Morbid.AI v0.0.1 - Mortality Prediction Model

Model Description

Morbid.AI is a specialized language model fine-tuned for mortality analysis and actuarial predictions. Based on Llama-2-7b, it's trained on the World Mortality Dataset to provide insights on:

• Life expectancy calculations
• Mortality trends analysis
• Death probability estimations
• Actuarial assessments
• Country-specific mortality comparisons

Intended Use

This model is designed for:

• Actuarial analysis
• Healthcare research
• Mortality trend analysis
• Educational purposes

Note: This model should NOT be used for personal medical advice or life insurance underwriting decisions.

Training Data

Fine-tuned on:

• World Mortality Dataset (2015-2024)
• 34,537 training examples
• Countries: 200+ nations
• Mortality metrics from official statistics

Usage

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("h3ir/morbid0.0.1")
model = AutoModelForCausalLM.from_pretrained("h3ir/morbid0.0.1")

prompt = "What are the mortality trends for Japan in 2023?"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_length=200)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)

API Usage

curl https://api-inference.huggingface.co/models/h3ir/morbid0.0.1 \
  -X POST \
  -d '{"inputs": "What is the life expectancy in France?"}' \
  -H "Authorization: Bearer YOUR_TOKEN"

Model Performance

• Training Loss: 0.42
• Validation Accuracy: 87%
• Specialization: Mortality & Actuarial Data

Limitations

• Data limited to 2015-2024
• Predictions are statistical estimates
• Should not replace professional actuarial advice
• May have biases from source data

Citation

@misc{morbidai2024,
  author = {h3ir},
  title = {Morbid.AI: Mortality Prediction Model},
  year = {2024},
  publisher = {HuggingFace},
  url = {https://huggingface.co/h3ir/morbid0.0.1}
}

Contact

For questions: Visit morbid.ai