bert_bilstm_crf-ner-weibo

This model is a fine-tuned version of google-bert/bert-base-chinese on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1945
• Precision: 0.6524
• Recall: 0.7429
• F1: 0.6947
• Accuracy: 0.9703

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.4272	1.0	22	0.3531	0.0	0.0	0.0	0.9330
0.2529	2.0	44	0.1587	0.4922	0.4884	0.4903	0.9613
0.1472	3.0	66	0.1171	0.5524	0.6915	0.6142	0.9681
0.0977	4.0	88	0.1057	0.5866	0.6967	0.6369	0.9714
0.065	5.0	110	0.1035	0.6336	0.7069	0.6683	0.9715
0.0538	6.0	132	0.1149	0.6307	0.7069	0.6667	0.9699
0.0413	7.0	154	0.1057	0.6315	0.7224	0.6739	0.9724
0.0344	8.0	176	0.1236	0.5979	0.7455	0.6636	0.9693
0.0296	9.0	198	0.1271	0.5958	0.7352	0.6582	0.9680
0.0297	10.0	220	0.1257	0.6442	0.6889	0.6658	0.9702
0.0212	11.0	242	0.1440	0.6037	0.7481	0.6682	0.9664
0.0208	12.0	264	0.1368	0.6284	0.7044	0.6642	0.9683
0.0165	13.0	286	0.1337	0.6545	0.7404	0.6948	0.9698
0.0164	14.0	308	0.1388	0.6514	0.7301	0.6885	0.9700
0.014	15.0	330	0.1403	0.6690	0.7275	0.6970	0.9701
0.0109	16.0	352	0.1467	0.6448	0.7326	0.6859	0.9694
0.0108	17.0	374	0.1488	0.6081	0.7301	0.6636	0.9670
0.0106	18.0	396	0.1564	0.6572	0.7147	0.6847	0.9687
0.0105	19.0	418	0.1620	0.6667	0.7147	0.6898	0.9691
0.01	20.0	440	0.1638	0.7046	0.6684	0.6860	0.9705
0.0106	21.0	462	0.1542	0.6709	0.6761	0.6735	0.9692
0.0092	22.0	484	0.1487	0.6683	0.7198	0.6931	0.9694
0.011	23.0	506	0.1502	0.6396	0.7301	0.6819	0.9691
0.0068	24.0	528	0.1534	0.6801	0.7378	0.7078	0.9705
0.0077	25.0	550	0.1600	0.6793	0.7352	0.7062	0.9710
0.0071	26.0	572	0.1644	0.6386	0.7404	0.6857	0.9676
0.0062	27.0	594	0.1714	0.6430	0.7224	0.6804	0.9688
0.006	28.0	616	0.1649	0.6461	0.7275	0.6844	0.9694
0.0072	29.0	638	0.1631	0.6643	0.7326	0.6968	0.9695
0.0122	30.0	660	0.1802	0.6054	0.7455	0.6682	0.9676
0.0062	31.0	682	0.1829	0.6154	0.7404	0.6721	0.9676
0.0075	32.0	704	0.1674	0.6313	0.7352	0.6793	0.9691
0.0048	33.0	726	0.1664	0.6422	0.7429	0.6889	0.9692
0.0045	34.0	748	0.1724	0.6374	0.7455	0.6872	0.9697
0.0055	35.0	770	0.1714	0.6636	0.7301	0.6952	0.9700
0.0071	36.0	792	0.1673	0.6316	0.7404	0.6817	0.9692
0.0039	37.0	814	0.1635	0.6620	0.7352	0.6967	0.9709
0.0036	38.0	836	0.1727	0.6584	0.7532	0.7026	0.9710
0.0051	39.0	858	0.1735	0.6509	0.7429	0.6939	0.9708
0.0033	40.0	880	0.1758	0.6949	0.7378	0.7157	0.9718
0.0045	41.0	902	0.1812	0.6309	0.7558	0.6877	0.9698
0.0035	42.0	924	0.1791	0.6729	0.7404	0.7050	0.9709
0.0043	43.0	946	0.1923	0.6532	0.7455	0.6963	0.9697
0.0045	44.0	968	0.1815	0.6492	0.7326	0.6884	0.9696
0.0037	45.0	990	0.1830	0.6493	0.7378	0.6907	0.9700
0.0045	46.0	1012	0.1809	0.6493	0.7378	0.6907	0.9700
0.0039	47.0	1034	0.1811	0.6545	0.7404	0.6948	0.9701
0.0046	48.0	1056	0.1740	0.6659	0.7172	0.6906	0.9708
0.0039	49.0	1078	0.1827	0.6318	0.7455	0.6840	0.9694
0.0036	50.0	1100	0.1762	0.6443	0.7404	0.6890	0.9698
0.0046	51.0	1122	0.1752	0.6538	0.7378	0.6932	0.9702
0.0036	52.0	1144	0.1856	0.6344	0.7404	0.6833	0.9692
0.0036	53.0	1166	0.1870	0.6350	0.7378	0.6825	0.9693
0.0049	54.0	1188	0.1840	0.6723	0.7121	0.6916	0.9699
0.0042	55.0	1210	0.1927	0.6220	0.7404	0.6761	0.9687
0.0039	56.0	1232	0.1854	0.6545	0.7352	0.6925	0.9704
0.0042	57.0	1254	0.1900	0.6523	0.7378	0.6924	0.9700
0.0028	58.0	1276	0.1894	0.6486	0.7404	0.6915	0.9697
0.0049	59.0	1298	0.1904	0.6366	0.7429	0.6856	0.9695
0.0031	60.0	1320	0.1844	0.6492	0.7326	0.6884	0.9698
0.0045	61.0	1342	0.1866	0.6429	0.7404	0.6882	0.9696
0.004	62.0	1364	0.1888	0.625	0.7326	0.6746	0.9686
0.0031	63.0	1386	0.1922	0.6875	0.7352	0.7106	0.9710
0.0044	64.0	1408	0.1918	0.6722	0.7326	0.7011	0.9706
0.0046	65.0	1430	0.1987	0.6475	0.7506	0.6952	0.9685
0.0044	66.0	1452	0.1868	0.6388	0.7455	0.6880	0.9698
0.0042	67.0	1474	0.1920	0.6356	0.7532	0.6894	0.9695
0.0038	68.0	1496	0.1852	0.6606	0.7506	0.7028	0.9705
0.0033	69.0	1518	0.1843	0.6476	0.7558	0.6975	0.9700
0.0034	70.0	1540	0.1797	0.6532	0.7506	0.6986	0.9707
0.0042	71.0	1562	0.1820	0.6332	0.7455	0.6848	0.9699
0.0033	72.0	1584	0.1874	0.6482	0.7532	0.6968	0.9704
0.0039	73.0	1606	0.1878	0.6636	0.7506	0.7045	0.9708
0.003	74.0	1628	0.1857	0.6553	0.7429	0.6964	0.9712
0.0038	75.0	1650	0.1889	0.6606	0.7404	0.6982	0.9709
0.004	76.0	1672	0.1880	0.6539	0.7481	0.6978	0.9709
0.0032	77.0	1694	0.1875	0.6590	0.7404	0.6973	0.9706
0.0034	78.0	1716	0.1868	0.6532	0.7455	0.6963	0.9710
0.0029	79.0	1738	0.1899	0.6545	0.7404	0.6948	0.9705
0.0032	80.0	1760	0.1899	0.6628	0.7429	0.7006	0.9709
0.0037	81.0	1782	0.1928	0.6545	0.7404	0.6948	0.9705
0.0039	82.0	1804	0.1916	0.6560	0.7404	0.6957	0.9705
0.0034	83.0	1826	0.1926	0.6560	0.7352	0.6933	0.9705
0.0032	84.0	1848	0.1931	0.6621	0.7455	0.7013	0.9709
0.0048	85.0	1870	0.1925	0.6659	0.7481	0.7046	0.9712
0.0039	86.0	1892	0.1903	0.6690	0.7326	0.6994	0.9709
0.0039	87.0	1914	0.1948	0.6538	0.7429	0.6955	0.9709
0.0032	88.0	1936	0.1949	0.6682	0.7558	0.7093	0.9710
0.003	89.0	1958	0.1948	0.6697	0.7609	0.7124	0.9710
0.0027	90.0	1980	0.1927	0.6489	0.7506	0.6961	0.9705
0.0029	91.0	2002	0.1931	0.6496	0.7481	0.6953	0.9706
0.003	92.0	2024	0.1932	0.6532	0.7455	0.6963	0.9712
0.0029	93.0	2046	0.1928	0.6539	0.7481	0.6978	0.9712
0.0036	94.0	2068	0.1935	0.6503	0.7506	0.6969	0.9710
0.0034	95.0	2090	0.1941	0.6607	0.7558	0.7050	0.9714
0.0035	96.0	2112	0.1940	0.6621	0.7455	0.7013	0.9711
0.0028	97.0	2134	0.1940	0.6553	0.7429	0.6964	0.9707
0.0032	98.0	2156	0.1944	0.6509	0.7429	0.6939	0.9704
0.0028	99.0	2178	0.1943	0.6509	0.7429	0.6939	0.9705
0.0021	100.0	2200	0.1945	0.6524	0.7429	0.6947	0.9703

Framework versions

• Transformers 4.46.1
• Pytorch 1.13.1+cu117
• Datasets 3.1.0
• Tokenizers 0.20.2