Skip to content
View original post on X: Perplexity DevelopersOfficial
Published
31/100

Perplexity trains embedding models on distilled chunk-level relevance scores

AIPerplexity says its contextual embedding models are trained differently from other contextual embeddings models. Instead of one gold chunk per query, chunk-level relevance scores are distilled from a query-aware context compression model in its answer pipeline, and supporting chunks are no longer treated as negatives.
Perplexity DevelopersVerified on XOfficial
@perplexitydevs
View on X

Part of a thread · earlier post

The models are trained differently from other contextual embeddings models.

Instead of one gold chunk per query, chunk-level relevance scores are distilled from a query-aware context compression model in Perplexity's answer pipeline. Supporting chunks are no longer negatives.

Source details

Perplexity Developers · x.comPublished · added here

Original title: The models are trained differently from other contextual embeddings models.

Topics

Data & trainingPapers & research

Tags

#RAG#Search