This model repo provides the Stable Diffusion model finetuned on the images from the Something-Something-v2, Epic Kitchen and the Bridge V2 datasets. This repo is related to the
stable-control-representations GitHub repo.
Vision- and language-guided embodied AI requires a fine-grained understanding of the physical world through language and visual inputs. Such capabilities are difficult to learn solely from task-specific data, which has led to the emergence of pre-trained vision-language models as a tool for transferring representations learned from internet-scale data to downstream tasks and new domains. However, commonly used contrastively trained representations such as in CLIP have been shown to fail at enabling embodied agents to gain a sufficiently fine-grained scene understanding—a capability vital for control. To address this shortcoming, we consider representations from pre-trained text-to-image diffusion models, which are explicitly optimized to generate images from text prompts and as such, contain text-conditioned representations that reflect highly fine-grained visuo-spatial information. Using pre-trained text-to-image diffusion models, we construct Stable Control Representations which allow learning downstream control policies that generalize to complex, open-ended environments. We show that policies learned using Stable Control Representations are competitive with state-of-the-art representation learning approaches across a broad range of simulated control settings, encompassing challenging manipulation and navigation tasks. Most notably, we show that Stable Control Representations enable learning policies that exhibit state-of-the-art performance on OVMM, a difficult open-vocabulary navigation benchmark
If you use SCR in your research, please cite
the following paper:
1@inproceedings{gupta2024scr,
2 title={Pre-trained Text-to-Image Diffusion Models Are Versatile Representation Learners for Control},
3 author={Gunshi Gupta and Karmesh Yadav and Yarin Gal and Dhruv Batra and Zsolt Kira and Cong Lu and Tim G. J. Rudner},
4 year={2024},
5 eprint={2405.05852},
6 archivePrefix={arXiv},
7 primaryClass={cs.CV}
8}
We are thankful to the creators of
Stable Diffusion for releasing the model, which has significantly contributed to the progress in the field. Additionally, we extend our thanks to the authors of
Visual Cortex for releasing the code for CortexBench evaluations.
