|
--- |
|
library_name: transformers |
|
license: other |
|
--- |
|
|
|
# Theia |
|
|
|
[The AI Institute](https://theaiinstitute.com/) |
|
|
|
Theia is a vision foundation model for robot learning that distills multiple off-the-shelf vision foundation models trained on varied vision tasks. Theia’s rich visual representations encode diverse visual knowledge, enhancing downstream robot learning. It was introduced in the paper [Theia: Distilling Diverse Vision Foundation Models for Robot Learning](https://arxiv.org/abs/2407.20179), which also includes experiments demonstrating that Theia outperforms its teacher |
|
models and prior robot learning models using less training data and smaller model sizes. Demo videos can be found on the [project page](http://theia.theaiinstitute.com/). |
|
|
|
<img src="https://raw.githubusercontent.com/bdaiinstitute/theia/main/doc/theia_overview.gif" height="300px"> |
|
|
|
|
|
## Model Details |
|
|
|
The `theia-small-patch16-224-cdiv` model, uses [DeiT-Small](https://huggingface.co/facebook/deit-small-patch16-224) as a backbone, and simulatenously distills [CLIP](https://github.com/openai/CLIP), [DINOv2](https://github.com/facebookresearch/dinov2), and [ViT](https://github.com/google-research/vision_transformer). For more information on usage, please visit the [Theia repository](https://github.com/bdaiinstitute/theia/tree/main). |
|
|
|
## Citation |
|
If you use Theia in your research, please use the following BibTeX entry: |
|
```bibtex |
|
@article{shang2024theia, |
|
author = {Shang, Jinghuan and Schmeckpeper, Karl and May, Brandon B. and Minniti, Maria Vittoria and Kelestemur, Tarik and Watkins, David and Herlant, Laura}, |
|
title = {Theia: Distilling Diverse Vision Foundation Models for Robot Learning}, |
|
journal = {arXiv}, |
|
year = {2024}, |
|
} |
|
``` |
|
|
|
## Usage |
|
|
|
The pre-trained model weights and code released with Theia are available for use under [The AI Institute License](https://raw.githubusercontent.com/bdaiinstitute/theia/main/LICENSE), reproduced in full below: |
|
``` |
|
Copyright (c) 2024 Boston Dynamics AI Institute LLC |
|
|
|
Redistribution and use in source and binary forms, with or without |
|
modification, are permitted provided that the following conditions are met: |
|
1. Redistributions of source code must retain the copyright notice included |
|
with the software, this list of conditions and the following disclaimer. |
|
2. Redistributions in binary form must reproduce the copyright notice, this |
|
list of conditions and the following disclaimer in the documentation and/or |
|
other materials provided with the distribution. |
|
3. Modified versions of the software must be conspicuously marked as such. |
|
4. The software may only be used for non-commercial research purposes. |
|
For profit enterprises may use the software, subject to this limitation. |
|
|
|
THIS SOFTWARE IS PROVIDED BY THE AI INSTITUTE AND CONTRIBUTORS "AS IS" AND |
|
ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, NON- |
|
INFRINGEMENT,TITLE, MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE |
|
DISCLAIMED. IN NO EVENT SHALL THE AI INSTITUTE OR CONTRIBUTORS BE LIABLE FOR |
|
ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, PUNITIVE OR CONSEQUENTIAL |
|
DAMAGES (INCLUDING, BUT NOT LIMITED TO, DAMAGES ARISING OUT OF CLAIMS OF |
|
INTELLECTUAL PROPERTY RIGHTS INFRINGEMENT; PROCUREMENT OF SUBSTITUTE GOODS OR |
|
SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER |
|
CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, |
|
OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
|
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
|
``` |
