README.md

---
base_model:
- google/gemma-2-2b-it
- Kukedlc/Gemma-2-2B-Spanish-1.0
tags:
- merge
- mergekit
- lazymergekit
- google/gemma-2-2b-it
- Kukedlc/Gemma-2-2B-Spanish-1.0
license: apache-2.0
datasets:
- Kukedlc/Big-Spanish-1.2M
language:
- es
library_name: transformers
---

# Kukedlc/NeuralGemma2-2b-Spanish

![image/png](https://cdn-uploads.huggingface.co/production/uploads/64d71ab4089bc502ceb44d29/C9ni8n8QufniYDHFwePby.png)


Kukedlc/NeuralGemma2-2b-Spanish is a merge of the following models using [LazyMergekit](https://colab.research.google.com/drive/1obulZ1ROXHjYLn6PPZJwRR6GzgQogxxb?usp=sharing):
* [google/gemma-2-2b-it](https://huggingface.co/google/gemma-2-2b-it)
* [Kukedlc/Gemma-2-2B-Spanish-1.0](https://huggingface.co/Kukedlc/Gemma-2-2B-Spanish-1.0)

## 🧩 Configuration

```yaml
models:
  - model: google/gemma-2-2b
    # No parameters necessary for base model
  - model: google/gemma-2-2b-it
    parameters:
      density: 0.55
      weight: 0.6
  - model: Kukedlc/Gemma-2-2B-Spanish-1.0
    parameters:
      density: 0.55
      weight: 0.4
merge_method: dare_ties
base_model: google/gemma-2-2b
parameters:
  int8_mask: true
dtype: float16
```

## 💻 Usage

```python
!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "Kukedlc/NeuralGemma2-2b-Spanish"
messages = [{"role": "user", "content": "Pueden las maquinas pensar?"}]

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"])
```