Bagel 8b V1.0 GGUF

🚀 Llamacpp imatrix Quantizations of bagel-8b-v1.0

This project provides LlamaCpp imatrix quantizations of the bagel-8b-v1.0 model. It offers various quantization options for different performance and quality requirements, making it easier to run the model on diverse hardware.

🚀 Quick Start

Downloading using huggingface-cli

First, ensure you have huggingface-cli installed:

pip install -U "huggingface_hub[cli]"

Then, you can target the specific file you want:

huggingface-cli download bartowski/bagel-8b-v1.0-GGUF --include "bagel-8b-v1.0-Q4_K_M.gguf" --local-dir ./ --local-dir-use-symlinks False

If the model is bigger than 50GB, it will have been split into multiple files. To download them all to a local folder, run:

huggingface-cli download bartowski/bagel-8b-v1.0-GGUF --include "bagel-8b-v1.0-Q8_0.gguf/*" --local-dir bagel-8b-v1.0-Q8_0 --local-dir-use-symlinks False

You can either specify a new local-dir (e.g., bagel-8b-v1.0-Q8_0) or download them all in place (./).

✨ Features

Quantization

• Multiple Quantization Options: Offers a wide range of quantization types, such as Q8_0, Q6_K, Q5_K_M, etc., to balance between model quality and file size.
• Using LlamaCpp: Utilizes the llama.cpp release b2854 for quantization.

Prompt Format

The specific prompt format is as follows:

<|begin_of_text|><|start_header_id|>system<|end_header_id|>

{system_prompt}<|eot_id|><|start_header_id|>user<|end_header_id|>

{prompt}<|eot_id|><|start_header_id|>assistant<|end_header_id|>

Downloadable Files

You can choose from a variety of quantized files based on your hardware and performance requirements. Here is a detailed table:

📚 Documentation

Which file should I choose?

A great write - up with charts showing various performances is provided by Artefact2 here

The first thing to figure out is how big a model you can run. To do this, you'll need to figure out how much RAM and/or VRAM you have.

If you want your model running as FAST as possible, you'll want to fit the whole thing on your GPU's VRAM. Aim for a quant with a file size 1 - 2GB smaller than your GPU's total VRAM.

If you want the absolute maximum quality, add both your system RAM and your GPU's VRAM together, then similarly grab a quant with a file size 1 - 2GB smaller than that total.

Next, you'll need to decide if you want to use an 'I - quant' or a 'K - quant'.

If you don't want to think too much, grab one of the K - quants. These are in format 'QX_K_X', like Q5_K_M.

If you want to get more into the weeds, you can check out this extremely useful feature chart:

llama.cpp feature matrix

But basically, if you're aiming for below Q4, and you're running cuBLAS (Nvidia) or rocBLAS (AMD), you should look towards the I - quants. These are in format IQX_X, like IQ3_M. These are newer and offer better performance for their size.

These I - quants can also be used on CPU and Apple Metal, but will be slower than their K - quant equivalent, so speed vs performance is a tradeoff you'll have to decide.

The I - quants are not compatible with Vulcan, which is also AMD, so if you have an AMD card double - check if you're using the rocBLAS build or the Vulcan build. At the time of writing this, LM Studio has a preview with ROCm support, and other inference engines have specific builds for ROCm.

Support

Want to support my work? Visit my ko - fi page here: https://ko-fi.com/bartowski

📄 License

This project uses the llama3 license.