Meta-Llama-3-8B-Instruct-old-GGUF

Llamacpp imatrix Quantizations of Meta-Llama-3-8B-Instruct (old)

<b>This conversion is based on the merged Llama 3 support in llama.cpp (release b2710)</b>

This model is being deprecated in favour of the incoming conversion/quant with BPE tokenizers fixed. Will be here: https://huggingface.co/bartowski/Meta-Llama-3-8B-Instruct-GGUF

Known working on:

• LM Studio 0.2.20*
• koboldcpp 1.63

Confirmed not working on (as of April 21):

• text-generation-webui master/dev

Any others unknown, feel free to comment

*: LM Studio 0.2.20 seems to work on Mac, but not on Windows, test and verify for yourself to see if this is the right version to use

Using <a href="https://github.com/ggerganov/llama.cpp/">llama.cpp</a> release <a href="https://github.com/ggerganov/llama.cpp/releases/tag/b2710">b2710</a> for quantization.

Original model: https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct

All quants made using imatrix option with dataset provided by Kalomaze here

Prompt format

<|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|>

<b>Warning: you will need to update your inference tool to be on at least version 2710 of llama.cpp, this will vary across tools</b>

Download a file (not the whole branch) from below:

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.

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