Foundation models for the real world
Abstract/Contents
- Abstract
- Foundation models are quickly moving from their origins in the lab into real world deployment and use. In this thesis, I discuss two connected lines of research that work towards bridging this gap, so that foundation models can be fruitfully used in real-world settings, e.g. in engineering, medicine, or the sciences. The first is making models more domain-agnostic: while techniques for training foundation models were developed for language and vision domains, we show that simple techniques can generalize these approaches to work across at least twelve different domains. The second is making models more useful in cases of task ambiguity, where the user's desired task may be vague or not-perfectly specified, as is often the case in real-world settings. Here we show how to measure and improve the performance of foundation models under task ambiguity, and explore how models themselves can aid in the process of disambiguating user intent. We close by discussing future directions and the broader outlook of challenges and opportunities ahead.
Description
Type of resource | text |
---|---|
Form | electronic resource; remote; computer; online resource |
Extent | 1 online resource. |
Place | California |
Place | [Stanford, California] |
Publisher | [Stanford University] |
Copyright date | 2023; ©2023 |
Publication date | 2023; 2023 |
Issuance | monographic |
Language | English |
Creators/Contributors
Author | Tamkin, Alexander |
---|---|
Degree supervisor | Goodman, Noah (Noah D.) |
Thesis advisor | Goodman, Noah (Noah D.) |
Thesis advisor | Hashimoto, Tatsunori |
Thesis advisor | Jurafsky, Dan, 1962- |
Degree committee member | Hashimoto, Tatsunori |
Degree committee member | Jurafsky, Dan, 1962- |
Associated with | Stanford University, School of Engineering |
Associated with | Stanford University, Computer Science Department |
Subjects
Genre | Theses |
---|---|
Genre | Text |
Bibliographic information
Statement of responsibility | Alex Tamkin. |
---|---|
Note | Submitted to the Computer Science Department. |
Thesis | Thesis Ph.D. Stanford University 2023. |
Location | https://purl.stanford.edu/gs024vd2209 |
Access conditions
- Copyright
- © 2023 by Alexander Tamkin
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...