Salvato in:
| Autori principali: | , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2509.13341 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914060538216448 |
|---|---|
| author | Güzel, Ahmet H. Jackson, Matthew Thomas Liesen, Jarek Luca Rocktäschel, Tim Foerster, Jakob Nicolaus Bogunovic, Ilija Parker-Holder, Jack |
| author_facet | Güzel, Ahmet H. Jackson, Matthew Thomas Liesen, Jarek Luca Rocktäschel, Tim Foerster, Jakob Nicolaus Bogunovic, Ilija Parker-Holder, Jack |
| contents | Training agents to act in embodied environments typically requires vast training data or access to accurate simulation, neither of which exists for many cases in the real world. Instead, world models are emerging as an alternative leveraging offline, passively collected data, they make it possible to generate diverse worlds for training agents in simulation. In this work, we harness world models to generate imagined environments to train robust agents capable of generalizing to novel task variations. One of the challenges in doing this is ensuring the agent trains on useful generated data. We thus propose a novel approach, IMAC (Imagined Autocurricula), leveraging Unsupervised Environment Design (UED), which induces an automatic curriculum over generated worlds. In a series of challenging, procedurally generated environments, we show it is possible to achieve strong transfer performance on held-out environments, having trained only inside a world model learned from a narrower dataset. We believe this opens the path to utilizing larger-scale, foundation world models for generally capable agents. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_13341 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Imagined Autocurricula Güzel, Ahmet H. Jackson, Matthew Thomas Liesen, Jarek Luca Rocktäschel, Tim Foerster, Jakob Nicolaus Bogunovic, Ilija Parker-Holder, Jack Artificial Intelligence Machine Learning Training agents to act in embodied environments typically requires vast training data or access to accurate simulation, neither of which exists for many cases in the real world. Instead, world models are emerging as an alternative leveraging offline, passively collected data, they make it possible to generate diverse worlds for training agents in simulation. In this work, we harness world models to generate imagined environments to train robust agents capable of generalizing to novel task variations. One of the challenges in doing this is ensuring the agent trains on useful generated data. We thus propose a novel approach, IMAC (Imagined Autocurricula), leveraging Unsupervised Environment Design (UED), which induces an automatic curriculum over generated worlds. In a series of challenging, procedurally generated environments, we show it is possible to achieve strong transfer performance on held-out environments, having trained only inside a world model learned from a narrower dataset. We believe this opens the path to utilizing larger-scale, foundation world models for generally capable agents. |
| title | Imagined Autocurricula |
| topic | Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2509.13341 |