Fine-grained text emotions analysis from a hierarchical perspective. The hierarchical label stucture is applied in our new disigned loss function.
Dataset labeled 58000 Reddit comments with 28 emotions
- admiration, amusement, anger, annoyance, approval, caring, confusion, curiosity, desire, disappointment, disapproval, disgust, embarrassment, excitement, fear, gratitude, grief, joy, love, nervousness, optimism, pride, realization, relief, remorse, sadness, surprise + neutral
- torch==1.4.0
- transformers==2.11.0
- attrdict==2.0.1
You can change the parameters from the json files in config
directory.
Parameter | |
---|---|
Learning rate | 5e-5 |
Warmup proportion | 0.1 |
Epochs | 10 |
Max Seq Length | 50 |
Batch size | 16 |