| --- |
| license: apache-2.0 |
| pipeline_tag: feature-extraction |
| --- |
| |
| # SOLAR: Self-supervised Joint Learning for Symmetric Multimodal Retrieval |
|
|
| Models of ICML 2026 paper [SOLAR: Self-supervised Joint Learning for Symmetric Multimodal Retrieval](https://arxiv.org/abs/2605.15868), including SOLAR-C and SOLAR-B+D (both after stage 2). |
|
|
| To inference, please refer to [our code](https://github.com/codefuse-ai/SOLAR). |
|
|
| For benchmark, please refer to [sym-MM2MM](https://huggingface.co/datasets/yangwj211/sym-MM2MM). |
|
|
| ## Citation |
|
|
| ```bibtex |
| @article{yang2026solar, |
| title={SOLAR: Self-supervised Joint Learning for Symmetric Multimodal Retrieval}, |
| author={Yang, Wenjie and Yu, Hang and Guo, Yuyu and Di, Peng}, |
| journal={arXiv preprint arXiv:2605.15868}, |
| year={2026} |
| } |
| ``` |
