|
--- |
|
language: en |
|
tags: |
|
- emotion-classification |
|
datasets: |
|
- go-emotions |
|
- bdotloh/empathetic-dialogues-contexts |
|
--- |
|
|
|
|
|
# Model Description |
|
|
|
We performed transfer learning experiments on a distilbert-base-uncased model fine-tuned on the GoEmotions dataset for the purpose of classifying [(emotional) contexts in the Empathetic Dialogues dataset](https://huggingface.co/datasets/bdotloh/empathetic-dialogues-contexts). |
|
|
|
The fine-tuned distilbert-base-uncased can be found [here](https://huggingface.co/bhadresh-savani/bert-base-go-emotion). |
|
|
|
### Limitations and bias |
|
|
|
GoEmotions: |
|
1) Demographics of Reddit Users |
|
2) Imbalanced class distribution |
|
3) ... |
|
|
|
EmpatheticDialogues: |
|
1) Unable to ascertain the degree of cultural specificity for the context that a respondent described when given an emotion label (i.e., p(description | emotion, *culture*)) |
|
2) ... |
|
|
|
|
|
## Training data |
|
|
|
## Training procedure |
|
|
|
### Preprocessing |
|
|
|
## Evaluation results |
|
|
|
|
