Perplexitypplx-embed v2
pplx-embed-v2-context-0.6b
This model creates contextual embeddings for text, images and visual documents.
What the release provides
It supports contextual and independent input modes. The 0.6B model shares the 1,024-dimensional prefix space of the 9B model. Contextual document chunks must not overlap, and queries use a separate encoding method.pplx-embed-v2-context-0.6b official model card
How to use it well
Practical guidance. Use the documented query encoder when searching an index. Retain document and chunk boundaries so surrounding context does not mix unrelated sources.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | perplexity-ai/pplx-embed-v2-context-0.6b |