---
model-index:
- name: EEVE-Instruct-Math-10.8B
  results:
  - task:
      type: text-generation
    dataset:
      name: gsm8k-ko
      type: gsm8k
    metrics:
    - name: pass@1
      type: pass@1
      value: 0.4845
      verified: false
base_model:
- yanolja/EEVE-Korean-Instruct-10.8B-v1.0
- kuotient/EEVE-Math-10.8B-SFT
tags:
- merge
license: cc-by-sa-4.0
language:
- ko
---
# EEVE-Instruct-Math-10.8B

`EEVE-Math` 프로젝트는
- Orca-Math-200k 번역 ([Orca-Math: Unlocking the potential of SLMs in Grade School Math](https://arxiv.org/pdf/2402.14830.pdf))
- gsm8k 번역, lm_eval 활용
- Mergekit을 이용한 dare-ties 사용 ([DARE](https://arxiv.org/abs/2311.03099))

에 대한 내용을 포괄하고 있습니다.

> 이 모델은 EEVE-Math와 EEVE-Instruct의 dare-ties로 병합한 병합 모델입니다. 이 프로젝트는 이런 과정을 통해 특화 모델의 EEVE-Math의 성능을 많이 잃지 않고 Instruct 모델의 사용성을 유지할 수 있음을 보여주는 Proof of concept의 성격을 가지고 있습니다.

| Model | gsm8k-ko(pass@1) |
|---|---|
| EEVE(Base) | 0.4049 |
| [EEVE-Math](https://huggingface.co/kuotient/EEVE-Math-10.8B) (epoch 1) | 0.508 |
| EEVE-Math (epoch 2) | **0.539** |
| [EEVE-Instruct](https://huggingface.co/yanolja/EEVE-Korean-Instruct-10.8B-v1.0) | 0.4511 |
| EEVE-Instruct + Math | **0.4845** |

## Merge Details
This model was merged using the [DARE](https://arxiv.org/abs/2311.03099) [TIES](https://arxiv.org/abs/2306.01708) merge method using [yanolja/EEVE-Korean-Instruct-10.8B-v1.0](https://huggingface.co/yanolja/EEVE-Korean-Instruct-10.8B-v1.0) as a base.

### Models Merged

The following models were included in the merge:
* [kuotient/EEVE-Math-10.8B](https://huggingface.co/kuotient/EEVE-Math-10.8B)

### Configuration

The following YAML configuration was used to produce this model:

```yaml
models:
  - model: yanolja/EEVE-Korean-10.8B-v1.0
    # no parameters necessary for base model
  - model: yanolja/EEVE-Korean-Instruct-10.8B-v1.0
    parameters:
      density: 0.53
      weight: 0.6
  - model: kuotient/EEVE-Math-10.8B
    parameters:
      density: 0.53
      weight: 0.4
merge_method: dare_ties
base_model: yanolja/EEVE-Korean-10.8B-v1.0
parameters:
  int8_mask: true
dtype: bfloat16
```

## Evaluation
[gsm8k-ko](https://huggingface.co/datasets/kuotient/gsm8k-ko), kobest
```
git clone https://github.com/kuotient/lm-evaluation-harness.git
cd lm-evaluation-harness
pip install -e .
```
```
lm_eval --model hf \
    --model_args pretrained=yanolja/EEVE-Korean-Instruct-2.8B-v1.0 \
    --tasks gsm8k-ko \
    --device cuda:0 \
    --batch_size auto:4
```

| Model | gsm8k(pass@1) | boolq(acc) | copa(acc) | hellaswag(acc) | Overall |
|---|---|---|---|---|---|
| yanolja/EEVE-Korean-10.8B-v1.0 | 0.4049 | - | - | - | - | - |
| yanolja/EEVE-Korean-Instruct-10.8B-v1.0 | 0.4511 | **0.8668** | **0.7450** | 0.4940 | 0.6392 |
| [**EEVE-Math-10.8B**](https://huggingface.co/kuotient/EEVE-Math-10.8B) | **0.5390** | 0.8027 | 0.7260 | 0.4760 | 0.6359 |
| **EEVE-Instruct-Math-10.8B** | 0.4845 | 0.8519 | 0.7410 | **0.4980** | **0.6439** |