spellcorrector_710_v8

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: 0.0144
• Precision: 0.9961
• Recall: 0.9962
• F1: 0.9962
• Accuracy: 0.9957

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
0.2237	1.0	1951	0.1844	0.9061	0.9700	0.9370	0.9557
0.1882	2.0	3902	0.1577	0.9144	0.9719	0.9423	0.9598
0.1669	3.0	5853	0.1389	0.9311	0.9689	0.9497	0.9633
0.154	4.0	7804	0.1234	0.9343	0.9751	0.9543	0.9669
0.141	5.0	9755	0.1076	0.9480	0.9734	0.9605	0.9711
0.1286	6.0	11706	0.0959	0.9584	0.9746	0.9664	0.9747
0.1131	7.0	13657	0.0799	0.9624	0.9792	0.9708	0.9780
0.1016	8.0	15608	0.0714	0.9696	0.9801	0.9748	0.9805
0.0915	9.0	17559	0.0627	0.9737	0.9821	0.9779	0.9825
0.0839	10.0	19510	0.0574	0.9781	0.9830	0.9806	0.9839
0.0761	11.0	21461	0.0500	0.9808	0.9849	0.9828	0.9858
0.069	12.0	23412	0.0437	0.9807	0.9887	0.9847	0.9873
0.0644	13.0	25363	0.0404	0.9849	0.9882	0.9866	0.9882
0.057	14.0	27314	0.0371	0.9871	0.9892	0.9881	0.9892
0.0555	15.0	29265	0.0343	0.9890	0.9895	0.9893	0.9900
0.0512	16.0	31216	0.0288	0.9899	0.9919	0.9909	0.9914
0.0464	17.0	33167	0.0265	0.9914	0.9922	0.9918	0.9920
0.0424	18.0	35118	0.0234	0.9924	0.9937	0.9931	0.9929
0.0391	19.0	37069	0.0215	0.9940	0.9936	0.9938	0.9936
0.0375	20.0	39020	0.0195	0.9944	0.9944	0.9944	0.9942
0.0344	21.0	40971	0.0178	0.9952	0.9949	0.9950	0.9947
0.032	22.0	42922	0.0160	0.9955	0.9957	0.9956	0.9952
0.0299	23.0	44873	0.0153	0.9957	0.9960	0.9958	0.9954
0.0291	24.0	46824	0.0145	0.9961	0.9961	0.9961	0.9957
0.0289	25.0	48775	0.0144	0.9961	0.9962	0.9962	0.9957

Framework versions

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