README.md

---
base_model:
- ertghiu256/qwen-3-4b-mixture-of-thought
- ertghiu256/qwen3-math-reasoner
- ertghiu256/qwen3-4b-code-reasoning
- ertghiu256/qwen3-multi-reasoner
library_name: transformers
tags:
- mergekit
- merge
- code
- reasoning
- thinking
- think
- reason
- math
---
# Qwen 3 ultimate reasoning 4b

This is a merge of 4 Qwen pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).

## Merge Details
This model aims for general reasoning by mergeing a few Qwen 3 4b model that is trained multiple reasoning datasets

### Merge Method

This model was merged using the [Linear](https://arxiv.org/abs/2203.05482) merge method using [ertghiu256/qwen3-multi-reasoner](https://huggingface.co/ertghiu256/qwen3-multi-reasoner) as a base.

### Models Merged

The following models were included in the merge:
* [ertghiu256/qwen-3-4b-mixture-of-thought](https://huggingface.co/ertghiu256/qwen-3-4b-mixture-of-thought)
* [ertghiu256/qwen3-math-reasoner](https://huggingface.co/ertghiu256/qwen3-math-reasoner)
* [ertghiu256/qwen3-4b-code-reasoning](https://huggingface.co/ertghiu256/qwen3-4b-code-reasoning)

### Configuration

The following YAML configuration was used to produce this model:

```yaml
base_model: ertghiu256/qwen3-multi-reasoner
dtype: float16
merge_method: linear
modules:
  default:
    slices:
    - sources:
      - layer_range: [0, 36]
        model: ertghiu256/qwen3-multi-reasoner
        parameters:
          weight: 0.7
      - layer_range: [0, 36]
        model: ertghiu256/qwen-3-4b-mixture-of-thought
        parameters:
          weight: 0.9
      - layer_range: [0, 36]
        model: ertghiu256/qwen3-4b-code-reasoning
        parameters:
          weight: 0.8
      - layer_range: [0, 36]
        model: ertghiu256/qwen3-math-reasoner
        parameters:
          weight: 0.6
parameters:
  int8_mask: 1.0
  normalize: 1.0
```