8942
+10	12:34:45	4	0.0000	0.03798332489249556	0.027400659397244453	0.9349	0.9388	0.9368	0.8943

training.log

@@ -0,0 +1,499 @@
+2022-02-05 01:08:47,419 ----------------------------------------------------------------------------------------------------
+2022-02-05 01:08:47,461 Model: "SequenceTagger(
+  (embeddings): TransformerWordEmbeddings(
+    (model): RobertaModel(
+      (embeddings): RobertaEmbeddings(
+        (word_embeddings): Embedding(32768, 768, padding_idx=1)
+        (position_embeddings): Embedding(514, 768, padding_idx=1)
+        (token_type_embeddings): Embedding(1, 768)
+        (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (encoder): RobertaEncoder(
+        (layer): ModuleList(
+          (0): RobertaLayer(
+            (attention): RobertaAttention(
+              (self): RobertaSelfAttention(
+                (query): Linear(in_features=768, out_features=768, bias=True)
+                (key): Linear(in_features=768, out_features=768, bias=True)
+                (value): Linear(in_features=768, out_features=768, bias=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+              (output): RobertaSelfOutput(
+                (dense): Linear(in_features=768, out_features=768, bias=True)
+                (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+            )
+            (intermediate): RobertaIntermediate(
+              (dense): Linear(in_features=768, out_features=3072, bias=True)
+            )
+            (output): RobertaOutput(
+              (dense): Linear(in_features=3072, out_features=768, bias=True)
+              (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+              (dropout): Dropout(p=0.1, inplace=False)
+            )
+          )
+          (1): RobertaLayer(
+            (attention): RobertaAttention(
+              (self): RobertaSelfAttention(
+                (query): Linear(in_features=768, out_features=768, bias=True)
+                (key): Linear(in_features=768, out_features=768, bias=True)
+                (value): Linear(in_features=768, out_features=768, bias=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+              (output): RobertaSelfOutput(
+                (dense): Linear(in_features=768, out_features=768, bias=True)
+                (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+            )
+            (intermediate): RobertaIntermediate(
+              (dense): Linear(in_features=768, out_features=3072, bias=True)
+            )
+            (output): RobertaOutput(
+              (dense): Linear(in_features=3072, out_features=768, bias=True)
+              (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+              (dropout): Dropout(p=0.1, inplace=False)
+            )
+          )
+          (2): RobertaLayer(
+            (attention): RobertaAttention(
+              (self): RobertaSelfAttention(
+                (query): Linear(in_features=768, out_features=768, bias=True)
+                (key): Linear(in_features=768, out_features=768, bias=True)
+                (value): Linear(in_features=768, out_features=768, bias=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+              (output): RobertaSelfOutput(
+                (dense): Linear(in_features=768, out_features=768, bias=True)
+                (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+            )
+            (intermediate): RobertaIntermediate(
+              (dense): Linear(in_features=768, out_features=3072, bias=True)
+            )
+            (output): RobertaOutput(
+              (dense): Linear(in_features=3072, out_features=768, bias=True)
+              (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+              (dropout): Dropout(p=0.1, inplace=False)
+            )
+          )
+          (3): RobertaLayer(
+            (attention): RobertaAttention(
+              (self): RobertaSelfAttention(
+                (query): Linear(in_features=768, out_features=768, bias=True)
+                (key): Linear(in_features=768, out_features=768, bias=True)
+                (value): Linear(in_features=768, out_features=768, bias=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+              (output): RobertaSelfOutput(
+                (dense): Linear(in_features=768, out_features=768, bias=True)
+                (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+            )
+            (intermediate): RobertaIntermediate(
+              (dense): Linear(in_features=768, out_features=3072, bias=True)
+            )
+            (output): RobertaOutput(
+              (dense): Linear(in_features=3072, out_features=768, bias=True)
+              (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+              (dropout): Dropout(p=0.1, inplace=False)
+            )
+          )
+          (4): RobertaLayer(
+            (attention): RobertaAttention(
+              (self): RobertaSelfAttention(
+                (query): Linear(in_features=768, out_features=768, bias=True)
+                (key): Linear(in_features=768, out_features=768, bias=True)
+                (value): Linear(in_features=768, out_features=768, bias=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+              (output): RobertaSelfOutput(
+                (dense): Linear(in_features=768, out_features=768, bias=True)
+                (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+            )
+            (intermediate): RobertaIntermediate(
+              (dense): Linear(in_features=768, out_features=3072, bias=True)
+            )
+            (output): RobertaOutput(
+              (dense): Linear(in_features=3072, out_features=768, bias=True)
+              (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+              (dropout): Dropout(p=0.1, inplace=False)
+            )
+          )
+          (5): RobertaLayer(
+            (attention): RobertaAttention(
+              (self): RobertaSelfAttention(
+                (query): Linear(in_features=768, out_features=768, bias=True)
+                (key): Linear(in_features=768, out_features=768, bias=True)
+                (value): Linear(in_features=768, out_features=768, bias=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+              (output): RobertaSelfOutput(
+                (dense): Linear(in_features=768, out_features=768, bias=True)
+                (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+            )
+            (intermediate): RobertaIntermediate(
+              (dense): Linear(in_features=768, out_features=3072, bias=True)
+            )
+            (output): RobertaOutput(
+              (dense): Linear(in_features=3072, out_features=768, bias=True)
+              (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+              (dropout): Dropout(p=0.1, inplace=False)
+            )
+          )
+          (6): RobertaLayer(
+            (attention): RobertaAttention(
+              (self): RobertaSelfAttention(
+                (query): Linear(in_features=768, out_features=768, bias=True)
+                (key): Linear(in_features=768, out_features=768, bias=True)
+                (value): Linear(in_features=768, out_features=768, bias=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+              (output): RobertaSelfOutput(
+                (dense): Linear(in_features=768, out_features=768, bias=True)
+                (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+            )
+            (intermediate): RobertaIntermediate(
+              (dense): Linear(in_features=768, out_features=3072, bias=True)
+            )
+            (output): RobertaOutput(
+              (dense): Linear(in_features=3072, out_features=768, bias=True)
+              (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+              (dropout): Dropout(p=0.1, inplace=False)
+            )
+          )
+          (7): RobertaLayer(
+            (attention): RobertaAttention(
+              (self): RobertaSelfAttention(
+                (query): Linear(in_features=768, out_features=768, bias=True)
+                (key): Linear(in_features=768, out_features=768, bias=True)
+                (value): Linear(in_features=768, out_features=768, bias=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+              (output): RobertaSelfOutput(
+                (dense): Linear(in_features=768, out_features=768, bias=True)
+                (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+            )
+            (intermediate): RobertaIntermediate(
+              (dense): Linear(in_features=768, out_features=3072, bias=True)
+            )
+            (output): RobertaOutput(
+              (dense): Linear(in_features=3072, out_features=768, bias=True)
+              (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+              (dropout): Dropout(p=0.1, inplace=False)
+            )
+          )
+          (8): RobertaLayer(
+            (attention): RobertaAttention(
+              (self): RobertaSelfAttention(
+                (query): Linear(in_features=768, out_features=768, bias=True)
+                (key): Linear(in_features=768, out_features=768, bias=True)
+                (value): Linear(in_features=768, out_features=768, bias=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+              (output): RobertaSelfOutput(
+                (dense): Linear(in_features=768, out_features=768, bias=True)
+                (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+            )
+            (intermediate): RobertaIntermediate(
+              (dense): Linear(in_features=768, out_features=3072, bias=True)
+            )
+            (output): RobertaOutput(
+              (dense): Linear(in_features=3072, out_features=768, bias=True)
+              (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+              (dropout): Dropout(p=0.1, inplace=False)
+            )
+          )
+          (9): RobertaLayer(
+            (attention): RobertaAttention(
+              (self): RobertaSelfAttention(
+                (query): Linear(in_features=768, out_features=768, bias=True)
+                (key): Linear(in_features=768, out_features=768, bias=True)
+                (value): Linear(in_features=768, out_features=768, bias=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+              (output): RobertaSelfOutput(
+                (dense): Linear(in_features=768, out_features=768, bias=True)
+                (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+            )
+            (intermediate): RobertaIntermediate(
+              (dense): Linear(in_features=768, out_features=3072, bias=True)
+            )
+            (output): RobertaOutput(
+              (dense): Linear(in_features=3072, out_features=768, bias=True)
+              (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+              (dropout): Dropout(p=0.1, inplace=False)
+            )
+          )
+          (10): RobertaLayer(
+            (attention): RobertaAttention(
+              (self): RobertaSelfAttention(
+                (query): Linear(in_features=768, out_features=768, bias=True)
+                (key): Linear(in_features=768, out_features=768, bias=True)
+                (value): Linear(in_features=768, out_features=768, bias=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+              (output): RobertaSelfOutput(
+                (dense): Linear(in_features=768, out_features=768, bias=True)
+                (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+            )
+            (intermediate): RobertaIntermediate(
+              (dense): Linear(in_features=768, out_features=3072, bias=True)
+            )
+            (output): RobertaOutput(
+              (dense): Linear(in_features=3072, out_features=768, bias=True)
+              (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+              (dropout): Dropout(p=0.1, inplace=False)
+            )
+          )
+          (11): RobertaLayer(
+            (attention): RobertaAttention(
+              (self): RobertaSelfAttention(
+                (query): Linear(in_features=768, out_features=768, bias=True)
+                (key): Linear(in_features=768, out_features=768, bias=True)
+                (value): Linear(in_features=768, out_features=768, bias=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+              (output): RobertaSelfOutput(
+                (dense): Linear(in_features=768, out_features=768, bias=True)
+                (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+            )
+            (intermediate): RobertaIntermediate(
+              (dense): Linear(in_features=768, out_features=3072, bias=True)
+            )
+            (output): RobertaOutput(
+              (dense): Linear(in_features=3072, out_features=768, bias=True)
+              (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+              (dropout): Dropout(p=0.1, inplace=False)
+            )
+          )
+        )
+      )
+      (pooler): RobertaPooler(
+        (dense): Linear(in_features=768, out_features=768, bias=True)
+        (activation): Tanh()
+      )
+    )
+  )
+  (word_dropout): WordDropout(p=0.05)
+  (locked_dropout): LockedDropout(p=0.5)
+  (linear): Linear(in_features=768, out_features=18, bias=True)
+  (beta): 1.0
+  (weights): None
+  (weight_tensor) None
+)"
+2022-02-05 01:08:47,466 ----------------------------------------------------------------------------------------------------
+2022-02-05 01:08:47,466 Corpus: "Corpus: 126973 train + 7037 dev + 7090 test sentences"
+2022-02-05 01:08:47,466 ----------------------------------------------------------------------------------------------------
+2022-02-05 01:08:47,466 Parameters:
+2022-02-05 01:08:47,466  - learning_rate: "5e-05"
+2022-02-05 01:08:47,466  - mini_batch_size: "16"
+2022-02-05 01:08:47,466  - patience: "3"
+2022-02-05 01:08:47,466  - anneal_factor: "0.5"
+2022-02-05 01:08:47,466  - max_epochs: "10"
+2022-02-05 01:08:47,466  - shuffle: "True"
+2022-02-05 01:08:47,466  - train_with_dev: "False"
+2022-02-05 01:08:47,466  - batch_growth_annealing: "False"
+2022-02-05 01:08:47,466 ----------------------------------------------------------------------------------------------------
+2022-02-05 01:08:47,466 Model training base path: "resources/taggers/ner-dalembert-2ndtry"
+2022-02-05 01:08:47,466 ----------------------------------------------------------------------------------------------------
+2022-02-05 01:08:47,466 Device: cuda:0
+2022-02-05 01:08:47,466 ----------------------------------------------------------------------------------------------------
+2022-02-05 01:08:47,467 Embeddings storage mode: none
+2022-02-05 01:08:47,469 ----------------------------------------------------------------------------------------------------
+2022-02-05 01:15:08,771 epoch 1 - iter 793/7936 - loss 0.78007372 - samples/sec: 33.28 - lr: 0.000005
+2022-02-05 01:22:45,940 epoch 1 - iter 1586/7936 - loss 0.41932043 - samples/sec: 27.76 - lr: 0.000010
+2022-02-05 01:29:23,897 epoch 1 - iter 2379/7936 - loss 0.33514542 - samples/sec: 31.89 - lr: 0.000015
+2022-02-05 01:35:24,915 epoch 1 - iter 3172/7936 - loss 0.30212998 - samples/sec: 35.15 - lr: 0.000020
+2022-02-05 01:42:28,297 epoch 1 - iter 3965/7936 - loss 0.27341208 - samples/sec: 29.97 - lr: 0.000025
+2022-02-05 01:49:23,543 epoch 1 - iter 4758/7936 - loss 0.25403588 - samples/sec: 30.56 - lr: 0.000030
+2022-02-05 01:55:46,783 epoch 1 - iter 5551/7936 - loss 0.24241496 - samples/sec: 33.11 - lr: 0.000035
+2022-02-05 02:01:45,654 epoch 1 - iter 6344/7936 - loss 0.23381719 - samples/sec: 35.36 - lr: 0.000040
+2022-02-05 02:07:29,407 epoch 1 - iter 7137/7936 - loss 0.22586308 - samples/sec: 36.92 - lr: 0.000045
+2022-02-05 02:13:54,603 epoch 1 - iter 7930/7936 - loss 0.21834611 - samples/sec: 32.94 - lr: 0.000050
+2022-02-05 02:13:57,692 ----------------------------------------------------------------------------------------------------
+2022-02-05 02:13:57,693 EPOCH 1 done: loss 0.2183 - lr 0.0000500
+2022-02-05 02:16:47,190 DEV : loss 0.0355144739151001 - f1-score (micro avg)  0.8254
+2022-02-05 02:16:47,244 BAD EPOCHS (no improvement): 4
+2022-02-05 02:16:47,244 ----------------------------------------------------------------------------------------------------
+2022-02-05 02:23:15,435 epoch 2 - iter 793/7936 - loss 0.14903310 - samples/sec: 32.69 - lr: 0.000049
+2022-02-05 02:30:06,605 epoch 2 - iter 1586/7936 - loss 0.14777394 - samples/sec: 30.86 - lr: 0.000049
+2022-02-05 02:36:48,570 epoch 2 - iter 2379/7936 - loss 0.14637300 - samples/sec: 31.57 - lr: 0.000048
+2022-02-05 02:43:37,172 epoch 2 - iter 3172/7936 - loss 0.14491485 - samples/sec: 31.06 - lr: 0.000048
+2022-02-05 02:50:13,040 epoch 2 - iter 3965/7936 - loss 0.14361996 - samples/sec: 32.06 - lr: 0.000047
+2022-02-05 02:56:49,904 epoch 2 - iter 4758/7936 - loss 0.14232123 - samples/sec: 31.98 - lr: 0.000047
+2022-02-05 03:03:34,383 epoch 2 - iter 5551/7936 - loss 0.14116820 - samples/sec: 31.38 - lr: 0.000046
+2022-02-05 03:10:09,778 epoch 2 - iter 6344/7936 - loss 0.14001072 - samples/sec: 32.10 - lr: 0.000046
+2022-02-05 03:16:43,847 epoch 2 - iter 7137/7936 - loss 0.13868572 - samples/sec: 32.20 - lr: 0.000045
+2022-02-05 03:23:28,994 epoch 2 - iter 7930/7936 - loss 0.13731517 - samples/sec: 31.33 - lr: 0.000044
+2022-02-05 03:23:31,622 ----------------------------------------------------------------------------------------------------
+2022-02-05 03:23:31,623 EPOCH 2 done: loss 0.1373 - lr 0.0000444
+2022-02-05 03:26:13,727 DEV : loss 0.015243684872984886 - f1-score (micro avg)  0.9132
+2022-02-05 03:26:13,788 BAD EPOCHS (no improvement): 4
+2022-02-05 03:26:13,806 ----------------------------------------------------------------------------------------------------
+2022-02-05 03:32:57,765 epoch 3 - iter 793/7936 - loss 0.11924788 - samples/sec: 31.42 - lr: 0.000044
+2022-02-05 03:39:33,229 epoch 3 - iter 1586/7936 - loss 0.11867811 - samples/sec: 32.09 - lr: 0.000043
+2022-02-05 03:46:09,619 epoch 3 - iter 2379/7936 - loss 0.11819415 - samples/sec: 32.01 - lr: 0.000043
+2022-02-05 03:52:49,510 epoch 3 - iter 3172/7936 - loss 0.11779082 - samples/sec: 31.74 - lr: 0.000042
+2022-02-05 03:59:27,917 epoch 3 - iter 3965/7936 - loss 0.11691604 - samples/sec: 31.85 - lr: 0.000042
+2022-02-05 04:06:01,365 epoch 3 - iter 4758/7936 - loss 0.11592267 - samples/sec: 32.26 - lr: 0.000041
+2022-02-05 04:12:41,174 epoch 3 - iter 5551/7936 - loss 0.11480043 - samples/sec: 31.74 - lr: 0.000041
+2022-02-05 04:19:14,243 epoch 3 - iter 6344/7936 - loss 0.11389582 - samples/sec: 32.29 - lr: 0.000040
+2022-02-05 04:25:45,192 epoch 3 - iter 7137/7936 - loss 0.11289267 - samples/sec: 32.46 - lr: 0.000039
+2022-02-05 04:32:26,310 epoch 3 - iter 7930/7936 - loss 0.11196899 - samples/sec: 31.64 - lr: 0.000039
+2022-02-05 04:32:29,352 ----------------------------------------------------------------------------------------------------
+2022-02-05 04:32:29,353 EPOCH 3 done: loss 0.1120 - lr 0.0000389
+2022-02-05 04:35:09,639 DEV : loss 0.016585879027843475 - f1-score (micro avg)  0.9229
+2022-02-05 04:35:09,698 BAD EPOCHS (no improvement): 4
+2022-02-05 04:35:09,698 ----------------------------------------------------------------------------------------------------
+2022-02-05 04:41:46,821 epoch 4 - iter 793/7936 - loss 0.09739851 - samples/sec: 31.96 - lr: 0.000038
+2022-02-05 04:48:23,504 epoch 4 - iter 1586/7936 - loss 0.09750632 - samples/sec: 31.99 - lr: 0.000038
+2022-02-05 04:55:05,833 epoch 4 - iter 2379/7936 - loss 0.09636659 - samples/sec: 31.54 - lr: 0.000037
+2022-02-05 05:01:34,951 epoch 4 - iter 3172/7936 - loss 0.09583742 - samples/sec: 32.61 - lr: 0.000037
+2022-02-05 05:08:07,163 epoch 4 - iter 3965/7936 - loss 0.09518243 - samples/sec: 32.36 - lr: 0.000036
+2022-02-05 05:14:50,781 epoch 4 - iter 4758/7936 - loss 0.09444265 - samples/sec: 31.44 - lr: 0.000036
+2022-02-05 05:21:24,983 epoch 4 - iter 5551/7936 - loss 0.09374740 - samples/sec: 32.19 - lr: 0.000035
+2022-02-05 05:27:54,052 epoch 4 - iter 6344/7936 - loss 0.09321236 - samples/sec: 32.62 - lr: 0.000034
+2022-02-05 05:34:32,228 epoch 4 - iter 7137/7936 - loss 0.09231997 - samples/sec: 31.87 - lr: 0.000034
+2022-02-05 05:41:08,580 epoch 4 - iter 7930/7936 - loss 0.09147929 - samples/sec: 32.02 - lr: 0.000033
+2022-02-05 05:41:11,479 ----------------------------------------------------------------------------------------------------
+2022-02-05 05:41:11,479 EPOCH 4 done: loss 0.0915 - lr 0.0000333
+2022-02-05 05:44:00,197 DEV : loss 0.016923826187849045 - f1-score (micro avg)  0.9213
+2022-02-05 05:44:00,256 BAD EPOCHS (no improvement): 4
+2022-02-05 05:44:00,270 ----------------------------------------------------------------------------------------------------
+2022-02-05 05:50:27,537 epoch 5 - iter 793/7936 - loss 0.07986125 - samples/sec: 32.77 - lr: 0.000033
+2022-02-05 05:56:56,203 epoch 5 - iter 1586/7936 - loss 0.08031745 - samples/sec: 32.65 - lr: 0.000032
+2022-02-05 06:03:34,109 epoch 5 - iter 2379/7936 - loss 0.07984185 - samples/sec: 31.89 - lr: 0.000032
+2022-02-05 06:10:03,550 epoch 5 - iter 3172/7936 - loss 0.07905074 - samples/sec: 32.59 - lr: 0.000031
+2022-02-05 06:16:30,085 epoch 5 - iter 3965/7936 - loss 0.07843193 - samples/sec: 32.83 - lr: 0.000031
+2022-02-05 06:23:10,671 epoch 5 - iter 4758/7936 - loss 0.07785540 - samples/sec: 31.68 - lr: 0.000030
+2022-02-05 06:29:45,063 epoch 5 - iter 5551/7936 - loss 0.07709413 - samples/sec: 32.18 - lr: 0.000029
+2022-02-05 06:36:23,513 epoch 5 - iter 6344/7936 - loss 0.07634510 - samples/sec: 31.85 - lr: 0.000029
+2022-02-05 06:42:51,615 epoch 5 - iter 7137/7936 - loss 0.07566508 - samples/sec: 32.70 - lr: 0.000028
+2022-02-05 06:49:23,409 epoch 5 - iter 7930/7936 - loss 0.07495508 - samples/sec: 32.39 - lr: 0.000028
+2022-02-05 06:49:26,372 ----------------------------------------------------------------------------------------------------
+2022-02-05 06:49:26,373 EPOCH 5 done: loss 0.0750 - lr 0.0000278
+2022-02-05 06:52:15,459 DEV : loss 0.017464155331254005 - f1-score (micro avg)  0.9311
+2022-02-05 06:52:15,518 BAD EPOCHS (no improvement): 4
+2022-02-05 06:52:15,518 ----------------------------------------------------------------------------------------------------
+2022-02-05 06:58:49,072 epoch 6 - iter 793/7936 - loss 0.06552824 - samples/sec: 32.25 - lr: 0.000027
+2022-02-05 07:05:27,796 epoch 6 - iter 1586/7936 - loss 0.06569517 - samples/sec: 31.83 - lr: 0.000027
+2022-02-05 07:11:58,162 epoch 6 - iter 2379/7936 - loss 0.06536467 - samples/sec: 32.51 - lr: 0.000026
+2022-02-05 07:18:25,878 epoch 6 - iter 3172/7936 - loss 0.06467146 - samples/sec: 32.73 - lr: 0.000026
+2022-02-05 07:25:10,562 epoch 6 - iter 3965/7936 - loss 0.06426965 - samples/sec: 31.36 - lr: 0.000025
+2022-02-05 07:31:39,437 epoch 6 - iter 4758/7936 - loss 0.06371305 - samples/sec: 32.63 - lr: 0.000024
+2022-02-05 07:38:08,323 epoch 6 - iter 5551/7936 - loss 0.06328229 - samples/sec: 32.63 - lr: 0.000024
+2022-02-05 07:44:52,176 epoch 6 - iter 6344/7936 - loss 0.06272143 - samples/sec: 31.42 - lr: 0.000023
+2022-02-05 07:51:20,507 epoch 6 - iter 7137/7936 - loss 0.06218937 - samples/sec: 32.68 - lr: 0.000023
+2022-02-05 07:57:52,828 epoch 6 - iter 7930/7936 - loss 0.06175113 - samples/sec: 32.35 - lr: 0.000022
+2022-02-05 07:57:55,686 ----------------------------------------------------------------------------------------------------
+2022-02-05 07:57:55,687 EPOCH 6 done: loss 0.0617 - lr 0.0000222
+2022-02-05 08:00:45,565 DEV : loss 0.01982131227850914 - f1-score (micro avg)  0.9358
+2022-02-05 08:00:45,625 BAD EPOCHS (no improvement): 4
+2022-02-05 08:00:45,644 ----------------------------------------------------------------------------------------------------
+2022-02-05 08:07:26,967 epoch 7 - iter 793/7936 - loss 0.05520420 - samples/sec: 31.62 - lr: 0.000022
+2022-02-05 08:13:58,782 epoch 7 - iter 1586/7936 - loss 0.05522964 - samples/sec: 32.39 - lr: 0.000021
+2022-02-05 08:20:32,705 epoch 7 - iter 2379/7936 - loss 0.05482898 - samples/sec: 32.21 - lr: 0.000021
+2022-02-05 08:27:14,353 epoch 7 - iter 3172/7936 - loss 0.05433105 - samples/sec: 31.59 - lr: 0.000020
+2022-02-05 08:33:45,236 epoch 7 - iter 3965/7936 - loss 0.05397125 - samples/sec: 32.47 - lr: 0.000019
+2022-02-05 08:40:14,072 epoch 7 - iter 4758/7936 - loss 0.05348281 - samples/sec: 32.64 - lr: 0.000019
+2022-02-05 08:46:52,674 epoch 7 - iter 5551/7936 - loss 0.05316673 - samples/sec: 31.84 - lr: 0.000018
+2022-02-05 08:53:20,653 epoch 7 - iter 6344/7936 - loss 0.05275831 - samples/sec: 32.71 - lr: 0.000018
+2022-02-05 08:59:52,741 epoch 7 - iter 7137/7936 - loss 0.05230036 - samples/sec: 32.37 - lr: 0.000017
+2022-02-05 09:06:38,983 epoch 7 - iter 7930/7936 - loss 0.05190552 - samples/sec: 31.24 - lr: 0.000017
+2022-02-05 09:06:41,639 ----------------------------------------------------------------------------------------------------
+2022-02-05 09:06:41,639 EPOCH 7 done: loss 0.0519 - lr 0.0000167
+2022-02-05 09:09:20,864 DEV : loss 0.02467426098883152 - f1-score (micro avg)  0.9355
+2022-02-05 09:09:20,924 BAD EPOCHS (no improvement): 4
+2022-02-05 09:09:20,939 ----------------------------------------------------------------------------------------------------
+2022-02-05 09:16:05,134 epoch 8 - iter 793/7936 - loss 0.04726178 - samples/sec: 31.40 - lr: 0.000016
+2022-02-05 09:22:33,870 epoch 8 - iter 1586/7936 - loss 0.04719666 - samples/sec: 32.64 - lr: 0.000016
+2022-02-05 09:29:02,929 epoch 8 - iter 2379/7936 - loss 0.04663752 - samples/sec: 32.62 - lr: 0.000015
+2022-02-05 09:35:42,369 epoch 8 - iter 3172/7936 - loss 0.04634901 - samples/sec: 31.77 - lr: 0.000014
+2022-02-05 09:42:14,843 epoch 8 - iter 3965/7936 - loss 0.04602895 - samples/sec: 32.33 - lr: 0.000014
+2022-02-05 09:48:48,062 epoch 8 - iter 4758/7936 - loss 0.04582764 - samples/sec: 32.27 - lr: 0.000013
+2022-02-05 09:55:28,863 epoch 8 - iter 5551/7936 - loss 0.04566599 - samples/sec: 31.66 - lr: 0.000013
+2022-02-05 10:01:52,699 epoch 8 - iter 6344/7936 - loss 0.04545939 - samples/sec: 33.06 - lr: 0.000012
+2022-02-05 10:08:33,137 epoch 8 - iter 7137/7936 - loss 0.04526206 - samples/sec: 31.69 - lr: 0.000012
+2022-02-05 10:15:07,241 epoch 8 - iter 7930/7936 - loss 0.04503385 - samples/sec: 32.20 - lr: 0.000011
+2022-02-05 10:15:10,600 ----------------------------------------------------------------------------------------------------
+2022-02-05 10:15:10,600 EPOCH 8 done: loss 0.0450 - lr 0.0000111
+2022-02-05 10:18:00,280 DEV : loss 0.02364770695567131 - f1-score (micro avg)  0.9371
+2022-02-05 10:18:00,339 BAD EPOCHS (no improvement): 4
+2022-02-05 10:18:00,358 ----------------------------------------------------------------------------------------------------
+2022-02-05 10:24:31,011 epoch 9 - iter 793/7936 - loss 0.04122325 - samples/sec: 32.48 - lr: 0.000011
+2022-02-05 10:31:00,279 epoch 9 - iter 1586/7936 - loss 0.04130931 - samples/sec: 32.60 - lr: 0.000010
+2022-02-05 10:37:40,369 epoch 9 - iter 2379/7936 - loss 0.04131112 - samples/sec: 31.72 - lr: 0.000009
+2022-02-05 10:44:11,067 epoch 9 - iter 3172/7936 - loss 0.04141124 - samples/sec: 32.48 - lr: 0.000009
+2022-02-05 10:50:41,270 epoch 9 - iter 3965/7936 - loss 0.04120608 - samples/sec: 32.52 - lr: 0.000008
+2022-02-05 10:57:24,718 epoch 9 - iter 4758/7936 - loss 0.04108655 - samples/sec: 31.45 - lr: 0.000008
+2022-02-05 11:04:00,581 epoch 9 - iter 5551/7936 - loss 0.04093370 - samples/sec: 32.06 - lr: 0.000007
+2022-02-05 11:10:31,042 epoch 9 - iter 6344/7936 - loss 0.04078404 - samples/sec: 32.50 - lr: 0.000007
+2022-02-05 11:17:13,751 epoch 9 - iter 7137/7936 - loss 0.04061073 - samples/sec: 31.51 - lr: 0.000006
+2022-02-05 11:23:44,231 epoch 9 - iter 7930/7936 - loss 0.04050638 - samples/sec: 32.50 - lr: 0.000006
+2022-02-05 11:23:47,941 ----------------------------------------------------------------------------------------------------
+2022-02-05 11:23:47,942 EPOCH 9 done: loss 0.0405 - lr 0.0000056
+2022-02-05 11:26:37,114 DEV : loss 0.026182951405644417 - f1-score (micro avg)  0.9361
+2022-02-05 11:26:37,173 BAD EPOCHS (no improvement): 4
+2022-02-05 11:26:37,186 ----------------------------------------------------------------------------------------------------
+2022-02-05 11:33:05,778 epoch 10 - iter 793/7936 - loss 0.03876526 - samples/sec: 32.66 - lr: 0.000005
+2022-02-05 11:39:45,501 epoch 10 - iter 1586/7936 - loss 0.03871561 - samples/sec: 31.75 - lr: 0.000004
+2022-02-05 11:46:18,242 epoch 10 - iter 2379/7936 - loss 0.03842790 - samples/sec: 32.31 - lr: 0.000004
+2022-02-05 11:52:48,370 epoch 10 - iter 3172/7936 - loss 0.03820246 - samples/sec: 32.53 - lr: 0.000003
+2022-02-05 11:59:28,420 epoch 10 - iter 3965/7936 - loss 0.03807900 - samples/sec: 31.72 - lr: 0.000003
+2022-02-05 12:05:57,882 epoch 10 - iter 4758/7936 - loss 0.03798954 - samples/sec: 32.58 - lr: 0.000002
+2022-02-05 12:12:25,766 epoch 10 - iter 5551/7936 - loss 0.03803371 - samples/sec: 32.72 - lr: 0.000002
+2022-02-05 12:19:03,411 epoch 10 - iter 6344/7936 - loss 0.03805844 - samples/sec: 31.91 - lr: 0.000001
+2022-02-05 12:25:27,539 epoch 10 - iter 7137/7936 - loss 0.03799490 - samples/sec: 33.04 - lr: 0.000001
+2022-02-05 12:31:55,442 epoch 10 - iter 7930/7936 - loss 0.03798541 - samples/sec: 32.71 - lr: 0.000000
+2022-02-05 12:31:58,461 ----------------------------------------------------------------------------------------------------
+2022-02-05 12:31:58,462 EPOCH 10 done: loss 0.0380 - lr 0.0000000
+2022-02-05 12:34:45,700 DEV : loss 0.027400659397244453 - f1-score (micro avg)  0.9368
+2022-02-05 12:34:45,760 BAD EPOCHS (no improvement): 4
+2022-02-05 12:34:46,755 ----------------------------------------------------------------------------------------------------
+2022-02-05 12:34:46,757 Testing using last state of model ...
+2022-02-05 12:37:34,421 0.9329	0.9323	0.9326	0.8893
+2022-02-05 12:37:34,422 
+Results:
+- F-score (micro) 0.9326
+- F-score (macro) 0.9111
+- Accuracy 0.8893
+
+By class:
+              precision    recall  f1-score   support
+
+        pers     0.9355    0.9279    0.9317      2734
+         loc     0.9242    0.9335    0.9288      1384
+      amount     0.9800    0.9800    0.9800       250
+        time     0.9456    0.9576    0.9516       236
+        func     0.9333    0.9000    0.9164       140
+         org     0.8148    0.8980    0.8544        49
+        prod     0.8621    0.9259    0.8929        27
+       event     0.8333    0.8333    0.8333        12
+
+   micro avg     0.9329    0.9323    0.9326      4832
+   macro avg     0.9036    0.9195    0.9111      4832
+weighted avg     0.9331    0.9323    0.9327      4832
+ samples avg     0.8893    0.8893    0.8893      4832
+
+2022-02-05 12:37:34,422 ----------------------------------------------------------------------------------------------------