moonshotai/Kimi-VL-A3B-Instruct

Note

This is a remotely-sourced model from the Kimi_VL_A3B 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.

Kimi-VL is an efficient open-source Mixture-of-Experts (MoE) vision-language model (VLM) that offers advanced multimodal reasoning, long-context understanding, and strong agent capabilities—all while activating only 2.8B parameters in its language decoder.

Details

• Model name: moonshotai/Kimi-VL-A3B-Instruct

• Model source: https://huggingface.co/moonshotai/Kimi-VL-A3B-Instruct

• Model author: Moonshot AI

• Model license: MIT license

• Exposes embeddings? no

• Tags: agentic, detection, ocr, VLM, classification, zero-shot, visual-agent

Requirements

• Packages: huggingface-hub, transformers, torch, torchvision, accelerate, tiktoken, protobuf, blobfile

• 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/harpreetsahota204/Kimi_VL_A3B")

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("moonshotai/Kimi-VL-A3B-Instruct")

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

session = fo.launch_app(dataset)