ludocomito_-_Minerva_3B_Ties_1.0-8bits

Quantization made by Richard Erkhov.

Github

Discord

Request more models

Minerva_3B_Ties_1.0 - bnb 8bits

• Model creator: https://huggingface.co/ludocomito/
• Original model: https://huggingface.co/ludocomito/Minerva_3B_Ties_1.0/

Original model description:

tags:

• merge
• mergekit
• lazymergekit
• mudler/Asinello-Minerva-3B-v0.1
• mii-llm/minerva-chat-v0.1-alpha-sft base_model:
• mudler/Asinello-Minerva-3B-v0.1
• mii-llm/minerva-chat-v0.1-alpha-sft license: apache-2.0 language:
• it

---

M_Moe_3x3B_TIES

M_Moe_3x3B_TIES is a merge of the following models using LazyMergekit:

• mudler/Asinello-Minerva-3B-v0.1
• mii-llm/minerva-chat-v0.1-alpha-sft

🧩 Configuration

models:
  - model: sapienzanlp/Minerva-3B-base-v1.0
    # no parameters necessary for base model
  - model: mudler/Asinello-Minerva-3B-v0.1
    parameters:
      density: 0.5
      weight: 0.5
  - model: mii-llm/minerva-chat-v0.1-alpha-sft
    parameters:
      density: 0.5
      weight: 0.3
merge_method: ties
base_model: sapienzanlp/Minerva-3B-base-v1.0
parameters:
  normalize: true
dtype: bfloat16

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "ludocomito/M_Moe_3x3B_TIES"
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"])