LLMs as Scalable, General-Purpose Simulators For Evolving Digital Agent Training

• URL: http://arxiv.org/abs/2510.14969v1
• Date: Thu, 16 Oct 2025 17:59:38 GMT
• Title: LLMs as Scalable, General-Purpose Simulators For Evolving Digital Agent Training
• Authors: Yiming Wang, Da Yin, Yuedong Cui, Ruichen Zheng, Zhiqian Li, Zongyu Lin, Di Wu, Xueqing Wu, Chenchen Ye, Yu Zhou, Kai-Wei Chang,
• Abstract summary: We introduce a scalable paradigm that generates structured UI states and transitions to synthesize training trajectories at scale.<n>Our paradigm integrates a digital world simulator for diverse UI states, a guided rollout process for coherent exploration, and a trajectory wrapper.<n>Experiments on WebArena and AndroidWorld show that UI-Simulator rivals or surpasses open-source agents trained on real UIs with significantly better robustness.
• Score: 55.72784274656801
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Digital agents require diverse, large-scale UI trajectories to generalize across real-world tasks, yet collecting such data is prohibitively expensive in both human annotation, infra and engineering perspectives. To this end, we introduce $\textbf{UI-Simulator}$, a scalable paradigm that generates structured UI states and transitions to synthesize training trajectories at scale. Our paradigm integrates a digital world simulator for diverse UI states, a guided rollout process for coherent exploration, and a trajectory wrapper that produces high-quality and diverse trajectories for agent training. We further propose $\textbf{UI-Simulator-Grow}$, a targeted scaling strategy that enables more rapid and data-efficient scaling by prioritizing high-impact tasks and synthesizes informative trajectory variants. Experiments on WebArena and AndroidWorld show that UI-Simulator rivals or surpasses open-source agents trained on real UIs with significantly better robustness, despite using weaker teacher models. Moreover, UI-Simulator-Grow matches the performance of Llama-3-70B-Instruct using only Llama-3-8B-Instruct as the base model, highlighting the potential of targeted synthesis scaling paradigm to continuously and efficiently enhance the digital agents.

Related papers

• ASTRA: Automated Synthesis of agentic Trajectories and Reinforcement Arenas [13.919124676472022]
  ASTRA is an end-to-end framework for training tool-augmented language model agents.<n>ASTRA integrates scalable data synthesis and verifiable reinforcement learning.<n> Experiments on multiple agentic tool-use benchmarks demonstrate that ASTRA-trained models achieve state-of-the-art performance.
  arXiv  Detail & Related papers  (2026-01-29T11:22:23Z)
• Watch and Learn: Learning to Use Computers from Online Videos [50.10702690339142]
  Watch & Learn (W&L) is a framework that converts human demonstration videos readily available on the Internet into executable UI trajectories at scale.<n>We develop an inverse dynamics labeling pipeline with task-aware video retrieval, generate over 53k high-quality trajectories from raw web videos.<n>These results highlight web-scale human demonstration videos as a practical and scalable foundation for advancing CUAs towards real-world deployment.
  arXiv  Detail & Related papers  (2025-10-06T10:29:00Z)
• EMMA: Generalizing Real-World Robot Manipulation via Generative Visual Transfer [35.27100635173712]
  Vision-language-action (VLA) models increasingly rely on diverse training data to achieve robust generalization.<n>We propose Embodied Manipulation Media Adaptation (EMMA), a VLA policy enhancement framework that integrates a generative data engine with an effective training pipeline.<n>DreamTransfer enables text-controlled visual editing of robot videos, transforming foreground, background, and lighting conditions without compromising 3D structure or geometrical plausibility.<n>AdaMix is a hard-sample-aware training strategy that dynamically reweights training batches to focus optimization on perceptually or kinematically challenging samples.
  arXiv  Detail & Related papers  (2025-09-26T14:34:44Z)
• UI-TARS-2 Technical Report: Advancing GUI Agent with Multi-Turn Reinforcement Learning [155.51875080423883]
  The development of autonomous agents for graphical user interfaces presents major challenges in artificial intelligence.<n>We present UI-TARS-2, a native GUI-centered agent model that addresses these challenges through a systematic training methodology.<n> Empirical evaluation demonstrates that UI-TARS-2 achieves significant improvements over its predecessor UI-TARS-1.5.
  arXiv  Detail & Related papers  (2025-09-02T17:44:45Z)
• Dreamland: Controllable World Creation with Simulator and Generative Models [32.427050300421115]
  Large-scale video generative models can synthesize diverse and realistic visual content for dynamic world creation.<n>But they often lack element-wise controllability, hindering their use in editing scenes and training embodied AI agents.<n>We propose Dreamland, a hybrid world generation framework combining the granular control of a physics-based simulator and the photorealistic content output of large-scale pretrained generative models.
  arXiv  Detail & Related papers  (2025-06-09T17:59:52Z)
• AgentTrek: Agent Trajectory Synthesis via Guiding Replay with Web Tutorials [53.376263056033046]
  Existing approaches rely on expensive human annotation, making them unsustainable at scale.<n>We propose AgentTrek, a scalable data synthesis pipeline that generates web agent trajectories by leveraging publicly available tutorials.<n>Our fully automated approach significantly reduces data collection costs, achieving a cost of just $0.55 per high-quality trajectory without human annotators.
  arXiv  Detail & Related papers  (2024-12-12T18:59:27Z)
• Synthesizing Post-Training Data for LLMs through Multi-Agent Simulation [51.20656279478878]
  MATRIX is a multi-agent simulator that automatically generates diverse text-based scenarios.<n>We introduce MATRIX-Gen for controllable and highly realistic data synthesis.<n>On AlpacaEval 2 and Arena-Hard benchmarks, Llama-3-8B-Base, post-trained on datasets synthesized by MATRIX-Gen with just 20K instruction-response pairs, outperforms Meta's Llama-3-8B-Instruct model.
  arXiv  Detail & Related papers  (2024-10-18T08:01:39Z)
• Gaussian Splatting to Real World Flight Navigation Transfer with Liquid Networks [93.38375271826202]
  We present a method to improve generalization and robustness to distribution shifts in sim-to-real visual quadrotor navigation tasks. We first build a simulator by integrating Gaussian splatting with quadrotor flight dynamics, and then, train robust navigation policies using Liquid neural networks. In this way, we obtain a full-stack imitation learning protocol that combines advances in 3D Gaussian splatting radiance field rendering, programming of expert demonstration training data, and the task understanding capabilities of Liquid networks.
  arXiv  Detail & Related papers  (2024-06-21T13:48:37Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.