We're releasing pplx-embed-v2-late, two late-interaction embedding models that retrieve text, images, and pages with a shared embedding space for cross-model querying.
Both models achieve frontier performance and are publicly available on Hugging Face.
https://t.co/AM7YEyQkKE
Late-interaction embeddingA list of numbers representing a piece of text's meaning, so that similar meanings end up numerically close and can be searched.Full definition → that cover text, images and document pages in one shared space let you build multimodalA model that works with more than text — reading images, audio, or video, and sometimes generating them too.Full definition → retrieval without separate indexes. Weights are public on Hugging Face.
Terms in this piece · Glossary
embedding — A list of numbers representing a piece of text's meaning, so that similar meanings end up numerically close and can be searched.
multimodal — A model that works with more than text — reading images, audio, or video, and sometimes generating them too.
postTwo 350M retrieval models: one dense, one late interactionLiquid AI
articleMulti-Vector (Late Interaction) Embedding Models with Sentence Transformershuggingface.co
articleDenseOn with the LateOn: Fully Open Dense and Late-Interaction Models for Multilingual, Long-Context, and Code SearchRapha\"el Sourty, Antoine Chaffin, Paulo Roberto Moura Junior, Am\'elie Chatelain