OPEA/gemma-3-27b-it-AutoRound-gguf-q4-0

Model Details

This model is an int4 model with group_size 32 and symmetric quantization of google/gemma-3-27b-it generated by intel/auto-round algorithm.

Please follow the license of the original model.

How To Use

Requirements

Please follow the Build llama.cpp locally to install the necessary dependencies.

INT4 Inference

This model has vision capabilities, more details here: https://github.com/ggml-org/llama.cpp/pull/12344

After building with Gemma 3 clip support, run the following command:

>>> wget https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg
>>> ./build/bin/llama-gemma3-cli -m gemma-3-27b-it-27B-Q4_0.gguf --mmproj mmproj-gemma-3-27b-it-27B-Q4_0.gguf

Output:

 Running in chat mode, available commands:
   /image <path>    load an image
   /clear           clear the chat history
   /quit or /exit   exit the program
> /image bee.jpg                
Encoding image bee.jpg

> Describe this image in detail.
Here's a detailed description of the image:

**Overall Impression:**

The image is a close-up shot of a vibrant garden scene, focusing on pink cosmos flowers and a busy bumblebee. The composition is natural and slightly wild, with a mix of blooming and fading flowers.

**Foreground:**

*   **Cosmos Flowers:** The most prominent feature is a cluster of pink cosmos flowers. They have delicate, ray-like petals in a soft, pastel pink hue. The central disc of the flower is a golden yellow.
*   **Bumblebee:** A bumblebee is actively foraging on the central disc of one of the cosmos flowers. Its body is fuzzy and striped with black and yellow. The bee is positioned in the center of the image, drawing the eye.
*   **Fading Flowers:** There are several spent or fading cosmos flowers surrounding the fresh blooms. These are brown and dried, adding texture and a sense of the natural life cycle.
*   **Greenery:** Green leaves and stems are interspersed among the flowers, providing a contrast in color and texture.

**Background:**

*   **Blurred Foliage:** The background is blurred, consisting of green leaves and hints of other flowers. This creates depth and focuses attention on the foreground.
*   **Red Flower:** A small, bright red flower is visible in the lower right corner, adding a pop of color.

**Lighting and Composition:**

*   **Natural Light:** The image appears to be taken in natural daylight.
*   **Shallow Depth of Field:** The shallow depth of field emphasizes the foreground flowers and bee, blurring the background.
*   **Natural Arrangement:** The flowers and bee are arranged in a natural, unposed manner, giving the image a sense of authenticity.

**Overall, the image is a charming depiction of a summer garden, highlighting the beauty of nature and the important role of pollinators like bumblebees.**

Generate the model

Here is the sample command to reproduce the model.

pip install git+https://github.com/intel/auto-round.git@main
auto-round-mllm \
--model google/gemma-3-27b-it \
--device 0 \
--bits 4 \
--group_size 32 \
--batch_size 1 \
--gradient_accumulate_steps 8 \
--format 'gguf:q4_0' \
--output_dir "./tmp_autoround"

Ethical Considerations and Limitations

The model can produce factually incorrect output, and should not be relied on to produce factually accurate information. Because of the limitations of the pretrained model and the finetuning datasets, it is possible that this model could generate lewd, biased or otherwise offensive outputs.

Therefore, before deploying any applications of the model, developers should perform safety testing.

Caveats and Recommendations

Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model.

Here are a couple of useful links to learn more about Intel's AI software:

• Intel Neural Compressor link

Disclaimer

The license on this model does not constitute legal advice. We are not responsible for the actions of third parties who use this model. Please consult an attorney before using this model for commercial purposes.

Cite

@article{cheng2023optimize, title={Optimize weight rounding via signed gradient descent for the quantization of llms}, author={Cheng, Wenhua and Zhang, Weiwei and Shen, Haihao and Cai, Yiyang and He, Xin and Lv, Kaokao and Liu, Yi}, journal={arXiv preprint arXiv:2309.05516}, year={2023} }

arxiv github