CoGen: Learning from Feedback with Coupled Comprehension and Generation
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866914928007315456 |
|---|---|
| author | Gul, Mustafa Omer Artzi, Yoav |
| author_facet | Gul, Mustafa Omer Artzi, Yoav |
| contents | Systems with both language comprehension and generation capabilities can benefit from the tight connection between the two. This work studies coupling comprehension and generation with focus on continually learning from interaction with users. We propose techniques to tightly integrate the two capabilities for both learning and inference. We situate our studies in two-player reference games, and deploy various models for thousands of interactions with human users, while learning from interaction feedback signals. We show dramatic improvements in performance over time, with comprehension-generation coupling leading to performance improvements up to 26% in absolute terms and up to 17% higher accuracies compared to a non-coupled system. Our analysis also shows coupling has substantial qualitative impact on the system's language, making it significantly more human-like. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2408_15992 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | CoGen: Learning from Feedback with Coupled Comprehension and Generation Gul, Mustafa Omer Artzi, Yoav Computation and Language Artificial Intelligence Computer Vision and Pattern Recognition Machine Learning Systems with both language comprehension and generation capabilities can benefit from the tight connection between the two. This work studies coupling comprehension and generation with focus on continually learning from interaction with users. We propose techniques to tightly integrate the two capabilities for both learning and inference. We situate our studies in two-player reference games, and deploy various models for thousands of interactions with human users, while learning from interaction feedback signals. We show dramatic improvements in performance over time, with comprehension-generation coupling leading to performance improvements up to 26% in absolute terms and up to 17% higher accuracies compared to a non-coupled system. Our analysis also shows coupling has substantial qualitative impact on the system's language, making it significantly more human-like. |
| title | CoGen: Learning from Feedback with Coupled Comprehension and Generation |
| topic | Computation and Language Artificial Intelligence Computer Vision and Pattern Recognition Machine Learning |
| url | https://arxiv.org/abs/2408.15992 |