Comparer des méthodes
Examinez les méthodes sélectionnées côte à côte ; les lignes qui diffèrent sont mises en évidence.
| Apprentissage par renforcement multilingue× | Apprentissage par renforcement affiné× | |
|---|---|---|
| Domaine | Apprentissage profond | Apprentissage profond |
| Famille | Machine learning | Machine learning |
| Année d'origine≠ | 2010s (applied to multilingual NLP settings) | 2017–2022 |
| Auteur d'origine≠ | Sutton, R. S. & Barto, A. G. (RL foundations); multilingual extensions emerged from the NLP/RL community in the 2010s | Christiano, P. et al.; Ouyang, L. et al. |
| Type≠ | Reinforcement learning applied to multilingual environments | Policy adaptation via fine-tuning |
| Source fondatrice≠ | Sutton, R. S., & Barto, A. G. (1998). Reinforcement Learning: An Introduction. MIT Press. ISBN: 978-0262193986 | Ouyang, L., Wu, J., Jiang, X., Almeida, D., Wainwright, C., Mishkin, P., Zhang, C., Agarwal, S., Slama, K., Ray, A., Schulman, J., Hilton, J., Kelton, F., Miller, L., Simens, M., Askell, A., Welinder, P., Christiano, P., Leike, J., & Lowe, R. (2022). Training language models to follow instructions with human feedback. Advances in Neural Information Processing Systems, 35, 27730–27744. link ↗ |
| Alias | Cross-Lingual RL, Multilingual RL, Multilingual Policy Learning, Cross-Lingual Reinforcement Learning | RL fine-tuning, policy fine-tuning, RLHF, reinforcement learning from human feedback |
| Apparentées | 5 | 5 |
| Résumé≠ | Multilingual Reinforcement Learning applies the RL paradigm — an agent learning by interaction and reward — to environments that involve multiple languages. The agent must interpret multilingual observations, follow cross-lingual instructions, or generalize policies trained in one language to new target languages, making it applicable to cross-lingual dialogue, multilingual game-playing agents, and language-grounded sequential decision tasks. | Fine-Tuned Reinforcement Learning adapts a pre-trained policy or model to a new task or behavioral objective using reinforcement signals — including human feedback — rather than retraining from scratch. Popularized by RLHF, it is the core technique behind aligning large language models and adapting deep RL agents to specialized environments with minimal additional data. |
| ScholarGateJeu de données ↗ |
|
|