Trace：一个分类学引导的多领域视觉推理环境

摘要

基于可验证奖励的强化学习（RLVR）已显著提升语言模型的推理能力，但其在视觉语言模型上的扩展仍受限于缺乏同时具备广泛性、精确可验证性和可复现性的训练数据。我们提出Trace——一种基于分类法指导的多领域视觉推理环境。Trace将任务构建分解为场景语法与可执行任务程序，从而将视觉实现与答案计算相分离。共享语义状态决定了渲染图像、提示词、类型化答案、验证器状态以及可回放的实例轨迹。该环境最终包含跨越277种场景语法和11个视觉领域的1000个任务，语义与视觉变异受控。在64000个Trace实例上进行的RLVR实验，使得Qwen2.5-VL-3B在24个外部基准测试中的宏观平均值提升了3.51个百分点，Qwen2.5-VL-7B提升了4.06个百分点，这表明广泛的程序化训练能够迁移至生成任务分布之外。项目页面：https://maveryn.github.io/trace/。

English

Reinforcement learning with verifiable rewards (RLVR) has substantially improved language-model reasoning, yet its extension to vision-language models remains constrained by the lack of training data that are simultaneously broad, exactly verifiable, and reproducible. We introduce Trace, a taxonomy-guided environment for multidomain visual reasoning. Trace factorizes task construction into a scene grammar and an executable task program, separating visual realization from answer computation. A shared semantic state determines the rendered image, prompt, typed answer, verifier state, and replayable instance trace. The resulting environment comprises 1,000 tasks over 277 scene grammars and 11 visual domains, with controlled semantic and visual variation. RLVR on 64,000 Trace instances improves the macro-average across 24 external benchmarks by 3.51 percentage points for Qwen2.5-VL-3B and 4.06 points for Qwen2.5-VL-7B, providing evidence that broad procedural training can transfer beyond the generated task distributions. Project page: https://maveryn.github.io/trace/.