Abstract
dc:description.abstractThe objective of the proposed research is to develop methods to leverage few-shot language feedback to improve language models on various tasks. The motivation for this work stems from the overarching goal in the artificial intelligence (AI) field of achieving alignment - that is, AI that advances the intended objectives of its users. Recent progress with instruction-tuned language models has given rise to capable chatbot agents that can solve myriad tasks described in natural language and engage with users in a conversational setting. Despite these advances, language models still face fundamental challenges in aligning with human objectives. In particular, they must continuously adapt and learn from human users. Current models can respond to direct feedback conversationally, but have limited abilities to generalize from that feedback to new situations and unseen data. This is the problem of few-shot language feedback. We seek to better understand the capabilities and limits of current language models in generalizing from language feedback and to develop methods to improve responses in the few-shot language feedback scenario.
Degree
thesis:*- Level thesis:degree_level
- Doctoral
- Department dc:contributor.department
- Electrical and Computer Engineering
- Grantor dc:publisher
- Georgia Institute of Technology
- Year dc:date.issued
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Richardson, Christopher
- Advisor dc:contributor.advisor
-
- Heck, Larry
- Committee members dc:contributor.committeemember
-
- Bloch, Matthieu
- Davenport, Mark
- Anderson, David
- Kira, Zsolt
Subjects
dc:subject × 6Rights
- Language dc:language.iso
- en_US
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/1853/77845