RoBERTa-base-finetuned-emotion

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

• Loss: 0.1629
• Accuracy: 0.933
• Precision: 0.8945
• Recall: 0.9002
• F1: 0.8968

Model description

This is a RoBERTa model fine-tuned on the emotion to determine whether a text is within any of the six categories: 'sadness', 'joy', 'love', 'anger', 'fear', 'surprise'. The Trainer API was used to train the model.

Intended uses & limitations

Training and evaluation data

🤗 load_dataset package was used to load the data from the hub.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.5693	1.0	500	0.2305	0.9215	0.8814	0.8854	0.8818
0.1946	2.0	1000	0.1923	0.9235	0.8698	0.9268	0.8899
0.1297	3.0	1500	0.1514	0.933	0.9060	0.8879	0.8913
0.1041	4.0	2000	0.1545	0.9265	0.9165	0.8567	0.8789
0.0826	5.0	2500	0.1629	0.933	0.8945	0.9002	0.8968

Framework versions

• Transformers 4.33.0
• Pytorch 2.0.0
• Datasets 2.1.0
• Tokenizers 0.13.3