models:
    - model: Trappu/Nemo-Picaro-V2-12B
      parameters:
        density: 0.7
        weight: 0.5
    - model: anthracite-org/magnum-v2-12b
      parameters:
        density: 0.3
        weight: 0.5

merge_method: ties
base_model: Trappu/Nemo-Picaro-V2-12B
parameters:
    normalize: true
    int8_mask: true
dtype: bfloat16

💻 Usage

python

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "Trappu/Magnum-Picaro-0.7-V3-12b"
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"])