Skills application
Unit
In ProgressCreating a model card for different audiences
In this skills application, you will practice creating documentation that serves both technical and non-technical audiences — a critical skill for long-term project success.
Context
You have just developed an ML model to classify customer support tickets (billing, technical, account access, etc.).
Model Details:
- Model: Fine-tuned BERT classifier.
- Data: 50,000 historical support tickets.
- Metrics: F1-score = 0.87, Precision = 0.85, Recall = 0.89.
- Stack: Python 3.10, PyTorch, Hugging Face Transformers.
- Goal: Faster ticket triage and improved customer satisfaction.
Success criteria
To complete this activity, you must:
- Create a model card with technical and non-technical sections.
- Write a short justification explaining the format choice.
- Describe your maintenance strategy.
Instructions and materials
Follow the instructions below and submit your answers in the form.
1. Create a model card
Draft a model card for your customer support ticket classifier.
- Technical section: Model architecture, dataset, metrics, and dependencies.
- Non-technical section: Business value, intended use, limitations, and ethical risks (e.g., bias in classification).
2. Justify choice of format
Explain why you chose the model card format and how it bridges the gap between different stakeholders.
3. Create a maintenance plan
Describe how you will ensure this documentation evolves with the model and remains a reliable source of truth.
Action item: Record your answers
Use the form below to document your model card, justification, and maintenance plan.