emotion-detector

Emotion Detection Model

---

license: mit tags:

• emotion-detection
• bert
• text-classification
• transformers
• huggingface
• crowdflower datasets:
• crowdflower language: en widget:
• text: "I am feeling loved and joyful today!"

---

🧠 Emotion Detection Model (BERT-based)

This is a BERT-based emotion classification model fine-tuned on the Crowdflower Emotion Dataset. The model can classify English text into one of six emotion categories:

• 😢 sadness
• 😀 joy
• ❤️ love
• 😠 anger
• 😨 fear
• 😲 surprise

---

📌 Use Cases

This model is best suited for:

• Emotion analysis in text messages
• Chatbot mood understanding
• Social media sentiment/emotion classification
• Customer feedback analysis

---

🧪 Model Details

---

💬 Example

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

model_id = "sickboi25/emotion-detector"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForSequenceClassification.from_pretrained(model_id)

text = "I am so happy and full of love!"
inputs = tokenizer(text, return_tensors="pt")
outputs = model(**inputs)
predicted_class = torch.argmax(outputs.logits).item()

emotion_labels = ['sadness', 'joy', 'love', 'anger', 'fear', 'surprise']
print("Predicted Emotion:", emotion_labels[predicted_class])
📂 Dataset
Name: Crowdflower Emotion Dataset

Size: ~40,000 labeled English tweets

Labels: sadness, joy, love, anger, fear, surprise

Link: Crowdflower on Kaggle

🧠 Performance
Metric	Score
Accuracy	~90%
Loss	Low after 3 epochs
F1-Score	High on all classes

📜 License
This model is released under the MIT License.

👤 Author
Name: Ahsaan Ullah

Hugging Face Profile: @sickboi25

🔗 Model Page: https://huggingface.co/sickboi25/emotion-detector