Fast SceneScript：基于多词元预测的快速准确语言驱动 3D 场景理解
Fast SceneScript: Fast and Accurate Language-Based 3D Scene Understanding via Multi-Token Prediction
对做 3D 场景理解的研究者，这篇给出了把自回归式结构化语言模型（SceneScript 类）的推理加速到每步最多 9 个词元的方案：多词元预测头 + 置信度引导的自推测解码，可直接迁移到任何以序列形式输出 3D 布局/检测框的模型上，且只增加 7.5% 参数。
原文依据 Fast SceneScript can generate up to 9 tokens per decoder inference step without compromising accuracy
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论近期基于语言模型的通用感知方法通过统一的架构与接口，在包括 3D 场景布局估计和 3D 目标检测在内的多种任务上取得了最优结果。然而，这些方法依赖自回归的下一词元预测，本质上速度较慢。本文提出 Fast SceneScript，一种用于准确且高效 3D 场景理解的新型结构化语言模型。该方法采用多词元预测（MTP）以减少自回归迭代次数，从而显著加速推理。尽管 MTP 提升了速度，但不可靠的词元预测会显著降低准确率。为过滤掉不可靠的词元，我们将自推测解码（SSD）适配到结构化语言模型，并引入置信度引导解码（CGD），配合改进的词元可靠性评分机制。此外，我们设计了一种参数高效机制，以降低 MTP 的参数开销。在合成与真实世界基准上的大量实验表明，Fast SceneScript 在解码器单次推理步骤中最多可生成 9 个词元而不损失准确率，且仅增加 7.5% 的额外参数。
英文原文摘要 1219 字符
Recent perception-generalist approaches based on language models have achieved state-of-the-art results across diverse tasks, including 3D scene layout estimation and 3D object detection, via unified architecture and interface. However, these approaches rely on autoregressive next-token prediction, which is inherently slow. In this work, we introduce Fast SceneScript, a novel structured language model for accurate and efficient 3D scene understanding. Our method employs multi-token prediction (MTP) to reduce the number of autoregressive iterations and significantly accelerate inference. While MTP improves speed, unreliable token predictions can significantly reduce accuracy. To filter out unreliable tokens, we adapt self-speculative decoding (SSD) for structured language models and introduce confidence-guided decoding (CGD) with an improved scoring mechanism for token reliability. Furthermore, we design a parameter-efficient mechanism that reduces the parameter overhead of MTP. Extensive experiments on synthetic and real-world benchmarks demonstrate that Fast SceneScript can generate up to 9 tokens per decoder inference step without compromising accuracy, while adding only 7.5% additional parameters.