gemma-3-4B_Test1

gemma-3-4B_Test1

gemma-3-4B_Test1 is a merge of the following models using LazyMergekit:

• unsloth/gemma-3-4b-it
• vanta-research/scout-4b

🧩 Configuration

models:
  - model: unsloth/gemma-3-4b-it
    #no parameters necessary for base model
  - model: unsloth/gemma-3-4b-it
    parameters:
      density: 0.477
      weight: 0.522

  - model: vanta-research/scout-4b
    parameters:
      density: 0.522
      weight: 0.477

merge_method: dare_ties
base_model: unsloth/gemma-3-4b-it
parameters:
  normalize: true
dtype: bfloat16

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "zelk12/gemma-3-4B_Test1"
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"])