README.md

---
base_model:
- open-r1/OlympicCoder-32B
- qihoo360/Light-R1-32B
- qihoo360/Light-R1-32B-DS
- Qwen/QwQ-32B
- Qwen/Qwen2.5-Coder-32B-Instruct
library_name: transformers
tags:
- mergekit
- merge
new_version: YOYO-AI/YOYO-O1-32B-V4
---
![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/64e174e202fa032de4143324/GNPQMcj3XRXlXy3ebsPRZ.jpeg)

# YOYO-O1-32B-V3

*Combined the most top-notch 32B **inference** model and **code** model in the entire open-source community.*

### Merge Method

This model was merged using the [SCE](https://arxiv.org/abs/2408.07990) merge method using [Qwen/Qwen2.5-Coder-32B-instruct](https://huggingface.co/Qwen/Qwen2.5-Coder-32B-instruct) as a base.

### Models Merged

The following models were included in the merge:
* [Qwen/QwQ-32B](https://huggingface.co/Qwen/QwQ-32B)
* [qihoo360/Light-R1-32B-DS](https://huggingface.co/qihoo360/Light-R1-32B-DS)
* [qihoo360/Light-R1-32B](https://huggingface.co/qihoo360/Light-R1-32B)
* [open-r1/OlympicCoder-32B](https://huggingface.co/open-r1/OlympicCoder-32B)

### Configuration

The following YAML configuration was used to produce this model:

```yaml
merge_method: sce
models:
  # Pivot model
  - model: Qwen/Qwen2.5-Coder-32B-instruct
  # Target models
  - model: qihoo360/Light-R1-32B-DS
  - model: qihoo360/Light-R1-32B
  - model: Qwen/QwQ-32B
  - model: open-r1/OlympicCoder-32B
base_model: Qwen/Qwen2.5-Coder-32B-instruct
parameters:
  select_topk: 1
dtype: bfloat16
tokenizer_source: Qwen/QwQ-32B
normalize: true
int8_mask: true
```