mistral-orpo-beta-NeuralBeagle14-7B-dare-ties

mistral-orpo-beta-NeuralBeagle14-7B-dare-ties is a merge of the following models using LazyMergekit:

• kaist-ai/mistral-orpo-beta
• mlabonne/NeuralBeagle14-7B

🧩 Configuration

models:
  - model: kaist-ai/mistral-orpo-beta
    parameters:
      density: 0.5
      weight: 0.6
    # No parameters necessary for base model
  - model: mlabonne/NeuralBeagle14-7B
    parameters:
      density: 0.5
      weight: 0.4
merge_method: dare_ties
base_model: kaist-ai/mistral-orpo-beta
parameters:
  int8_mask: true
dtype: bfloat16

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "saucam/mistral-orpo-beta-NeuralBeagle14-7B-dare-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"])

Evaluation results for openllm benchmark via llm-autoeval

https://gist.github.com/saucam/dcc1f43acce8179f476afc2d91be53ff