Mistral-MBX-7B-slerp

Research & Development for AutoSynthetix AI

🌐 Website https://autosynthetix.com/

📨 Discord https://discord.gg/pAKqENStQr

📦 GitHub https://github.com/jdwebprogrammer

📦 GitLab https://gitlab.com/jdwebprogrammer

🏆 Patreon https://patreon.com/jdwebprogrammer

📷 YouTube https://www.youtube.com/@jdwebprogrammer

📺 Twitch https://www.twitch.tv/jdwebprogrammer

🐦 Twitter(X) https://twitter.com/jdwebprogrammer

• License includes the license of the model derivatives:

• MergeKit LGPL-3.0 https://github.com/arcee-ai/mergekit?tab=LGPL-3.0-1-ov-file#readme
• Mistral Apache 2.0 https://huggingface.co/mistralai/Mistral-7B-v0.1
• CultriX Apache 2.0 https://huggingface.co/flemmingmiguel/MBX-7B-v3

Mistral-MBX-7B-slerp is a merge of the following models using LazyMergekit:

• mistralai/Mistral-7B-v0.1
• flemmingmiguel/MBX-7B-v3

🧩 Configuration

slices:
  - sources:
      - model: mistralai/Mistral-7B-v0.1
        layer_range: [0, 32]
      - model: flemmingmiguel/MBX-7B-v3
        layer_range: [0, 32]
merge_method: slerp
base_model: mistralai/Mistral-7B-v0.1
parameters:
  t:
    - filter: self_attn
      value: [0, 0.5, 0.3, 0.7, 1]
    - filter: mlp
      value: [1, 0.5, 0.7, 0.3, 0]
    - value: 0.5
dtype: bfloat16

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "JDWebProgrammer/Mistral-MBX-7B-slerp"
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"])