ernie-gram-zh-ner

This model is a fine-tuned version of nghuyong/ernie-gram-zh on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0309
• Overall Precision: 0.8935
• Overall Recall: 0.9325
• Overall F1: 0.9126
• Overall Accuracy: 0.9909
• Ucm: 0.9085
• Lcm: 0.8989

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: 1e-05
• train_batch_size: 512
• eval_batch_size: 512
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Overall Precision	Overall Recall	Overall F1	Overall Accuracy	Ucm	Lcm
1.8192	0.08	10	1.7428	0.0007	0.0036	0.0011	0.7089	0.0299	0.0299
0.6082	0.16	20	0.5770	0.0	0.0	0.0	0.8993	0.4668	0.4668
0.4604	0.24	30	0.4031	0.0171	0.0039	0.0063	0.9029	0.4668	0.4659
0.3272	0.32	40	0.2515	0.1391	0.1987	0.1636	0.9310	0.5393	0.5281
0.2219	0.4	50	0.1789	0.3743	0.5158	0.4338	0.9517	0.6002	0.5885
0.1959	0.48	60	0.1358	0.4535	0.5968	0.5154	0.9651	0.6606	0.6470
0.1474	0.56	70	0.1071	0.5905	0.7118	0.6455	0.9713	0.7258	0.7092
0.13	0.64	80	0.0892	0.6481	0.7664	0.7023	0.9748	0.7560	0.7400
0.1014	0.72	90	0.0764	0.6914	0.7869	0.7361	0.9775	0.7811	0.7654
0.0987	0.8	100	0.0674	0.7185	0.8150	0.7637	0.9802	0.7889	0.7745
0.0944	0.88	110	0.0587	0.7703	0.8486	0.8075	0.9827	0.8077	0.7920
0.0851	0.96	120	0.0560	0.7857	0.8675	0.8245	0.9834	0.8143	0.7986
0.0857	1.04	130	0.0527	0.8021	0.8854	0.8417	0.9850	0.8315	0.8167
0.0883	1.12	140	0.0480	0.8207	0.8867	0.8524	0.9863	0.8545	0.8397
0.0666	1.2	150	0.0478	0.8207	0.8946	0.8561	0.9862	0.8481	0.8342
0.0881	1.28	160	0.0452	0.8321	0.8968	0.8632	0.9869	0.8542	0.8406
0.0734	1.3600	170	0.0437	0.8337	0.8961	0.8638	0.9872	0.8596	0.8454
0.0734	1.44	180	0.0412	0.8412	0.8961	0.8678	0.9877	0.8626	0.8499
0.0721	1.52	190	0.0407	0.8398	0.9	0.8688	0.9879	0.8620	0.8475
0.0888	1.6	200	0.0398	0.8488	0.9043	0.8757	0.9880	0.8656	0.8524
0.0643	1.6800	210	0.0397	0.8503	0.9058	0.8771	0.9883	0.8690	0.8548
0.077	1.76	220	0.0379	0.8576	0.9051	0.8807	0.9885	0.8804	0.8662
0.0645	1.8400	230	0.0381	0.8554	0.9079	0.8809	0.9885	0.8762	0.8623
0.056	1.92	240	0.0368	0.8615	0.9099	0.8850	0.9890	0.8819	0.8684
0.0612	2.0	250	0.0372	0.8571	0.9118	0.8836	0.9890	0.8819	0.8675
0.054	2.08	260	0.0362	0.8723	0.9141	0.8927	0.9893	0.8877	0.8747
0.0584	2.16	270	0.0355	0.8773	0.9171	0.8968	0.9896	0.8922	0.8786
0.0581	2.24	280	0.0357	0.8713	0.9158	0.8930	0.9894	0.8907	0.8774
0.0541	2.32	290	0.0352	0.8804	0.9171	0.8984	0.9897	0.8946	0.8807
0.0507	2.4	300	0.0351	0.8757	0.9176	0.8962	0.9896	0.8937	0.8801
0.0574	2.48	310	0.0351	0.8803	0.9197	0.8996	0.9897	0.8949	0.8813
0.0528	2.56	320	0.0352	0.8771	0.9203	0.8982	0.9897	0.8907	0.8771
0.0488	2.64	330	0.0345	0.8804	0.9218	0.9006	0.9898	0.8934	0.8804
0.0582	2.7200	340	0.0335	0.8802	0.9218	0.9005	0.9901	0.8979	0.8844
0.063	2.8	350	0.0332	0.8788	0.9238	0.9007	0.9902	0.8976	0.8841
0.059	2.88	360	0.0328	0.8778	0.9242	0.9004	0.9902	0.8967	0.8853
0.052	2.96	370	0.0319	0.8837	0.9238	0.9033	0.9905	0.9025	0.8895
0.0687	3.04	380	0.0328	0.8832	0.9263	0.9043	0.9901	0.8986	0.8865
0.0536	3.12	390	0.0323	0.8848	0.9261	0.9050	0.9904	0.9019	0.8904
0.0608	3.2	400	0.0324	0.8860	0.9287	0.9068	0.9904	0.9004	0.8898
0.0538	3.2800	410	0.0329	0.8885	0.9302	0.9089	0.9903	0.8982	0.8889
0.0524	3.36	420	0.0328	0.8915	0.9304	0.9105	0.9903	0.9010	0.8913
0.0483	3.44	430	0.0322	0.8889	0.9300	0.9090	0.9904	0.9019	0.8919
0.0615	3.52	440	0.0318	0.8849	0.9300	0.9069	0.9906	0.9034	0.8925
0.0583	3.6	450	0.0329	0.8833	0.9317	0.9068	0.9901	0.8992	0.8877
0.0495	3.68	460	0.0317	0.8876	0.9285	0.9076	0.9905	0.9034	0.8925
0.0539	3.76	470	0.0320	0.8853	0.9289	0.9066	0.9904	0.9031	0.8916
0.0522	3.84	480	0.0312	0.8892	0.9278	0.9081	0.9907	0.9067	0.8943
0.0457	3.92	490	0.0313	0.8905	0.9296	0.9096	0.9906	0.9055	0.8940
0.0532	4.0	500	0.0305	0.8971	0.9293	0.9129	0.9911	0.9082	0.8986
0.0514	4.08	510	0.0307	0.8919	0.9298	0.9105	0.9909	0.9067	0.8967
0.0571	4.16	520	0.0310	0.8897	0.9306	0.9097	0.9908	0.9073	0.8961
0.0491	4.24	530	0.0317	0.8846	0.9308	0.9071	0.9905	0.9031	0.8922
0.0512	4.32	540	0.0309	0.8921	0.9317	0.9115	0.9909	0.9082	0.8976
0.0484	4.4	550	0.0313	0.8918	0.9317	0.9113	0.9907	0.9070	0.8973
0.0458	4.48	560	0.0315	0.8913	0.9321	0.9112	0.9907	0.9064	0.8967
0.052	4.5600	570	0.0314	0.8917	0.9323	0.9115	0.9908	0.9088	0.8992
0.0511	4.64	580	0.0310	0.8931	0.9323	0.9123	0.9908	0.9088	0.8992
0.049	4.72	590	0.0307	0.8942	0.9319	0.9127	0.9909	0.9088	0.8992
0.0432	4.8	600	0.0307	0.8946	0.9323	0.9131	0.9909	0.9094	0.8998
0.0489	4.88	610	0.0308	0.8929	0.9323	0.9122	0.9909	0.9082	0.8982
0.0544	4.96	620	0.0309	0.8935	0.9325	0.9126	0.9909	0.9085	0.8989

Framework versions

• Transformers 4.50.0.dev0
• Pytorch 2.6.0+cu126
• Datasets 3.4.1
• Tokenizers 0.21.1