2023-10-11 14:33:43,977 ---------------------------------------------------------------------------------------------------- 2023-10-11 14:33:43,979 Model: "SequenceTagger( (embeddings): ByT5Embeddings( (model): T5EncoderModel( (shared): Embedding(384, 1472) (encoder): T5Stack( (embed_tokens): Embedding(384, 1472) (block): ModuleList( (0): T5Block( (layer): ModuleList( (0): T5LayerSelfAttention( (SelfAttention): T5Attention( (q): Linear(in_features=1472, out_features=384, bias=False) (k): Linear(in_features=1472, out_features=384, bias=False) (v): Linear(in_features=1472, out_features=384, bias=False) (o): Linear(in_features=384, out_features=1472, bias=False) (relative_attention_bias): Embedding(32, 6) ) (layer_norm): FusedRMSNorm(torch.Size([1472]), eps=1e-06, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (1): T5LayerFF( (DenseReluDense): T5DenseGatedActDense( (wi_0): Linear(in_features=1472, out_features=3584, bias=False) (wi_1): Linear(in_features=1472, out_features=3584, bias=False) (wo): Linear(in_features=3584, out_features=1472, bias=False) (dropout): Dropout(p=0.1, inplace=False) (act): NewGELUActivation() ) (layer_norm): FusedRMSNorm(torch.Size([1472]), eps=1e-06, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) ) ) (1-11): 11 x T5Block( (layer): ModuleList( (0): T5LayerSelfAttention( (SelfAttention): T5Attention( (q): Linear(in_features=1472, out_features=384, bias=False) (k): Linear(in_features=1472, out_features=384, bias=False) (v): Linear(in_features=1472, out_features=384, bias=False) (o): Linear(in_features=384, out_features=1472, bias=False) ) (layer_norm): FusedRMSNorm(torch.Size([1472]), eps=1e-06, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) (1): T5LayerFF( (DenseReluDense): T5DenseGatedActDense( (wi_0): Linear(in_features=1472, out_features=3584, bias=False) (wi_1): Linear(in_features=1472, out_features=3584, bias=False) (wo): Linear(in_features=3584, out_features=1472, bias=False) (dropout): Dropout(p=0.1, inplace=False) (act): NewGELUActivation() ) (layer_norm): FusedRMSNorm(torch.Size([1472]), eps=1e-06, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) ) ) ) (final_layer_norm): FusedRMSNorm(torch.Size([1472]), eps=1e-06, elementwise_affine=True) (dropout): Dropout(p=0.1, inplace=False) ) ) ) (locked_dropout): LockedDropout(p=0.5) (linear): Linear(in_features=1472, out_features=17, bias=True) (loss_function): CrossEntropyLoss() )" 2023-10-11 14:33:43,980 ---------------------------------------------------------------------------------------------------- 2023-10-11 14:33:43,980 MultiCorpus: 20847 train + 1123 dev + 3350 test sentences - NER_HIPE_2022 Corpus: 20847 train + 1123 dev + 3350 test sentences - /root/.flair/datasets/ner_hipe_2022/v2.1/newseye/de/with_doc_seperator 2023-10-11 14:33:43,980 ---------------------------------------------------------------------------------------------------- 2023-10-11 14:33:43,980 Train: 20847 sentences 2023-10-11 14:33:43,980 (train_with_dev=False, train_with_test=False) 2023-10-11 14:33:43,980 ---------------------------------------------------------------------------------------------------- 2023-10-11 14:33:43,980 Training Params: 2023-10-11 14:33:43,980 - learning_rate: "0.00016" 2023-10-11 14:33:43,980 - mini_batch_size: "4" 2023-10-11 14:33:43,981 - max_epochs: "10" 2023-10-11 14:33:43,981 - shuffle: "True" 2023-10-11 14:33:43,981 ---------------------------------------------------------------------------------------------------- 2023-10-11 14:33:43,981 Plugins: 2023-10-11 14:33:43,981 - TensorboardLogger 2023-10-11 14:33:43,981 - LinearScheduler | warmup_fraction: '0.1' 2023-10-11 14:33:43,981 ---------------------------------------------------------------------------------------------------- 2023-10-11 14:33:43,981 Final evaluation on model from best epoch (best-model.pt) 2023-10-11 14:33:43,981 - metric: "('micro avg', 'f1-score')" 2023-10-11 14:33:43,981 ---------------------------------------------------------------------------------------------------- 2023-10-11 14:33:43,981 Computation: 2023-10-11 14:33:43,981 - compute on device: cuda:0 2023-10-11 14:33:43,981 - embedding storage: none 2023-10-11 14:33:43,982 ---------------------------------------------------------------------------------------------------- 2023-10-11 14:33:43,982 Model training base path: "hmbench-newseye/de-hmbyt5-preliminary/byt5-small-historic-multilingual-span20-flax-bs4-wsFalse-e10-lr0.00016-poolingfirst-layers-1-crfFalse-3" 2023-10-11 14:33:43,982 ---------------------------------------------------------------------------------------------------- 2023-10-11 14:33:43,982 ---------------------------------------------------------------------------------------------------- 2023-10-11 14:33:43,982 Logging anything other than scalars to TensorBoard is currently not supported. 2023-10-11 14:36:06,547 epoch 1 - iter 521/5212 - loss 2.76662736 - time (sec): 142.56 - samples/sec: 262.14 - lr: 0.000016 - momentum: 0.000000 2023-10-11 14:38:27,452 epoch 1 - iter 1042/5212 - loss 2.28624082 - time (sec): 283.47 - samples/sec: 270.79 - lr: 0.000032 - momentum: 0.000000 2023-10-11 14:40:44,639 epoch 1 - iter 1563/5212 - loss 1.79645371 - time (sec): 420.66 - samples/sec: 269.24 - lr: 0.000048 - momentum: 0.000000 2023-10-11 14:43:01,808 epoch 1 - iter 2084/5212 - loss 1.46859051 - time (sec): 557.82 - samples/sec: 267.94 - lr: 0.000064 - momentum: 0.000000 2023-10-11 14:45:26,500 epoch 1 - iter 2605/5212 - loss 1.27159482 - time (sec): 702.52 - samples/sec: 266.51 - lr: 0.000080 - momentum: 0.000000 2023-10-11 14:47:48,763 epoch 1 - iter 3126/5212 - loss 1.12283149 - time (sec): 844.78 - samples/sec: 264.51 - lr: 0.000096 - momentum: 0.000000 2023-10-11 14:50:12,100 epoch 1 - iter 3647/5212 - loss 1.01175361 - time (sec): 988.12 - samples/sec: 261.77 - lr: 0.000112 - momentum: 0.000000 2023-10-11 14:52:34,157 epoch 1 - iter 4168/5212 - loss 0.93063939 - time (sec): 1130.17 - samples/sec: 259.51 - lr: 0.000128 - momentum: 0.000000 2023-10-11 14:54:53,773 epoch 1 - iter 4689/5212 - loss 0.85556421 - time (sec): 1269.79 - samples/sec: 260.71 - lr: 0.000144 - momentum: 0.000000 2023-10-11 14:57:19,976 epoch 1 - iter 5210/5212 - loss 0.79086304 - time (sec): 1415.99 - samples/sec: 259.36 - lr: 0.000160 - momentum: 0.000000 2023-10-11 14:57:20,519 ---------------------------------------------------------------------------------------------------- 2023-10-11 14:57:20,519 EPOCH 1 done: loss 0.7907 - lr: 0.000160 2023-10-11 14:57:56,736 DEV : loss 0.13207665085792542 - f1-score (micro avg) 0.3237 2023-10-11 14:57:56,789 saving best model 2023-10-11 14:57:57,698 ---------------------------------------------------------------------------------------------------- 2023-10-11 15:00:22,791 epoch 2 - iter 521/5212 - loss 0.19332337 - time (sec): 145.09 - samples/sec: 254.59 - lr: 0.000158 - momentum: 0.000000 2023-10-11 15:02:45,528 epoch 2 - iter 1042/5212 - loss 0.18215094 - time (sec): 287.83 - samples/sec: 254.97 - lr: 0.000156 - momentum: 0.000000 2023-10-11 15:05:10,335 epoch 2 - iter 1563/5212 - loss 0.18444577 - time (sec): 432.63 - samples/sec: 260.62 - lr: 0.000155 - momentum: 0.000000 2023-10-11 15:07:31,990 epoch 2 - iter 2084/5212 - loss 0.18224841 - time (sec): 574.29 - samples/sec: 260.31 - lr: 0.000153 - momentum: 0.000000 2023-10-11 15:09:52,786 epoch 2 - iter 2605/5212 - loss 0.17760403 - time (sec): 715.09 - samples/sec: 258.27 - lr: 0.000151 - momentum: 0.000000 2023-10-11 15:12:16,111 epoch 2 - iter 3126/5212 - loss 0.17321912 - time (sec): 858.41 - samples/sec: 258.53 - lr: 0.000149 - momentum: 0.000000 2023-10-11 15:14:34,077 epoch 2 - iter 3647/5212 - loss 0.17155961 - time (sec): 996.38 - samples/sec: 256.43 - lr: 0.000148 - momentum: 0.000000 2023-10-11 15:16:56,628 epoch 2 - iter 4168/5212 - loss 0.16731547 - time (sec): 1138.93 - samples/sec: 256.39 - lr: 0.000146 - momentum: 0.000000 2023-10-11 15:19:19,074 epoch 2 - iter 4689/5212 - loss 0.16554814 - time (sec): 1281.37 - samples/sec: 257.92 - lr: 0.000144 - momentum: 0.000000 2023-10-11 15:21:37,400 epoch 2 - iter 5210/5212 - loss 0.16192292 - time (sec): 1419.70 - samples/sec: 258.75 - lr: 0.000142 - momentum: 0.000000 2023-10-11 15:21:37,838 ---------------------------------------------------------------------------------------------------- 2023-10-11 15:21:37,838 EPOCH 2 done: loss 0.1619 - lr: 0.000142 2023-10-11 15:22:17,665 DEV : loss 0.16001686453819275 - f1-score (micro avg) 0.3627 2023-10-11 15:22:17,722 saving best model 2023-10-11 15:22:20,336 ---------------------------------------------------------------------------------------------------- 2023-10-11 15:24:36,175 epoch 3 - iter 521/5212 - loss 0.09560052 - time (sec): 135.83 - samples/sec: 257.42 - lr: 0.000140 - momentum: 0.000000 2023-10-11 15:26:53,189 epoch 3 - iter 1042/5212 - loss 0.10144802 - time (sec): 272.85 - samples/sec: 260.58 - lr: 0.000139 - momentum: 0.000000 2023-10-11 15:29:13,216 epoch 3 - iter 1563/5212 - loss 0.09999435 - time (sec): 412.88 - samples/sec: 258.87 - lr: 0.000137 - momentum: 0.000000 2023-10-11 15:31:36,953 epoch 3 - iter 2084/5212 - loss 0.10591845 - time (sec): 556.61 - samples/sec: 261.87 - lr: 0.000135 - momentum: 0.000000 2023-10-11 15:33:57,505 epoch 3 - iter 2605/5212 - loss 0.11024910 - time (sec): 697.16 - samples/sec: 263.88 - lr: 0.000133 - momentum: 0.000000 2023-10-11 15:36:17,912 epoch 3 - iter 3126/5212 - loss 0.10967581 - time (sec): 837.57 - samples/sec: 262.61 - lr: 0.000132 - momentum: 0.000000 2023-10-11 15:38:38,424 epoch 3 - iter 3647/5212 - loss 0.10727296 - time (sec): 978.08 - samples/sec: 261.21 - lr: 0.000130 - momentum: 0.000000 2023-10-11 15:40:59,896 epoch 3 - iter 4168/5212 - loss 0.10769948 - time (sec): 1119.55 - samples/sec: 261.35 - lr: 0.000128 - momentum: 0.000000 2023-10-11 15:43:20,757 epoch 3 - iter 4689/5212 - loss 0.10703189 - time (sec): 1260.42 - samples/sec: 260.75 - lr: 0.000126 - momentum: 0.000000 2023-10-11 15:45:46,414 epoch 3 - iter 5210/5212 - loss 0.10585857 - time (sec): 1406.07 - samples/sec: 261.24 - lr: 0.000124 - momentum: 0.000000 2023-10-11 15:45:46,886 ---------------------------------------------------------------------------------------------------- 2023-10-11 15:45:46,886 EPOCH 3 done: loss 0.1060 - lr: 0.000124 2023-10-11 15:46:27,270 DEV : loss 0.21690955758094788 - f1-score (micro avg) 0.3633 2023-10-11 15:46:27,322 saving best model 2023-10-11 15:46:29,931 ---------------------------------------------------------------------------------------------------- 2023-10-11 15:48:54,172 epoch 4 - iter 521/5212 - loss 0.07319839 - time (sec): 144.24 - samples/sec: 244.92 - lr: 0.000123 - momentum: 0.000000 2023-10-11 15:51:17,181 epoch 4 - iter 1042/5212 - loss 0.07369046 - time (sec): 287.25 - samples/sec: 250.39 - lr: 0.000121 - momentum: 0.000000 2023-10-11 15:53:41,453 epoch 4 - iter 1563/5212 - loss 0.07241720 - time (sec): 431.52 - samples/sec: 253.60 - lr: 0.000119 - momentum: 0.000000 2023-10-11 15:56:04,766 epoch 4 - iter 2084/5212 - loss 0.07187332 - time (sec): 574.83 - samples/sec: 251.80 - lr: 0.000117 - momentum: 0.000000 2023-10-11 15:58:30,038 epoch 4 - iter 2605/5212 - loss 0.07110536 - time (sec): 720.10 - samples/sec: 256.81 - lr: 0.000116 - momentum: 0.000000 2023-10-11 16:00:49,154 epoch 4 - iter 3126/5212 - loss 0.07215527 - time (sec): 859.22 - samples/sec: 256.21 - lr: 0.000114 - momentum: 0.000000 2023-10-11 16:03:11,988 epoch 4 - iter 3647/5212 - loss 0.07322367 - time (sec): 1002.05 - samples/sec: 257.20 - lr: 0.000112 - momentum: 0.000000 2023-10-11 16:05:38,070 epoch 4 - iter 4168/5212 - loss 0.07331747 - time (sec): 1148.14 - samples/sec: 259.47 - lr: 0.000110 - momentum: 0.000000 2023-10-11 16:07:57,178 epoch 4 - iter 4689/5212 - loss 0.07365219 - time (sec): 1287.24 - samples/sec: 257.98 - lr: 0.000108 - momentum: 0.000000 2023-10-11 16:10:18,604 epoch 4 - iter 5210/5212 - loss 0.07301371 - time (sec): 1428.67 - samples/sec: 257.16 - lr: 0.000107 - momentum: 0.000000 2023-10-11 16:10:19,011 ---------------------------------------------------------------------------------------------------- 2023-10-11 16:10:19,011 EPOCH 4 done: loss 0.0730 - lr: 0.000107 2023-10-11 16:10:58,573 DEV : loss 0.2578122615814209 - f1-score (micro avg) 0.4037 2023-10-11 16:10:58,626 saving best model 2023-10-11 16:11:01,254 ---------------------------------------------------------------------------------------------------- 2023-10-11 16:13:24,831 epoch 5 - iter 521/5212 - loss 0.04596828 - time (sec): 143.57 - samples/sec: 250.03 - lr: 0.000105 - momentum: 0.000000 2023-10-11 16:15:52,154 epoch 5 - iter 1042/5212 - loss 0.05176257 - time (sec): 290.89 - samples/sec: 253.07 - lr: 0.000103 - momentum: 0.000000 2023-10-11 16:18:26,212 epoch 5 - iter 1563/5212 - loss 0.05301496 - time (sec): 444.95 - samples/sec: 245.48 - lr: 0.000101 - momentum: 0.000000 2023-10-11 16:20:56,560 epoch 5 - iter 2084/5212 - loss 0.05310164 - time (sec): 595.30 - samples/sec: 244.83 - lr: 0.000100 - momentum: 0.000000 2023-10-11 16:23:29,574 epoch 5 - iter 2605/5212 - loss 0.05219774 - time (sec): 748.32 - samples/sec: 246.05 - lr: 0.000098 - momentum: 0.000000 2023-10-11 16:25:52,444 epoch 5 - iter 3126/5212 - loss 0.05195458 - time (sec): 891.19 - samples/sec: 246.25 - lr: 0.000096 - momentum: 0.000000 2023-10-11 16:28:14,834 epoch 5 - iter 3647/5212 - loss 0.05186622 - time (sec): 1033.58 - samples/sec: 247.88 - lr: 0.000094 - momentum: 0.000000 2023-10-11 16:30:31,671 epoch 5 - iter 4168/5212 - loss 0.05189342 - time (sec): 1170.41 - samples/sec: 248.98 - lr: 0.000092 - momentum: 0.000000 2023-10-11 16:32:51,894 epoch 5 - iter 4689/5212 - loss 0.05074979 - time (sec): 1310.64 - samples/sec: 251.02 - lr: 0.000091 - momentum: 0.000000 2023-10-11 16:35:18,427 epoch 5 - iter 5210/5212 - loss 0.05240335 - time (sec): 1457.17 - samples/sec: 252.09 - lr: 0.000089 - momentum: 0.000000 2023-10-11 16:35:18,890 ---------------------------------------------------------------------------------------------------- 2023-10-11 16:35:18,891 EPOCH 5 done: loss 0.0524 - lr: 0.000089 2023-10-11 16:35:57,847 DEV : loss 0.3427276015281677 - f1-score (micro avg) 0.3733 2023-10-11 16:35:57,906 ---------------------------------------------------------------------------------------------------- 2023-10-11 16:38:16,413 epoch 6 - iter 521/5212 - loss 0.03453889 - time (sec): 138.50 - samples/sec: 243.67 - lr: 0.000087 - momentum: 0.000000 2023-10-11 16:40:36,448 epoch 6 - iter 1042/5212 - loss 0.03594338 - time (sec): 278.54 - samples/sec: 245.57 - lr: 0.000085 - momentum: 0.000000 2023-10-11 16:42:56,379 epoch 6 - iter 1563/5212 - loss 0.03623246 - time (sec): 418.47 - samples/sec: 250.12 - lr: 0.000084 - momentum: 0.000000 2023-10-11 16:45:13,717 epoch 6 - iter 2084/5212 - loss 0.03614444 - time (sec): 555.81 - samples/sec: 253.09 - lr: 0.000082 - momentum: 0.000000 2023-10-11 16:47:33,364 epoch 6 - iter 2605/5212 - loss 0.03645808 - time (sec): 695.46 - samples/sec: 255.18 - lr: 0.000080 - momentum: 0.000000 2023-10-11 16:49:50,937 epoch 6 - iter 3126/5212 - loss 0.03572045 - time (sec): 833.03 - samples/sec: 255.72 - lr: 0.000078 - momentum: 0.000000 2023-10-11 16:52:13,879 epoch 6 - iter 3647/5212 - loss 0.03557647 - time (sec): 975.97 - samples/sec: 258.76 - lr: 0.000076 - momentum: 0.000000 2023-10-11 16:54:39,082 epoch 6 - iter 4168/5212 - loss 0.03568794 - time (sec): 1121.17 - samples/sec: 258.91 - lr: 0.000075 - momentum: 0.000000 2023-10-11 16:57:09,848 epoch 6 - iter 4689/5212 - loss 0.03525994 - time (sec): 1271.94 - samples/sec: 259.16 - lr: 0.000073 - momentum: 0.000000 2023-10-11 16:59:33,202 epoch 6 - iter 5210/5212 - loss 0.03514270 - time (sec): 1415.29 - samples/sec: 259.42 - lr: 0.000071 - momentum: 0.000000 2023-10-11 16:59:33,838 ---------------------------------------------------------------------------------------------------- 2023-10-11 16:59:33,838 EPOCH 6 done: loss 0.0351 - lr: 0.000071 2023-10-11 17:00:14,726 DEV : loss 0.4193594753742218 - f1-score (micro avg) 0.3791 2023-10-11 17:00:14,781 ---------------------------------------------------------------------------------------------------- 2023-10-11 17:02:44,618 epoch 7 - iter 521/5212 - loss 0.02269233 - time (sec): 149.83 - samples/sec: 267.66 - lr: 0.000069 - momentum: 0.000000 2023-10-11 17:05:04,738 epoch 7 - iter 1042/5212 - loss 0.02635316 - time (sec): 289.95 - samples/sec: 262.02 - lr: 0.000068 - momentum: 0.000000 2023-10-11 17:07:27,032 epoch 7 - iter 1563/5212 - loss 0.02308396 - time (sec): 432.25 - samples/sec: 262.49 - lr: 0.000066 - momentum: 0.000000 2023-10-11 17:09:48,465 epoch 7 - iter 2084/5212 - loss 0.02400360 - time (sec): 573.68 - samples/sec: 264.91 - lr: 0.000064 - momentum: 0.000000 2023-10-11 17:12:06,814 epoch 7 - iter 2605/5212 - loss 0.02542856 - time (sec): 712.03 - samples/sec: 260.99 - lr: 0.000062 - momentum: 0.000000 2023-10-11 17:14:31,094 epoch 7 - iter 3126/5212 - loss 0.02631394 - time (sec): 856.31 - samples/sec: 260.96 - lr: 0.000060 - momentum: 0.000000 2023-10-11 17:16:53,608 epoch 7 - iter 3647/5212 - loss 0.02640215 - time (sec): 998.82 - samples/sec: 259.79 - lr: 0.000059 - momentum: 0.000000 2023-10-11 17:19:14,821 epoch 7 - iter 4168/5212 - loss 0.02608981 - time (sec): 1140.04 - samples/sec: 258.59 - lr: 0.000057 - momentum: 0.000000 2023-10-11 17:21:40,650 epoch 7 - iter 4689/5212 - loss 0.02628534 - time (sec): 1285.87 - samples/sec: 257.56 - lr: 0.000055 - momentum: 0.000000 2023-10-11 17:24:06,529 epoch 7 - iter 5210/5212 - loss 0.02607886 - time (sec): 1431.75 - samples/sec: 256.60 - lr: 0.000053 - momentum: 0.000000 2023-10-11 17:24:06,965 ---------------------------------------------------------------------------------------------------- 2023-10-11 17:24:06,965 EPOCH 7 done: loss 0.0261 - lr: 0.000053 2023-10-11 17:24:47,377 DEV : loss 0.4072570204734802 - f1-score (micro avg) 0.3823 2023-10-11 17:24:47,431 ---------------------------------------------------------------------------------------------------- 2023-10-11 17:27:15,714 epoch 8 - iter 521/5212 - loss 0.01684084 - time (sec): 148.28 - samples/sec: 248.06 - lr: 0.000052 - momentum: 0.000000 2023-10-11 17:29:41,615 epoch 8 - iter 1042/5212 - loss 0.01921364 - time (sec): 294.18 - samples/sec: 252.08 - lr: 0.000050 - momentum: 0.000000 2023-10-11 17:32:06,810 epoch 8 - iter 1563/5212 - loss 0.02074029 - time (sec): 439.38 - samples/sec: 250.98 - lr: 0.000048 - momentum: 0.000000 2023-10-11 17:34:30,337 epoch 8 - iter 2084/5212 - loss 0.01866275 - time (sec): 582.90 - samples/sec: 252.36 - lr: 0.000046 - momentum: 0.000000 2023-10-11 17:36:55,146 epoch 8 - iter 2605/5212 - loss 0.01912678 - time (sec): 727.71 - samples/sec: 253.80 - lr: 0.000044 - momentum: 0.000000 2023-10-11 17:39:18,799 epoch 8 - iter 3126/5212 - loss 0.01960685 - time (sec): 871.37 - samples/sec: 253.85 - lr: 0.000043 - momentum: 0.000000 2023-10-11 17:41:42,158 epoch 8 - iter 3647/5212 - loss 0.01902745 - time (sec): 1014.72 - samples/sec: 253.30 - lr: 0.000041 - momentum: 0.000000 2023-10-11 17:44:07,681 epoch 8 - iter 4168/5212 - loss 0.01851736 - time (sec): 1160.25 - samples/sec: 253.42 - lr: 0.000039 - momentum: 0.000000 2023-10-11 17:46:34,426 epoch 8 - iter 4689/5212 - loss 0.01823475 - time (sec): 1306.99 - samples/sec: 253.91 - lr: 0.000037 - momentum: 0.000000 2023-10-11 17:48:56,985 epoch 8 - iter 5210/5212 - loss 0.01875215 - time (sec): 1449.55 - samples/sec: 253.25 - lr: 0.000036 - momentum: 0.000000 2023-10-11 17:48:57,706 ---------------------------------------------------------------------------------------------------- 2023-10-11 17:48:57,707 EPOCH 8 done: loss 0.0187 - lr: 0.000036 2023-10-11 17:49:37,133 DEV : loss 0.41436994075775146 - f1-score (micro avg) 0.4115 2023-10-11 17:49:37,186 saving best model 2023-10-11 17:49:39,805 ---------------------------------------------------------------------------------------------------- 2023-10-11 17:52:05,310 epoch 9 - iter 521/5212 - loss 0.01196532 - time (sec): 145.50 - samples/sec: 263.89 - lr: 0.000034 - momentum: 0.000000 2023-10-11 17:54:27,376 epoch 9 - iter 1042/5212 - loss 0.01333594 - time (sec): 287.57 - samples/sec: 264.07 - lr: 0.000032 - momentum: 0.000000 2023-10-11 17:56:48,142 epoch 9 - iter 1563/5212 - loss 0.01211283 - time (sec): 428.33 - samples/sec: 260.34 - lr: 0.000030 - momentum: 0.000000 2023-10-11 17:59:07,321 epoch 9 - iter 2084/5212 - loss 0.01222749 - time (sec): 567.51 - samples/sec: 257.43 - lr: 0.000028 - momentum: 0.000000 2023-10-11 18:01:27,565 epoch 9 - iter 2605/5212 - loss 0.01192194 - time (sec): 707.76 - samples/sec: 259.45 - lr: 0.000027 - momentum: 0.000000 2023-10-11 18:03:45,920 epoch 9 - iter 3126/5212 - loss 0.01186385 - time (sec): 846.11 - samples/sec: 259.03 - lr: 0.000025 - momentum: 0.000000 2023-10-11 18:06:04,727 epoch 9 - iter 3647/5212 - loss 0.01241909 - time (sec): 984.92 - samples/sec: 260.05 - lr: 0.000023 - momentum: 0.000000 2023-10-11 18:08:24,518 epoch 9 - iter 4168/5212 - loss 0.01204146 - time (sec): 1124.71 - samples/sec: 261.02 - lr: 0.000021 - momentum: 0.000000 2023-10-11 18:10:47,837 epoch 9 - iter 4689/5212 - loss 0.01273801 - time (sec): 1268.03 - samples/sec: 260.93 - lr: 0.000020 - momentum: 0.000000 2023-10-11 18:13:07,309 epoch 9 - iter 5210/5212 - loss 0.01307196 - time (sec): 1407.50 - samples/sec: 260.86 - lr: 0.000018 - momentum: 0.000000 2023-10-11 18:13:07,901 ---------------------------------------------------------------------------------------------------- 2023-10-11 18:13:07,901 EPOCH 9 done: loss 0.0131 - lr: 0.000018 2023-10-11 18:13:46,547 DEV : loss 0.4808931350708008 - f1-score (micro avg) 0.3967 2023-10-11 18:13:46,607 ---------------------------------------------------------------------------------------------------- 2023-10-11 18:16:05,453 epoch 10 - iter 521/5212 - loss 0.00668555 - time (sec): 138.84 - samples/sec: 262.74 - lr: 0.000016 - momentum: 0.000000 2023-10-11 18:18:22,229 epoch 10 - iter 1042/5212 - loss 0.00706135 - time (sec): 275.62 - samples/sec: 260.81 - lr: 0.000014 - momentum: 0.000000 2023-10-11 18:20:40,291 epoch 10 - iter 1563/5212 - loss 0.00669620 - time (sec): 413.68 - samples/sec: 262.77 - lr: 0.000012 - momentum: 0.000000 2023-10-11 18:22:57,132 epoch 10 - iter 2084/5212 - loss 0.00716732 - time (sec): 550.52 - samples/sec: 261.09 - lr: 0.000011 - momentum: 0.000000 2023-10-11 18:25:18,034 epoch 10 - iter 2605/5212 - loss 0.00737559 - time (sec): 691.43 - samples/sec: 264.83 - lr: 0.000009 - momentum: 0.000000 2023-10-11 18:27:35,108 epoch 10 - iter 3126/5212 - loss 0.00746666 - time (sec): 828.50 - samples/sec: 264.40 - lr: 0.000007 - momentum: 0.000000 2023-10-11 18:29:52,239 epoch 10 - iter 3647/5212 - loss 0.00801610 - time (sec): 965.63 - samples/sec: 264.55 - lr: 0.000005 - momentum: 0.000000 2023-10-11 18:32:08,536 epoch 10 - iter 4168/5212 - loss 0.00797516 - time (sec): 1101.93 - samples/sec: 263.54 - lr: 0.000004 - momentum: 0.000000 2023-10-11 18:34:30,744 epoch 10 - iter 4689/5212 - loss 0.00813383 - time (sec): 1244.13 - samples/sec: 265.19 - lr: 0.000002 - momentum: 0.000000 2023-10-11 18:36:51,864 epoch 10 - iter 5210/5212 - loss 0.00803504 - time (sec): 1385.26 - samples/sec: 265.22 - lr: 0.000000 - momentum: 0.000000 2023-10-11 18:36:52,264 ---------------------------------------------------------------------------------------------------- 2023-10-11 18:36:52,265 EPOCH 10 done: loss 0.0080 - lr: 0.000000 2023-10-11 18:37:30,395 DEV : loss 0.47644102573394775 - f1-score (micro avg) 0.4056 2023-10-11 18:37:31,324 ---------------------------------------------------------------------------------------------------- 2023-10-11 18:37:31,326 Loading model from best epoch ... 2023-10-11 18:37:35,775 SequenceTagger predicts: Dictionary with 17 tags: O, S-LOC, B-LOC, E-LOC, I-LOC, S-PER, B-PER, E-PER, I-PER, S-ORG, B-ORG, E-ORG, I-ORG, S-HumanProd, B-HumanProd, E-HumanProd, I-HumanProd 2023-10-11 18:39:14,485 Results: - F-score (micro) 0.4756 - F-score (macro) 0.3235 - Accuracy 0.317 By class: precision recall f1-score support LOC 0.4876 0.5972 0.5368 1214 PER 0.4418 0.4790 0.4596 808 ORG 0.2923 0.3031 0.2976 353 HumanProd 0.0000 0.0000 0.0000 15 micro avg 0.4455 0.5100 0.4756 2390 macro avg 0.3054 0.3448 0.3235 2390 weighted avg 0.4402 0.5100 0.4720 2390 2023-10-11 18:39:14,485 ----------------------------------------------------------------------------------------------------