distilbert-base-uncased-finetuned-emotion
This model is a fine-tuned version of distilbert-base-uncased on the emotion dataset. It achieves the following results on the evaluation set:
- Loss: 0.5342
- Accuracy: 0.885
- Balanced accuracy: 0.8457
- F1: 0.8861
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=1e-08
- lr_scheduler_type: linear
- num_epochs: 20
Training results
Training Loss | Epoch | Step | Validation Loss | Accuracy | Balanced accuracy | F1 |
---|---|---|---|---|---|---|
1.6372 | 1.0 | 25 | 1.4302 | 0.53 | 0.2661 | 0.4247 |
1.3152 | 2.0 | 50 | 1.1864 | 0.57 | 0.2892 | 0.4648 |
1.0588 | 3.0 | 75 | 1.0524 | 0.605 | 0.3390 | 0.5390 |
0.8495 | 4.0 | 100 | 0.8517 | 0.76 | 0.5691 | 0.7315 |
0.6198 | 5.0 | 125 | 0.6699 | 0.79 | 0.6073 | 0.7671 |
0.4309 | 6.0 | 150 | 0.5773 | 0.835 | 0.7656 | 0.8382 |
0.2887 | 7.0 | 175 | 0.5278 | 0.84 | 0.7435 | 0.8391 |
0.203 | 8.0 | 200 | 0.4942 | 0.865 | 0.8268 | 0.8669 |
0.1459 | 9.0 | 225 | 0.4451 | 0.885 | 0.8189 | 0.8847 |
0.1053 | 10.0 | 250 | 0.4940 | 0.865 | 0.7809 | 0.8641 |
0.0786 | 11.0 | 275 | 0.5234 | 0.865 | 0.7746 | 0.8629 |
0.0659 | 12.0 | 300 | 0.5266 | 0.86 | 0.7944 | 0.8601 |
0.0591 | 13.0 | 325 | 0.5427 | 0.845 | 0.7628 | 0.8461 |
0.0456 | 14.0 | 350 | 0.5309 | 0.86 | 0.8072 | 0.8620 |
0.0352 | 15.0 | 375 | 0.5377 | 0.87 | 0.8119 | 0.8711 |
0.032 | 16.0 | 400 | 0.5320 | 0.87 | 0.7908 | 0.8690 |
0.0274 | 17.0 | 425 | 0.5240 | 0.87 | 0.8119 | 0.8698 |
0.0247 | 18.0 | 450 | 0.5326 | 0.88 | 0.8429 | 0.8812 |
0.0231 | 19.0 | 475 | 0.5309 | 0.88 | 0.8384 | 0.8802 |
0.0227 | 20.0 | 500 | 0.5342 | 0.885 | 0.8457 | 0.8861 |
Framework versions
- Transformers 4.38.1
- Pytorch 2.1.0+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2
- Downloads last month
- 5
Finetuned from
Dataset used to train carver63/distilbert-base-uncased-finetuned-emotion
Evaluation results
- Accuracy on emotionself-reported0.885
- F1 on emotionself-reported0.886