Permalink
Cannot retrieve contributors at this time
Name already in use
A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?
web-data/gluonnlp/logs/bert/finetune_STS-B_base_mx1.6.0rc1.log
Go to fileThis commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
210 lines (210 sloc)
16.5 KB
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| INFO:root:04:26:58 Namespace(accumulate=None, batch_size=32, bert_dataset='book_corpus_wiki_en_uncased', bert_model='bert_12_768_12', dev_batch_size=8, dtype='float32', early_stop=None, epochs=5, epsilon=1e-06, gpu=0, log_interval=10, lr=2e-05, max_len=128, model_parameters=None, only_inference=False, optimizer='bertadam', output_dir='./output_dir', pad=False, pretrained_bert_parameters=None, seed=24, task_name='STS-B', training_steps=None, warmup_ratio=0.1) | |
| INFO:root:04:27:03 processing dataset... | |
| INFO:root:04:27:04 Now we are doing BERT classification training on gpu(0)! | |
| INFO:root:04:27:04 training steps=898 | |
| INFO:root:04:27:07 [Epoch 1 Batch 10/186] loss=3.6108, lr=0.0000020, metrics:pearsonr:0.1008 | |
| INFO:root:04:27:08 [Epoch 1 Batch 20/186] loss=2.6426, lr=0.0000043, metrics:pearsonr:0.0770 | |
| INFO:root:04:27:09 [Epoch 1 Batch 30/186] loss=2.3315, lr=0.0000065, metrics:pearsonr:0.0852 | |
| INFO:root:04:27:10 [Epoch 1 Batch 40/186] loss=1.1828, lr=0.0000088, metrics:pearsonr:0.0428 | |
| INFO:root:04:27:11 [Epoch 1 Batch 50/186] loss=0.5126, lr=0.0000110, metrics:pearsonr:0.1694 | |
| INFO:root:04:27:12 [Epoch 1 Batch 60/186] loss=0.4436, lr=0.0000133, metrics:pearsonr:0.2465 | |
| INFO:root:04:27:13 [Epoch 1 Batch 70/186] loss=0.4791, lr=0.0000155, metrics:pearsonr:0.3107 | |
| INFO:root:04:27:14 [Epoch 1 Batch 80/186] loss=0.4374, lr=0.0000178, metrics:pearsonr:0.3517 | |
| INFO:root:04:27:15 [Epoch 1 Batch 90/186] loss=0.4406, lr=0.0000200, metrics:pearsonr:0.3972 | |
| INFO:root:04:27:17 [Epoch 1 Batch 100/186] loss=0.3405, lr=0.0000198, metrics:pearsonr:0.4323 | |
| INFO:root:04:27:18 [Epoch 1 Batch 110/186] loss=0.3176, lr=0.0000195, metrics:pearsonr:0.4679 | |
| INFO:root:04:27:19 [Epoch 1 Batch 120/186] loss=0.3588, lr=0.0000193, metrics:pearsonr:0.4952 | |
| INFO:root:04:27:20 [Epoch 1 Batch 130/186] loss=0.3048, lr=0.0000190, metrics:pearsonr:0.5134 | |
| INFO:root:04:27:21 [Epoch 1 Batch 140/186] loss=0.3876, lr=0.0000188, metrics:pearsonr:0.5301 | |
| INFO:root:04:27:22 [Epoch 1 Batch 150/186] loss=0.3090, lr=0.0000185, metrics:pearsonr:0.5488 | |
| INFO:root:04:27:23 [Epoch 1 Batch 160/186] loss=0.2981, lr=0.0000183, metrics:pearsonr:0.5699 | |
| INFO:root:04:27:24 [Epoch 1 Batch 170/186] loss=0.2682, lr=0.0000180, metrics:pearsonr:0.5855 | |
| INFO:root:04:27:25 [Epoch 1 Batch 180/186] loss=0.2268, lr=0.0000178, metrics:pearsonr:0.6013 | |
| INFO:root:04:27:26 Now we are doing evaluation on dev with gpu(0). | |
| INFO:root:04:27:26 [Batch 10/188] loss=0.1964, metrics:pearsonr:0.9240 | |
| INFO:root:04:27:26 [Batch 20/188] loss=0.2029, metrics:pearsonr:0.9312 | |
| INFO:root:04:27:27 [Batch 30/188] loss=0.1690, metrics:pearsonr:0.9320 | |
| INFO:root:04:27:27 [Batch 40/188] loss=0.2815, metrics:pearsonr:0.9231 | |
| INFO:root:04:27:27 [Batch 50/188] loss=0.2446, metrics:pearsonr:0.9188 | |
| INFO:root:04:27:27 [Batch 60/188] loss=0.1665, metrics:pearsonr:0.9205 | |
| INFO:root:04:27:27 [Batch 70/188] loss=0.3285, metrics:pearsonr:0.9133 | |
| INFO:root:04:27:27 [Batch 80/188] loss=0.2654, metrics:pearsonr:0.9111 | |
| INFO:root:04:27:28 [Batch 90/188] loss=0.3748, metrics:pearsonr:0.9015 | |
| INFO:root:04:27:28 [Batch 100/188] loss=0.2416, metrics:pearsonr:0.8989 | |
| INFO:root:04:27:28 [Batch 110/188] loss=0.2666, metrics:pearsonr:0.8932 | |
| INFO:root:04:27:28 [Batch 120/188] loss=0.2792, metrics:pearsonr:0.8904 | |
| INFO:root:04:27:28 [Batch 130/188] loss=0.2010, metrics:pearsonr:0.8909 | |
| INFO:root:04:27:28 [Batch 140/188] loss=0.1779, metrics:pearsonr:0.8930 | |
| INFO:root:04:27:29 [Batch 150/188] loss=0.1559, metrics:pearsonr:0.8940 | |
| INFO:root:04:27:29 [Batch 160/188] loss=0.2053, metrics:pearsonr:0.8939 | |
| INFO:root:04:27:29 [Batch 170/188] loss=0.2128, metrics:pearsonr:0.8932 | |
| INFO:root:04:27:29 [Batch 180/188] loss=0.2666, metrics:pearsonr:0.8912 | |
| INFO:root:04:27:29 validation metrics:pearsonr:0.8910 | |
| INFO:root:04:27:29 Time cost=3.22s, throughput=467.47 samples/s | |
| INFO:root:04:27:30 params saved in: ./output_dir/model_bert_STS-B_0.params | |
| INFO:root:04:27:30 Time cost=26.37s | |
| INFO:root:04:27:31 [Epoch 2 Batch 10/186] loss=0.2125, lr=0.0000174, metrics:pearsonr:0.9066 | |
| INFO:root:04:27:33 [Epoch 2 Batch 20/186] loss=0.1993, lr=0.0000171, metrics:pearsonr:0.8988 | |
| INFO:root:04:27:34 [Epoch 2 Batch 30/186] loss=0.1767, lr=0.0000169, metrics:pearsonr:0.9088 | |
| INFO:root:04:27:35 [Epoch 2 Batch 40/186] loss=0.1865, lr=0.0000166, metrics:pearsonr:0.9086 | |
| INFO:root:04:27:36 [Epoch 2 Batch 50/186] loss=0.1736, lr=0.0000164, metrics:pearsonr:0.9092 | |
| INFO:root:04:27:37 [Epoch 2 Batch 60/186] loss=0.1796, lr=0.0000161, metrics:pearsonr:0.9097 | |
| INFO:root:04:27:38 [Epoch 2 Batch 70/186] loss=0.1830, lr=0.0000159, metrics:pearsonr:0.9110 | |
| INFO:root:04:27:39 [Epoch 2 Batch 80/186] loss=0.2054, lr=0.0000156, metrics:pearsonr:0.9093 | |
| INFO:root:04:27:40 [Epoch 2 Batch 90/186] loss=0.2154, lr=0.0000154, metrics:pearsonr:0.9069 | |
| INFO:root:04:27:42 [Epoch 2 Batch 100/186] loss=0.1643, lr=0.0000152, metrics:pearsonr:0.9085 | |
| INFO:root:04:27:43 [Epoch 2 Batch 110/186] loss=0.1976, lr=0.0000149, metrics:pearsonr:0.9088 | |
| INFO:root:04:27:44 [Epoch 2 Batch 120/186] loss=0.1669, lr=0.0000147, metrics:pearsonr:0.9094 | |
| INFO:root:04:27:45 [Epoch 2 Batch 130/186] loss=0.1826, lr=0.0000144, metrics:pearsonr:0.9088 | |
| INFO:root:04:27:46 [Epoch 2 Batch 140/186] loss=0.1740, lr=0.0000142, metrics:pearsonr:0.9083 | |
| INFO:root:04:27:47 [Epoch 2 Batch 150/186] loss=0.1869, lr=0.0000139, metrics:pearsonr:0.9079 | |
| INFO:root:04:27:48 [Epoch 2 Batch 160/186] loss=0.2017, lr=0.0000137, metrics:pearsonr:0.9074 | |
| INFO:root:04:27:49 [Epoch 2 Batch 170/186] loss=0.1681, lr=0.0000134, metrics:pearsonr:0.9078 | |
| INFO:root:04:27:50 [Epoch 2 Batch 180/186] loss=0.1615, lr=0.0000132, metrics:pearsonr:0.9090 | |
| INFO:root:04:27:51 Now we are doing evaluation on dev with gpu(0). | |
| INFO:root:04:27:51 [Batch 10/188] loss=0.1701, metrics:pearsonr:0.9377 | |
| INFO:root:04:27:51 [Batch 20/188] loss=0.1612, metrics:pearsonr:0.9442 | |
| INFO:root:04:27:51 [Batch 30/188] loss=0.1650, metrics:pearsonr:0.9415 | |
| INFO:root:04:27:51 [Batch 40/188] loss=0.1976, metrics:pearsonr:0.9369 | |
| INFO:root:04:27:52 [Batch 50/188] loss=0.2091, metrics:pearsonr:0.9331 | |
| INFO:root:04:27:52 [Batch 60/188] loss=0.1240, metrics:pearsonr:0.9358 | |
| INFO:root:04:27:52 [Batch 70/188] loss=0.2863, metrics:pearsonr:0.9288 | |
| INFO:root:04:27:52 [Batch 80/188] loss=0.2148, metrics:pearsonr:0.9267 | |
| INFO:root:04:27:52 [Batch 90/188] loss=0.3524, metrics:pearsonr:0.9163 | |
| INFO:root:04:27:52 [Batch 100/188] loss=0.2587, metrics:pearsonr:0.9114 | |
| INFO:root:04:27:53 [Batch 110/188] loss=0.2672, metrics:pearsonr:0.9050 | |
| INFO:root:04:27:53 [Batch 120/188] loss=0.2762, metrics:pearsonr:0.9013 | |
| INFO:root:04:27:53 [Batch 130/188] loss=0.1914, metrics:pearsonr:0.9015 | |
| INFO:root:04:27:53 [Batch 140/188] loss=0.1941, metrics:pearsonr:0.9030 | |
| INFO:root:04:27:53 [Batch 150/188] loss=0.1684, metrics:pearsonr:0.9034 | |
| INFO:root:04:27:53 [Batch 160/188] loss=0.2332, metrics:pearsonr:0.9026 | |
| INFO:root:04:27:54 [Batch 170/188] loss=0.2447, metrics:pearsonr:0.9015 | |
| INFO:root:04:27:54 [Batch 180/188] loss=0.2437, metrics:pearsonr:0.9007 | |
| INFO:root:04:27:54 validation metrics:pearsonr:0.9004 | |
| INFO:root:04:27:54 Time cost=2.92s, throughput=515.63 samples/s | |
| INFO:root:04:27:55 params saved in: ./output_dir/model_bert_STS-B_1.params | |
| INFO:root:04:27:55 Time cost=24.59s | |
| INFO:root:04:27:56 [Epoch 3 Batch 10/186] loss=0.1064, lr=0.0000128, metrics:pearsonr:0.9422 | |
| INFO:root:04:27:57 [Epoch 3 Batch 20/186] loss=0.0958, lr=0.0000125, metrics:pearsonr:0.9489 | |
| INFO:root:04:27:58 [Epoch 3 Batch 30/186] loss=0.1009, lr=0.0000123, metrics:pearsonr:0.9503 | |
| INFO:root:04:28:00 [Epoch 3 Batch 40/186] loss=0.0887, lr=0.0000120, metrics:pearsonr:0.9521 | |
| INFO:root:04:28:01 [Epoch 3 Batch 50/186] loss=0.0993, lr=0.0000118, metrics:pearsonr:0.9525 | |
| INFO:root:04:28:02 [Epoch 3 Batch 60/186] loss=0.0940, lr=0.0000115, metrics:pearsonr:0.9542 | |
| INFO:root:04:28:03 [Epoch 3 Batch 70/186] loss=0.1059, lr=0.0000113, metrics:pearsonr:0.9542 | |
| INFO:root:04:28:04 [Epoch 3 Batch 80/186] loss=0.0964, lr=0.0000111, metrics:pearsonr:0.9546 | |
| INFO:root:04:28:05 [Epoch 3 Batch 90/186] loss=0.0899, lr=0.0000108, metrics:pearsonr:0.9546 | |
| INFO:root:04:28:06 [Epoch 3 Batch 100/186] loss=0.0925, lr=0.0000106, metrics:pearsonr:0.9551 | |
| INFO:root:04:28:07 [Epoch 3 Batch 110/186] loss=0.0828, lr=0.0000103, metrics:pearsonr:0.9551 | |
| INFO:root:04:28:08 [Epoch 3 Batch 120/186] loss=0.0868, lr=0.0000101, metrics:pearsonr:0.9552 | |
| INFO:root:04:28:09 [Epoch 3 Batch 130/186] loss=0.0894, lr=0.0000098, metrics:pearsonr:0.9551 | |
| INFO:root:04:28:11 [Epoch 3 Batch 140/186] loss=0.1161, lr=0.0000096, metrics:pearsonr:0.9539 | |
| INFO:root:04:28:12 [Epoch 3 Batch 150/186] loss=0.0878, lr=0.0000093, metrics:pearsonr:0.9541 | |
| INFO:root:04:28:13 [Epoch 3 Batch 160/186] loss=0.0973, lr=0.0000091, metrics:pearsonr:0.9542 | |
| INFO:root:04:28:14 [Epoch 3 Batch 170/186] loss=0.0834, lr=0.0000088, metrics:pearsonr:0.9547 | |
| INFO:root:04:28:15 [Epoch 3 Batch 180/186] loss=0.1055, lr=0.0000086, metrics:pearsonr:0.9545 | |
| INFO:root:04:28:16 Now we are doing evaluation on dev with gpu(0). | |
| INFO:root:04:28:16 [Batch 10/188] loss=0.1523, metrics:pearsonr:0.9427 | |
| INFO:root:04:28:16 [Batch 20/188] loss=0.1635, metrics:pearsonr:0.9471 | |
| INFO:root:04:28:16 [Batch 30/188] loss=0.1505, metrics:pearsonr:0.9459 | |
| INFO:root:04:28:16 [Batch 40/188] loss=0.1863, metrics:pearsonr:0.9421 | |
| INFO:root:04:28:16 [Batch 50/188] loss=0.2061, metrics:pearsonr:0.9378 | |
| INFO:root:04:28:16 [Batch 60/188] loss=0.1104, metrics:pearsonr:0.9405 | |
| INFO:root:04:28:17 [Batch 70/188] loss=0.3029, metrics:pearsonr:0.9325 | |
| INFO:root:04:28:17 [Batch 80/188] loss=0.2330, metrics:pearsonr:0.9305 | |
| INFO:root:04:28:17 [Batch 90/188] loss=0.3750, metrics:pearsonr:0.9198 | |
| INFO:root:04:28:17 [Batch 100/188] loss=0.2723, metrics:pearsonr:0.9152 | |
| INFO:root:04:28:17 [Batch 110/188] loss=0.2558, metrics:pearsonr:0.9086 | |
| INFO:root:04:28:17 [Batch 120/188] loss=0.3045, metrics:pearsonr:0.9039 | |
| INFO:root:04:28:18 [Batch 130/188] loss=0.1875, metrics:pearsonr:0.9042 | |
| INFO:root:04:28:18 [Batch 140/188] loss=0.1634, metrics:pearsonr:0.9059 | |
| INFO:root:04:28:18 [Batch 150/188] loss=0.1392, metrics:pearsonr:0.9067 | |
| INFO:root:04:28:18 [Batch 160/188] loss=0.2007, metrics:pearsonr:0.9061 | |
| INFO:root:04:28:18 [Batch 170/188] loss=0.2210, metrics:pearsonr:0.9048 | |
| INFO:root:04:28:18 [Batch 180/188] loss=0.2268, metrics:pearsonr:0.9036 | |
| INFO:root:04:28:18 validation metrics:pearsonr:0.9034 | |
| INFO:root:04:28:18 Time cost=2.91s, throughput=517.01 samples/s | |
| INFO:root:04:28:20 params saved in: ./output_dir/model_bert_STS-B_2.params | |
| INFO:root:04:28:20 Time cost=24.59s | |
| INFO:root:04:28:21 [Epoch 4 Batch 10/186] loss=0.0695, lr=0.0000082, metrics:pearsonr:0.9696 | |
| INFO:root:04:28:22 [Epoch 4 Batch 20/186] loss=0.0697, lr=0.0000079, metrics:pearsonr:0.9689 | |
| INFO:root:04:28:23 [Epoch 4 Batch 30/186] loss=0.0672, lr=0.0000077, metrics:pearsonr:0.9682 | |
| INFO:root:04:28:24 [Epoch 4 Batch 40/186] loss=0.0736, lr=0.0000074, metrics:pearsonr:0.9673 | |
| INFO:root:04:28:25 [Epoch 4 Batch 50/186] loss=0.0571, lr=0.0000072, metrics:pearsonr:0.9679 | |
| INFO:root:04:28:26 [Epoch 4 Batch 60/186] loss=0.0775, lr=0.0000069, metrics:pearsonr:0.9679 | |
| INFO:root:04:28:28 [Epoch 4 Batch 70/186] loss=0.0653, lr=0.0000067, metrics:pearsonr:0.9681 | |
| INFO:root:04:28:29 [Epoch 4 Batch 80/186] loss=0.0702, lr=0.0000065, metrics:pearsonr:0.9677 | |
| INFO:root:04:28:30 [Epoch 4 Batch 90/186] loss=0.0641, lr=0.0000062, metrics:pearsonr:0.9672 | |
| INFO:root:04:28:31 [Epoch 4 Batch 100/186] loss=0.0613, lr=0.0000060, metrics:pearsonr:0.9678 | |
| INFO:root:04:28:32 [Epoch 4 Batch 110/186] loss=0.0684, lr=0.0000057, metrics:pearsonr:0.9681 | |
| INFO:root:04:28:33 [Epoch 4 Batch 120/186] loss=0.0604, lr=0.0000055, metrics:pearsonr:0.9681 | |
| INFO:root:04:28:34 [Epoch 4 Batch 130/186] loss=0.0662, lr=0.0000052, metrics:pearsonr:0.9682 | |
| INFO:root:04:28:35 [Epoch 4 Batch 140/186] loss=0.0642, lr=0.0000050, metrics:pearsonr:0.9685 | |
| INFO:root:04:28:36 [Epoch 4 Batch 150/186] loss=0.0709, lr=0.0000047, metrics:pearsonr:0.9685 | |
| INFO:root:04:28:37 [Epoch 4 Batch 160/186] loss=0.0623, lr=0.0000045, metrics:pearsonr:0.9687 | |
| INFO:root:04:28:39 [Epoch 4 Batch 170/186] loss=0.0628, lr=0.0000042, metrics:pearsonr:0.9686 | |
| INFO:root:04:28:40 [Epoch 4 Batch 180/186] loss=0.0719, lr=0.0000040, metrics:pearsonr:0.9684 | |
| INFO:root:04:28:40 Now we are doing evaluation on dev with gpu(0). | |
| INFO:root:04:28:40 [Batch 10/188] loss=0.1636, metrics:pearsonr:0.9383 | |
| INFO:root:04:28:40 [Batch 20/188] loss=0.1687, metrics:pearsonr:0.9433 | |
| INFO:root:04:28:41 [Batch 30/188] loss=0.1242, metrics:pearsonr:0.9450 | |
| INFO:root:04:28:41 [Batch 40/188] loss=0.1781, metrics:pearsonr:0.9419 | |
| INFO:root:04:28:41 [Batch 50/188] loss=0.1929, metrics:pearsonr:0.9386 | |
| INFO:root:04:28:41 [Batch 60/188] loss=0.1110, metrics:pearsonr:0.9415 | |
| INFO:root:04:28:41 [Batch 70/188] loss=0.2980, metrics:pearsonr:0.9333 | |
| INFO:root:04:28:41 [Batch 80/188] loss=0.2274, metrics:pearsonr:0.9307 | |
| INFO:root:04:28:42 [Batch 90/188] loss=0.3642, metrics:pearsonr:0.9196 | |
| INFO:root:04:28:42 [Batch 100/188] loss=0.2594, metrics:pearsonr:0.9148 | |
| INFO:root:04:28:42 [Batch 110/188] loss=0.2700, metrics:pearsonr:0.9078 | |
| INFO:root:04:28:42 [Batch 120/188] loss=0.3053, metrics:pearsonr:0.9028 | |
| INFO:root:04:28:42 [Batch 130/188] loss=0.1881, metrics:pearsonr:0.9029 | |
| INFO:root:04:28:42 [Batch 140/188] loss=0.1650, metrics:pearsonr:0.9047 | |
| INFO:root:04:28:43 [Batch 150/188] loss=0.1393, metrics:pearsonr:0.9054 | |
| INFO:root:04:28:43 [Batch 160/188] loss=0.2182, metrics:pearsonr:0.9046 | |
| INFO:root:04:28:43 [Batch 170/188] loss=0.2394, metrics:pearsonr:0.9032 | |
| INFO:root:04:28:43 [Batch 180/188] loss=0.2298, metrics:pearsonr:0.9025 | |
| INFO:root:04:28:43 validation metrics:pearsonr:0.9024 | |
| INFO:root:04:28:43 Time cost=2.92s, throughput=514.64 samples/s | |
| INFO:root:04:28:44 params saved in: ./output_dir/model_bert_STS-B_3.params | |
| INFO:root:04:28:44 Time cost=24.74s | |
| INFO:root:04:28:45 [Epoch 5 Batch 10/186] loss=0.0482, lr=0.0000036, metrics:pearsonr:0.9784 | |
| INFO:root:04:28:47 [Epoch 5 Batch 20/186] loss=0.0478, lr=0.0000033, metrics:pearsonr:0.9767 | |
| INFO:root:04:28:48 [Epoch 5 Batch 30/186] loss=0.0455, lr=0.0000031, metrics:pearsonr:0.9760 | |
| INFO:root:04:28:49 [Epoch 5 Batch 40/186] loss=0.0504, lr=0.0000028, metrics:pearsonr:0.9762 | |
| INFO:root:04:28:50 [Epoch 5 Batch 50/186] loss=0.0415, lr=0.0000026, metrics:pearsonr:0.9764 | |
| INFO:root:04:28:51 [Epoch 5 Batch 60/186] loss=0.0537, lr=0.0000023, metrics:pearsonr:0.9763 | |
| INFO:root:04:28:53 [Epoch 5 Batch 70/186] loss=0.0599, lr=0.0000021, metrics:pearsonr:0.9759 | |
| INFO:root:04:28:55 [Epoch 5 Batch 80/186] loss=0.0568, lr=0.0000019, metrics:pearsonr:0.9760 | |
| INFO:root:04:28:56 [Epoch 5 Batch 90/186] loss=0.0459, lr=0.0000016, metrics:pearsonr:0.9761 | |
| INFO:root:04:28:57 [Epoch 5 Batch 100/186] loss=0.0482, lr=0.0000014, metrics:pearsonr:0.9762 | |
| INFO:root:04:28:58 [Epoch 5 Batch 110/186] loss=0.0441, lr=0.0000011, metrics:pearsonr:0.9768 | |
| INFO:root:04:28:59 [Epoch 5 Batch 120/186] loss=0.0477, lr=0.0000009, metrics:pearsonr:0.9769 | |
| INFO:root:04:29:00 [Epoch 5 Batch 130/186] loss=0.0480, lr=0.0000006, metrics:pearsonr:0.9770 | |
| INFO:root:04:29:01 [Epoch 5 Batch 140/186] loss=0.0437, lr=0.0000004, metrics:pearsonr:0.9771 | |
| INFO:root:04:29:02 [Epoch 5 Batch 150/186] loss=0.0581, lr=0.0000001, metrics:pearsonr:0.9766 | |
| INFO:root:04:29:03 Finish training step: 898 | |
| INFO:root:04:29:03 Now we are doing evaluation on dev with gpu(0). | |
| INFO:root:04:29:03 [Batch 10/188] loss=0.1655, metrics:pearsonr:0.9379 | |
| INFO:root:04:29:03 [Batch 20/188] loss=0.1691, metrics:pearsonr:0.9431 | |
| INFO:root:04:29:03 [Batch 30/188] loss=0.1231, metrics:pearsonr:0.9452 | |
| INFO:root:04:29:03 [Batch 40/188] loss=0.1781, metrics:pearsonr:0.9421 | |
| INFO:root:04:29:04 [Batch 50/188] loss=0.2032, metrics:pearsonr:0.9384 | |
| INFO:root:04:29:04 [Batch 60/188] loss=0.1151, metrics:pearsonr:0.9412 | |
| INFO:root:04:29:04 [Batch 70/188] loss=0.3041, metrics:pearsonr:0.9330 | |
| INFO:root:04:29:04 [Batch 80/188] loss=0.2338, metrics:pearsonr:0.9303 | |
| INFO:root:04:29:04 [Batch 90/188] loss=0.3693, metrics:pearsonr:0.9192 | |
| INFO:root:04:29:04 [Batch 100/188] loss=0.2605, metrics:pearsonr:0.9144 | |
| INFO:root:04:29:05 [Batch 110/188] loss=0.2713, metrics:pearsonr:0.9075 | |
| INFO:root:04:29:05 [Batch 120/188] loss=0.3024, metrics:pearsonr:0.9027 | |
| INFO:root:04:29:05 [Batch 130/188] loss=0.1898, metrics:pearsonr:0.9027 | |
| INFO:root:04:29:05 [Batch 140/188] loss=0.1653, metrics:pearsonr:0.9046 | |
| INFO:root:04:29:05 [Batch 150/188] loss=0.1373, metrics:pearsonr:0.9054 | |
| INFO:root:04:29:05 [Batch 160/188] loss=0.2164, metrics:pearsonr:0.9047 | |
| INFO:root:04:29:06 [Batch 170/188] loss=0.2402, metrics:pearsonr:0.9033 | |
| INFO:root:04:29:06 [Batch 180/188] loss=0.2323, metrics:pearsonr:0.9026 | |
| INFO:root:04:29:06 validation metrics:pearsonr:0.9023 | |
| INFO:root:04:29:06 Time cost=2.91s, throughput=516.48 samples/s | |
| INFO:root:04:29:07 params saved in: ./output_dir/model_bert_STS-B_4.params | |
| INFO:root:04:29:07 Time cost=22.60s | |
| INFO:root:04:29:07 Best model at epoch 2. Validation metrics:pearsonr:0.9034 | |
| INFO:root:04:29:07 Now we are doing testing on test with gpu(0). | |
| INFO:root:04:29:10 Time cost=2.43s, throughput=568.96 samples/s |