spellcorrector_20_02_050_qwerty_v11

This model is a fine-tuned version of Buseak/spellcorrector_20_02_050_qwerty_v7 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0001
• Precision: 0.9989
• Recall: 0.9984
• F1: 0.9987
• Accuracy: 1.0000

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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.0558	1.0	967	0.0281	0.9919	0.9909	0.9914	0.9917
0.0391	2.0	1934	0.0200	0.9941	0.9930	0.9936	0.9940
0.0295	3.0	2901	0.0141	0.9952	0.9941	0.9946	0.9957
0.0239	4.0	3868	0.0107	0.9957	0.9952	0.9954	0.9968
0.0201	5.0	4835	0.0075	0.9957	0.9962	0.9960	0.9977
0.0174	6.0	5802	0.0054	0.9984	0.9968	0.9976	0.9984
0.0138	7.0	6769	0.0042	0.9979	0.9973	0.9976	0.9988
0.0117	8.0	7736	0.0028	0.9984	0.9979	0.9981	0.9992
0.0098	9.0	8703	0.0021	0.9979	0.9979	0.9979	0.9994
0.0081	10.0	9670	0.0014	0.9984	0.9984	0.9984	0.9996
0.0073	11.0	10637	0.0011	0.9989	0.9984	0.9987	0.9997
0.006	12.0	11604	0.0009	0.9989	0.9984	0.9987	0.9998
0.0056	13.0	12571	0.0006	0.9989	0.9984	0.9987	0.9999
0.0045	14.0	13538	0.0004	0.9989	0.9984	0.9987	0.9999
0.0039	15.0	14505	0.0003	0.9989	0.9984	0.9987	0.9999
0.0033	16.0	15472	0.0003	0.9989	0.9984	0.9987	1.0000
0.0029	17.0	16439	0.0002	0.9989	0.9984	0.9987	1.0000
0.0026	18.0	17406	0.0002	0.9989	0.9984	0.9987	1.0000
0.0023	19.0	18373	0.0002	0.9989	0.9984	0.9987	1.0000
0.0021	20.0	19340	0.0001	0.9989	0.9984	0.9987	1.0000

Framework versions

• Transformers 4.37.2
• Pytorch 2.1.0+cu121
• Datasets 2.17.1
• Tokenizers 0.15.2