rubert-base-cased-1-third

This model is a fine-tuned version of DeepPavlov/rubert-base-cased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.2935
• Accuracy: 0.919

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: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.6641	1.0	1500	0.3723	0.9029
0.285	2.0	3000	0.3000	0.9154
0.1981	3.0	4500	0.2935	0.919
0.1488	4.0	6000	0.3073	0.9194
0.1139	5.0	7500	0.3401	0.9177
0.0902	6.0	9000	0.3662	0.9166
0.077	7.0	10500	0.3955	0.9175
0.0633	8.0	12000	0.4064	0.916
0.0548	9.0	13500	0.4286	0.9173
0.0487	10.0	15000	0.4429	0.916
0.0405	11.0	16500	0.4777	0.9195
0.0367	12.0	18000	0.4836	0.9202
0.0314	13.0	19500	0.4854	0.9194
0.0271	14.0	21000	0.5018	0.9175
0.023	15.0	22500	0.5123	0.9191

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Datasets 2.16.0
• Tokenizers 0.15.0