distilbert-base-uncased-fine-tuned-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: 0.2156
• Accuracy: 0.9255
• F1: 0.9254
• Recall: 0.9255

Model description

This is the resuls of fine-tuning a distilbert-base-uncased trained on a NVIDIA GeForce GTX 1650, using a WSL with 7 gb of ram on windows 11.

The fine-tuning was obtained by following the book Natural Language Processing with Tranformers: Building Languaje Applications with Hugging Fabe, By Lewis Tunstall, Leandro von Werra & Thomas Wolf

Labels are associated to:

1. LABEL_0 is sadness
2. LABEL_1 is joy
3. LABEL_2 is love
4. LABEL_3 is anger
5. LABEL_4 is fear
6. LABEL_5 is surprise

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: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Recall
0.7838	1.0	250	0.2995	0.906	0.9039	0.906
0.237	2.0	500	0.2156	0.9255	0.9254	0.9255

Framework versions

• Transformers 4.30.2
• Pytorch 1.13.1+cu117
• Datasets 2.13.2
• Tokenizers 0.12.1