|
- Unparsed args: ['--lstm_layer', '2']
- fix_gaz_emb : True
- fix_entity_emb : True
- contrast : False
- number_layer : 1
- bilstm_flag : True
- gat_nhidden : 300
- gat_nhead : 4
- gat_layer : 2
- pre_model : BERT
- ner_encode_type : lstm
- strategy : n
- model_type : lstm
- entity : True
- use_bert : True
- attention : True
- tri_fuse : False
- alpha : 0.1
- dropout : 0.6
- droplstm : 0.0
- dropbert : 0.1
- dropgat : 0.0
- gaz_dropout : 0.2
- kg_dropout : 0.2
- use_lstm : False
- dataset_name : meizhou_all
- train_file : data/train_ch.txt
- test_file : data/train_ch.txt
- dev_file : data/train_ch.txt
- gaz_file : None
- kg_file : None
- char_embedding_path : None
- data_stored_directory : data/generated_data/
- param_stored_directory : data/model_param/
- entity_type_embdding_path : None
- language : CH
- norm_char_emb : True
- norm_gaz_emb : True
- norm_kg_emb : False
- number_normalized : True
- max_sentence_length : 200
- batch_size : 80
- max_epoch : 8
- lr : 2e-05
- lr_other : 0.0001
- lr_decay : 0.005
- use_clip : True
- clip : 5.0
- gradient_accumulation_steps : 1
- optimizer : Adam
- l2_penalty : 5e-08
- refresh : False
- use_gpu : True
- visible_gpu : 0
- random_seed : 2022
- Data setting loaded from file: data/generated_data/meizhou_all_dataset.dset
- DATA SUMMARY START:
- Dataset name: meizhou_all
- Tag scheme: BIO
- Max Sentence Length: 200
- Char alphabet size: 2066
- Gaz alphabet size: 2
- Label alphabet size: 18
- Char embedding size: 200
- Gaz embedding size: 100
- Number normalized: True
- Norm char emb: True
- Norm gaz emb: True
- Train instance number: 31991
- Dev instance number: 31991
- Test instance number: 31991
- Train cut number: 0
- Dev cut number: 0
- Test cut number: 0
- DATA SUMMARY END.
- build BLSTM_GCN_CRF model...
- using model lstm......
- Epoch: 0/8
- Learning rate is setted as: 2e-05
- Instance: 2000; Time: 223.97s; loss: 219854.5942
- Instance: 4000; Time: 139.18s; loss: 44202.5361
- Instance: 6000; Time: 143.15s; loss: 18476.4727
- Instance: 8000; Time: 141.03s; loss: 10638.6191
- Instance: 10000; Time: 140.94s; loss: 9246.6426
- Instance: 12000; Time: 144.74s; loss: 9196.0605
- Instance: 14000; Time: 140.51s; loss: 7530.8477
- Instance: 16000; Time: 135.69s; loss: 7965.3477
- Instance: 18000; Time: 139.25s; loss: 6334.6465
- Instance: 20000; Time: 141.28s; loss: 7122.3301
- Instance: 22000; Time: 141.31s; loss: 6345.6621
- Instance: 24000; Time: 146.90s; loss: 6569.0664
- Instance: 26000; Time: 147.21s; loss: 6463.4922
- Instance: 28000; Time: 144.98s; loss: 6626.8770
- Instance: 30000; Time: 145.67s; loss: 7052.2539
- Instance: 31991; Time: 144.84s; loss: 7291.8633
- Epoch: 0 training finished. Time: 2360.66s, speed: 13.55st/s, total loss: 380917.31201171875
- $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
- 31991
- gold_num = 53249 pred_num = 54898 right_num = 49039
- Dev: time: 1480.37s, speed: 21.62st/s; acc: 0.9815, p: 0.8933, r: 0.9209, f: 0.9069
- Exceed previous best f score: -1
- Epoch: 1/8
- Learning rate is setted as: 1.9900000000000003e-05
- Instance: 2000; Time: 144.74s; loss: 5149.1094
- Instance: 4000; Time: 141.63s; loss: 5394.8320
- Instance: 6000; Time: 144.42s; loss: 5766.7930
- Instance: 8000; Time: 148.32s; loss: 5077.2871
- Instance: 10000; Time: 142.46s; loss: 5404.7305
- Instance: 12000; Time: 147.97s; loss: 5512.8984
- Instance: 14000; Time: 148.85s; loss: 4779.3047
- Instance: 16000; Time: 147.25s; loss: 5765.6523
- Instance: 18000; Time: 148.51s; loss: 5586.5625
- Instance: 20000; Time: 148.87s; loss: 4654.5117
- Instance: 22000; Time: 145.09s; loss: 4991.3438
- Instance: 24000; Time: 143.83s; loss: 5233.4297
- Instance: 26000; Time: 148.26s; loss: 5393.7305
- Instance: 28000; Time: 148.56s; loss: 5434.0195
- Instance: 30000; Time: 147.24s; loss: 5670.1055
- Instance: 31991; Time: 148.50s; loss: 5148.2930
- Epoch: 1 training finished. Time: 2344.49s, speed: 13.65st/s, total loss: 84962.603515625
- $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
- 31991
- gold_num = 53249 pred_num = 54823 right_num = 50019
- Dev: time: 1489.68s, speed: 21.49st/s; acc: 0.9845, p: 0.9124, r: 0.9393, f: 0.9257
- Exceed previous best f score: 0.9068952444358143
- Epoch: 2/8
- Learning rate is setted as: 1.98005e-05
- Instance: 2000; Time: 142.77s; loss: 4367.0938
- Instance: 4000; Time: 145.81s; loss: 4394.8594
- Instance: 6000; Time: 142.55s; loss: 4540.9922
- Instance: 8000; Time: 141.22s; loss: 4349.7461
- Instance: 10000; Time: 144.03s; loss: 4182.9062
- Instance: 12000; Time: 147.38s; loss: 4337.2070
- Instance: 14000; Time: 144.41s; loss: 4526.6016
- Instance: 16000; Time: 142.63s; loss: 4411.9258
- Instance: 18000; Time: 143.46s; loss: 4336.0898
- Instance: 20000; Time: 143.08s; loss: 4275.6484
- Instance: 22000; Time: 145.35s; loss: 4509.1406
- Instance: 24000; Time: 138.87s; loss: 4273.8359
- Instance: 26000; Time: 141.82s; loss: 4229.5117
- Instance: 28000; Time: 146.36s; loss: 4766.3398
- Instance: 30000; Time: 147.27s; loss: 4407.0625
- Instance: 31991; Time: 146.17s; loss: 4306.4297
- Epoch: 2 training finished. Time: 2303.17s, speed: 13.89st/s, total loss: 70215.390625
- $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
- 31991
- gold_num = 53249 pred_num = 53905 right_num = 50214
- Dev: time: 1488.64s, speed: 21.50st/s; acc: 0.9869, p: 0.9315, r: 0.9430, f: 0.9372
- Exceed previous best f score: 0.9256606706640018
- Epoch: 3/8
- Learning rate is setted as: 1.97014975e-05
- Instance: 2000; Time: 141.09s; loss: 3715.4141
- Instance: 4000; Time: 147.64s; loss: 3744.5117
- Instance: 6000; Time: 142.11s; loss: 3397.3711
- Instance: 8000; Time: 143.11s; loss: 3984.7227
- Instance: 10000; Time: 144.99s; loss: 3288.6406
- Instance: 12000; Time: 140.59s; loss: 3897.9102
- Instance: 14000; Time: 146.98s; loss: 3624.1758
- Instance: 16000; Time: 145.83s; loss: 3926.2266
- Instance: 18000; Time: 141.96s; loss: 3943.3164
- Instance: 20000; Time: 141.86s; loss: 4099.5000
- Instance: 22000; Time: 144.40s; loss: 4224.2461
- Instance: 24000; Time: 141.80s; loss: 3501.4180
- Instance: 26000; Time: 140.42s; loss: 3819.6914
- Instance: 28000; Time: 144.38s; loss: 3805.3555
- Instance: 30000; Time: 140.94s; loss: 3446.1484
- Instance: 31991; Time: 145.87s; loss: 3878.2109
- Epoch: 3 training finished. Time: 2293.95s, speed: 13.95st/s, total loss: 60296.859375
- $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
- 31991
- gold_num = 53249 pred_num = 54277 right_num = 50578
- Dev: time: 1479.52s, speed: 21.64st/s; acc: 0.9885, p: 0.9318, r: 0.9498, f: 0.9408
- Exceed previous best f score: 0.9372305280250853
- Epoch: 4/8
- Learning rate is setted as: 1.9602990012500002e-05
- Instance: 2000; Time: 139.89s; loss: 3108.8242
- Instance: 4000; Time: 146.49s; loss: 3294.8398
- Instance: 6000; Time: 143.74s; loss: 3436.0820
- Instance: 8000; Time: 141.32s; loss: 3013.4766
- Instance: 10000; Time: 142.61s; loss: 3615.6914
- Instance: 12000; Time: 144.47s; loss: 3732.9062
- Instance: 14000; Time: 141.40s; loss: 2980.2188
- Instance: 16000; Time: 145.92s; loss: 3266.6445
- Instance: 18000; Time: 146.21s; loss: 3288.1836
- Instance: 20000; Time: 143.83s; loss: 3352.2734
- Instance: 22000; Time: 140.19s; loss: 3133.9375
- Instance: 24000; Time: 144.07s; loss: 3662.1523
- Instance: 26000; Time: 144.87s; loss: 3482.4414
- Instance: 28000; Time: 146.02s; loss: 3809.6445
- Instance: 30000; Time: 143.78s; loss: 3549.6602
- Instance: 31991; Time: 139.61s; loss: 2872.3125
- Epoch: 4 training finished. Time: 2294.41s, speed: 13.94st/s, total loss: 53599.2890625
- $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
- 31991
- gold_num = 53249 pred_num = 54250 right_num = 50840
- Dev: time: 1503.94s, speed: 21.29st/s; acc: 0.9894, p: 0.9371, r: 0.9548, f: 0.9459
- Exceed previous best f score: 0.9407585142198166
- Epoch: 5/8
- Learning rate is setted as: 1.95049750624375e-05
- Instance: 2000; Time: 145.28s; loss: 3062.4570
- Instance: 4000; Time: 144.18s; loss: 3031.4102
- Instance: 6000; Time: 146.71s; loss: 2906.7383
- Instance: 8000; Time: 142.81s; loss: 3244.6172
- Instance: 10000; Time: 145.30s; loss: 3178.4180
- Instance: 12000; Time: 144.44s; loss: 2904.4375
- Instance: 14000; Time: 143.65s; loss: 3298.3008
- Instance: 16000; Time: 144.08s; loss: 3086.2148
- Instance: 18000; Time: 143.79s; loss: 2656.9727
- Instance: 20000; Time: 145.80s; loss: 3174.3438
- Instance: 22000; Time: 141.75s; loss: 2635.3594
- Instance: 24000; Time: 140.58s; loss: 2874.7500
- Instance: 26000; Time: 146.70s; loss: 2918.7656
- Instance: 28000; Time: 142.53s; loss: 3079.8438
- Instance: 30000; Time: 142.09s; loss: 3762.9688
- Instance: 31991; Time: 143.87s; loss: 2811.3711
- Epoch: 5 training finished. Time: 2303.55s, speed: 13.89st/s, total loss: 48626.96875
- $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
- 31991
- gold_num = 53249 pred_num = 53699 right_num = 50860
- Dev: time: 1480.31s, speed: 21.63st/s; acc: 0.9904, p: 0.9471, r: 0.9551, f: 0.9511
- Exceed previous best f score: 0.9458692639001293
- Epoch: 6/8
- Learning rate is setted as: 1.9407450187125313e-05
- Instance: 2000; Time: 146.27s; loss: 3014.0117
- Instance: 4000; Time: 143.83s; loss: 2758.7891
- Instance: 6000; Time: 145.18s; loss: 2538.3398
- Instance: 8000; Time: 145.04s; loss: 2666.7031
- Instance: 10000; Time: 147.32s; loss: 2806.5469
- Instance: 12000; Time: 145.32s; loss: 2754.9102
- Instance: 14000; Time: 139.50s; loss: 2835.4648
- Instance: 16000; Time: 142.03s; loss: 3431.2578
- Instance: 18000; Time: 137.83s; loss: 2918.4219
- Instance: 20000; Time: 145.35s; loss: 2635.7695
- Instance: 22000; Time: 143.30s; loss: 3393.3398
- Instance: 24000; Time: 140.34s; loss: 2397.8008
- Instance: 26000; Time: 143.66s; loss: 2561.0977
- Instance: 28000; Time: 146.64s; loss: 3002.3477
- Instance: 30000; Time: 144.50s; loss: 2559.7539
- Instance: 31991; Time: 145.85s; loss: 2569.6055
- Epoch: 6 training finished. Time: 2301.95s, speed: 13.90st/s, total loss: 44844.16015625
- $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
- 31991
- gold_num = 53249 pred_num = 54304 right_num = 51167
- Dev: time: 1486.95s, speed: 21.53st/s; acc: 0.9907, p: 0.9422, r: 0.9609, f: 0.9515
- Exceed previous best f score: 0.9511164304147811
- Epoch: 7/8
- Learning rate is setted as: 1.9310412936189687e-05
- Instance: 2000; Time: 145.41s; loss: 2525.0898
- Instance: 4000; Time: 144.43s; loss: 2175.6758
- Instance: 6000; Time: 144.55s; loss: 2585.5625
- Instance: 8000; Time: 144.31s; loss: 2615.1367
- Instance: 10000; Time: 146.44s; loss: 2616.1523
- Instance: 12000; Time: 142.69s; loss: 2946.4766
- Instance: 14000; Time: 143.01s; loss: 2693.7109
- Instance: 16000; Time: 141.71s; loss: 2813.1797
- Instance: 18000; Time: 144.83s; loss: 3049.9219
- Instance: 20000; Time: 145.74s; loss: 2467.6250
- Instance: 22000; Time: 147.38s; loss: 2947.2930
- Instance: 24000; Time: 145.95s; loss: 2329.8594
- Instance: 26000; Time: 147.08s; loss: 2424.7578
- Instance: 28000; Time: 145.71s; loss: 2715.2617
- Instance: 30000; Time: 143.31s; loss: 2122.9375
- Instance: 31991; Time: 146.41s; loss: 2702.8047
- Epoch: 7 training finished. Time: 2318.94s, speed: 13.80st/s, total loss: 41731.4453125
- $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
- 31991
- gold_num = 53249 pred_num = 54450 right_num = 51269
- Dev: time: 1491.82s, speed: 21.46st/s; acc: 0.9908, p: 0.9416, r: 0.9628, f: 0.9521
- Exceed previous best f score: 0.9514750867014401
|