ner-german-legal / training.log
alanakbik's picture
initial model commit
cf27b9e
2020-09-11 07:40:36,037 ----------------------------------------------------------------------------------------------------
2020-09-11 07:40:36,038 Model: "SequenceTagger(
(embeddings): StackedEmbeddings(
(list_embedding_0): WordEmbeddings('de')
(list_embedding_1): FlairEmbeddings(
(lm): LanguageModel(
(drop): Dropout(p=0.25, inplace=False)
(encoder): Embedding(275, 100)
(rnn): LSTM(100, 2048)
(decoder): Linear(in_features=2048, out_features=275, bias=True)
)
)
(list_embedding_2): FlairEmbeddings(
(lm): LanguageModel(
(drop): Dropout(p=0.25, inplace=False)
(encoder): Embedding(275, 100)
(rnn): LSTM(100, 2048)
(decoder): Linear(in_features=2048, out_features=275, bias=True)
)
)
)
(word_dropout): WordDropout(p=0.05)
(locked_dropout): LockedDropout(p=0.5)
(embedding2nn): Linear(in_features=4396, out_features=4396, bias=True)
(rnn): LSTM(4396, 256, batch_first=True, bidirectional=True)
(linear): Linear(in_features=512, out_features=78, bias=True)
(beta): 1.0
(weights): None
(weight_tensor) None
)"
2020-09-11 07:40:36,038 ----------------------------------------------------------------------------------------------------
2020-09-11 07:40:36,038 Corpus: "Corpus: 51986 train + 5770 dev + 6419 test sentences"
2020-09-11 07:40:36,038 ----------------------------------------------------------------------------------------------------
2020-09-11 07:40:36,038 Parameters:
2020-09-11 07:40:36,038 - learning_rate: "0.1"
2020-09-11 07:40:36,038 - mini_batch_size: "32"
2020-09-11 07:40:36,038 - patience: "3"
2020-09-11 07:40:36,038 - anneal_factor: "0.5"
2020-09-11 07:40:36,038 - max_epochs: "150"
2020-09-11 07:40:36,038 - shuffle: "True"
2020-09-11 07:40:36,038 - train_with_dev: "True"
2020-09-11 07:40:36,038 - batch_growth_annealing: "False"
2020-09-11 07:40:36,038 ----------------------------------------------------------------------------------------------------
2020-09-11 07:40:36,038 Model training base path: "resources/taggers/legal-ner"
2020-09-11 07:40:36,038 ----------------------------------------------------------------------------------------------------
2020-09-11 07:40:36,038 Device: cuda:0
2020-09-11 07:40:36,038 ----------------------------------------------------------------------------------------------------
2020-09-11 07:40:36,038 Embeddings storage mode: cpu
2020-09-11 07:40:36,042 ----------------------------------------------------------------------------------------------------
2020-09-11 07:42:10,892 epoch 1 - iter 180/1805 - loss 13.80402097 - samples/sec: 60.75 - lr: 0.100000
2020-09-11 07:43:46,710 epoch 1 - iter 360/1805 - loss 10.27129066 - samples/sec: 60.14 - lr: 0.100000
2020-09-11 07:45:28,099 epoch 1 - iter 540/1805 - loss 8.53650223 - samples/sec: 56.83 - lr: 0.100000
2020-09-11 07:47:02,566 epoch 1 - iter 720/1805 - loss 7.46493234 - samples/sec: 61.00 - lr: 0.100000
2020-09-11 07:48:36,770 epoch 1 - iter 900/1805 - loss 6.67135013 - samples/sec: 61.16 - lr: 0.100000
2020-09-11 07:50:10,723 epoch 1 - iter 1080/1805 - loss 6.07681235 - samples/sec: 61.33 - lr: 0.100000
2020-09-11 07:51:51,829 epoch 1 - iter 1260/1805 - loss 5.58949287 - samples/sec: 56.99 - lr: 0.100000
2020-09-11 07:53:25,569 epoch 1 - iter 1440/1805 - loss 5.17579757 - samples/sec: 61.47 - lr: 0.100000
2020-09-11 07:55:00,098 epoch 1 - iter 1620/1805 - loss 4.84291735 - samples/sec: 60.95 - lr: 0.100000
2020-09-11 07:56:31,968 epoch 1 - iter 1800/1805 - loss 4.56184329 - samples/sec: 62.72 - lr: 0.100000
2020-09-11 07:56:34,557 ----------------------------------------------------------------------------------------------------
2020-09-11 07:56:34,557 EPOCH 1 done: loss 4.5526 - lr 0.1000000
2020-09-11 07:56:34,557 BAD EPOCHS (no improvement): 0
2020-09-11 07:56:34,563 ----------------------------------------------------------------------------------------------------
2020-09-11 07:57:32,527 epoch 2 - iter 180/1805 - loss 1.83645120 - samples/sec: 99.42 - lr: 0.100000
2020-09-11 07:58:31,892 epoch 2 - iter 360/1805 - loss 1.81463750 - samples/sec: 97.07 - lr: 0.100000
2020-09-11 07:59:30,134 epoch 2 - iter 540/1805 - loss 1.71575339 - samples/sec: 98.94 - lr: 0.100000
2020-09-11 08:00:28,619 epoch 2 - iter 720/1805 - loss 1.66278592 - samples/sec: 98.53 - lr: 0.100000
2020-09-11 08:01:27,950 epoch 2 - iter 900/1805 - loss 1.61247091 - samples/sec: 97.13 - lr: 0.100000
2020-09-11 08:02:27,326 epoch 2 - iter 1080/1805 - loss 1.58761564 - samples/sec: 97.05 - lr: 0.100000
2020-09-11 08:03:25,601 epoch 2 - iter 1260/1805 - loss 1.54939332 - samples/sec: 98.89 - lr: 0.100000
2020-09-11 08:04:24,827 epoch 2 - iter 1440/1805 - loss 1.51264721 - samples/sec: 97.30 - lr: 0.100000
2020-09-11 08:05:23,611 epoch 2 - iter 1620/1805 - loss 1.49112630 - samples/sec: 98.03 - lr: 0.100000
2020-09-11 08:06:22,409 epoch 2 - iter 1800/1805 - loss 1.46093700 - samples/sec: 98.01 - lr: 0.100000
2020-09-11 08:06:23,921 ----------------------------------------------------------------------------------------------------
2020-09-11 08:06:23,922 EPOCH 2 done: loss 1.4598 - lr 0.1000000
2020-09-11 08:06:23,922 BAD EPOCHS (no improvement): 0
2020-09-11 08:06:23,925 ----------------------------------------------------------------------------------------------------
2020-09-11 08:07:22,373 epoch 3 - iter 180/1805 - loss 1.15581222 - samples/sec: 98.60 - lr: 0.100000