marsouuuu-beyond-mergekit

marsouuuu-beyond-mergekit is a Mixture of Experts (MoE) made with the following models using LazyMergekit:

• microsoft/Phi-3-mini-4k-instruct
• unsloth/Phi-3-mini-4k-instruct
• microsoft/Phi-3-mini-128k-instruct
• microsoft/Phi-3.5-mini-instruct

🧩 Configuration

base_model: jpacifico/Chocolatine-3B-Instruct-DPO-Revised
experts:
  - source_model: microsoft/Phi-3-mini-4k-instruct
    positive_prompts:
    - "chat"
    - "assistant"
    - "tell me"
    - "explain"
    - "I want"
  - source_model: unsloth/Phi-3-mini-4k-instruct
    positive_prompts:
    - "code"
    - "python"
    - "javascript"
    - "programming"
    - "algorithm"
  - source_model: microsoft/Phi-3-mini-128k-instruct
    positive_prompts:
    - "storywriting"
    - "write"
    - "scene"
    - "story"
    - "character"
  - source_model: microsoft/Phi-3.5-mini-instruct
    positive_prompts:
    - "reason"
    - "math"
    - "mathematics"
    - "solve"
    - "count"

💻 Usage

!pip install -qU transformers bitsandbytes accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "Marsouuu/marsouuuu-beyond-mergekit"

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"])