Alibaba releases MIPR: reasoning-intensive multimodal intent-to-product retrieval benchmark
_reachsumit · x · 2026-11-18
Alibaba introduces MIPR, a benchmark for multimodal intent-to-product retrieval targeting real e-commerce queries that express intent without naming a product and can be solved by items from multiple categories.
- 3,522 multimodal queries over a 30,731-product corpus, with reasoning-intensive retrieval, multi-category solutions and fine-grained relevance labels
- Best embedding retriever reaches only 49.23% nDCG@10 overall and 55.86% solution-category recall@10 on multi-solution queries; reranking lifts these to 60.11% and 67.00%, still far from the oracle
- Data and code are publicly available, making MIPR a challenging diagnostic benchmark
More from Research
- Voyager 1 Reaches One Light-Day From Earth, Signals Now Take 24 Hours — snowclipsed · 2026-11-18
- TCS Hiring Statement Urges Quality Over Quantity as AI Fuels Paper Arms Race — fortnow · 2026-11-18
- New paper: a sociocultural approach to linguistic creativity — abenitezburraco · 2026-11-18
- 27B Model Rewrites Its Own Weights Without Training or Gradients, Paper Claims — udmrzn · 2026-11-18
- From 'impossible' to a working Bluetooth beacon: open-source EDA chip design is real — IgorCarron · 2026-11-18
- AURA 2.0: a streaming video LLM that speaks first, calls tools, and grounds visually — Shuhan Li · 2026-11-18