scenario-kd-pre-ner-full_data-univner_full55

This model is a fine-tuned version of FacebookAI/xlm-roberta-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4412
• Precision: 0.8211
• Recall: 0.8091
• F1: 0.8151
• Accuracy: 0.9808

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 55
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
1.3575	0.2910	500	0.9103	0.6705	0.6653	0.6679	0.9689
0.7321	0.5821	1000	0.7344	0.7247	0.7273	0.7260	0.9738
0.6472	0.8731	1500	0.6643	0.7405	0.7642	0.7522	0.9759
0.5635	1.1641	2000	0.6244	0.7627	0.7720	0.7673	0.9775
0.4932	1.4552	2500	0.6102	0.7445	0.7855	0.7644	0.9760
0.4871	1.7462	3000	0.5773	0.7682	0.7847	0.7764	0.9778
0.4543	2.0373	3500	0.5692	0.7888	0.7834	0.7861	0.9786
0.4077	2.3283	4000	0.5501	0.7671	0.8003	0.7834	0.9785
0.3882	2.6193	4500	0.5512	0.7822	0.7831	0.7827	0.9784
0.3826	2.9104	5000	0.5284	0.7860	0.7934	0.7897	0.9789
0.3527	3.2014	5500	0.5283	0.7854	0.7984	0.7919	0.9793
0.3353	3.4924	6000	0.5180	0.7964	0.8023	0.7993	0.9794
0.3336	3.7835	6500	0.5079	0.7831	0.8042	0.7935	0.9792
0.3176	4.0745	7000	0.4999	0.7927	0.8140	0.8032	0.9798
0.2974	4.3655	7500	0.4975	0.8068	0.8044	0.8056	0.9797
0.2932	4.6566	8000	0.5007	0.7983	0.7917	0.7950	0.9792
0.291	4.9476	8500	0.5011	0.7919	0.7979	0.7949	0.9788
0.2684	5.2386	9000	0.5011	0.8014	0.8032	0.8023	0.9801
0.2636	5.5297	9500	0.4938	0.8079	0.7943	0.8010	0.9796
0.2636	5.8207	10000	0.4924	0.8067	0.8009	0.8038	0.9800
0.255	6.1118	10500	0.4796	0.7997	0.8075	0.8036	0.9804
0.2417	6.4028	11000	0.4982	0.8030	0.7990	0.8010	0.9796
0.2423	6.6938	11500	0.4827	0.7932	0.8129	0.8029	0.9797
0.2377	6.9849	12000	0.4774	0.8135	0.8080	0.8107	0.9805
0.2208	7.2759	12500	0.4759	0.8157	0.8078	0.8117	0.9809
0.2228	7.5669	13000	0.4669	0.8140	0.8139	0.8139	0.9808
0.2224	7.8580	13500	0.4762	0.8111	0.8088	0.8099	0.9806
0.2154	8.1490	14000	0.4756	0.8163	0.8085	0.8124	0.9806
0.2057	8.4400	14500	0.4751	0.8127	0.8097	0.8112	0.9805
0.2072	8.7311	15000	0.4678	0.8035	0.8146	0.8090	0.9803
0.2023	9.0221	15500	0.4678	0.8213	0.8065	0.8139	0.9805
0.1951	9.3132	16000	0.4665	0.7996	0.8096	0.8046	0.9802
0.1928	9.6042	16500	0.4695	0.8157	0.8106	0.8131	0.9805
0.1925	9.8952	17000	0.4607	0.8112	0.8127	0.8120	0.9805
0.1876	10.1863	17500	0.4573	0.8087	0.8247	0.8166	0.9811
0.1825	10.4773	18000	0.4520	0.8147	0.8293	0.8220	0.9817
0.1796	10.7683	18500	0.4566	0.8137	0.8146	0.8141	0.9807
0.1809	11.0594	19000	0.4524	0.8231	0.8137	0.8184	0.9810
0.1704	11.3504	19500	0.4593	0.8130	0.8156	0.8143	0.9809
0.1729	11.6414	20000	0.4549	0.8225	0.8075	0.8149	0.9809
0.173	11.9325	20500	0.4620	0.8166	0.8166	0.8166	0.9809
0.1656	12.2235	21000	0.4467	0.8015	0.8070	0.8042	0.9804
0.1623	12.5146	21500	0.4504	0.8139	0.8247	0.8193	0.9813
0.1651	12.8056	22000	0.4496	0.8208	0.8142	0.8175	0.9809
0.1595	13.0966	22500	0.4448	0.8141	0.8172	0.8157	0.9810
0.1561	13.3877	23000	0.4496	0.8187	0.8162	0.8174	0.9811
0.1576	13.6787	23500	0.4509	0.8198	0.8124	0.8161	0.9810
0.1563	13.9697	24000	0.4445	0.8205	0.8119	0.8162	0.9809
0.15	14.2608	24500	0.4398	0.8179	0.8152	0.8165	0.9812
0.153	14.5518	25000	0.4460	0.8281	0.8071	0.8175	0.9811
0.1482	14.8428	25500	0.4480	0.8246	0.8145	0.8195	0.9809
0.1485	15.1339	26000	0.4438	0.8199	0.8175	0.8187	0.9810
0.1449	15.4249	26500	0.4423	0.8216	0.8106	0.8160	0.9808
0.1455	15.7159	27000	0.4440	0.8181	0.8078	0.8129	0.9807
0.1438	16.0070	27500	0.4437	0.8298	0.8119	0.8207	0.9812
0.1396	16.2980	28000	0.4412	0.8211	0.8091	0.8151	0.9808

Framework versions

• Transformers 4.44.2
• Pytorch 2.1.1+cu121
• Datasets 2.14.5
• Tokenizers 0.19.1