Terrarium: Growing Personal Assistants via Scalable Synthesis of App Worlds
To address this challenge, we introduce Terrarium, a framework that synthesizes executable App Worlds from real-world application information, generates training task packages with evaluator-side rubrics from these worlds, and iteratively validates and refines both through agent rollouts.
ProofPaper ↗
Key points
- Advances in large language models (LLMs) have brought the vision of a general-purpose personal assistant closer to reality.
- For App World synthesis, Terrarium organizes public Web information and in-application observations into structured app specifications that guide the construction of self-contained runtimes with callable operations and mutable, resettable state.
- We construct a training suite of 275 App Worlds and 13,344 tasks for outcome-based reinforcement learning across three Qwen3 model scales (4B, 8B, and 14B parameters).
- These results suggest that app-grounded world and task synthesis offers a scalable path toward general-purpose assistants capable of operating across broad application ecosystems.
Sources (1)
- [1]Terrarium: Growing Personal Assistants via Scalable Synthesis of App WorldsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 10, 12:30 PM
To address this challenge, we introduce Terrarium, a framework that synthesizes executable App Worlds from real-world application information, generates training task packages with evaluator-side rubrics from these worlds, and iteratively validates and refines both through agent rollouts.
Advances in large language models (LLMs) have brought the vision of a general-purpose personal assistant closer to reality.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 10, 2026SAGE: Pushing 2-Bit Vector-Quantized LLMs toward Real-World Usability
- Oct 10, 2026Vela: User Behavior as Supervision Sparks LLM Intelligence
- Oct 10, 2026UltraBERT: Beyond the Limit of BERT Pre-Training
- Oct 9, 2026SAT-RL: Training Language Models for Logical Reasoning with Symbolic Solvers
- Oct 9, 2026ASSET: Adaptive Speech Segmentation via Reinforcement Learning
- Oct 9, 2026The Thinking Leakage in Reinforcement Learning for LLMs