README.md

metadata

license: apache-2.0
tags:
  - moe
  - frankenmoe
  - merge
  - mergekit
  - lazymergekit
  - unsloth/mistral-7b-v0.2
  - mistralai/Mistral-7B-Instruct-v0.2
base_model:
  - unsloth/mistral-7b-v0.2
  - mistralai/Mistral-7B-Instruct-v0.2

Mini-Mixtral-v0.2

Mini-Mixtral-v0.2 is a Mixture of Experts (MoE) made with the following models using LazyMergekit:

• unsloth/mistral-7b-v0.2
• mistralai/Mistral-7B-Instruct-v0.2

🧩 Configuration

base_model: unsloth/mistral-7b-v0.2
gate_mode: hidden
dtype: bfloat16
experts:
  - source_model: unsloth/mistral-7b-v0.2
    positive_prompts:
      - "Answer this question from the ARC (Argument Reasoning Comprehension)."
      - "Use common sense and logical reasoning skills."
    negative_prompts:
      - "nonsense"
      - "irrational"
      - "math"
      - "code"
  - source_model: mistralai/Mistral-7B-Instruct-v0.2
    positive_prompts:
      - "Calculate the answer to this math problem"
      - "My mathematical capabilities are strong, allowing me to handle complex mathematical queries"
      - "solve for"
    negative_prompts:
      - "incorrect"
      - "inaccurate"
      - "creativity"

💻 Usage

!pip install -qU transformers bitsandbytes accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "NeuralNovel/Mini-Mixtral-v0.2"

tokenizer = AutoTokenizer.from_pretrained(model)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    model_kwargs={"torch_dtype": torch.float16, "load_in_4bit": True},
)

messages = [{"role": "user", "content": "Explain what a Mixture of Experts is in less than 100 words."}]
prompt = pipeline.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
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"])