CoolFace
Modelpublic

bartowski/codegemma-7b-GGUF

sourceHugging Facegemmaupdated 2y agoView on Hugging Face
5likes413downloads
Model Card

Llamacpp Quantizations of codegemma-7b

Using <a href="https://github.com/ggerganov/llama.cpp/">llama.cpp</a> release <a href="https://github.com/ggerganov/llama.cpp/releases/tag/b2589">b2589</a> for quantization.

Original model: https://huggingface.co/google/codegemma-7b

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

Prompt format

No prompt template for this model.

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

FilenameQuant typeFile SizeDescription
codegemma-7b-Q8_0.ggufQ8_09.07GBExtremely high quality, generally unneeded but max available quant.
codegemma-7b-Q6_K.ggufQ6_K7.01GBVery high quality, near perfect, recommended.
codegemma-7b-Q5_K_M.ggufQ5KM6.14GBHigh quality, recommended.
codegemma-7b-Q5_K_S.ggufQ5KS5.98GBHigh quality, recommended.
codegemma-7b-Q4_K_M.ggufQ4KM5.32GBGood quality, uses about 4.83 bits per weight, recommended.
codegemma-7b-Q4_K_S.ggufQ4KS5.04GBSlightly lower quality with more space savings, recommended.
codegemma-7b-IQ4_NL.ggufIQ4_NL5.01GBDecent quality, slightly smaller than Q4KS with similar performance recommended.
codegemma-7b-IQ4_XS.ggufIQ4_XS4.76GBDecent quality, smaller than Q4KS with similar performance, recommended.
codegemma-7b-Q3_K_L.ggufQ3KL4.70GBLower quality but usable, good for low RAM availability.
codegemma-7b-Q3_K_M.ggufQ3KM4.36GBEven lower quality.
codegemma-7b-IQ3_M.ggufIQ3_M4.10GBMedium-low quality, new method with decent performance comparable to Q3KM.
codegemma-7b-IQ3_S.ggufIQ3_S3.98GBLower quality, new method with decent performance, recommended over Q3KS quant, same size with better performance.
codegemma-7b-Q3_K_S.ggufQ3KS3.98GBLow quality, not recommended.
codegemma-7b-IQ3_XS.ggufIQ3_XS3.80GBLower quality, new method with decent performance, slightly better than Q3KS.
codegemma-7b-IQ3_XXS.ggufIQ3_XXS3.48GBLower quality, new method with decent performance, comparable to Q3 quants.
codegemma-7b-Q2_K.ggufQ2_K3.48GBVery low quality but surprisingly usable.
codegemma-7b-IQ2_M.ggufIQ2_M3.13GBVery low quality, uses SOTA techniques to also be surprisingly usable.
codegemma-7b-IQ2_S.ggufIQ2_S2.91GBVery low quality, uses SOTA techniques to be usable.
codegemma-7b-IQ2_XS.ggufIQ2_XS2.81GBVery low quality, uses SOTA techniques to be usable.
codegemma-7b-IQ2_XXS.ggufIQ2_XXS2.58GBLower quality, uses SOTA techniques to be usable.
codegemma-7b-IQ1_M.ggufIQ1_M2.32GBExtremely low quality, not recommended.
codegemma-7b-IQ1_S.ggufIQ1_S2.16GBExtremely 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 'QXKX', like Q5KM.

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 IQXX, like IQ3M. 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