RoPEを超えて：STRINGによる2D・3D位置エンコーディングの改善
Learning the RoPEs: Better 2D and 3D Position Encodings with STRING
大規模言語モデルで広く使われるRotary Position Encodingsを統一的な理論枠組みで拡張し、任意次元のトークン座標に対して厳密な並進不変性を保ちつつ低計算コストで2D・3D位置エンコーディングを実現するSTRINGを提案。Vision Transformerに統合し、オープンボキャブラリ物体検出やロボティクス制御で性能向上を示した。
著者: Connor Schenck, Isaac Reid, Mithun George Jacob, Alex Bewley, Joshua Ainslie, David Rendleman, Deepali Jain, Mohit Sharma, Avinava Dubey, Ayzaan Wahid, Sumeet Singh, René Wagner, Tianli Ding, Chuyuan Fu, Arunkumar Byravan, Jake Varley, Alexey Gritsenko, Matthias Minderer, Dmitry Kalashnikov, Jonathan Tompson, Vikas Sindhwani, Krzysztof Choromanski
分類: cs.LG, cs.AI, cs.CV, cs.RO, stat.ML
原文アブストラクト
We introduce STRING: Separable Translationally Invariant Position Encodings. STRING extends Rotary Position Encodings, a recently proposed and widely used algorithm in large language models, via a unifying theoretical framework. Importantly, STRING still provides exact translation invariance, including token coordinates of arbitrary dimensionality, whilst maintaining a low computational footprint. These properties are especially important in robotics, where efficient 3D token representation is key. We integrate STRING into Vision Transformers with RGB(-D) inputs (color plus optional depth), showing substantial gains, e.g. in open-vocabulary object detection and for robotics controllers. We complement our experiments with a rigorous mathematical analysis, proving the universality of our methods.
