{"version":1,"type":"story","url":"https://digestai.news/story/researchers-find-instruction-tuning-sharpens-number-choice-heuristics","json":"https://digestai.news/story/researchers-find-instruction-tuning-sharpens-number-choice-heuristics.json","markdown":"https://digestai.news/story/researchers-find-instruction-tuning-sharpens-number-choice-heuristics.md","slug":"researchers-find-instruction-tuning-sharpens-number-choice-heuristics","headline":"Researchers find Instruction tuning sharpens Number-Choice heuristics in language models","summary":"A new study on arXiv examines how instruction tuning affects the way language models pick numbers. The authors measured the probability of models selecting integers from 1 to N without sampling error, testing 18 open-weight model lines across five ranges up to N = 100. They found that instruction tuning reduces uniform distribution in all 15 models with matched base and tuned checkpoints, reinforcing two heuristics: avoiding round numbers and endpoints like 1 or N.\n\nThe study notes that base models do not avoid round numbers once endpoints are excluded, while tuned models nearly eliminate them. Tuned models also rarely pick endpoints, and a preference for odd numbers appears only at N = 100. The findings align partially with human number-choice behavior but differ in key ways, such as the absence of midpoint avoidance. The study concludes that tuning does not create a universal favorite number across models, as no integer exceeds 42% preference in tuned conditions.","keyPoints":["Instruction tuning reduces uniform number selection in 15 of 18 tested models, reinforcing heuristics like endpoint and round-number avoidance","Tuned models nearly eliminate round numbers and endpoints (1 or N) as choices, while base models do not show this behavior","Preference for odd numbers appears only at N = 100, with no shared favorite integer across tuned models"],"whyItMatters":"This research clarifies how instruction tuning influences decision-making in language models, offering insights into their biases and potential limitations in real-world applications like surveys or automated number generation.","category":{"slug":"research","name":"Research","url":"https://digestai.news/category/research"},"entities":{"companies":[],"models":[],"people":[]},"firstPublishedAt":"2026-10-28T04:00:00Z","updatedAt":"2026-10-28T04:00:00Z","sourceCount":1,"hasPrimarySource":true,"sources":[{"outlet":"arXiv cs.CL","title":"3, 7, 14, 17, 42: Instruction Tuning Installs and Sharpens Number-Choice Heuristics in Language Models","url":"https://arxiv.org/abs/2610.36178","publishedAt":"2026-10-28T04:00:00Z","type":"primary","primary":true,"lead":true}],"sourceNotes":null,"discussions":[],"thread":null,"cite":{"text":"Digest AI, \"Researchers find Instruction tuning sharpens Number-Choice heuristics in language models\", 28 October 2026, https://digestai.news/story/researchers-find-instruction-tuning-sharpens-number-choice-heuristics","publisher":"Digest AI","title":"Researchers find Instruction tuning sharpens Number-Choice heuristics in language models","datePublished":"2026-10-28T04:00:00Z","url":"https://digestai.news/story/researchers-find-instruction-tuning-sharpens-number-choice-heuristics"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}