{"slug": "reducing-representational-interference-in-retrieval-embeddings-with-trainable", "title": "Reducing Representational Interference in Retrieval Embeddings with Trainable Prefix Tokens", "summary": "A proposed method prepends trainable [DUMP] tokens to input text to reduce representational interference in retrieval embedding models, addressing how fixed-size vector compression can entangle semantic content with residual token-level information such as artifacts from attention sink behavior. The approach targets the separation of semantic meaning from token-level noise in embeddings used for retrieval.", "body_md": "Embedding models compress variable-length text into fixed-size vector representations, but this compression can entangle semantic content with residual token-level information, such as artifacts arising from attention sink behavior. We propose a simple method that prepends trainable [DUMP] tokens to", "url": "https://wpnews.pro/news/reducing-representational-interference-in-retrieval-embeddings-with-trainable", "canonical_source": "https://aiflash.com/news/146758/", "published_at": "2026-11-06 12:30:22+00:00", "updated_at": "2026-11-06 12:52:51.743650+00:00", "lang": "en", "topics": ["machine-learning", "natural-language-processing", "ai-research", "artificial-intelligence"], "entities": [], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/reducing-representational-interference-in-retrieval-embeddings-with-trainable", "markdown": "https://wpnews.pro/news/reducing-representational-interference-in-retrieval-embeddings-with-trainable.md", "text": "https://wpnews.pro/news/reducing-representational-interference-in-retrieval-embeddings-with-trainable.txt", "jsonld": "https://wpnews.pro/news/reducing-representational-interference-in-retrieval-embeddings-with-trainable.jsonld"}}