Edit model card

Llamacpp Quantizations of CodeQwen1.5-7B-Chat

Using llama.cpp PR 6707 for quantization.

Original model: https://huggingface.co/Qwen/CodeQwen1.5-7B-Chat

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

Prompt format

<|im_start|>system
{system_prompt}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant

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

Filename Quant type File Size Description
CodeQwen1.5-7B-Chat-Q8_0.gguf Q8_0 7.70GB Extremely high quality, generally unneeded but max available quant.
CodeQwen1.5-7B-Chat-Q6_K.gguf Q6_K 6.37GB Very high quality, near perfect, recommended.
CodeQwen1.5-7B-Chat-Q5_K_M.gguf Q5_K_M 5.42GB High quality, recommended.
CodeQwen1.5-7B-Chat-Q5_K_S.gguf Q5_K_S 5.14GB High quality, recommended.
CodeQwen1.5-7B-Chat-Q4_K_M.gguf Q4_K_M 4.73GB Good quality, uses about 4.83 bits per weight, recommended.
CodeQwen1.5-7B-Chat-Q4_K_S.gguf Q4_K_S 4.41GB Slightly lower quality with more space savings, recommended.
CodeQwen1.5-7B-Chat-IQ4_NL.gguf IQ4_NL 4.18GB Decent quality, slightly smaller than Q4_K_S with similar performance recommended.
CodeQwen1.5-7B-Chat-IQ4_XS.gguf IQ4_XS 4.03GB Decent quality, smaller than Q4_K_S with similar performance, recommended.
CodeQwen1.5-7B-Chat-Q3_K_L.gguf Q3_K_L 3.98GB Lower quality but usable, good for low RAM availability.
CodeQwen1.5-7B-Chat-Q3_K_M.gguf Q3_K_M 3.80GB Even lower quality.
CodeQwen1.5-7B-Chat-IQ3_M.gguf IQ3_M 3.60GB Medium-low quality, new method with decent performance comparable to Q3_K_M.
CodeQwen1.5-7B-Chat-IQ3_S.gguf IQ3_S 3.50GB Lower quality, new method with decent performance, recommended over Q3_K_S quant, same size with better performance.
CodeQwen1.5-7B-Chat-Q3_K_S.gguf Q3_K_S 3.50GB Low quality, not recommended.
CodeQwen1.5-7B-Chat-IQ3_XS.gguf IQ3_XS 3.35GB Lower quality, new method with decent performance, slightly better than Q3_K_S.
CodeQwen1.5-7B-Chat-IQ3_XXS.gguf IQ3_XXS 3.22GB Lower quality, new method with decent performance, comparable to Q3 quants.
CodeQwen1.5-7B-Chat-Q2_K.gguf Q2_K 3.05GB Very low quality but surprisingly usable.
CodeQwen1.5-7B-Chat-IQ2_M.gguf IQ2_M 3.00GB Very low quality, uses SOTA techniques to also be surprisingly usable.
CodeQwen1.5-7B-Chat-IQ2_S.gguf IQ2_S 2.87GB Very low quality, uses SOTA techniques to be usable.
CodeQwen1.5-7B-Chat-IQ2_XS.gguf IQ2_XS 2.76GB Very low quality, uses SOTA techniques to be usable.
CodeQwen1.5-7B-Chat-IQ2_XXS.gguf IQ2_XXS 2.61GB Lower quality, uses SOTA techniques to be usable.
CodeQwen1.5-7B-Chat-IQ1_M.gguf IQ1_M 2.45GB Extremely low quality, not recommended.
CodeQwen1.5-7B-Chat-IQ1_S.gguf IQ1_S 2.36GB Extremely low quality, not recommended.

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
437
GGUF
Model size
7.25B params
Architecture
qwen2

1-bit

2-bit

3-bit

4-bit

5-bit

6-bit

8-bit

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