Llamacpp imatrix Quantizations of granite-34b-code-instruct

Using llama.cpp release b2940 for quantization.

Original model: https://huggingface.co/ibm-granite/granite-34b-code-instruct

All quants made using imatrix option with dataset from here

Prompt format

System:
{system_prompt}

Question:
{prompt}

Answer:

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

Filename	Quant type	File Size	Description
granite-34b-code-instruct-Q8_0.gguf	Q8_0	35.99GB	Extremely high quality, generally unneeded but max available quant.
granite-34b-code-instruct-Q6_K.gguf	Q6_K	27.83GB	Very high quality, near perfect, recommended.
granite-34b-code-instruct-Q5_K_M.gguf	Q5_K_M	24.74GB	High quality, recommended.
granite-34b-code-instruct-Q5_K_S.gguf	Q5_K_S	23.40GB	High quality, recommended.
granite-34b-code-instruct-Q4_K_M.gguf	Q4_K_M	21.38GB	Good quality, uses about 4.83 bits per weight, recommended.
granite-34b-code-instruct-Q4_K_S.gguf	Q4_K_S	19.44GB	Slightly lower quality with more space savings, recommended.
granite-34b-code-instruct-IQ4_NL.gguf	IQ4_NL	19.23GB	Decent quality, slightly smaller than Q4_K_S with similar performance recommended.
granite-34b-code-instruct-IQ4_XS.gguf	IQ4_XS	18.19GB	Decent quality, smaller than Q4_K_S with similar performance, recommended.
granite-34b-code-instruct-Q3_K_L.gguf	Q3_K_L	19.54GB	Lower quality but usable, good for low RAM availability.
granite-34b-code-instruct-Q3_K_M.gguf	Q3_K_M	17.56GB	Even lower quality.
granite-34b-code-instruct-IQ3_M.gguf	IQ3_M	15.92GB	Medium-low quality, new method with decent performance comparable to Q3_K_M.
granite-34b-code-instruct-IQ3_S.gguf	IQ3_S	14.80GB	Lower quality, new method with decent performance, recommended over Q3_K_S quant, same size with better performance.
granite-34b-code-instruct-Q3_K_S.gguf	Q3_K_S	14.80GB	Low quality, not recommended.
granite-34b-code-instruct-IQ3_XS.gguf	IQ3_XS	14.34GB	Lower quality, new method with decent performance, slightly better than Q3_K_S.
granite-34b-code-instruct-IQ3_XXS.gguf	IQ3_XXS	13.35GB	Lower quality, new method with decent performance, comparable to Q3 quants.
granite-34b-code-instruct-Q2_K.gguf	Q2_K	13.10GB	Very low quality but surprisingly usable.
granite-34b-code-instruct-IQ2_M.gguf	IQ2_M	11.66GB	Very low quality, uses SOTA techniques to also be surprisingly usable.
granite-34b-code-instruct-IQ2_S.gguf	IQ2_S	10.77GB	Very low quality, uses SOTA techniques to be usable.
granite-34b-code-instruct-IQ2_XS.gguf	IQ2_XS	10.14GB	Very low quality, uses SOTA techniques to be usable.
granite-34b-code-instruct-IQ2_XXS.gguf	IQ2_XXS	9.15GB	Lower quality, uses SOTA techniques to be usable.
granite-34b-code-instruct-IQ1_M.gguf	IQ1_M	8.04GB	Extremely low quality, not recommended.
granite-34b-code-instruct-IQ1_S.gguf	IQ1_S	7.37GB	Extremely low quality, not recommended.

Downloading using huggingface-cli

First, make sure you have hugginface-cli installed:

pip install -U "huggingface_hub[cli]"

Then, you can target the specific file you want:

huggingface-cli download bartowski/granite-34b-code-instruct-GGUF --include "granite-34b-code-instruct-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. In order to download them all to a local folder, run:

huggingface-cli download bartowski/granite-34b-code-instruct-GGUF --include "granite-34b-code-instruct-Q8_0.gguf/*" --local-dir granite-34b-code-instruct-Q8_0 --local-dir-use-symlinks False

You can either specify a new local-dir (granite-34b-code-instruct-Q8_0) or download them all in place (./)

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