Edit model card

Heads up: currently CUDA offloading is broken unless you enable flash attention

Llamacpp imatrix Quantizations of Qwen2-72B-Instruct

Using llama.cpp commit ee459f40f65810a810151b24eba5b8bd174ceffe for quantization.

Original model: https://huggingface.co/Qwen/Qwen2-72B-Instruct

All quants made using imatrix option with dataset from here

Prompt format


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

Filename Quant type File Size Description
Qwen2-72B-Instruct-Q8_0.gguf Q8_0 77.26GB Extremely high quality, generally unneeded but max available quant.
Qwen2-72B-Instruct-Q5_K_M.gguf Q5_K_M 54.44GB High quality, recommended.
Qwen2-72B-Instruct-Q4_K_M.gguf Q4_K_M 47.41GB Good quality, uses about 4.83 bits per weight, recommended.
Qwen2-72B-Instruct-IQ4_XS.gguf IQ4_XS 39.70GB Decent quality, smaller than Q4_K_S with similar performance, recommended.
Qwen2-72B-Instruct-Q3_K_M.gguf Q3_K_M 37.69GB Even lower quality.
Qwen2-72B-Instruct-IQ3_M.gguf IQ3_M 35.50GB Medium-low quality, new method with decent performance comparable to Q3_K_M.
Qwen2-72B-Instruct-Q3_K_S.gguf Q3_K_S 34.48GB Low quality, not recommended.
Qwen2-72B-Instruct-IQ3_XXS.gguf IQ3_XXS 31.84GB Lower quality, new method with decent performance, comparable to Q3 quants.
Qwen2-72B-Instruct-Q2_K.gguf Q2_K 29.81GB Very low quality but surprisingly usable.
Qwen2-72B-Instruct-IQ2_M.gguf IQ2_M 29.33GB Very low quality, uses SOTA techniques to also be surprisingly usable.
Qwen2-72B-Instruct-IQ2_XXS.gguf IQ2_XXS 25.49GB Lower quality, uses SOTA techniques to be usable.
Qwen2-72B-Instruct-IQ1_M.gguf IQ1_M 23.74GB 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/Qwen2-72B-Instruct-GGUF --include "Qwen2-72B-Instruct-Q4_K_M.gguf" --local-dir ./

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/Qwen2-72B-Instruct-GGUF --include "Qwen2-72B-Instruct-Q8_0.gguf/*" --local-dir Qwen2-72B-Instruct-Q8_0

You can either specify a new local-dir (Qwen2-72B-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

Downloads last month
Model size
72.7B params





Unable to determine this model's library. Check the docs .