← 研踪 PaperTrailarXiv:2611.25653 · v1
ColorRef: Calibrating Verbal Feedback for Color Revision
首次提交 2026-11-19 15:35
版本更新 2026-11-19 15:35
雷达观测 2026-11-20 19:58
摘要
People rarely specify an exact color when asking someone to revise a design: they might say \textit{a little darker} or \textit{more blue} after seeing a proposal. A language-model assistant must then infer not only which way to change the color, but how far. We study this distinction with \textsc{ColorRef}, a color reference game in which a model guesses a color from a description and a teacher, knowing a recorded target hidden from the guesser, supplies verbal corrections. On 1{,}000 crowd-sourced description--target pairs, three rounds of deterministic oracle feedback reduce mean target distance from 42.33 to 22.15 . However, those corrections provide additional information, and improvement alone does not establish well-calibrated feedback following. A controlled Qwen3-14B diagnostic reveals a more specific failure: following a requested direction and ordering graded phrases do not guarantee a suitable step size; strong modifiers can induce excessive changes. We therefore measure each receiver's typical response to \textit{a little}, \textit{somewhat}, and \textit{much} on separate colors and select wording matched to its observed steps. On 128 held-out starting colors, calibration reduces one-step mean error from 12.36 to 6.86 for Qwen3-14B and from 11.06 to 5.89 for Gemma 4 12B against a fixed graded-wording rule. The benefit is concentrated on large corrections, with many unchanged cases and no consistent increase in threshold success. Our findings show that following the direction of feedback is not the same as making an appropriately sized revision.
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-19 15:35 · 最新观测版本
ColorRef: Calibrating Verbal Feedback for Color Revision
People rarely specify an exact color when asking someone to revise a design: they might say \textit{a little darker} or \textit{more blue} after seeing a proposal. A language-model assistant must then infer not only which way to change the color, but how far. We study this distinction with \textsc{ColorRef}, a color reference game in which a model guesses a color from a description and a teacher, knowing a recorded target hidden from the guesser, supplies verbal corrections. On 1{,}000 crowd-sourced description--target pairs, three rounds of deterministic oracle feedback reduce mean target distance from 42.33 to 22.15 . However, those corrections provide additional information, and improvement alone does not establish well-calibrated feedback following. A controlled Qwen3-14B diagnostic reveals a more specific failure: following a requested direction and ordering graded phrases do not guarantee a suitable step size; strong modifiers can induce excessive changes. We therefore measure each receiver's typical response to \textit{a little}, \textit{somewhat}, and \textit{much} on separate colors and select wording matched to its observed steps. On 128 held-out starting colors, calibration reduces one-step mean error from 12.36 to 6.86 for Qwen3-14B and from 11.06 to 5.89 for Gemma 4 12B against a fixed graded-wording rule. The benefit is concentrated on large corrections, with many unchanged cases and no consistent increase in threshold success. Our findings show that following the direction of feedback is not the same as making an appropriately sized revision.
查看此版本解读 →查看此版本原文 ↗