deberta-v3-base-finetuned-ner

sohamtiwari3120/deberta-v3-base-finetuned-ner

deberta-v3-base-finetuned-ner is a token classification model from sohamtiwari3120. Use it when you need labels on individual words, such as names. It is set up for transformers. The card lists the license as mit.

should probably proofread and complete it, then remove this comment. --

Downloads · 30 days

1% of all-time downloads

All-time downloads

777

Public

Repo size

48.5 GB

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.bin735 MB · 99%

At a glance

Task: Token Classification
Library: transformers
License: mit
Model type: deberta-v2
Access: Public
Created: Oct 24, 2022
Updated: Oct 31, 2022
SHA: e7deaf0b

Try a prompt

Datasets

generator

Model card

From the Hugging Face model README

deberta-v3-base-finetuned-ner

This model is a fine-tuned version of microsoft/deberta-v3-base on the generator dataset. It achieves the following results on the evaluation set:

Loss: 0.7679
Overall Precision: 0.4915
Overall Recall: 0.6463
Overall F1: 0.5584
Overall Accuracy: 0.9555
Datasetname F1: 0.3304
Hyperparametername F1: 0.6341
Hyperparametervalue F1: 0.7463
Methodname F1: 0.6093
Metricname F1: 0.7089
Metricvalue F1: 0.7500
Taskname F1: 0.4426

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: 8
eval_batch_size: 8
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	Overall Precision	Overall Recall	Overall F1	Overall Accuracy	Datasetname F1	Hyperparametername F1	Hyperparametervalue F1	Methodname F1	Metricname F1	Metricvalue F1	Taskname F1
No log	1.0	132	0.5046	0.2771	0.5041	0.3576	0.9356	0.2405	0.1988	0.4545	0.4638	0.4539	0.6486	0.2793
No log	2.0	264	0.3928	0.3344	0.6463	0.4407	0.9376	0.2449	0.3968	0.6292	0.5641	0.5373	0.4583	0.3359
No log	3.0	396	0.4714	0.4419	0.6179	0.5153	0.9533	0.3822	0.5310	0.7536	0.6262	0.6328	0.6857	0.3291
0.5663	4.0	528	0.3741	0.4493	0.7114	0.5507	0.9509	0.4717	0.7241	0.6353	0.5918	0.5714	0.6275	0.4372
0.5663	5.0	660	0.4202	0.3930	0.6870	0.5	0.9458	0.2759	0.6525	0.65	0.5596	0.7097	0.7368	0.3573
0.5663	6.0	792	0.4676	0.4244	0.6850	0.5241	0.9473	0.3333	0.5949	0.7397	0.5653	0.6988	0.7568	0.3652
0.5663	7.0	924	0.5744	0.4328	0.5955	0.5013	0.9517	0.2585	0.6167	0.5915	0.5825	0.6386	0.7500	0.3824
0.1503	8.0	1056	0.5340	0.4309	0.6585	0.5209	0.9499	0.2976	0.6299	0.7105	0.6140	0.6708	0.7568	0.3544
0.1503	9.0	1188	0.5229	0.4628	0.6829	0.5517	0.9531	0.4630	0.5103	0.6087	0.625	0.6541	0.7778	0.4493
0.1503	10.0	1320	0.6287	0.4978	0.6748	0.5729	0.9563	0.4314	0.6500	0.7463	0.6413	0.7432	0.7568	0.4108
0.1503	11.0	1452	0.5163	0.4571	0.7033	0.5540	0.9519	0.3925	0.5256	0.6024	0.6828	0.6626	0.7368	0.4466
0.0735	12.0	1584	0.6737	0.5046	0.6687	0.5752	0.9555	0.3883	0.6615	0.6757	0.6074	0.7051	0.7778	0.4577
0.0735	13.0	1716	0.5849	0.44	0.6931	0.5383	0.9480	0.3770	0.6555	0.6479	0.5922	0.6957	0.6512	0.4071
0.0735	14.0	1848	0.8314	0.5018	0.5793	0.5377	0.9539	0.3	0.6549	0.6667	0.5613	0.7361	0.7368	0.4294
0.0735	15.0	1980	0.5986	0.4549	0.6768	0.5441	0.9506	0.3793	0.6000	0.6667	0.6181	0.7089	0.6829	0.3978
0.0408	16.0	2112	0.7579	0.4900	0.6443	0.5566	0.9541	0.4103	0.6032	0.6765	0.6238	0.7123	0.6667	0.4217
0.0408	17.0	2244	0.9175	0.5285	0.6037	0.5636	0.9565	0.4	0.6789	0.7692	0.5949	0.7101	0.6857	0.4122
0.0408	18.0	2376	0.7771	0.5041	0.6179	0.5553	0.9562	0.3684	0.6207	0.7246	0.5842	0.7383	0.6667	0.4353
0.0226	19.0	2508	0.7992	0.5213	0.6463	0.5771	0.9569	0.32	0.6724	0.7353	0.6485	0.7114	0.7179	0.4510
0.0226	20.0	2640	0.7679	0.4915	0.6463	0.5584	0.9555	0.3304	0.6341	0.7463	0.6093	0.7089	0.7500	0.4426

Framework versions

Transformers 4.23.1
Pytorch 1.12.1+cu102
Datasets 2.6.1
Tokenizers 0.13.1

Task: Token Classification
Library: transformers
Type: deberta-v2
License: mit
Created: Oct 24, 2022
Updated: Oct 31, 2022