---
license: llama3.2
base_model:
  - ValiantLabs/Llama3.2-3B-Enigma
  - deep-div/MediLlama-3.2
  - prithivMLmods/Llama-3.2-3B-Math-Oct
  - Saxo/Linkbricks-Llama3.2-Korean-cpt-3b
  - GetSoloTech/Llama-3.2-3B-Reasoning
tags:
  - merge
  - llama
  - llama-3.2
---

# Llama3.2-3B-ModelStock-5way

A merge of 5 specialist Llama 3.2 3B models using **Model Stock** merge method.

## Merged Models
1. **ValiantLabs/Llama3.2-3B-Enigma** - Code specialist
2. **deep-div/MediLlama-3.2** - Medical specialist
3. **prithivMLmods/Llama-3.2-3B-Math-Oct** - Math specialist
4. **Saxo/Linkbricks-Llama3.2-Korean-cpt-3b** - Korean language specialist
5. **GetSoloTech/Llama-3.2-3B-Reasoning** - Reasoning specialist

## Evaluation
- **ARC-Challenge**: 47.7%

## Usage

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("Bhavkumar21/Llama3.2-3B-ModelStock-5way")
tokenizer = AutoTokenizer.from_pretrained("Bhavkumar21/Llama3.2-3B-ModelStock-5way")
```