--- title: PyTerrier SPLADE emoji: 🐕 colorFrom: green colorTo: green sdk: gradio sdk_version: 3.7 app_file: app.py pinned: false --- This is a demonstration of [PyTerrier's SPLADE package](https://github.com/cmacdonald/pyt_splade). The SPLADE model encodes queries and documents into sparse representations, which can then be used for indexing and retrieval. ### Query Encoding Let's start by exploring SPLADE's query encoder. The query encoder is a `Q→Q` (query rewriting, query-to-query) transformer, and can be used in pipelines accordingly. It maps a query string into [MatchOp](https://terrier-core.readthedocs.io/en/latest/querylanguage.html#matching-op-query-language) query with terms from the query re-weighted and weighted expansion terms added.
Q
SPLADE
Q