Qwen3-4b-tcomanr-merge-v2.2/README.md

---
base_model:
- huihui-ai/Huihui-Qwen3-4B-Thinking-2507-abliterated
- Tesslate/UIGEN-T3-4B-Preview-MAX
- ValiantLabs/Qwen3-4B-Esper3
- ValiantLabs/Qwen3-4B-ShiningValiant3
- ertghiu256/Qwen3-Hermes-4b
- ertghiu256/qwen3-math-reasoner
- ertghiu256/deepseek-r1-0528-distilled-qwen3
- ertghiu256/qwen-3-4b-mixture-of-thought
- Qwen/Qwen3-4B-Thinking-2507
- POLARIS-Project/Polaris-4B-Preview
- ertghiu256/qwen3-multi-reasoner
- ertghiu256/qwen3-4b-code-reasoning
library_name: transformers
tags:
- mergekit
- merge
- thinking
- think
- reasoning
- reason
- code
- math
- qwen
- qwen3
new_version: ertghiu256/Qwen3-4b-tcomanr-merge-v2.3
---
# Ties merged COde MAth aNd Reasoning model
This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).

## Merge Details
This model is a revision of the [ertghiu256/Qwen3-4b-tcomanr-merge-v2](https://huggingface.co/ertghiu256/Qwen3-4b-tcomanr-merge-v2/)

This model aims to combine the code and math capabilities by merging Qwen 3 2507 with multiple Qwen 3 finetunes. 

# How to run 
You can run this model by using multiple interface choices 

## Transformers
As the qwen team suggested to use
```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "ertghiu256/Qwen3-4b-tcomanr-merge-v2.2"

# load the tokenizer and the model
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto"
)

# prepare the model input
prompt = "Give me a short introduction to large language model."
messages = [
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
)
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)

# conduct text completion
generated_ids = model.generate(
    **model_inputs,
    max_new_tokens=32768
)
output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist() 

# parsing thinking content
try:
    # rindex finding 151668 (</think>)
    index = len(output_ids) - output_ids[::-1].index(151668)
except ValueError:
    index = 0

thinking_content = tokenizer.decode(output_ids[:index], skip_special_tokens=True).strip("\n")
content = tokenizer.decode(output_ids[index:], skip_special_tokens=True).strip("\n")

print("thinking content:", thinking_content) # no opening <think> tag
print("content:", content)
```

## Vllm
Run this command
```bash
vllm serve ertghiu256/Qwen3-4b-tcomanr-merge-v2.2 --enable-reasoning --reasoning-parser deepseek_r1
```

## Sglang
Run this command
```bash
python -m sglang.launch_server --model-path ertghiu256/Qwen3-4b-tcomanr-merge-v2.2 --reasoning-parser deepseek-r1
```

## llama.cpp
Run this command
```bash
llama-server --hf-repo ertghiu256/Qwen3-4b-tcomanr-merge-v2.2
```
or
```bash
llama-cli --hf ertghiu256/Qwen3-4b-tcomanr-merge-v2.2
```

## Ollama
Run this command 
```bash
ollama run hf.co/ertghiu256/Qwen3-4b-tcomanr-merge-v2.2:Q8_0
```
or 
```bash
ollama run hf.co/ertghiu256/Qwen3-4b-tcomanr-merge-v2.2:Q5_K_M
```

## LM Studio 
Search 
```
ertghiu256/Qwen3-4b-tcomanr-merge-v2.2
```
in the lm studio model search list then download

### Recomended parameters
```
temp: 0.7
num_ctx: ≥8192
top_p: 0.95
top_k: 40
Repeat Penalty: 1.1
```

### Merge Method

This model was merged using the [TIES](https://arxiv.org/abs/2306.01708) merge method using [Qwen/Qwen3-4B-Thinking-2507](https://huggingface.co/Qwen/Qwen3-4B-Thinking-2507) as a base.

### Models Merged

The following models were included in the merge:
* [huihui-ai/Huihui-Qwen3-4B-Thinking-2507-abliterated](https://huggingface.co/huihui-ai/Huihui-Qwen3-4B-Thinking-2507-abliterated)
* [Tesslate/UIGEN-T3-4B-Preview-MAX](https://huggingface.co/Tesslate/UIGEN-T3-4B-Preview-MAX)
* [ValiantLabs/Qwen3-4B-Esper3](https://huggingface.co/ValiantLabs/Qwen3-4B-Esper3)
* [ValiantLabs/Qwen3-4B-ShiningValiant3](https://huggingface.co/ValiantLabs/Qwen3-4B-ShiningValiant3)
* [ertghiu256/Qwen3-Hermes-4b](https://huggingface.co/ertghiu256/Qwen3-Hermes-4b)
* [ertghiu256/qwen3-math-reasoner](https://huggingface.co/ertghiu256/qwen3-math-reasoner)
* [ertghiu256/deepseek-r1-0528-distilled-qwen3](https://huggingface.co/ertghiu256/deepseek-r1-0528-distilled-qwen3)
* [ertghiu256/qwen-3-4b-mixture-of-thought](https://huggingface.co/ertghiu256/qwen-3-4b-mixture-of-thought)
* [POLARIS-Project/Polaris-4B-Preview](https://huggingface.co/POLARIS-Project/Polaris-4B-Preview)
* [ertghiu256/qwen3-multi-reasoner](https://huggingface.co/ertghiu256/qwen3-multi-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
models:
  - model: ertghiu256/qwen3-math-reasoner
    parameters:
      weight: 0.8
  - model: ertghiu256/qwen3-4b-code-reasoning
    parameters:
      weight: 0.9
  - model: ertghiu256/qwen-3-4b-mixture-of-thought
    parameters:
      weight: 0.9
  - model: POLARIS-Project/Polaris-4B-Preview
    parameters:
      weight: 0.9
  - model: ertghiu256/qwen3-multi-reasoner
    parameters:
      weight: 0.8
  - model: ertghiu256/Qwen3-Hermes-4b
    parameters:
      weight: 0.8
  - model: ValiantLabs/Qwen3-4B-Esper3
    parameters:
      weight: 0.8
  - model: Tesslate/UIGEN-T3-4B-Preview-MAX
    parameters:
      weight: 0.9
  - model: ValiantLabs/Qwen3-4B-ShiningValiant3
    parameters:
      weight: 0.6
  - model: ertghiu256/deepseek-r1-0528-distilled-qwen3
    parameters:
      weight: 0.1
  - model: huihui-ai/Huihui-Qwen3-4B-Thinking-2507-abliterated
    parameters:
      weight: 0.6
  - model: Qwen/Qwen3-4B-Thinking-2507
    parameters:
      weight: 0.9
merge_method: ties
base_model: Qwen/Qwen3-4B-Thinking-2507
parameters:
  normalize: true
  int8_mask: true
  lambda: 1.0
dtype: float16

```