Merge-Mixtral-Prometheus-8x7B

Merge-Mixtral-Prometheus-8x7B is a merge of the following models using LazyMergekit:

• prometheus-eval/prometheus-8x7b-v2.0
• mistralai/Mixtral-8x7B-Instruct-v0.1

🧩 Configuration

models:
  - model: prometheus-eval/prometheus-8x7b-v2.0
    parameters:
      weight: 1.0
  - model: mistralai/Mixtral-8x7B-Instruct-v0.1
    parameters:
      weight: 1.0
merge_method: linear
dtype: bfloat16

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "vicgalle/test-merge-3"
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"])

Paper citation

Paper: https://arxiv.org/abs/2406.07188

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	24.61
IFEval (0-Shot)	57.44
BBH (3-Shot)	34.65
MATH Lvl 5 (4-Shot)	8.31
GPQA (0-shot)	7.83
MuSR (0-shot)	9.59
MMLU-PRO (5-shot)	29.82