It also provides the Q4_0_8_8, Q4_0_4_8 and Q4_0_4_4 ggufs for edge devices that were otherwise not made by bartowski. These models should also be good for edge devices with 16GB RAM.

Prompt format

<start_of_turn>user
{prompt}<end_of_turn>
<start_of_turn>model
<end_of_turn>
<start_of_turn>model

Note that this model does not support a System prompt.

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

ELIZA-Tasks-100 is pretty standard benchmark for Japanese LLMs. The perfect score is 5.00. As a reference, bartowski's gemma-2-27b-it.Q6_K.gguf scores 4.04.

Filename	Quant type	File Size	ELIZA-Tasks-100	Nvidia 3090	Description
gemma-2-9b-it.f16.gguf	f16	18.5GB	3.75	31.9t/s	Full F16 weights.
gemma-2-9b-it.Q8_0.gguf	Q8_0	9.83GB	3.66	56.1t/s	Extremely high quality, recommended for edge devices with 16GB RAM.
gemma-2-9b-it-imatrix.Q4_0.gguf	Q4_0	5.44GB	3.76	80.6t/s	Good quality, recommended for edge devices wth 8GB RAM.
gemma-2-9b-it-imatrix.Q4_0_8_8.gguf	Q4_0_8_8	5.44GB	3.74	0.7t/s	Good quality, recommended for edge devices with 8GB RAM.
gemma-2-9b-it-imatrix.Q4_0_4_8.gguf	Q4_0_4_8	5.44GB	3.64	0.7t/s	Good quality, recommended for edge devices with 8GB RAM.
gemma-2-9b-it-imatrix.Q4_0_4_4.gguf	Q4_0_4_4	5.44GB	3.72	0.72t/s	Good quality, recommended for edge devices with 8GB RAM.
gemma-2-9b-it.Q4_0.gguf	Q4_0	5.44GB	3.64	65.1t/s	Good quality, recommended for edge device with 8GB RAM
gemma-2-9b-it.Q4_0_8_8.gguf	Q4_0_8_8	5.44GB	3.64	0.57t/s	Good quality but imatrix version seems better.
gemma-2-9b-it.Q4_0_4_8.gguf	Q4_0_4_8	5.44GB	3.68	0.61t/s	Good quality but imatrix version seems better.
gemma-2-9b-it.Q4_0_4_4.gguf	Q4_0_4_4	5.44GB	3.63	0.76t/s	Good quality but imatrix version seems better.

How to check i8mm and sve support for ARM devices

ARM i8mm support is necessary to take advantage of Q4_0_4_8 gguf. All ARM architecture >= ARMv8.6-A supports i8mm.

ARM sve support is necessary to take advantage of Q4_0_8_8 gguf. sve is an optional feature that starts from ARMv8.2-A but majority of ARM chips doesn't implement it.

For ARM devices without both, it is recommended to use Q4_0_4_4. However, in reality, Q4_0 can perform better for some phones, so you better try both and see which one is better.

With these support, the inference speed should be faster in the order of Q4_0_8_8 > Q4_0_4_8 > Q4_0_4_4 > Q4_0 without much effect on the quality of response.

This is a list of ARM CPUs that support different ARM instructions. Another list. Apparently, they only cover limited number of ARM CPUs. It is better you check for i8mm and sve support by yourself.

For Apple devices,

sysctl hw

For other ARM devices (ie most Android devices),

cat /proc/cpuinfo

There are also android apps that can display /proc/cpuinfo.

I was told that for Intel/AMD CPU inference, support for AVX2/AVX512 can also improve the performance of Q4_0_8_8.

On the other hand, Nvidia 3090 inference speed is significantly faster for Q4_0 than the other ggufs. That means for GPU inference, you better off using Q4_0.

Which Q4_0 model to use for ARM devices

Brand	Series	Model	i8mm	sve	Quant Type
Apple	A	A4 to A14	No	No	Q4_0_4_4
Apple	A	A15 to A18	Yes	No	Q4_0_4_8
Apple	M	M1	No	No	Q4_0_4_4
Apple	M	M2/M3/M4	Yes	No	Q4_0_4_8
Google	Tensor	G1,G2	No	No	Q4_0_4_4
Google	Tensor	G3,G4	Yes	Yes	Q4_0_8_8
Samsung	Exynos	2200,2400	Yes	Yes	Q4_0_8_8
Mediatek	Dimensity	9000,9000+	Yes	Yes	Q4_0_8_8
Mediatek	Dimensity	9300	Yes	No	Q4_0_4_8
Qualcomm	Snapdragon	7+ Gen 2,8/8+ Gen 1	Yes	Yes	Q4_0_8_8
Qualcomm	Snapdragon	8 Gen 2,8 Gen 3,X Elite	Yes	No	Q4_0_4_8

imatrix quantization

According to this blog, adding imatrix to low bit quant can significantly improve performance. The best dataset for Japanese is MTFMC/imatrix-dataset-for-japanese-llm. Therefore, I also created the imatrix versions of different Q4_0 quants.

However, based on my benchmarking results, it seems like imatrix does improve the performance of a non-Japanese optimized model but doesn't do much for a Japanese optimized model like gemma-2-2b-jpn-it.

Convert safetensors to f16 gguf

Make sure you have llama.cpp git cloned:

python3 convert_hf_to_gguf.py gemma-2-9b-it/ --outfile gemma-2-9b-it.f16.gguf --outtype f16

Convert f16 gguf to Q8_0 gguf without imatrix

Make sure you have llama.cpp compiled:

./llama-quantize gemma-2-9b-it.f16.gguf gemma-2-9b-it.Q8_0.gguf q8_0

Convert f16 gguf to other ggufs with imatrix

First, prepare imatrix from f16 gguf and c4_en_ja_imatrix.txt

./llama-imatrix -m gemma-2-9b-it.f16.gguf -f c4_en_ja_imatrix.txt -o gemma-2-9b-it.imatrix --chunks 32

Then, convert f16 gguf with imatrix to create imatrix gguf

./llama-quantize --imatrix gemma-2-9b-it.imatrix gemma-2-9b-it.f16.gguf gemma-2-9b-it-imatrix.Q4_0_8_8.gguf q4_0_8_8

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 ymcki/gemma-2-9b-it-GGUF --include "gemma-2-9b-it.Q8_0.gguf" --local-dir ./

Credits

Thank you bartowski for providing a README.md to get me started.

Thank you YoutechA320U for the ELYZA-tasks-100 auto evaluation tool.