CNEC_1_1_slavicbert

This model is a fine-tuned version of DeepPavlov/bert-base-bg-cs-pl-ru-cased on the cnec dataset. It achieves the following results on the evaluation set:

• Loss: 0.3720
• Precision: 0.8513
• Recall: 0.8671
• F1: 0.8591
• Accuracy: 0.9509

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: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.3658	0.85	1000	0.2671	0.8101	0.8172	0.8136	0.9366
0.227	1.7	2000	0.2624	0.8190	0.8172	0.8181	0.9380
0.141	2.56	3000	0.2474	0.8317	0.8424	0.8370	0.9448
0.092	3.41	4000	0.2498	0.8412	0.8534	0.8472	0.9460
0.0839	4.26	5000	0.2689	0.8438	0.8583	0.8510	0.9489
0.0698	5.11	6000	0.2830	0.8420	0.8539	0.8479	0.9473
0.0507	5.96	7000	0.2902	0.8359	0.8503	0.8431	0.9468
0.0344	6.81	8000	0.3221	0.8310	0.8512	0.8410	0.9478
0.0249	7.67	9000	0.3262	0.8444	0.8508	0.8476	0.9478
0.0185	8.52	10000	0.3214	0.8458	0.8525	0.8492	0.9502
0.0151	9.37	11000	0.3399	0.8382	0.8578	0.8479	0.9499
0.01	10.22	12000	0.3348	0.8385	0.8574	0.8478	0.9492
0.0086	11.07	13000	0.3636	0.8395	0.8543	0.8468	0.9479
0.0092	11.93	14000	0.3644	0.8419	0.8578	0.8498	0.9485
0.0058	12.78	15000	0.3624	0.8450	0.8618	0.8533	0.9503
0.0032	13.63	16000	0.3703	0.8483	0.8614	0.8548	0.9507
0.003	14.48	17000	0.3720	0.8513	0.8671	0.8591	0.9509

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.2+cu121
• Datasets 2.16.1
• Tokenizers 0.15.0