rubert-tiny2_best_finetuned_emotion_experiment_augmented_anger_fear

This model is a fine-tuned version of cointegrated/rubert-tiny2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3902
• Accuracy: 0.8727
• F1: 0.8720
• Precision: 0.8718
• Recall: 0.8727

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: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=0.0001
• lr_scheduler_type: linear
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
1.3497	1.0	69	1.2944	0.5376	0.4665	0.6374	0.5376
1.2023	2.0	138	1.0370	0.7056	0.6745	0.7458	0.7056
0.9289	3.0	207	0.7437	0.8121	0.8082	0.8117	0.8121
0.6932	4.0	276	0.5717	0.8445	0.8428	0.8434	0.8445
0.5613	5.0	345	0.4888	0.8580	0.8572	0.8573	0.8580
0.469	6.0	414	0.4401	0.8633	0.8625	0.8623	0.8633
0.4176	7.0	483	0.4156	0.8653	0.8646	0.8644	0.8653
0.3724	8.0	552	0.4001	0.8706	0.8700	0.8699	0.8706
0.3427	9.0	621	0.3972	0.8706	0.8698	0.8701	0.8706
0.3243	10.0	690	0.3898	0.8737	0.8729	0.8736	0.8737
0.3039	11.0	759	0.3887	0.8716	0.8710	0.8717	0.8716
0.2803	12.0	828	0.3841	0.8716	0.8709	0.8709	0.8716
0.264	13.0	897	0.3872	0.8758	0.8753	0.8758	0.8758
0.2607	14.0	966	0.3837	0.8747	0.8743	0.8741	0.8747
0.2437	15.0	1035	0.3893	0.8716	0.8710	0.8712	0.8716
0.2358	16.0	1104	0.3867	0.8695	0.8691	0.8690	0.8695
0.2278	17.0	1173	0.3886	0.8737	0.8732	0.8732	0.8737
0.2143	18.0	1242	0.3902	0.8727	0.8720	0.8718	0.8727

Framework versions

• Transformers 4.19.2
• Pytorch 1.11.0+cu113
• Datasets 2.2.2
• Tokenizers 0.12.1