winnow-12b-torch#

12B multimodal decision model that answers yes-or-no, multiple-choice and rating questions about an image with a probability for every option.

Details

  • Model name: winnow-12b-torch

  • Model source: https://huggingface.co/EldanRing/Winnow-12B

  • Model author: EldanRing

  • Model license: Apache 2.0

  • Model size: 22.39 GB

  • Exposes embeddings? no

  • Tags: classification, vlm, torch, transformer, zero-shot

Requirements

  • Packages: torch, torchvision, transformers>=5.17, gguf>=0.10

  • CPU support

    • yes

  • GPU support

    • yes

Example usage

 1import fiftyone as fo
 2import fiftyone.zoo as foz
 3
 4dataset = foz.load_zoo_dataset(
 5    "coco-2017",
 6    split="validation",
 7    dataset_name=fo.get_default_dataset_name(),
 8    max_samples=50,
 9    shuffle=True,
10)
11
12model = foz.load_zoo_model("winnow-12b-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)