# ReMoMask-2: Latent Retrieval-Augmented Masked Motion Generation

arXiv: https://arxiv.org/abs/2609.08365 · HTML: https://mlsanity.com/paper/2609.08365/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Multimodal RAG, one of 15 directions in Retrieval-augmented generation.
- Builds on 25 of its 36 references that are in the map.
- First posted on arXiv 8 Sep 2026.

## Placed in

- Retrieval-augmented generation / [Multimodal RAG](https://mlsanity.com/rag/multimodal-rag/)

## Builds on (most cited first, 25 of 25)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [Denoising Diffusion Implicit Models](https://mlsanity.com/paper/2010.02502/) (2020)
- [Generative Adversarial Networks](https://mlsanity.com/paper/2203.00667/) (2022)
- [Human Motion Diffusion Model](https://mlsanity.com/paper/2209.14916/) (2022)
- [T2M-GPT: Generating Human Motion from Textual Descriptions with Discrete Representations](https://mlsanity.com/paper/2301.06052/) (2023)
- [MotionGPT: Human Motion as a Foreign Language](https://mlsanity.com/paper/2306.14795/) (2023)
- [TEMOS: Generating diverse human motions from textual descriptions](https://mlsanity.com/paper/2204.14109/) (2022)
- [MoMask: Generative Masked Modeling of 3D Human Motions](https://mlsanity.com/paper/2312.00063/) (2023)
- [TM2T: Stochastic and Tokenized Modeling for the Reciprocal Generation of 3D Human Motions and Texts](https://mlsanity.com/paper/2207.01696/) (2022)
- [The KIT Motion-Language Dataset](https://mlsanity.com/paper/1607.03827/) (2016)
- [ReMoDiffuse: Retrieval-Augmented Motion Diffusion Model](https://mlsanity.com/paper/2304.01116/) (2023)
- [TMR: Text-to-Motion Retrieval Using Contrastive 3D Human Motion Synthesis](https://mlsanity.com/paper/2305.00976/) (2023)
- [MMM: Generative Masked Motion Model](https://mlsanity.com/paper/2312.03596/) (2023)
- [LaMP: Language-Motion Pretraining for Motion Generation, Retrieval, and Captioning](https://mlsanity.com/paper/2410.07093/) (2024)
- [ParCo: Part-Coordinating Text-to-Motion Synthesis](https://mlsanity.com/paper/2403.18512/) (2024)
- [SnapMoGen: Human Motion Generation from Expressive Texts](https://mlsanity.com/paper/2507.09122/) (2025)
- [MoGenTS: Motion Generation based on Spatial-Temporal Joint Modeling](https://mlsanity.com/paper/2409.17686/) (2024)
- [Exploring Vision Transformers for 3D Human Motion-Language Models with Motion Patches](https://mlsanity.com/paper/2405.04771/) (2024)
- [Motion Avatar: Generate Human and Animal Avatars with Arbitrary Motion](https://mlsanity.com/paper/2405.11286/) (2024)
- [MoRAG -- Multi-Fusion Retrieval Augmented Generation for Human Motion](https://mlsanity.com/paper/2409.12140/) (2024)
- [ReMoMask: Retrieval-Augmented Masked Motion Generation](https://mlsanity.com/paper/2508.02605/) (2025)
- [RMD: A Simple Baseline for More General Human Motion Generation via Training-free Retrieval-Augmented Motion Diffuse](https://mlsanity.com/paper/2412.04343/) (2024)
- [ChromouVQA: Benchmarking Vision-Language Models under Chromatic Camouflaged Images](https://mlsanity.com/paper/2512.05137/) (2025)

## Abstract (arXiv)

Text-to-motion (T2M) generation maps natural language to human joint movements, aiding gaming, VR, and robotics. Retrieval-Augmented Text-to-Motion (RAG-T2M) improves generation on complex descriptions by conditioning on retrieved motion-text pairs. However, existing RAG-T2M models face two challenges: coarse-grained retrieval and fusion mechanisms overlook the hierarchical, spatial-temporal topology of human motion, and a representation gap exists because retrieved evidence resides in a semantic space separate from the generator's latents. To address the first, we present ReMoMask, a structure-aware RAG framework coupling Hierarchical Bidirectional Momentum (HBM) contrastive learning to align global and part-level features with text; Semantic Spatial-Temporal Attention (SSTA) for topology-aware fusion; and Topology Structured Masking (TSM) to force robust part-level grounding via adaptive masking. To address the second, we introduce ReMoMask-2, which rebuilds the retrieval database directly within the generator's pre-quantization latent space and aligns text queries via a distilled lightweight projector, allowing the generator to directly consume the retrieved motion's semantic content. Extensive experiments on HumanML3D, KIT-ML, and SnapMoGen demonstrate our retriever achieves state-of-the-art accuracy, while ReMoMask-2 attains the lowest FID on KIT-ML and SnapMoGen; notably, its single mask-transformer stage surpasses ReMoMask's full two-stage pipeline and delivers the fastest inference.