distilrubert-tiny-cased-conversational-v1_finetuned_emotion_experiment_augmented_anger_fear

This model is a fine-tuned version of DeepPavlov/distilrubert-tiny-cased-conversational-v1 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3760
• Accuracy: 0.8758
• F1: 0.8750
• Precision: 0.8753
• Recall: 0.8758

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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
1.2636	1.0	69	1.0914	0.6013	0.5599	0.5780	0.6013
1.029	2.0	138	0.9180	0.6514	0.6344	0.6356	0.6514
0.904	3.0	207	0.8235	0.6827	0.6588	0.6904	0.6827
0.8084	4.0	276	0.7272	0.7537	0.7477	0.7564	0.7537
0.7242	5.0	345	0.6435	0.7860	0.7841	0.7861	0.7860
0.6305	6.0	414	0.5543	0.8173	0.8156	0.8200	0.8173
0.562	7.0	483	0.4860	0.8392	0.8383	0.8411	0.8392
0.5042	8.0	552	0.4474	0.8528	0.8514	0.8546	0.8528
0.4535	9.0	621	0.4213	0.8580	0.8579	0.8590	0.8580
0.4338	10.0	690	0.4106	0.8591	0.8578	0.8605	0.8591
0.4026	11.0	759	0.4064	0.8622	0.8615	0.8632	0.8622
0.3861	12.0	828	0.3874	0.8737	0.8728	0.8733	0.8737
0.3709	13.0	897	0.3841	0.8706	0.8696	0.8701	0.8706
0.3592	14.0	966	0.3841	0.8716	0.8709	0.8714	0.8716
0.3475	15.0	1035	0.3834	0.8737	0.8728	0.8732	0.8737
0.3537	16.0	1104	0.3805	0.8727	0.8717	0.8722	0.8727
0.3317	17.0	1173	0.3775	0.8747	0.8739	0.8741	0.8747
0.323	18.0	1242	0.3759	0.8727	0.8718	0.8721	0.8727
0.3327	19.0	1311	0.3776	0.8758	0.8750	0.8756	0.8758
0.3339	20.0	1380	0.3760	0.8758	0.8750	0.8753	0.8758

Framework versions

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