google/gemma-4-31B-it

Note

This is a remotely-sourced model from the gemma4 plugin, maintained by the community. It is not part of FiftyOne core and may have special installation requirements. Please review the plugin documentation and license before use.

Gemma 4 31B is a 30.7B dense multimodal model supporting text and image. Image operations: detect, point, classify, vqa. No video support..

Details

• Model name: google/gemma-4-31B-it

• Model source: https://huggingface.co/google/gemma-4-31B-it

• Model author: Google DeepMind

• Model license: gemma (https://ai.google.dev/gemma/terms)

• Exposes embeddings? no

• Tags: detection, classification, keypoints, vqa, torch, zero-shot, image, VLM

Requirements

• Packages: huggingface-hub, transformers>=4.52.0, torch, torchvision, accelerate, torchcodec

• CPU support

  • yes

• GPU support

  • yes

Example usage

import fiftyone as fo
import fiftyone.zoo as foz

foz.register_zoo_model_source("https://github.com/Burhan-Q/gemma4")

dataset = foz.load_zoo_dataset(
    "coco-2017",
    split="validation",
    dataset_name=fo.get_default_dataset_name(),
    max_samples=50,
    shuffle=True,
)

model = foz.load_zoo_model("google/gemma-4-31B-it")

dataset.apply_model(model, label_field="predictions")

session = fo.launch_app(dataset)