helizac/Novus-7b-tr_v1-GGUF

Novus-7b-tr_v1 - GGUF

• Model creator: mlabonne

• Original model: Daredevil-7B

• Model Fine-Tuner: Novus Research

• Fine-tuned model: Novus-7b-tr_v1

Description

This repo contains GGUF format model files for mlabonne's Daredevil-7B model, fine-tuned to create Novus-7b-tr_v1 by Novus Research.

About GGUF

GGUF is a new format introduced by the llama.cpp team on August 21st 2023. It is a replacement for GGML, which is no longer supported by llama.cpp.

Here is an incomplete list of clients and libraries that are known to support GGUF:

• llama.cpp. The source project for GGUF. Offers a CLI and a server option.
• text-generation-webui, the most widely used web UI, with many features and powerful extensions. Supports GPU acceleration.
• KoboldCpp, a fully featured web UI, with GPU accel across all platforms and GPU architectures. Especially good for storytelling.
• GPT4All, a free and open-source local running GUI, supporting Windows, Linux, and macOS with full GPU accel.
• LM Studio, an easy-to-use and powerful local GUI for Windows and macOS (Silicon), with GPU acceleration. Linux available, in beta as of 27/11/2023.
• LoLLMS Web UI, a great web UI with many interesting and unique features, including a full model library for easy model selection.
• Faraday.dev, an attractive and easy-to-use character-based chat GUI for Windows and macOS (both Silicon and Intel), with GPU acceleration.
• llama-cpp-python, a Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
• candle, a Rust ML framework with a focus on performance, including GPU support, and ease of use.
• ctransformers, a Python library with GPU accel, LangChain support, and OpenAI-compatible AI server. Note, as of time of writing (November 27th 2023), ctransformers has not been updated in a long time and does not support many recent models.

Novus Research

Novus Research is committed to pushing the boundaries in natural language processing by collaborating with the open-source community through innovative research. This commitment is coupled with our focus on empowering businesses with tailored, on-site AI and large language model solutions.

Compatibility

These quantized GGUF files are compatible with candle from Hugging Face.

Provided Files

Name	Bit	Quant Method	Size	Use case
Novus-7b-tr_v1.Q2_K.gguf	2	Q2_K	2.72G	Smallest size, lowest precision
Novus-7b-tr_v1.Q3_K.gguf	3	Q3_K	3.16G	Very low precision
Novus-7b-tr_v1.Q3_K_S.gguf	3	Q3_K_S	3.52G	Low precision, level 0
Novus-7b-tr_v1.Q3_K_M.gguf	3	Q3_K_M	3.82G	Slightly better than Q4_0
Novus-7b-tr_v1.Q3_K_L.gguf	3	Q3_K_L	3.47G	Kernel optimized, low precision
Novus-7b-tr_v1.Q4_0.gguf	4	Q4_0	4.11G	Moderate precision, level 0
Novus-7b-tr_v1.Q4_K_M.gguf	4	Q4_K_M	4.37G	Better than Q5_0
Novus-7b-tr_v1.Q5_0.gguf	5	Q5_0	5.00G	Kernel optimized, moderate precision
Novus-7b-tr_v1.Q5_K_S.gguf	5	Q5_K_S	5.00G	Higher precision than Q5_K
Novus-7b-tr_v1.Q5_K_M.gguf	5	Q5_K_M	5.13G	Higher precision, level 0
Novus-7b-tr_v1.Q6_K.gguf	6	Q6_K	5.94G	Highest precision, level 1
Novus-7b-tr_v1.Q8_0.gguf	8	Q8_0	7.77G	Kernel optimized, high precision
Novus-7b-tr_v1.F32.gguf	32	F32	29.00G	Single-precision floating point

How to Download

To download the models, you can use the huggingface-cli command or the equivalent Python code with hf_hub_download.

Using huggingface-cli command:

huggingface-cli download helizac/Novus-7b-tr_v1-GGUF <model_file>

For example, to download the Q2_K model:

huggingface-cli download helizac/Novus-7b-tr_v1-GGUF Novus-7b-tr_v1_Q2_K.gguf

Downloading all models:

huggingface-cli download helizac/Novus-7b-tr_v1-GGUF Novus-7b-tr_v1_Q2_K.gguf
huggingface-cli download helizac/Novus-7b-tr_v1-GGUF Novus-7b-tr_v1_Q3_K.gguf
huggingface-cli download helizac/Novus-7b-tr_v1-GGUF Novus-7b-tr_v1_Q4_0.gguf
huggingface-cli download helizac/Novus-7b-tr_v1-GGUF Novus-7b-tr_v1_Q4_1.gguf
huggingface-cli download helizac/Novus-7b-tr_v1-GGUF Novus-7b-tr_v1_Q4_K.gguf
huggingface-cli download helizac/Novus-7b-tr_v1-GGUF Novus-7b-tr_v1_Q5_0.gguf
huggingface-cli download helizac/Novus-7b-tr_v1-GGUF Novus-7b-tr_v1_Q5_1.gguf
huggingface-cli download helizac/Novus-7b-tr_v1-GGUF Novus-7b-tr_v1_Q5_K.gguf
huggingface-cli download helizac/Novus-7b-tr_v1-GGUF Novus-7b-tr_v1_Q6_K.gguf
huggingface-cli download helizac/Novus-7b-tr_v1-GGUF Novus-7b-tr_v1_Q8_0.gguf
huggingface-cli download helizac/Novus-7b-tr_v1-GGUF Novus-7b-tr_v1_Q8_1.gguf
huggingface-cli download helizac/Novus-7b-tr_v1-GGUF Novus-7b-tr_v1_Q8_K.gguf
huggingface-cli download helizac/Novus-7b-tr_v1-GGUF Novus-7b-tr_v1_F16.gguf
huggingface-cli download helizac/Novus-7b-tr_v1-GGUF Novus-7b-tr_v1_F32.gguf

Using Python:

from huggingface_hub import hf_hub_download

hf_hub_download("helizac/Novus-7b-tr_v1-GGUF", "<model_file>")

To download all models, you can run:

model_files = [
    "Novus-7b-tr_v1_Q2_K.gguf",
    "Novus-7b-tr_v1_Q3_K.gguf",
    "Novus-7b-tr_v1_Q4_0.gguf",
    "Novus-7b-tr_v1_Q4_1.gguf",
    "Novus-7b-tr_v1_Q4_K.gguf",
    "Novus-7b-tr_v1_Q5_0.gguf",
    "Novus-7b-tr_v1_Q5_1.gguf",
    "Novus-7b-tr_v1_Q5_K.gguf",
    "Novus-7b-tr_v1_Q6_K.gguf",
    "Novus-7b-tr_v1_Q8_0.gguf",
    "Novus-7b-tr_v1_Q8_1.gguf",
    "Novus-7b-tr_v1_Q8_K.gguf",
    "Novus-7b-tr_v1_F32.gguf"
]

for model_file in model_files:
    hf_hub_download("helizac/Novus-7b-tr_v1-GGUF", model_file)

You can also specify a folder to download the file(s) to:

hf_hub_download("helizac/Novus-7b-tr_v1-GGUF", "<model_file>", local_dir="<output_directory>")

Usage

!pip install llama-cpp-python

from llama_cpp import Llama

# Download the model from Hugging Face (replace URL with the actual one)
model_url = "https://huggingface.co./helizac/Novus-7b-tr_v1-GGUF/blob/main/Novus-7b-tr_v1.Q8_0.gguf"
model_path = "Novus-7b-tr_v1.gguf"  # Local filename

# Function to download the model (optional)
def download_model(url, filename):
  import urllib.request
  if not os.path.isfile(filename):
    urllib.request.urlretrieve(url, filename)
    print(f"Downloaded model: {filename}")

download_model(model_url, model_path)

# Load the model
llm = Llama(model_path=model_path)

prompt = "Büyük dil modelleri nelerdir?"

# Adjust these parameters for different outputs
max_tokens = 256
temperature = 0.7

output = llm(prompt, max_tokens=max_tokens, temperature=temperature)
output_text = output["choices"][0]["text"].strip()

print(output_text)

Acknowledgements

This model is built on top of the efforts from the NovusResearch and mlabonne teams, and we appreciate their contribution to the AI community.

GGUF model card:

{Furkan Erdi}