README.md

---
base_model:
- Theros/Qwen2.5-ColdBrew-R1
- Theros/Qwen2.5-ColdBrew-R1
- Theros/Qwen2.5-ColdBrew-R1
- Theros/Qwen2.5-ColdBrew-R1
tags:
- merge
- mergekit
- lazymergekit
- Theros/Qwen2.5-ColdBrew-R1
---

# Q2.5-ColdBrew-R1-Indigo

Q2.5-ColdBrew-R1-Indigo is a merge of the following models using [LazyMergekit](https://colab.research.google.com/drive/1obulZ1ROXHjYLn6PPZJwRR6GzgQogxxb?usp=sharing):
* [Theros/Qwen2.5-ColdBrew-R1](https://huggingface.co/Theros/Qwen2.5-ColdBrew-R1)
* [Theros/Qwen2.5-ColdBrew-R1](https://huggingface.co/Theros/Qwen2.5-ColdBrew-R1)
* [Theros/Qwen2.5-ColdBrew-R1](https://huggingface.co/Theros/Qwen2.5-ColdBrew-R1)
* [Theros/Qwen2.5-ColdBrew-R1](https://huggingface.co/Theros/Qwen2.5-ColdBrew-R1)

## 🧩 Configuration

```yaml
name: Q2.5-ColdBrew-R1-Indigo
const_tag: &scale_factor 0.7071067812  # 1/sqrt(2) scaling for stability

attenuate-env: &attenuated_env
  parameters:
    scale:
      - filter: q_proj
        value: *scale_factor
      - filter: k_proj
        value: *scale_factor
      - value: 1.0

slices:
  - sources:
      - model: Theros/Qwen2.5-ColdBrew-R1
        layer_range: [0, 8]  # Retaining foundational knowledge and language structure.

  - sources:
      - model: Theros/Qwen2.5-ColdBrew-R1
        layer_range: [9, 19]  # Full-strength duplication of mid-range reasoning layers.

  - sources:
      - model: Theros/Qwen2.5-ColdBrew-R1
        layer_range: [10, 19]  # Targeted reinforcement, slightly attenuated to avoid over-dominance.
        <<: *attenuated_env

  - sources:
      - model: Theros/Qwen2.5-ColdBrew-R1
        layer_range: [20, 28]  # Keeping higher-level abstract processing untouched for stability.

merge_method: passthrough
dtype: bfloat16
normalize: true
int8_mask: true
```

## 💻 Usage

```python
!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "SvalTek/Q2.5-ColdBrew-R1-Indigo"
messages = [{"role": "user", "content": "What is a large language model?"}]

tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])
```