Thought Completion Classifier
This model classifies whether a conversation between an AI assistant and a customer represents a complete or incomplete thought.
Model Description
- Model Type: DistilBERT fine-tuned for sequence classification
- Task: Binary classification (Complete/Incomplete thought)
- Training Data: 900+ labeled conversation examples
- Input Format: "AI_Utterance [SEP] CX_Utterance"
Usage
from transformers import pipeline
# Load the model
classifier = pipeline("text-classification", model="samurai9776/thought-classifier")
# Example usage
result = classifier("What else can I get you? [SEP] That's all for now")
print(result)
# Output: [{'label': 'COMPLETE', 'score': 0.95}]
Training Details
- Base Model: distilbert-base-uncased
- Training epochs: 3
- Batch size: 16
- Learning rate: 2e-5
Limitations
- Trained on English conversations only
- Optimized for customer service/order-taking scenarios
- Downloads last month
- 8