ElasticTok-VAE Model Card

Model details

Model type: ElasticTok-VAE is an open-source vision tokenizer train on various internet images and video.

The model is a Jax checkpoint. Inference code and instructions can be found at: https://github.com/LargeWorldModel/ElasticTok

Model date: ElasticTok-VAE was trained in August 2024.

Paper or resources for more information: https://largeworldmodel.github.io/elastictok

This model is licensed under Apache 2.0.

Where to send questions or comments about the model: https://github.com/LargeWorldModel/ElasticTok/issues