Upload model files

final-model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ce3dd5c362bfc062655fbff2535cf84c7d53e403db2a1fb5a549b49f9594d81
+size 445100077

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	18:41:01	4	0.0000	3.633058030082742	2.0775277614593506	0.5698	0.5698	0.5698	0.5698
+2	18:43:46	4	0.0000	1.019771994275212	0.23464356362819672	0.9443	0.9443	0.9443	0.9443
+3	18:46:34	4	0.0000	0.410349326583172	0.140821173787117	0.9632	0.9632	0.9632	0.9632
+4	18:49:20	4	0.0000	0.34302561318631913	0.11640190333127975	0.9703	0.9703	0.9703	0.9703
+5	18:52:05	4	0.0000	0.3097532451655346	0.10135460644960403	0.9729	0.9729	0.9729	0.9729
+6	18:54:49	4	0.0000	0.2965522425579126	0.09480294585227966	0.974	0.974	0.974	0.974
+7	18:57:34	4	0.0000	0.28957056620947336	0.09033482521772385	0.9743	0.9743	0.9743	0.9743
+8	19:00:20	4	0.0000	0.28135947487172785	0.08581043034791946	0.9745	0.9745	0.9745	0.9745
+9	19:03:08	4	0.0000	0.2826198891036253	0.08502506464719772	0.974	0.974	0.974	0.974
+10	19:05:54	4	0.0000	0.28319776187525647	0.08448906987905502	0.974	0.974	0.974	0.974

training.log

@@ -0,0 +1,538 @@
+2022-01-16 18:38:17,520 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:38:17,523 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=51, bias=True)
+  (beta): 1.0
+  (weights): None
+  (weight_tensor) None
+)"
+2022-01-16 18:38:17,526 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:38:17,526 Corpus: "Corpus: 5642 train + 195 dev + 649 test sentences"
+2022-01-16 18:38:17,526 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:38:17,527 Parameters:
+2022-01-16 18:38:17,527  - learning_rate: "5e-06"
+2022-01-16 18:38:17,527  - mini_batch_size: "32"
+2022-01-16 18:38:17,527  - patience: "3"
+2022-01-16 18:38:17,528  - anneal_factor: "0.5"
+2022-01-16 18:38:17,528  - max_epochs: "10"
+2022-01-16 18:38:17,528  - shuffle: "True"
+2022-01-16 18:38:17,528  - train_with_dev: "False"
+2022-01-16 18:38:17,529  - batch_growth_annealing: "False"
+2022-01-16 18:38:17,529 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:38:17,529 Model training base path: "resources/taggers/pos-transformer"
+2022-01-16 18:38:17,530 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:38:17,530 Device: cuda:0
+2022-01-16 18:38:17,530 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:38:17,530 Embeddings storage mode: none
+2022-01-16 18:38:17,534 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:38:34,359 epoch 1 - iter 17/177 - loss 4.21719545 - samples/sec: 32.34 - lr: 0.000000
+2022-01-16 18:38:49,400 epoch 1 - iter 34/177 - loss 4.19345430 - samples/sec: 36.17 - lr: 0.000001
+2022-01-16 18:39:05,256 epoch 1 - iter 51/177 - loss 4.15633603 - samples/sec: 34.31 - lr: 0.000001
+2022-01-16 18:39:19,936 epoch 1 - iter 68/177 - loss 4.11811385 - samples/sec: 37.07 - lr: 0.000002
+2022-01-16 18:39:35,631 epoch 1 - iter 85/177 - loss 4.06705216 - samples/sec: 34.68 - lr: 0.000002
+2022-01-16 18:39:49,539 epoch 1 - iter 102/177 - loss 4.01162833 - samples/sec: 39.12 - lr: 0.000003
+2022-01-16 18:40:04,517 epoch 1 - iter 119/177 - loss 3.95117440 - samples/sec: 36.33 - lr: 0.000003
+2022-01-16 18:40:18,637 epoch 1 - iter 136/177 - loss 3.88391044 - samples/sec: 38.53 - lr: 0.000004
+2022-01-16 18:40:34,602 epoch 1 - iter 153/177 - loss 3.78662706 - samples/sec: 34.08 - lr: 0.000004
+2022-01-16 18:40:50,297 epoch 1 - iter 170/177 - loss 3.66565316 - samples/sec: 34.67 - lr: 0.000005
+2022-01-16 18:40:55,405 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:40:55,406 EPOCH 1 done: loss 3.6331 - lr 0.0000050
+2022-01-16 18:41:01,071 DEV : loss 2.0775277614593506 - f1-score (micro avg)  0.5698
+2022-01-16 18:41:01,073 BAD EPOCHS (no improvement): 4
+2022-01-16 18:41:01,075 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:41:14,873 epoch 2 - iter 17/177 - loss 2.20805337 - samples/sec: 39.44 - lr: 0.000005
+2022-01-16 18:41:29,867 epoch 2 - iter 34/177 - loss 1.96658974 - samples/sec: 36.29 - lr: 0.000005
+2022-01-16 18:41:45,607 epoch 2 - iter 51/177 - loss 1.75508128 - samples/sec: 34.57 - lr: 0.000005
+2022-01-16 18:42:01,386 epoch 2 - iter 68/177 - loss 1.58575541 - samples/sec: 34.48 - lr: 0.000005
+2022-01-16 18:42:16,804 epoch 2 - iter 85/177 - loss 1.45429547 - samples/sec: 35.29 - lr: 0.000005
+2022-01-16 18:42:32,178 epoch 2 - iter 102/177 - loss 1.34526502 - samples/sec: 35.39 - lr: 0.000005
+2022-01-16 18:42:48,735 epoch 2 - iter 119/177 - loss 1.23724431 - samples/sec: 32.86 - lr: 0.000005
+2022-01-16 18:43:03,310 epoch 2 - iter 136/177 - loss 1.16223838 - samples/sec: 37.33 - lr: 0.000005
+2022-01-16 18:43:18,304 epoch 2 - iter 153/177 - loss 1.09870495 - samples/sec: 36.29 - lr: 0.000005
+2022-01-16 18:43:34,956 epoch 2 - iter 170/177 - loss 1.03855466 - samples/sec: 32.67 - lr: 0.000004
+2022-01-16 18:43:40,722 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:43:40,723 EPOCH 2 done: loss 1.0198 - lr 0.0000044
+2022-01-16 18:43:46,405 DEV : loss 0.23464356362819672 - f1-score (micro avg)  0.9443
+2022-01-16 18:43:46,407 BAD EPOCHS (no improvement): 4
+2022-01-16 18:43:46,408 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:44:01,387 epoch 3 - iter 17/177 - loss 0.46476740 - samples/sec: 36.33 - lr: 0.000004
+2022-01-16 18:44:17,394 epoch 3 - iter 34/177 - loss 0.46233323 - samples/sec: 33.99 - lr: 0.000004
+2022-01-16 18:44:32,304 epoch 3 - iter 51/177 - loss 0.45235428 - samples/sec: 36.49 - lr: 0.000004
+2022-01-16 18:44:46,826 epoch 3 - iter 68/177 - loss 0.44547326 - samples/sec: 37.47 - lr: 0.000004
+2022-01-16 18:45:03,857 epoch 3 - iter 85/177 - loss 0.43503033 - samples/sec: 31.95 - lr: 0.000004
+2022-01-16 18:45:20,043 epoch 3 - iter 102/177 - loss 0.42734805 - samples/sec: 33.63 - lr: 0.000004
+2022-01-16 18:45:36,060 epoch 3 - iter 119/177 - loss 0.42237100 - samples/sec: 33.97 - lr: 0.000004
+2022-01-16 18:45:51,576 epoch 3 - iter 136/177 - loss 0.41700412 - samples/sec: 35.07 - lr: 0.000004
+2022-01-16 18:46:07,252 epoch 3 - iter 153/177 - loss 0.41455352 - samples/sec: 34.71 - lr: 0.000004
+2022-01-16 18:46:23,597 epoch 3 - iter 170/177 - loss 0.41134424 - samples/sec: 33.29 - lr: 0.000004
+2022-01-16 18:46:29,222 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:46:29,223 EPOCH 3 done: loss 0.4103 - lr 0.0000039
+2022-01-16 18:46:34,899 DEV : loss 0.140821173787117 - f1-score (micro avg)  0.9632
+2022-01-16 18:46:34,901 BAD EPOCHS (no improvement): 4
+2022-01-16 18:46:34,902 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:46:49,649 epoch 4 - iter 17/177 - loss 0.34770276 - samples/sec: 36.90 - lr: 0.000004
+2022-01-16 18:47:05,137 epoch 4 - iter 34/177 - loss 0.34449519 - samples/sec: 35.13 - lr: 0.000004
+2022-01-16 18:47:20,666 epoch 4 - iter 51/177 - loss 0.35038471 - samples/sec: 35.04 - lr: 0.000004
+2022-01-16 18:47:35,593 epoch 4 - iter 68/177 - loss 0.34965167 - samples/sec: 36.45 - lr: 0.000004
+2022-01-16 18:47:51,537 epoch 4 - iter 85/177 - loss 0.35074386 - samples/sec: 34.13 - lr: 0.000004
+2022-01-16 18:48:06,575 epoch 4 - iter 102/177 - loss 0.34919573 - samples/sec: 36.18 - lr: 0.000004
+2022-01-16 18:48:22,671 epoch 4 - iter 119/177 - loss 0.34906482 - samples/sec: 33.80 - lr: 0.000004
+2022-01-16 18:48:38,152 epoch 4 - iter 136/177 - loss 0.34645574 - samples/sec: 35.15 - lr: 0.000003
+2022-01-16 18:48:53,425 epoch 4 - iter 153/177 - loss 0.34515747 - samples/sec: 35.63 - lr: 0.000003
+2022-01-16 18:49:08,614 epoch 4 - iter 170/177 - loss 0.34411478 - samples/sec: 35.82 - lr: 0.000003
+2022-01-16 18:49:14,556 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:49:14,557 EPOCH 4 done: loss 0.3430 - lr 0.0000033
+2022-01-16 18:49:20,294 DEV : loss 0.11640190333127975 - f1-score (micro avg)  0.9703
+2022-01-16 18:49:20,297 BAD EPOCHS (no improvement): 4
+2022-01-16 18:49:20,297 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:49:36,057 epoch 5 - iter 17/177 - loss 0.31027747 - samples/sec: 34.53 - lr: 0.000003
+2022-01-16 18:49:51,823 epoch 5 - iter 34/177 - loss 0.31176440 - samples/sec: 34.51 - lr: 0.000003
+2022-01-16 18:50:06,630 epoch 5 - iter 51/177 - loss 0.31452075 - samples/sec: 36.75 - lr: 0.000003
+2022-01-16 18:50:22,294 epoch 5 - iter 68/177 - loss 0.31209996 - samples/sec: 34.73 - lr: 0.000003
+2022-01-16 18:50:36,301 epoch 5 - iter 85/177 - loss 0.31357991 - samples/sec: 38.85 - lr: 0.000003
+2022-01-16 18:50:52,962 epoch 5 - iter 102/177 - loss 0.31496866 - samples/sec: 32.66 - lr: 0.000003
+2022-01-16 18:51:08,260 epoch 5 - iter 119/177 - loss 0.31294977 - samples/sec: 35.57 - lr: 0.000003
+2022-01-16 18:51:24,158 epoch 5 - iter 136/177 - loss 0.31189665 - samples/sec: 34.22 - lr: 0.000003
+2022-01-16 18:51:39,145 epoch 5 - iter 153/177 - loss 0.31138881 - samples/sec: 36.31 - lr: 0.000003
+2022-01-16 18:51:54,700 epoch 5 - iter 170/177 - loss 0.30960234 - samples/sec: 34.98 - lr: 0.000003
+2022-01-16 18:51:59,742 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:51:59,743 EPOCH 5 done: loss 0.3098 - lr 0.0000028
+2022-01-16 18:52:05,466 DEV : loss 0.10135460644960403 - f1-score (micro avg)  0.9729
+2022-01-16 18:52:05,468 BAD EPOCHS (no improvement): 4
+2022-01-16 18:52:05,469 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:52:20,458 epoch 6 - iter 17/177 - loss 0.30154787 - samples/sec: 36.30 - lr: 0.000003
+2022-01-16 18:52:34,917 epoch 6 - iter 34/177 - loss 0.30197436 - samples/sec: 37.63 - lr: 0.000003
+2022-01-16 18:52:49,618 epoch 6 - iter 51/177 - loss 0.30167136 - samples/sec: 37.01 - lr: 0.000003
+2022-01-16 18:53:04,988 epoch 6 - iter 68/177 - loss 0.30196611 - samples/sec: 35.40 - lr: 0.000003
+2022-01-16 18:53:20,297 epoch 6 - iter 85/177 - loss 0.30182940 - samples/sec: 35.54 - lr: 0.000003
+2022-01-16 18:53:35,734 epoch 6 - iter 102/177 - loss 0.30003109 - samples/sec: 35.25 - lr: 0.000002
+2022-01-16 18:53:51,701 epoch 6 - iter 119/177 - loss 0.30091205 - samples/sec: 34.08 - lr: 0.000002
+2022-01-16 18:54:06,831 epoch 6 - iter 136/177 - loss 0.30099483 - samples/sec: 35.96 - lr: 0.000002
+2022-01-16 18:54:22,486 epoch 6 - iter 153/177 - loss 0.29848715 - samples/sec: 34.76 - lr: 0.000002
+2022-01-16 18:54:37,203 epoch 6 - iter 170/177 - loss 0.29689481 - samples/sec: 36.97 - lr: 0.000002
+2022-01-16 18:54:44,337 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:54:44,338 EPOCH 6 done: loss 0.2966 - lr 0.0000022
+2022-01-16 18:54:49,620 DEV : loss 0.09480294585227966 - f1-score (micro avg)  0.974
+2022-01-16 18:54:49,623 BAD EPOCHS (no improvement): 4
+2022-01-16 18:54:49,623 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:55:05,515 epoch 7 - iter 17/177 - loss 0.28239213 - samples/sec: 34.24 - lr: 0.000002
+2022-01-16 18:55:20,295 epoch 7 - iter 34/177 - loss 0.28557506 - samples/sec: 36.81 - lr: 0.000002
+2022-01-16 18:55:35,660 epoch 7 - iter 51/177 - loss 0.28541785 - samples/sec: 35.41 - lr: 0.000002
+2022-01-16 18:55:51,758 epoch 7 - iter 68/177 - loss 0.29320767 - samples/sec: 33.80 - lr: 0.000002
+2022-01-16 18:56:06,783 epoch 7 - iter 85/177 - loss 0.29339894 - samples/sec: 36.21 - lr: 0.000002
+2022-01-16 18:56:22,815 epoch 7 - iter 102/177 - loss 0.29253486 - samples/sec: 33.94 - lr: 0.000002
+2022-01-16 18:56:39,028 epoch 7 - iter 119/177 - loss 0.29145637 - samples/sec: 33.56 - lr: 0.000002
+2022-01-16 18:56:54,361 epoch 7 - iter 136/177 - loss 0.29111952 - samples/sec: 35.49 - lr: 0.000002
+2022-01-16 18:57:09,548 epoch 7 - iter 153/177 - loss 0.29113036 - samples/sec: 35.83 - lr: 0.000002
+2022-01-16 18:57:23,584 epoch 7 - iter 170/177 - loss 0.29066532 - samples/sec: 38.76 - lr: 0.000002
+2022-01-16 18:57:29,584 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:57:29,585 EPOCH 7 done: loss 0.2896 - lr 0.0000017
+2022-01-16 18:57:34,894 DEV : loss 0.09033482521772385 - f1-score (micro avg)  0.9743
+2022-01-16 18:57:34,896 BAD EPOCHS (no improvement): 4
+2022-01-16 18:57:34,898 ----------------------------------------------------------------------------------------------------
+2022-01-16 18:57:50,623 epoch 8 - iter 17/177 - loss 0.28329047 - samples/sec: 34.60 - lr: 0.000002
+2022-01-16 18:58:06,213 epoch 8 - iter 34/177 - loss 0.28096448 - samples/sec: 34.90 - lr: 0.000002
+2022-01-16 18:58:22,737 epoch 8 - iter 51/177 - loss 0.28201738 - samples/sec: 32.93 - lr: 0.000002
+2022-01-16 18:58:37,507 epoch 8 - iter 68/177 - loss 0.28137267 - samples/sec: 36.84 - lr: 0.000001
+2022-01-16 18:58:52,962 epoch 8 - iter 85/177 - loss 0.28405564 - samples/sec: 35.21 - lr: 0.000001
+2022-01-16 18:59:08,711 epoch 8 - iter 102/177 - loss 0.28496531 - samples/sec: 34.55 - lr: 0.000001
+2022-01-16 18:59:23,238 epoch 8 - iter 119/177 - loss 0.28466528 - samples/sec: 37.46 - lr: 0.000001
+2022-01-16 18:59:38,520 epoch 8 - iter 136/177 - loss 0.28246598 - samples/sec: 35.60 - lr: 0.000001
+2022-01-16 18:59:53,789 epoch 8 - iter 153/177 - loss 0.28078088 - samples/sec: 35.63 - lr: 0.000001
+2022-01-16 19:00:09,934 epoch 8 - iter 170/177 - loss 0.28075535 - samples/sec: 33.70 - lr: 0.000001
+2022-01-16 19:00:15,100 ----------------------------------------------------------------------------------------------------
+2022-01-16 19:00:15,101 EPOCH 8 done: loss 0.2814 - lr 0.0000011
+2022-01-16 19:00:20,403 DEV : loss 0.08581043034791946 - f1-score (micro avg)  0.9745
+2022-01-16 19:00:20,406 BAD EPOCHS (no improvement): 4
+2022-01-16 19:00:20,406 ----------------------------------------------------------------------------------------------------
+2022-01-16 19:00:36,469 epoch 9 - iter 17/177 - loss 0.27366042 - samples/sec: 33.87 - lr: 0.000001
+2022-01-16 19:00:51,042 epoch 9 - iter 34/177 - loss 0.27417563 - samples/sec: 37.34 - lr: 0.000001
+2022-01-16 19:01:06,968 epoch 9 - iter 51/177 - loss 0.27908066 - samples/sec: 34.16 - lr: 0.000001
+2022-01-16 19:01:21,551 epoch 9 - iter 68/177 - loss 0.27815091 - samples/sec: 37.31 - lr: 0.000001
+2022-01-16 19:01:38,409 epoch 9 - iter 85/177 - loss 0.27855783 - samples/sec: 32.28 - lr: 0.000001
+2022-01-16 19:01:53,547 epoch 9 - iter 102/177 - loss 0.28336618 - samples/sec: 35.94 - lr: 0.000001
+2022-01-16 19:02:09,188 epoch 9 - iter 119/177 - loss 0.28196400 - samples/sec: 34.79 - lr: 0.000001
+2022-01-16 19:02:25,112 epoch 9 - iter 136/177 - loss 0.28112997 - samples/sec: 34.17 - lr: 0.000001
+2022-01-16 19:02:41,122 epoch 9 - iter 153/177 - loss 0.28271008 - samples/sec: 33.99 - lr: 0.000001
+2022-01-16 19:02:57,003 epoch 9 - iter 170/177 - loss 0.28254205 - samples/sec: 34.26 - lr: 0.000001
+2022-01-16 19:03:02,602 ----------------------------------------------------------------------------------------------------
+2022-01-16 19:03:02,603 EPOCH 9 done: loss 0.2826 - lr 0.0000006
+2022-01-16 19:03:08,344 DEV : loss 0.08502506464719772 - f1-score (micro avg)  0.974
+2022-01-16 19:03:08,347 BAD EPOCHS (no improvement): 4
+2022-01-16 19:03:08,348 ----------------------------------------------------------------------------------------------------
+2022-01-16 19:03:22,683 epoch 10 - iter 17/177 - loss 0.29810598 - samples/sec: 37.96 - lr: 0.000001
+2022-01-16 19:03:38,044 epoch 10 - iter 34/177 - loss 0.29633129 - samples/sec: 35.42 - lr: 0.000000
+2022-01-16 19:03:54,399 epoch 10 - iter 51/177 - loss 0.28500408 - samples/sec: 33.27 - lr: 0.000000
+2022-01-16 19:04:09,802 epoch 10 - iter 68/177 - loss 0.28305573 - samples/sec: 35.32 - lr: 0.000000
+2022-01-16 19:04:25,641 epoch 10 - iter 85/177 - loss 0.28663575 - samples/sec: 34.35 - lr: 0.000000
+2022-01-16 19:04:40,354 epoch 10 - iter 102/177 - loss 0.28653115 - samples/sec: 36.98 - lr: 0.000000
+2022-01-16 19:04:56,702 epoch 10 - iter 119/177 - loss 0.28579694 - samples/sec: 33.28 - lr: 0.000000
+2022-01-16 19:05:12,070 epoch 10 - iter 136/177 - loss 0.28590446 - samples/sec: 35.40 - lr: 0.000000
+2022-01-16 19:05:27,377 epoch 10 - iter 153/177 - loss 0.28533742 - samples/sec: 35.55 - lr: 0.000000
+2022-01-16 19:05:42,603 epoch 10 - iter 170/177 - loss 0.28333786 - samples/sec: 35.73 - lr: 0.000000
+2022-01-16 19:05:48,443 ----------------------------------------------------------------------------------------------------
+2022-01-16 19:05:48,444 EPOCH 10 done: loss 0.2832 - lr 0.0000000
+2022-01-16 19:05:54,211 DEV : loss 0.08448906987905502 - f1-score (micro avg)  0.974
+2022-01-16 19:05:54,214 BAD EPOCHS (no improvement): 4
+2022-01-16 19:05:55,439 ----------------------------------------------------------------------------------------------------
+2022-01-16 19:05:55,440 Testing using last state of model ...
+2022-01-16 19:06:15,179 0.9788	0.9788	0.9788	0.9788
+2022-01-16 19:06:15,180 
+Results:
+- F-score (micro) 0.9788
+- F-score (macro) 0.7527
+- Accuracy 0.9788
+
+By class:
+              precision    recall  f1-score   support
+
+      NOMcom     0.9850    0.9840    0.9845      2130
+      VERcjg     0.9974    0.9954    0.9964      1535
+      PROper     0.9912    0.9920    0.9916      1368
+      PONfbl     1.0000    0.9993    0.9996      1341
+         PRE     0.9881    0.9955    0.9918      1331
+      ADVgen     0.9713    0.9263    0.9483       841
+      PONfrt     0.9895    1.0000    0.9947       662
+      DETdef     0.9983    0.9983    0.9983       606
+      ADJqua     0.9259    0.9500    0.9378       500
+      VERinf     0.9920    1.0000    0.9960       497
+      DETpos     1.0000    0.9957    0.9979       469
+      CONcoo     0.9957    0.9935    0.9946       465
+      CONsub     0.9337    0.9409    0.9373       389
+      VERppe     0.9659    0.9720    0.9689       321
+      ADVneg     0.9476    1.0000    0.9731       271
+      PROrel     0.9194    0.9296    0.9245       270
+      NOMpro     0.9634    0.9925    0.9777       265
+      DETndf     0.9958    0.9715    0.9835       246
+      PROind     0.9526    0.9628    0.9577       188
+  PRE.DETdef     0.9785    0.9945    0.9864       183
+      DETdem     1.0000    0.9806    0.9902       155
+      PROdem     0.9675    1.0000    0.9835       119
+      PROadv     0.9083    0.9820    0.9437       111
+      DETind     0.9223    0.9694    0.9453        98
+      VERppa     0.9683    0.9104    0.9385        67
+      PROimp     0.8333    0.8333    0.8333        54
+      DETcar     0.7381    1.0000    0.8493        31
+         INJ     1.0000    0.8571    0.9231        35
+      ADJind     0.9310    0.9000    0.9153        30
+      PROint     0.6957    0.7273    0.7111        22
+      ADJcar     0.8333    0.4762    0.6061        21
+      PROcar     0.7333    0.6111    0.6667        18
+      PONpga     1.0000    1.0000    1.0000        16
+      PROpos     0.9231    0.8571    0.8889        14
+      DETrel     0.6364    0.4375    0.5185        16
+      DETint     0.4706    0.8000    0.5926        10
+      PONpdr     1.0000    1.0000    1.0000        13
+      ADJord     0.8889    0.5000    0.6400        16
+      ADVint     1.0000    0.8000    0.8889         5
+      PONpxx     0.0000    0.0000    0.0000         6
+  PRE.PROrel     0.0000    0.0000    0.0000         2
+       latin     0.0000    0.0000    0.0000         2
+      PROord     0.0000    0.0000    0.0000         1
+  PRE.PROdem     0.0000    0.0000    0.0000         1
+  PRE.NOMcom     0.0000    0.0000    0.0000         1
+         ETR     0.0000    0.0000    0.0000         1
+      ADVsub     0.0000    0.0000    0.0000         1
+
+   micro avg     0.9788    0.9788    0.9788     14744
+   macro avg     0.7647    0.7497    0.7527     14744
+weighted avg     0.9781    0.9788    0.9782     14744
+ samples avg     0.9788    0.9788    0.9788     14744
+
+2022-01-16 19:06:15,180 ----------------------------------------------------------------------------------------------------