sembr2023-bert-mini

This model is a fine-tuned version of prajjwal1/bert-mini on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.1844
Precision: 0.7925
Recall: 0.7950
F1: 0.7938
Iou: 0.6581
Accuracy: 0.9620
Balanced Accuracy: 0.8870
Overall Accuracy: 0.9443

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: 0.0001
train_batch_size: 64
eval_batch_size: 128
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Iou	Accuracy	Balanced Accuracy	Overall Accuracy
0.5028	0.06	10	0.5036	0	0.0	0.0	0.0	0.9080	0.5	0.9080
0.417	0.12	20	0.4416	0	0.0	0.0	0.0	0.9080	0.5	0.9080
0.4292	0.18	30	0.4298	0	0.0	0.0	0.0	0.9080	0.5	0.9080
0.3964	0.24	40	0.4121	0	0.0	0.0	0.0	0.9080	0.5	0.9080
0.3611	0.3	50	0.3619	0.5833	0.0080	0.0158	0.0080	0.9082	0.5037	0.9081
0.3201	0.36	60	0.3103	0.6414	0.5085	0.5673	0.3960	0.9287	0.7399	0.9146
0.2983	0.42	70	0.2575	0.7866	0.6423	0.7072	0.5470	0.9511	0.8123	0.9333
0.2621	0.48	80	0.2402	0.7879	0.6824	0.7314	0.5765	0.9539	0.8319	0.9354
0.2464	0.55	90	0.2262	0.8304	0.6660	0.7391	0.5862	0.9568	0.8261	0.9384
0.21	0.61	100	0.2186	0.7990	0.7131	0.7536	0.6047	0.9571	0.8475	0.9382
0.2238	0.67	110	0.2042	0.8242	0.6860	0.7487	0.5984	0.9577	0.8356	0.9408
0.1849	0.73	120	0.1990	0.8799	0.6422	0.7425	0.5904	0.9590	0.8166	0.9432
0.1702	0.79	130	0.1959	0.7862	0.7436	0.7643	0.6185	0.9578	0.8615	0.9392
0.2033	0.85	140	0.1939	0.7996	0.7414	0.7694	0.6252	0.9591	0.8613	0.9404
0.1686	0.91	150	0.1916	0.7982	0.7437	0.7700	0.6260	0.9591	0.8623	0.9410
0.1484	0.97	160	0.1855	0.7898	0.7597	0.7745	0.6320	0.9593	0.8696	0.9408
0.1717	1.03	170	0.1864	0.8231	0.7301	0.7738	0.6310	0.9607	0.8571	0.9434
0.1584	1.09	180	0.1844	0.7919	0.7676	0.7796	0.6388	0.9601	0.8736	0.9414
0.1455	1.15	190	0.1807	0.8255	0.7496	0.7857	0.6470	0.9624	0.8668	0.9437
0.1521	1.21	200	0.1746	0.8277	0.7428	0.7829	0.6433	0.9621	0.8636	0.9450
0.1385	1.27	210	0.1833	0.8099	0.7573	0.7828	0.6431	0.9613	0.8697	0.9430
0.1246	1.33	220	0.1857	0.7400	0.8164	0.7763	0.6344	0.9567	0.8937	0.9366
0.1312	1.39	230	0.1738	0.8345	0.7441	0.7867	0.6484	0.9629	0.8646	0.9458
0.1139	1.45	240	0.1750	0.7964	0.7861	0.7912	0.6545	0.9618	0.8829	0.9428
0.1181	1.52	250	0.1728	0.8201	0.7578	0.7877	0.6498	0.9624	0.8705	0.9448
0.1248	1.58	260	0.1774	0.8148	0.7727	0.7932	0.6573	0.9629	0.8775	0.9441
0.1384	1.64	270	0.1748	0.7840	0.7925	0.7882	0.6505	0.9608	0.8852	0.9418
0.1068	1.7	280	0.1744	0.7943	0.7950	0.7947	0.6593	0.9622	0.8871	0.9430
0.114	1.76	290	0.1749	0.7916	0.7904	0.7910	0.6543	0.9616	0.8847	0.9428
0.1214	1.82	300	0.1778	0.7551	0.8233	0.7877	0.6498	0.9592	0.8981	0.9392
0.1139	1.88	310	0.1764	0.7897	0.7839	0.7868	0.6485	0.9609	0.8814	0.9425
0.1254	1.94	320	0.1771	0.7908	0.7925	0.7916	0.6551	0.9616	0.8856	0.9427
0.1001	2.0	330	0.1715	0.8057	0.7829	0.7942	0.6586	0.9627	0.8819	0.9445
0.0989	2.06	340	0.1705	0.8099	0.7803	0.7948	0.6595	0.9629	0.8809	0.9446
0.1222	2.12	350	0.1761	0.7843	0.7991	0.7916	0.6551	0.9613	0.8884	0.9422
0.1032	2.18	360	0.1754	0.7961	0.7864	0.7912	0.6545	0.9618	0.8830	0.9432
0.0799	2.24	370	0.1753	0.7867	0.7879	0.7873	0.6492	0.9608	0.8831	0.9432
0.099	2.3	380	0.1751	0.8101	0.7738	0.7915	0.6549	0.9625	0.8777	0.9451
0.0993	2.36	390	0.1699	0.8073	0.7791	0.7929	0.6569	0.9626	0.8801	0.9454
0.1025	2.42	400	0.1662	0.8203	0.7764	0.7978	0.6636	0.9638	0.8796	0.9465
0.1081	2.48	410	0.1762	0.8005	0.7893	0.7949	0.6596	0.9625	0.8847	0.9444
0.1118	2.55	420	0.1720	0.8130	0.7755	0.7938	0.6582	0.9630	0.8787	0.9458
0.0779	2.61	430	0.1712	0.8131	0.7797	0.7961	0.6612	0.9633	0.8808	0.9454
0.0944	2.67	440	0.1788	0.7754	0.8094	0.7921	0.6557	0.9609	0.8928	0.9419
0.1053	2.73	450	0.1696	0.7980	0.7901	0.7940	0.6584	0.9623	0.8849	0.9450
0.0889	2.79	460	0.1719	0.8215	0.7736	0.7968	0.6623	0.9637	0.8783	0.9465
0.0879	2.85	470	0.1712	0.8091	0.7828	0.7957	0.6608	0.9630	0.8820	0.9457
0.0867	2.91	480	0.1769	0.8021	0.78	0.7909	0.6541	0.9621	0.8803	0.9447
0.0787	2.97	490	0.1788	0.8044	0.7831	0.7936	0.6578	0.9625	0.8819	0.9447
0.0945	3.03	500	0.1736	0.8055	0.7820	0.7936	0.6578	0.9626	0.8815	0.9445
0.1011	3.09	510	0.1823	0.7881	0.7962	0.7921	0.6558	0.9616	0.8873	0.9432
0.0914	3.15	520	0.1819	0.7958	0.7939	0.7948	0.6595	0.9623	0.8866	0.9438
0.0837	3.21	530	0.1738	0.8129	0.7857	0.7991	0.6654	0.9637	0.8837	0.9460
0.0776	3.27	540	0.1828	0.7921	0.7961	0.7941	0.6585	0.9620	0.8874	0.9437
0.0916	3.33	550	0.1776	0.7835	0.7994	0.7913	0.6547	0.9612	0.8885	0.9433
0.081	3.39	560	0.1784	0.7784	0.8033	0.7907	0.6538	0.9609	0.8901	0.9428
0.0867	3.45	570	0.1793	0.7728	0.8074	0.7897	0.6525	0.9605	0.8917	0.9425
0.0816	3.52	580	0.1789	0.7829	0.8017	0.7922	0.6559	0.9613	0.8896	0.9433
0.0808	3.58	590	0.1791	0.7890	0.7941	0.7916	0.6550	0.9615	0.8863	0.9435
0.07	3.64	600	0.1844	0.7697	0.8071	0.7880	0.6501	0.9600	0.8913	0.9420
0.0775	3.7	610	0.1795	0.7849	0.7957	0.7902	0.6532	0.9612	0.8868	0.9433
0.0722	3.76	620	0.1772	0.7993	0.7814	0.7903	0.6532	0.9619	0.8808	0.9449
0.0786	3.82	630	0.1775	0.8159	0.7763	0.7956	0.6606	0.9633	0.8793	0.9457
0.0768	3.88	640	0.1823	0.8015	0.7848	0.7931	0.6571	0.9623	0.8826	0.9442
0.0728	3.94	650	0.1806	0.7918	0.7885	0.7901	0.6531	0.9615	0.8838	0.9438
0.0762	4.0	660	0.1831	0.7881	0.7935	0.7908	0.6540	0.9614	0.8859	0.9435
0.0776	4.06	670	0.1788	0.8015	0.7847	0.7930	0.6570	0.9623	0.8825	0.9453
0.0843	4.12	680	0.1824	0.8009	0.7876	0.7942	0.6587	0.9625	0.8839	0.9445
0.066	4.18	690	0.1843	0.7921	0.7918	0.7920	0.6556	0.9617	0.8854	0.9440
0.0832	4.24	700	0.1781	0.7957	0.7893	0.7925	0.6563	0.9620	0.8844	0.9447
0.0761	4.3	710	0.1871	0.7817	0.8017	0.7916	0.6550	0.9612	0.8895	0.9428
0.0696	4.36	720	0.1813	0.7957	0.7924	0.7940	0.6584	0.9622	0.8859	0.9446
0.0734	4.42	730	0.1827	0.7934	0.7938	0.7936	0.6578	0.9620	0.8864	0.9444
0.0823	4.48	740	0.1856	0.7956	0.7913	0.7935	0.6576	0.9621	0.8854	0.9443
0.0662	4.55	750	0.1790	0.7890	0.7952	0.7921	0.6557	0.9616	0.8868	0.9444
0.0775	4.61	760	0.1858	0.7899	0.7953	0.7926	0.6564	0.9617	0.8869	0.9439
0.0764	4.67	770	0.1853	0.7852	0.8011	0.7931	0.6572	0.9616	0.8895	0.9436
0.0689	4.73	780	0.1804	0.7964	0.7924	0.7944	0.6589	0.9623	0.8859	0.9445
0.0785	4.79	790	0.1817	0.7937	0.7921	0.7929	0.6569	0.9619	0.8856	0.9444
0.075	4.85	800	0.1856	0.7912	0.7929	0.7920	0.6556	0.9617	0.8858	0.9440
0.0691	4.91	810	0.1844	0.7805	0.8	0.7901	0.6531	0.9609	0.8886	0.9432
0.0835	4.97	820	0.1829	0.7984	0.7911	0.7947	0.6594	0.9624	0.8854	0.9448
0.0712	5.03	830	0.1820	0.7906	0.7927	0.7917	0.6552	0.9616	0.8857	0.9443
0.0594	5.09	840	0.1841	0.7902	0.7948	0.7925	0.6563	0.9617	0.8867	0.9441
0.0775	5.15	850	0.1834	0.7927	0.7936	0.7932	0.6572	0.9619	0.8863	0.9444
0.0755	5.21	860	0.1833	0.7924	0.7944	0.7934	0.6575	0.9619	0.8867	0.9444
0.0717	5.27	870	0.1838	0.7902	0.7958	0.7930	0.6570	0.9618	0.8872	0.9443
0.0694	5.33	880	0.1834	0.7918	0.7939	0.7928	0.6568	0.9618	0.8864	0.9444
0.0759	5.39	890	0.1826	0.7905	0.7954	0.7929	0.6569	0.9618	0.8870	0.9443
0.0666	5.45	900	0.1821	0.7945	0.7922	0.7933	0.6575	0.9620	0.8857	0.9446
0.08	5.52	910	0.1829	0.7924	0.7953	0.7938	0.6581	0.9620	0.8871	0.9444
0.0816	5.58	920	0.1837	0.7918	0.7954	0.7936	0.6578	0.9619	0.8871	0.9443
0.0762	5.64	930	0.1837	0.7922	0.7954	0.7938	0.6581	0.9620	0.8871	0.9443
0.0655	5.7	940	0.1843	0.7906	0.7962	0.7934	0.6575	0.9619	0.8874	0.9442
0.0737	5.76	950	0.1846	0.7904	0.7964	0.7934	0.6576	0.9619	0.8875	0.9441
0.0717	5.82	960	0.1846	0.7905	0.7961	0.7933	0.6574	0.9618	0.8873	0.9441
0.0829	5.88	970	0.1845	0.7917	0.7954	0.7935	0.6577	0.9619	0.8871	0.9443
0.0766	5.94	980	0.1844	0.7924	0.7952	0.7938	0.6581	0.9620	0.8870	0.9443
0.0704	6.0	990	0.1844	0.7925	0.7950	0.7938	0.6581	0.9620	0.8870	0.9443
0.0755	6.06	1000	0.1844	0.7925	0.7950	0.7938	0.6581	0.9620	0.8870	0.9443

Framework versions

Transformers 4.34.1
Pytorch 2.0.1
Datasets 2.14.6
Tokenizers 0.14.1

admko/sembr2023-bert-mini

sembr2023-bert-mini

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions