modelcard.md

Random Baseline Model Card

Model Description

Model Type: Random Baseline Classifier
Task: Climate Change Disinformation Classification
Version: 1.0.0
Last Updated: 2024

Overview

This is a random baseline model for climate change disinformation classification. It randomly assigns labels to text inputs, serving as a baseline for comparing more sophisticated models.

Intended Use

• Primary Use: Baseline comparison for climate disinformation classification models
• Intended Users: Researchers and developers working on climate disinformation detection
• Out-of-Scope Uses: Not intended for production or real-world classification tasks

Training Data

Dataset: QuotaClimat/frugalaichallenge-text-train

• Size: ~6000 examples
• Split: 80% train, 20% test
• Labels: 8 categories of climate disinformation claims

Labels

0. No relevant claim detected
1. Global warming is not happening
2. Not caused by humans
3. Not bad or beneficial
4. Solutions harmful/unnecessary
5. Science is unreliable
6. Proponents are biased
7. Fossil fuels are needed

Performance

Metrics

• Accuracy: ~12.5% (random chance)
• Environmental Impact:
  • Emissions (kgCO2eq)
  • Energy Consumed (kWh)

Limitations

• Random predictions with no learning
• No consideration of input text
• Serves only as a baseline reference

Ethical Considerations

• Model makes random predictions and should not be used for actual classification
• Dataset contains sensitive topics related to climate disinformation
• Environmental impact is tracked to promote awareness of AI's carbon footprint

Environmental Impact

This model tracks its environmental impact using CodeCarbon, measuring:

• Carbon emissions
• Energy consumption

Caveats and Recommendations

• Use only as a baseline comparison
• Not suitable for production use
• Consider environmental impact when running experiments