emotion-analysis-8-categories

Emotion Analysis Model - 8 Categories

This model performs multi-label emotion classification, detecting 8 basic emotions:

• joy
• trust
• fear
• surprise
• sadness
• disgust
• anger
• anticipation

Training Data

This model was trained on a processed version of the GoEmotions dataset. The original 27 emotion categories were mapped to 8 basic emotions.

The processed dataset is available at: https://huggingface.co/datasets/bulpara/emotion-analysis-8-categories-dataset and can be loaded with:

from datasets import load_dataset
dataset = load_dataset("bulpara/emotion-analysis-8-categories-dataset")

Model Details

• Architecture: DistilBERT for Sequence Classification
• Type: Multi-label classification
• Number of labels: 8

Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

# Load model and tokenizer
model_name = "{model_name}"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name)

# Prepare text
text = "I'm so excited about this new project!"
inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=128)

# Make prediction
model.eval()
with torch.no_grad():
    outputs = model(**inputs)

# Process outputs
import numpy as np
probs = torch.sigmoid(outputs.logits).squeeze().numpy()
emotions = ['joy', 'trust', 'fear', 'surprise', 'sadness', 'disgust', 'anger', 'anticipation']

# Print results
for emotion, prob in zip(emotions, probs):
    if prob > 0.3:  # threshold can be adjusted
        print(f"{emotion}: {prob:.4f}")

Citation

@inproceedings{demszky2020goemotions,
 author = {Demszky, Dorottya and Movshovitz-Attias, Dana and Ko, Jeongwoo and Cowen, Alan and Nemade, Gaurav and Ravi, Sujith},
 booktitle = {58th Annual Meeting of the Association for Computational Linguistics (ACL)},
 title = {{GoEmotions: A Dataset of Fine-Grained Emotions}},
 year = {2020}
}