bert_emo_classifier

This model is a fine-tuned version of bert-base-uncased on the emotion dataset. It achieves the following results on the evaluation set:

• Loss: 0.3768

Target Labels

label: a classification label, with possible values including

• sadness : 0
• joy : 1
• love : 2
• anger : 3
• fear : 4
• surprise : 5

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: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 4

Training results

Training Loss	Epoch	Step	Validation Loss
0.1497	0.25	500	0.2911
0.1221	0.5	1000	0.3190
0.108	0.75	1500	0.3343
0.1296	1.0	2000	0.2803
0.0611	1.25	2500	0.3392
0.0651	1.5	3000	0.3400
0.0588	1.75	3500	0.3733
0.0993	2.0	4000	0.3672
0.0385	2.25	4500	0.4041
0.0509	2.5	5000	0.3906
0.0651	2.75	5500	0.3809
0.0693	3.0	6000	0.3944
0.0471	3.25	6500	0.3926
0.0462	3.5	7000	0.3837
0.0326	3.75	7500	0.3752
0.0233	4.0	8000	0.3768

Framework versions

• Transformers 4.15.0
• Pytorch 1.12.1+cu113
• Datasets 2.4.0
• Tokenizers 0.10.3