HopitAI/moda-duo
Feature Extraction • Updated
Text-to-image retrieval
Find products from a written description.
Note Routes each query to MODA Pro Lite+ or MODA by its shape. Beats either alone on a mixed workload; one encoder, one ANN query per search.
HopitAI/moda-pro-lite-plus
Feature Extraction • UpdatedNote MODA Pro Lite+: the encoder with its calibrated serving recipe. Leads the ≤250M class on catalogue and title search.
HopitAI/moda-pro-lite
Feature Extraction • Updated • 111Note MODA Pro Lite: the bare 213M encoder. Open weights.
HopitAI/moda-fashionsiglip-multiview-203m
Image Feature Extraction • UpdatedNote MODA: FashionSigLIP with a serving recipe, zero new parameters. Strongest open model on long descriptions.