emotion-bert / README.md
hrrbb's picture
Create README.md
8b6e13e verified
|
Raw History Blame Contribute Delete
1.38 kB
metadata
license: apache-2.0
language:
  - en
tags:
  - text-classification
  - emotion-classification
  - bert
  - transformers
base_model: bert-base-uncased
pipeline_tag: text-classification
library_name: transformers

Emotion BERT – Islamic Guidance Emotion Classifier

A fine-tuned BERT model for detecting emotions in English text, built as part of RUHI — an AI-powered Islamic guidance platform that recommends Quranic verses, Hadith, and Duas based on a user's emotional state.

Emotion Classes

Label Description
neutral No strong emotion
sadness Sorrow or unhappiness
fear_anxiety Fear, stress, or nervousness
happiness_gratitude Joy, thankfulness, contentment
anger Irritation or rage
regret_repentance Guilt or seeking forgiveness

Model Details

  • Base model: bert-base-uncased
  • Task: Multi-label emotion classification (6 labels)
  • Framework: Hugging Face Transformers (PyTorch)
  • Input: English text (max 128 tokens)
  • Output: Probability score for each emotion

Usage

from transformers import pipeline

classifier = pipeline(
    "text-classification",
    model="hrrbb/emotion-bert",
    top_k=None,
    function_to_apply="sigmoid"
)

text = "I feel so lost and overwhelmed."
for r in classifier(text):
    print(f"{r['label']}: {r['score']:.4f}")