README.md

---
tags:
- merge
- mergekit
- lazymergekit
- jondurbin/bagel-dpo-34b-v0.2
- abacusai/MetaMath-Bagel-DPO-34B
base_model:
- jondurbin/bagel-dpo-34b-v0.2
- abacusai/MetaMath-Bagel-DPO-34B
license: apache-2.0
language:
- en
library_name: transformers
pipeline_tag: text-generation
---

# Pearl-34B-ties

Pearl-34B-ties is a merge of the following models:
* [jondurbin/bagel-dpo-34b-v0.2](https://huggingface.co/jondurbin/bagel-dpo-34b-v0.2)
* [abacusai/MetaMath-Bagel-DPO-34B](https://huggingface.co/abacusai/MetaMath-Bagel-DPO-34B)

## Configuration

```yaml
models:
  - model: abacusai/Smaug-34B-v0.1
  - model: jondurbin/bagel-dpo-34b-v0.2
    parameters:
      density: 0.45
      weight: 0.5
  - model: abacusai/MetaMath-Bagel-DPO-34B
    parameters:
      density: 0.48
      weight: 0.5
merge_method: ties
base_model: abacusai/Smaug-34B-v0.1
parameters:
  normalize: true
  int8_mask: true
dtype: bfloat16
```

## Usage

```python
!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "louisbrulenaudet/Pearl-34B-ties"
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"])
```