spellcorrector_910_v13_autoregressive

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

• Loss: 0.0170
• Precision: 0.9963
• Recall: 0.9957
• F1: 0.9960
• Accuracy: 0.9953

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
0.2531	1.0	1951	0.2224	0.9056	0.9748	0.9389	0.9590
0.208	2.0	3902	0.1857	0.9143	0.9734	0.9429	0.9613
0.1784	3.0	5853	0.1583	0.9213	0.9743	0.9470	0.9637
0.1604	4.0	7804	0.1380	0.9310	0.9735	0.9517	0.9662
0.1468	5.0	9755	0.1190	0.9399	0.9765	0.9579	0.9698
0.1301	6.0	11706	0.1028	0.9540	0.9744	0.9641	0.9737
0.1157	7.0	13657	0.0841	0.9610	0.9788	0.9698	0.9782
0.101	8.0	15608	0.0762	0.9702	0.9764	0.9733	0.9800
0.0892	9.0	17559	0.0621	0.9738	0.9823	0.9780	0.9835
0.0797	10.0	19510	0.0539	0.9756	0.9855	0.9805	0.9855
0.0734	11.0	21461	0.0490	0.9802	0.9848	0.9825	0.9868
0.0647	12.0	23412	0.0428	0.9826	0.9871	0.9849	0.9882
0.0606	13.0	25363	0.0400	0.9856	0.9871	0.9863	0.9890
0.0556	14.0	27314	0.0356	0.9874	0.9881	0.9877	0.9900
0.0506	15.0	29265	0.0334	0.9891	0.9896	0.9893	0.9907
0.0471	16.0	31216	0.0292	0.9902	0.9910	0.9906	0.9917
0.0442	17.0	33167	0.0270	0.9913	0.9915	0.9914	0.9923
0.0406	18.0	35118	0.0246	0.9924	0.9931	0.9928	0.9931
0.0383	19.0	37069	0.0230	0.9935	0.9935	0.9935	0.9936
0.0354	20.0	39020	0.0209	0.9944	0.9942	0.9943	0.9942
0.0337	21.0	40971	0.0204	0.9950	0.9942	0.9946	0.9943
0.0316	22.0	42922	0.0183	0.9956	0.9952	0.9954	0.9949
0.0302	23.0	44873	0.0177	0.9960	0.9954	0.9957	0.9951
0.0297	24.0	46824	0.0171	0.9961	0.9957	0.9959	0.9953
0.0284	25.0	48775	0.0170	0.9963	0.9957	0.9960	0.9953

Framework versions

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