Google Gemma 3 12b It Qat GGUF

quantized_by: bartowski pipeline_tag: image-text-to-text tags:

• gemma3
• gemma
• google license: gemma extra_gated_button_content: Acknowledge license base_model_relation: quantized extra_gated_heading: Access Gemma on Hugging Face extra_gated_prompt: >- To access Gemma on Hugging Face, you’re required to review and agree to Google’s usage license. To do this, please ensure you’re logged in to Hugging Face and click below. Requests are processed immediately. base_model: google/gemma-3-12b-it-qat-q4_0-unquantized

Llamacpp imatrix Quantizations of gemma-3-12b-it-qat by google

These are derived from the QAT (quantized aware training) weights provided by Google

ONLY Q4_0 is expected to be better, but figured while I'm at it I might as well make others to see what happens?

gemma-3-12b-it-qat-Q4_0.gguf | Q4_0 | 6.91GB | false | Should be improved due to QAT, offers online repacking for ARM and AVX CPU inference.

Using llama.cpp release b5147 for quantization.

Original model: https://huggingface.co/google/gemma-3-12b-it-qat-q4_0-unquantized

All quants made using imatrix option with dataset from here

Run them in LM Studio

Run them directly with llama.cpp, or any other llama.cpp based project

Prompt format

<bos><start_of_turn>user
{system_prompt}

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

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

Embed/output weights

Some of these quants (Q3_K_XL, Q4_K_L etc) are the standard quantization method with the embeddings and output weights quantized to Q8_0 instead of what they would normally default to.

Downloading using huggingface-cli

pip install -U "huggingface_hub[cli]"

huggingface-cli download bartowski/google_gemma-3-12b-it-qat-GGUF --include "google_gemma-3-12b-it-qat-Q4_K_M.gguf" --local-dir ./

huggingface-cli download bartowski/google_gemma-3-12b-it-qat-GGUF --include "google_gemma-3-12b-it-qat-Q8_0/*" --local-dir ./

ARM/AVX information

Previously, you would download Q4_0_4_4/4_8/8_8, and these would have their weights interleaved in memory in order to improve performance on ARM and AVX machines by loading up more data in one pass.

Now, however, there is something called "online repacking" for weights. details in this PR. If you use Q4_0 and your hardware would benefit from repacking weights, it will do it automatically on the fly.

As of llama.cpp build b4282 you will not be able to run the Q4_0_X_X files and will instead need to use Q4_0.

Additionally, if you want to get slightly better quality for , you can use IQ4_NL thanks to this PR which will also repack the weights for ARM, though only the 4_4 for now. The loading time may be slower but it will result in an overall speed incrase.

Which file should I choose?

Credits

Thank you kalomaze and Dampf for assistance in creating the imatrix calibration dataset.

Thank you ZeroWw for the inspiration to experiment with embed/output.

Thank you to LM Studio for sponsoring my work.

Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski