| Model Type | | large language model, finetuned, LoRA |
|
| Use Cases |
| Areas: | | mental health analysis, research |
|
| Applications: | | mental health condition analysis, explanation generation |
|
| Primary Use Cases: | | Interpretable mental health analysis with instruction-following |
|
| Limitations: | | Not for clinical use, Potential biases and incorrectness |
|
| Considerations: | | Use results for non-clinical research; ensure professional oversight in clinical settings. |
|
|
| Additional Notes | | MentaLLaMA is part of a series of models designed for mental health analysis. |
|
| Supported Languages | |
| Training Details |
| Data Sources: | | IMHI instruction tuning data |
|
| Data Volume: | | 75K high-quality natural language instructions |
|
| Methodology: | |
|
| Safety Evaluation |
| Methodologies: | | comprehensive evaluation on the IMHI benchmark |
|
| Findings: | | approaches state-of-the-art discriminative methods in correctness, generates high-quality explanations |
|
| Risk Categories: | | potential bias, gender gaps, incorrect predictions, inappropriate explanations |
|
| Ethical Considerations: | | Results should be used for non-clinical research. Professional assistance is recommended for clinical contexts. There are challenges in real-scenario applications. |
|
|
| Responsible Ai Considerations |
| Fairness: | | Potential gender gap bias addressed. |
|
|
| Input Output |
| Input Format: | | Natural language instructions |
|
| Accepted Modalities: | |
| Output Format: | | Mental health analyses and explanations |
|
|
