spellcorrector_810_v12

This model is a fine-tuned version of google/canine-s on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0041
• Precision: 0.9992
• Recall: 0.9990
• F1: 0.9991
• Accuracy: 0.9990

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: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.1941	1.0	1951	0.1593	0.9157	0.9720	0.9431	0.9616
0.1594	2.0	3902	0.1279	0.9313	0.9746	0.9524	0.9672
0.1334	3.0	5853	0.1078	0.9423	0.9772	0.9594	0.9713
0.1216	4.0	7804	0.0901	0.9537	0.9770	0.9652	0.9752
0.1061	5.0	9755	0.0745	0.9600	0.9804	0.9701	0.9789
0.092	6.0	11706	0.0600	0.9703	0.9826	0.9764	0.9830
0.0809	7.0	13657	0.0492	0.9755	0.9866	0.9810	0.9862
0.0671	8.0	15608	0.0449	0.9827	0.9837	0.9832	0.9874
0.062	9.0	17559	0.0365	0.9848	0.9878	0.9863	0.9896
0.0534	10.0	19510	0.0325	0.9873	0.9885	0.9879	0.9907
0.0474	11.0	21461	0.0267	0.9887	0.9918	0.9902	0.9922
0.042	12.0	23412	0.0228	0.9904	0.9932	0.9918	0.9933
0.0384	13.0	25363	0.0216	0.9929	0.9925	0.9927	0.9937
0.0338	14.0	27314	0.0201	0.9939	0.9935	0.9937	0.9943
0.0298	15.0	29265	0.0150	0.9949	0.9954	0.9951	0.9956
0.0262	16.0	31216	0.0128	0.9959	0.9961	0.9960	0.9962
0.0232	17.0	33167	0.0109	0.9970	0.9966	0.9968	0.9968
0.0222	18.0	35118	0.0090	0.9976	0.9977	0.9976	0.9974
0.0193	19.0	37069	0.0079	0.9979	0.9980	0.9980	0.9978
0.0185	20.0	39020	0.0068	0.9984	0.9982	0.9983	0.9981
0.016	21.0	40971	0.0057	0.9988	0.9985	0.9986	0.9985
0.0145	22.0	42922	0.0053	0.9989	0.9985	0.9987	0.9985
0.0136	23.0	44873	0.0045	0.9991	0.9988	0.9990	0.9988
0.0136	24.0	46824	0.0043	0.9992	0.9990	0.9991	0.9989
0.0116	25.0	48775	0.0041	0.9992	0.9990	0.9991	0.9990

Framework versions

• Transformers 4.28.0
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3