|
|
--- |
|
|
license: apache-2.0 |
|
|
language: |
|
|
- en |
|
|
library_name: transformers |
|
|
tags: |
|
|
- mortality |
|
|
- actuary |
|
|
- healthcare |
|
|
- llama |
|
|
- text-generation |
|
|
datasets: |
|
|
- world-mortality |
|
|
widget: |
|
|
- text: "What is the life expectancy in United States for 2024?" |
|
|
--- |
|
|
|
|
|
# Morbid.AI v0.0.4 - 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 |
|
|
|
|
|
```python |
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM |
|
|
|
|
|
tokenizer = AutoTokenizer.from_pretrained("h3ir/morbid0.0.4") |
|
|
model = AutoModelForCausalLM.from_pretrained("h3ir/morbid0.0.4") |
|
|
|
|
|
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 |
|
|
|
|
|
```bash |
|
|
curl https://api-inference.huggingface.co/models/h3ir/morbid0.0.4 \ |
|
|
-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 |
|
|
|
|
|
```bibtex |
|
|
@misc{morbidai2024, |
|
|
author = {h3ir}, |
|
|
title = {Morbid.AI: Mortality Prediction Model}, |
|
|
year = {2024}, |
|
|
publisher = {HuggingFace}, |
|
|
url = {https://huggingface.co/h3ir/morbid0.0.4} |
|
|
} |
|
|
``` |
|
|
|
|
|
## Contact |
|
|
|
|
|
For questions: Visit [morbid.ai](https://morbid.ai) |
