Test open-source AI models - Free deployment for classification and tagging tasks, accurate and efficient!

Test

Developed by vegetable

This model is a fine-tuned version based on hfl/chinese-bert-wwm-ext on the conll2003 dataset, designed for token classification tasks.

Sequence Labeling

Transformers

Open Source License:Apache-2.0 #Named Entity Recognition #BERT Fine-tuning #High F1 Score

Downloads 15

Release Time : 4/28/2022

Model Overview

This model is a token classification model based on the BERT architecture, primarily used for sequence labeling tasks such as named entity recognition.

Model Features

High Accuracy

Achieved an accuracy of 88.47% on the conll2003 evaluation set.

Balanced Performance

Precision (76.96%) and recall (83.96%) are well-balanced, with an F1 score reaching 80.31%.

Chinese Optimization

Fine-tuned based on the Chinese pre-trained model hfl/chinese-bert-wwm-ext.

Model Capabilities

Named Entity Recognition

Sequence Labeling

Text Token Classification

Use Cases

Natural Language Processing

Chinese Named Entity Recognition

Identify entities such as person names, locations, and organization names in Chinese text.

F1 score reaches 80.31%.

Information Extraction

Extract structured information from unstructured text.

🚀 Test

This model is a fine - tuned version of hfl/chinese-bert-wwm-ext on the conll2003 dataset. It can be used for token classification tasks and achieves good results on the evaluation set.

📚 Documentation

Model Evaluation Results

It achieves the following results on the evaluation set:

Loss: 0.7372
Precision: 0.7696
Recall: 0.8396
F1: 0.8031
Accuracy: 0.8847

Model Index

Property	Details
Model Name	test
Task	Token Classification
Dataset	conll2003
Precision	0.7696078431372549
Recall	0.839572192513369
F1	0.8030690537084398
Accuracy	0.8847040737893928

🔧 Technical Details

Training Hyperparameters

The following hyperparameters were used during training:

learning_rate: 2e-05
train_batch_size: 16
eval_batch_size: 16
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
num_epochs: 100

Training Results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	2	1.9496	0.0	0.0	0.0	0.4889
No log	2.0	4	1.6137	0.0	0.0	0.0	0.4919
No log	3.0	6	1.3906	0.0	0.0	0.0	0.5650
No log	4.0	8	1.2273	0.0652	0.0481	0.0554	0.6856
No log	5.0	10	1.0565	0.2051	0.1711	0.1866	0.7125
No log	6.0	12	0.9150	0.5094	0.4332	0.4682	0.7540
No log	7.0	14	0.8051	0.5988	0.5187	0.5559	0.7679
No log	8.0	16	0.7151	0.6707	0.5989	0.6328	0.7763
No log	9.0	18	0.6334	0.6685	0.6364	0.6521	0.8086
No log	10.0	20	0.5693	0.6957	0.6845	0.6900	0.8201
No log	11.0	22	0.5192	0.7166	0.7166	0.7166	0.8363
No log	12.0	24	0.4736	0.7135	0.7326	0.7230	0.8524
No log	13.0	26	0.4448	0.6938	0.7754	0.7323	0.8555
No log	14.0	28	0.4280	0.7177	0.8021	0.7576	0.8586
No log	15.0	30	0.4179	0.7588	0.8075	0.7824	0.8663
No log	16.0	32	0.4214	0.7356	0.8182	0.7747	0.8593
No log	17.0	34	0.4070	0.7391	0.8182	0.7766	0.8616
No log	18.0	36	0.4112	0.7586	0.8235	0.7897	0.8724
No log	19.0	38	0.4530	0.7330	0.8075	0.7684	0.8693
No log	20.0	40	0.4719	0.7766	0.8182	0.7969	0.8732
No log	21.0	42	0.4886	0.7260	0.8075	0.7646	0.8632
No log	22.0	44	0.5007	0.7217	0.8182	0.7669	0.8701
No log	23.0	46	0.5169	0.7321	0.8182	0.7727	0.8762
No log	24.0	48	0.5531	0.7238	0.8128	0.7657	0.8724
No log	25.0	50	0.5895	0.7311	0.8289	0.7769	0.8655
No log	26.0	52	0.5482	0.7330	0.8075	0.7684	0.8778
No log	27.0	54	0.5361	0.7488	0.8128	0.7795	0.8832
No log	28.0	56	0.5378	0.7427	0.8182	0.7786	0.8847
No log	29.0	58	0.5543	0.7371	0.8396	0.7850	0.8824
No log	30.0	60	0.5564	0.7585	0.8396	0.7970	0.8839
No log	31.0	62	0.5829	0.7235	0.8396	0.7772	0.8724
No log	32.0	64	0.5974	0.7269	0.8396	0.7792	0.8716
No log	33.0	66	0.5750	0.7610	0.8342	0.7959	0.8839
No log	34.0	68	0.5887	0.7723	0.8342	0.8021	0.8878
No log	35.0	70	0.6219	0.7441	0.8396	0.7889	0.8747
No log	36.0	72	0.6676	0.7269	0.8396	0.7792	0.8632
No log	37.0	74	0.6517	0.7452	0.8289	0.7848	0.8693
No log	38.0	76	0.6346	0.7828	0.8289	0.8052	0.8862
No log	39.0	78	0.6239	0.7839	0.8342	0.8083	0.8855
No log	40.0	80	0.6360	0.7277	0.8289	0.775	0.8762
No log	41.0	82	0.6645	0.7336	0.8396	0.7830	0.8701
No log	42.0	84	0.6611	0.7406	0.8396	0.7870	0.8747
No log	43.0	86	0.6707	0.7488	0.8289	0.7868	0.8762
No log	44.0	88	0.6901	0.7277	0.8289	0.775	0.8709
No log	45.0	90	0.6911	0.7393	0.8342	0.7839	0.8709
No log	46.0	92	0.6540	0.7761	0.8342	0.8041	0.8878
No log	47.0	94	0.6381	0.7761	0.8342	0.8041	0.8916
No log	48.0	96	0.6285	0.7745	0.8449	0.8082	0.8885
No log	49.0	98	0.6449	0.7692	0.8556	0.8101	0.8862
No log	50.0	100	0.6809	0.7442	0.8556	0.7960	0.8732
No log	51.0	102	0.6898	0.7395	0.8503	0.7910	0.8716
No log	52.0	104	0.6897	0.75	0.8503	0.7970	0.8762
No log	53.0	106	0.6714	0.7656	0.8556	0.8081	0.8855
No log	54.0	108	0.6612	0.7692	0.8556	0.8101	0.8855
No log	55.0	110	0.6583	0.7692	0.8556	0.8101	0.8855
No log	56.0	112	0.6648	0.7692	0.8556	0.8101	0.8855
No log	57.0	114	0.6757	0.7656	0.8556	0.8081	0.8832
No log	58.0	116	0.6803	0.7656	0.8556	0.8081	0.8839
No log	59.0	118	0.6834	0.7692	0.8556	0.8101	0.8862
No log	60.0	120	0.6889	0.7833	0.8503	0.8154	0.8878
No log	61.0	122	0.6963	0.7772	0.8396	0.8072	0.8862
No log	62.0	124	0.7057	0.7772	0.8396	0.8072	0.8862
No log	63.0	126	0.7212	0.7910	0.8503	0.8196	0.8862
No log	64.0	128	0.7334	0.7833	0.8503	0.8154	0.8824
No log	65.0	130	0.7398	0.7833	0.8503	0.8154	0.8801
No log	66.0	132	0.7400	0.7833	0.8503	0.8154	0.8809
No log	67.0	134	0.7345	0.7783	0.8449	0.8103	0.8855
No log	68.0	136	0.7270	0.79	0.8449	0.8165	0.8870
No log	69.0	138	0.7245	0.7839	0.8342	0.8083	0.8862
No log	70.0	140	0.7260	0.7868	0.8289	0.8073	0.8847
No log	71.0	142	0.7275	0.7817	0.8235	0.8021	0.8839
No log	72.0	144	0.7283	0.7778	0.8235	0.8000	0.8832
No log	73.0	146	0.7296	0.78	0.8342	0.8062	0.8847
No log	74.0	148	0.7344	0.7734	0.8396	0.8051	0.8832
No log	75.0	150	0.7314	0.7745	0.8449	0.8082	0.8824
No log	76.0	152	0.7299	0.7794	0.8503	0.8133	0.8832
No log	77.0	154	0.7282	0.7794	0.8503	0.8133	0.8839
No log	78.0	156	0.7252	0.7783	0.8449	0.8103	0.8839
No log	79.0	158	0.7216	0.7756	0.8503	0.8112	0.8855
No log	80.0	160	0.7194	0.7756	0.8503	0.8112	0.8870
No log	81.0	162	0.7191	0.7756	0.8503	0.8112	0.8878
No log	82.0	164	0.7201	0.7696	0.8396	0.8031	0.8862
No log	83.0	166	0.7211	0.7696	0.8396	0.8031	0.8862
No log	84.0	168	0.7222	0.7696	0.8396	0.8031	0.8862
No log	85.0	170	0.7220	0.7696	0.8396	0.8031	0.8862
No log	86.0	172	0.7239	0.7734	0.8396	0.8051	0.8870
No log	87.0	174	0.7291	0.7772	0.8396	0.8072	0.8847
No log	88.0	176	0.7344	0.7745	0.8449	0.8082	0.8824
No log	89.0	178	0.7373	0.7745	0.8449	0.8082	0.8824
No log	90.0	180	0.7391	0.7707	0.8449	0.8061	0.8832
No log	91.0	182	0.7403	0.7745	0.8449	0.8082	0.8824
No log	92.0	184	0.7412	0.7745	0.8449	0.8082	0.8832
No log	93.0	186	0.7417	0.7707	0.8449	0.8061	0.8832
No log	94.0	188	0.7402	0.7745	0.8449	0.8082	0.8839
No log	95.0	190	0.7389	0.7745	0.8449	0.8082	0.8847
No log	96.0	192	0.7381	0.7696	0.8396	0.8031	0.8839
No log	97.0	194	0.7377	0.7696	0.8396	0.8031	0.8847
No log	98.0	196	0.7374	0.7696	0.8396	0.8031	0.8847
No log	99.0	198	0.7372	0.7696	0.8396	0.8031	0.8847
No log	100.0	200	0.7372	0.7696	0.8396	0.8031	0.8847

Framework Versions

Transformers 4.18.0
Pytorch 1.11.0+cu113
Datasets 2.1.0
Tokenizers 0.12.1

📄 License

This project is licensed under the Apache - 2.0 license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご