README.md

---
tags:
- merge
- mergekit
- lazymergekit
- mlabonne/AlphaMonarch-7B
- mlabonne/NeuralMonarch-7B
- Kukedlc/NeuralMaxime-7B-slerp
- eren23/dpo-binarized-NeuralTrix-7B
base_model:
- mlabonne/AlphaMonarch-7B
- mlabonne/NeuralMonarch-7B
- Kukedlc/NeuralMaxime-7B-slerp
- eren23/dpo-binarized-NeuralTrix-7B
---

# NeMoTrix-v2

NeMoTrix-v2 is a merge of the following models using [LazyMergekit](https://colab.research.google.com/drive/1obulZ1ROXHjYLn6PPZJwRR6GzgQogxxb?usp=sharing):
* [mlabonne/AlphaMonarch-7B](https://huggingface.co/mlabonne/AlphaMonarch-7B)
* [mlabonne/NeuralMonarch-7B](https://huggingface.co/mlabonne/NeuralMonarch-7B)
* [Kukedlc/NeuralMaxime-7B-slerp](https://huggingface.co/Kukedlc/NeuralMaxime-7B-slerp)
* [eren23/dpo-binarized-NeuralTrix-7B](https://huggingface.co/eren23/dpo-binarized-NeuralTrix-7B)

## 🧩 Configuration

```yaml
models:
  - model: CultriX/MonaTrix-v4
    # No parameters necessary for base model, chosen for its strong performance and presence in high-scoring configurations
  - model: mlabonne/AlphaMonarch-7B
    parameters:
      density: 0.65
      weight: 0.4
  - model: mlabonne/NeuralMonarch-7B
    parameters:
      density: 0.65
      weight: 0.35
  - model: Kukedlc/NeuralMaxime-7B-slerp
    parameters:
      density: 0.6
      weight: 0.35
  - model: eren23/dpo-binarized-NeuralTrix-7B
    parameters:
      density: 0.55
      weight: 0.25
merge_method: dare_ties
base_model: CultriX/MonaTrix-v4
parameters:
  int8_mask: true
dtype: bfloat16
random_seed: 0
```

## 💻 Usage

```python
!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "CultriX/NeMoTrix-v2"
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"])
```