bdotloh's picture
Create README.md
286c714
|
raw
history blame
918 Bytes
metadata
language: en
tags:
  - emotion-classification
datasets:
  - bdotloh/empathetic-dialogues-contexts

Model Description

This is a distilbert-base-uncased model fine-tuned for the purpose of classifying (emotional) contexts in the Empathetic Dialogues dataset.

The fine-tuned distilbert-base-uncased can be found here.

Limitations and bias

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

see dataset

Training procedure

Preprocessing

Evaluation results

Test results

Top 1 accuracy: 53.4 Top 5 accuracy: 86.1