README.md

---

datasets:
- yirenc/4_ethics_all
language:
- en
metrics:
- accuracy
- f1
pipeline_tag: text-generation
---

# Fine-tuned Llama-3-8B on AI Ethics

<!-- Provide a quick summary of what the model is/does. -->

This is a fine-tuned version of Meta-Llama-3-8B model on AI Ethics dataset. The usage of this model should comply with the Llama 3 Community License and the Acceptable Use Policy.


## Evaluation

<!-- This section describes the evaluation protocols and provides the results. -->


### Original model: Meta-Llama-3-8B

| Category | F1 Score | Accuracy | Correct Formatting |
|----------|----------|----------|--------------------|
| Deontology | 0.258 | 0.180 | 0.340 |
| Justice | 0.100 | 0.060 | 0.100 |
| Utilitarianism | 0.123 | 0.070 | 0.150 |
| Virtue | 0.057 | 0.030 | 0.040 |

### Baseline model: Meta-Llama-3-8B on 4_ethics_100

| Category | F1 Score | Accuracy | Correct Formatting |
|----------|----------|----------|--------------------|
| Deontology | 0.428 | 0.490 | 1.000 |
| Justice | 0.366 | 0.520 | 1.000 |
| Utilitarianism | 0.449 | 0.540 | 1.000 |
| Virtue | 0.476 | 0.600 | 1.000 |

### Fine-tuned model: Meta-Llama-3-8B on 4_ethics_all

| Category | F1 Score | Accuracy | Correct Formatting |
|----------|----------|----------|--------------------|
| Deontology | 0.930 | 0.930 | 1.000 |
| Justice | 0.960 | 0.960 | 1.000 |
| Utilitarianism | 0.750 | 0.750 | 1.000 |
| Virtue | 0.891 | 0.890 | 1.000 |


## Model Card Contact

yiren.cao@swisscom.com

yiren.cao@epfl.ch

yiren.cao@outlook.com