The latest Data Phoenix news tagged information-retrieval.
Hugging Face published a method that trains retrieval models on full and pooled document representations, aiming to preserve more search quality…
Cohere's RCP-nDCG@10 combines rubric judgments, pairwise preferences and cross-query calibration to address gaps in conventional relevance labels.
Google Research's Retrieve-for-Train shifts reward-driven query fan-out into offline training, then serves results through a compact diffusion…
Perplexity's Q2D-Web benchmark tests first-stage retrieval across a company-reported 190 million web documents and 69,721 agent-generated queries,…