Multimodal Language ProcessingPEOPLE

To create a technology capable of comprehensively searching and explaining the real world, we are developing methods based on foundation models, such as large language models. These methods are being applied across various domains, including robotics, image analysis with textual data, autonomous driving, fashion, art, and more. We achieved first place in the CVPR2023 Embodied AI Workshop competition with our dialogue-enabled embodied instruction following model.