[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-spader-tackles-the-harder-qa-problem-finding-every-answer":10,"sections":34},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":24,"tags":25,"sources":29,"feedback":33,"feedback_at":22,"cost_usd":33,"total_tokens":33},3636,"spader-tackles-the-harder-qa-problem-finding-every-answer","SPADER Tackles the Harder QA Problem: Finding Every Answer","A new RL framework helps AI agents discover rare, long-tail answers instead of stopping at the most obvious ones.","Most AI benchmarks reward finding *an* answer. SPADER is built for finding *all* of them.\n\nResearchers have released SPADER, a reinforcement learning framework designed for Multi-Answer QA — the task of retrieving a comprehensive set of valid answers to a single question, not just the first plausible hit. The system addresses two problems that trip up existing tool-using agents: figuring out which steps in a long search actually deserved credit, and keeping the agent hunting for rare answers instead of piling up duplicates of common ones. SPADER's Step-wise Peer Advantage mechanism compares parallel search trajectories step-by-step and estimates how good each decision was relative to its peers — no separate critic model required. A second component, a diversity-aware exploration reward, pushes the agent toward obscure entities by boosting the value of rare finds and penalizing redundant ones.\n\nThe gap between single-answer and multi-answer retrieval matters more than benchmarks suggest. Real queries — \"list every country that ratified X\" or \"name all films directed by Y\" — need exhaustive recall, and current RL approaches optimized for single correct answers quietly fail at the tail. SPADER's experiments across four datasets (QAMPARI, Mintaka, WebQSP, and QUEST) show consistent recall and F1 gains over prompting baselines, outcome-supervised RL, and recent step-level methods.\n\nCode and model weights are public on GitHub, which is the right move — claims about recall improvements on academic benchmarks are worth nothing until the community stress-tests them on messier real-world data.","[\"ai\",\"reinforcement-learning\",\"research\",\"nlp\"]","2026-07-03T04:00:00.000Z","2026-07-03T10:40:12.983Z","2026-07-03T10:40:15.866Z","published",null,[],"ai",[24,26,27,28],"reinforcement-learning","research","nlp",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2606.00593",0,{"sections":35},[36,40,45,50,55,60,65,70,75,80,85,89,94,99],{"name":37,"slug":24,"count":38,"latest_published_at":39},"AI",2590,"2026-07-16T04:00:00.000Z",{"name":41,"slug":42,"count":43,"latest_published_at":44},"Security","security",294,"2026-07-15T19:59:48.000Z",{"name":46,"slug":47,"count":48,"latest_published_at":49},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":51,"slug":52,"count":53,"latest_published_at":54},"Policy","policy",158,"2026-07-16T00:02:48.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":59},"Hardware","hardware",122,"2026-07-14T19:46:26.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":64},"Consumer Tech","consumer-tech",93,"2026-07-13T13:20:48.000Z",{"name":66,"slug":67,"count":68,"latest_published_at":69},"Software","software",70,"2026-07-13T19:52:25.000Z",{"name":71,"slug":72,"count":73,"latest_published_at":74},"Science","science",66,"2026-07-10T10:29:37.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":79},"Dev Tools","dev-tools",59,"2026-07-07T04:00:00.000Z",{"name":81,"slug":82,"count":83,"latest_published_at":84},"Gaming","gaming",41,"2026-07-09T04:00:00.000Z",{"name":86,"slug":87,"count":83,"latest_published_at":88},"Startups","startups","2026-06-29T20:55:50.000Z",{"name":90,"slug":91,"count":92,"latest_published_at":93},"General","general",29,"2026-07-10T22:28:58.000Z",{"name":95,"slug":96,"count":97,"latest_published_at":98},"Reviews","reviews",20,"2026-06-24T12:00:01.000Z",{"name":100,"slug":101,"count":102,"latest_published_at":103},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]