LeVJEPA

LeVJEPA produces clip embeddings and spatial patch tokens from video.

Tasks: embeddings
Sizes: l at 224 px
Install: pip install libreyolo
Support tier: Inference only, since v1.6.0. Predict, validate and export only. Training features do not apply.
Upstream: LeVJEPA by MLO Lab, MIT code; CC-BY-NC-4.0 weights. Paper, source
Licenses: Code MIT, weights MIT code; CC-BY-NC-4.0 weights. Commercial use

Install

bash

pip install "libreyolo"

Predict

Python

from libreyolo import LibreYOLO, SAMPLE_IMAGE model = LibreYOLO("LibreLeVJEPAl-embed.pt", device="cpu")result = model(SAMPLE_IMAGE)print(result.embeddings)print(model.embed_tokens(SAMPLE_IMAGE).shape)

The encoder uses 16 frames at 224 pixels. Finite videos use a centered window sampled at approximately 7.5 FPS. embed_tokens() exposes patch tokens. Training is not supported. The TorchScript graph requires direct clip input with batch 1. Pretrained weights are CC-BY-NC-4.0.

Export

Task	ONNX	TorchScript	ExecuTorch	TensorRT	OpenVINO	Paddle	MNN	RKNN	ncnn	TFLite	CoreML	Core AI
Embeddings	Embeddings to ONNX: supported	Embeddings to TorchScript: supported	Embeddings to ExecuTorch: supported	Embeddings to TensorRT: supported	Embeddings to OpenVINO: supported	Embeddings to Paddle: not supported	Embeddings to MNN: not supported	Embeddings to RKNN: not supported	Embeddings to ncnn: not supported	Embeddings to TFLite: not supported	Embeddings to CoreML: not supported	Embeddings to Core AI: not supported

Python

from libreyolo import LibreYOLO, SAMPLE_IMAGE model = LibreYOLO("LibreLeVJEPAl-embed.pt", device="cpu")model.export(format="onnx")

Export setup lists format dependencies and loading exported artifacts.

Checkpoints

File	Weights license
Embeddings
LibreLeVJEPAl-embed.pt	cc-by-nc-4.0

Every file above exists in the LibreYOLO org today and downloads on first use.

Licensing

Check the license on the Hugging Face repository of the specific weights you download. Every checkpoint in the LibreYOLO org carries one, and they are not always the same across a family. That repository is the authoritative source; the summary below describes what applied when this page was last verified.

This is a description of the licenses involved, not legal advice. If the answer matters commercially, read the licenses yourself and take your own counsel.

Original work: LeVJEPA, MLO Lab
Upstream license: MIT code; CC-BY-NC-4.0 weights
Upstream source: github.com/MLO-lab/LeVJEPA
LibreYOLO code: MIT
Weights: MIT code; CC-BY-NC-4.0 weights, republished at huggingface.co/LibreYOLO
Interpretation: The repository declares an explicit CC-BY-NC-4.0 exception for pretrained weights. Commercial use of those weights is excluded.

Citation

@misc{kuhn2026levjepaefficientscalable,
      title={LeVJEPA: Efficient & Scalable Video Pretraining without the Heuristics}, 
      author={Lukas Kuhn and Lucas Maes and Giuseppe Serra and Quentin Le Lidec and Yann LeCun and Randall Balestriero and Florian Buettner},
      year={2026},
      eprint={2608.27395},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2608.27395}, 
}

Copied from the authors' citation block at raw.githubusercontent.com/MLO-lab/LeVJEPA/main/README.md.