de-ru-lid

This model is a fine-tuned version of papluca/xlm-roberta-base-language-detection on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5382
• Precision: 0.9088
• Recall: 0.7660
• F1: 0.8313
• Accuracy: 0.9458

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 12

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.1176	1.0	510	0.0314	0.9819	0.9690	0.9754	0.9941
0.0159	2.0	1020	0.0134	0.9935	0.9854	0.9894	0.9976
0.0074	3.0	1530	0.0134	0.9932	0.9882	0.9907	0.9978
0.0045	4.0	2040	0.0097	0.9953	0.9913	0.9933	0.9985
0.004	5.0	2550	0.0076	0.9954	0.9941	0.9948	0.9988
0.0022	6.0	3060	0.0088	0.9942	0.9920	0.9931	0.9985
0.0013	7.0	3570	0.0089	0.9969	0.9924	0.9946	0.9987
0.0009	8.0	4080	0.0096	0.9949	0.9936	0.9942	0.9986
0.0009	9.0	4590	0.0084	0.9973	0.9943	0.9958	0.9990
0.0005	10.0	5100	0.0076	0.9965	0.9938	0.9951	0.9990
0.0001	11.0	5610	0.0082	0.9965	0.9932	0.9949	0.9990
0.0003	12.0	6120	0.0080	0.9966	0.9939	0.9952	0.9990

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.0+cu118
• Datasets 2.12.0
• Tokenizers 0.13.3