---
base_model:
- beomi/Llama-3-Open-Ko-8B
- ProbeMedicalYonseiMAILab/medllama3-v20
library_name: transformers
tags:
- mergekit
- merge
license: apache-2.0
---
# BioLlama-Ko-8B


![image/png](https://cdn-uploads.huggingface.co/production/uploads/64c61e724399efa2fdac0375/9zF_PWSgjxRtWI-3dtwDC.png)

This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).

## 🏆 Evaluation

### [kormedmcqa(한국어 의학 벤치마크)](https://huggingface.co/datasets/sean0042/KorMedMCQA)

| Model                                    | Doctor | Nurse | Pharm | Avg |
|------------------------------------------|-------|-------|-------|-------|
| gpt-4-0613                               | 75.09 | 85.86 | 83.22 | 81.39 |
| **iRASC/BioLlama-Ko-8B**                 | 45.26 | 63.37 | 58.47 | 55.70 |
| gpt-3.5-turbo-0613                       | 41.75 | 62.18 | 56.35 | 53.43 |
| llama2-70b                               | 42.46 | 63.54 | 53.26 | 53.09 |
| SOLAR-10.7B-v1.0                         | 37.19 | 55.54 | 54.07 | 48.93 |
| ProbeMedicalYonseiMAILab/medllama3-v20   | 37.19 | 54.68 | 50.65 | 47.51 |
| beomi/Llama-3-Open-Ko-8B                 | 38.95 | 53.49 | 46.09 | 46.18 |


## Merge Details
### Merge Method

This model was merged using the [DARE](https://arxiv.org/abs/2311.03099) [TIES](https://arxiv.org/abs/2306.01708) merge method using [ProbeMedicalYonseiMAILab/medllama3-v20](https://huggingface.co/ProbeMedicalYonseiMAILab/medllama3-v20) as a base.

### Models Merged

The following models were included in the merge:
* [beomi/Llama-3-Open-Ko-8B](https://huggingface.co/beomi/Llama-3-Open-Ko-8B)

### Configuration

The following YAML configuration was used to produce this model:

```yaml
models:
  - model: ProbeMedicalYonseiMAILab/medllama3-v20
  - model: beomi/Llama-3-Open-Ko-8B
    parameters:
      density: 0.8
      weight: 0.45
merge_method: dare_ties
base_model: ProbeMedicalYonseiMAILab/medllama3-v20
parameters:
  int8_mask: true
dtype: bfloat16
```