Vintix: Action Model via In-Context Reinforcement Learning
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866909813999403008 |
|---|---|
| author | Polubarov, Andrey Lyubaykin, Nikita Derevyagin, Alexander Zisman, Ilya Tarasov, Denis Nikulin, Alexander Kurenkov, Vladislav |
| author_facet | Polubarov, Andrey Lyubaykin, Nikita Derevyagin, Alexander Zisman, Ilya Tarasov, Denis Nikulin, Alexander Kurenkov, Vladislav |
| contents | In-Context Reinforcement Learning (ICRL) represents a promising paradigm for developing generalist agents that learn at inference time through trial-and-error interactions, analogous to how large language models adapt contextually, but with a focus on reward maximization. However, the scalability of ICRL beyond toy tasks and single-domain settings remains an open challenge. In this work, we present the first steps toward scaling ICRL by introducing a fixed, cross-domain model capable of learning behaviors through in-context reinforcement learning. Our results demonstrate that Algorithm Distillation, a framework designed to facilitate ICRL, offers a compelling and competitive alternative to expert distillation to construct versatile action models. These findings highlight the potential of ICRL as a scalable approach for generalist decision-making systems. Code released at https://github.com/dunnolab/vintix |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2501_19400 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Vintix: Action Model via In-Context Reinforcement Learning Polubarov, Andrey Lyubaykin, Nikita Derevyagin, Alexander Zisman, Ilya Tarasov, Denis Nikulin, Alexander Kurenkov, Vladislav Machine Learning Artificial Intelligence Robotics In-Context Reinforcement Learning (ICRL) represents a promising paradigm for developing generalist agents that learn at inference time through trial-and-error interactions, analogous to how large language models adapt contextually, but with a focus on reward maximization. However, the scalability of ICRL beyond toy tasks and single-domain settings remains an open challenge. In this work, we present the first steps toward scaling ICRL by introducing a fixed, cross-domain model capable of learning behaviors through in-context reinforcement learning. Our results demonstrate that Algorithm Distillation, a framework designed to facilitate ICRL, offers a compelling and competitive alternative to expert distillation to construct versatile action models. These findings highlight the potential of ICRL as a scalable approach for generalist decision-making systems. Code released at https://github.com/dunnolab/vintix |
| title | Vintix: Action Model via In-Context Reinforcement Learning |
| topic | Machine Learning Artificial Intelligence Robotics |
| url | https://arxiv.org/abs/2501.19400 |