We provide hyper-parameters of our models in Table A.1. Table A.1: Hyper-parameters used for training our VisualCSE and AudioCSE. Vision, we use Dropout augmentation (the same strategy in SimCSE) for AudioCSE. We compare unsup-SimCSE and unsup-VisualCSE on a small scale retrieval test. As shown in Table C.1, VisualCSE generally retrieves qualitatively different sentences than SimCSE.

Semantic representation learning for sentences is an important and well-studied problem in NLP. The current trend for this task involves training a Transformer-based sentence encoder through a contrastive objective with text, i.e., clustering sentences with semantically similar meanings and scattering others. In this work, we find the performance of Transformer models as sentence encoders can be improved by training with multi-modal multi-task losses, using unpaired examples from another modality (e.g., sentences and unrelated image/audio data). In particular, besides learning by the contrastive loss on text, our model clusters examples from a non-linguistic domain (e.g., visual/audio) with a similar contrastive loss at the same time. The reliance of our framework on unpaired non-linguistic data makes it language-agnostic, enabling it to be widely applicable beyond English NLP. Experiments on 7 semantic textual similarity benchmarks reveal that models trained with the additional non-linguistic (images/audio) contrastive objective lead to higher quality sentence embeddings. This indicates that Transformer models are able to generalize better by doing a similar task (i.e., clustering) with unpaired examples from different modalities in a multi-task fashion.