Detectors for Safe and Reliable LLMs: Implementations, Uses, and Limitations
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866929463632068608 |
|---|---|
| author | Achintalwar, Swapnaja Garcia, Adriana Alvarado Anaby-Tavor, Ateret Baldini, Ioana Berger, Sara E. Bhattacharjee, Bishwaranjan Bouneffouf, Djallel Chaudhury, Subhajit Chen, Pin-Yu Chiazor, Lamogha Daly, Elizabeth M. DB, Kirushikesh de Paula, Rogério Abreu Dognin, Pierre Farchi, Eitan Ghosh, Soumya Hind, Michael Horesh, Raya Kour, George Lee, Ja Young Madaan, Nishtha Mehta, Sameep Miehling, Erik Murugesan, Keerthiram Nagireddy, Manish Padhi, Inkit Piorkowski, David Rawat, Ambrish Raz, Orna Sattigeri, Prasanna Strobelt, Hendrik Swaminathan, Sarathkrishna Tillmann, Christoph Trivedi, Aashka Varshney, Kush R. Wei, Dennis Witherspooon, Shalisha Zalmanovici, Marcel |
| author_facet | Achintalwar, Swapnaja Garcia, Adriana Alvarado Anaby-Tavor, Ateret Baldini, Ioana Berger, Sara E. Bhattacharjee, Bishwaranjan Bouneffouf, Djallel Chaudhury, Subhajit Chen, Pin-Yu Chiazor, Lamogha Daly, Elizabeth M. DB, Kirushikesh de Paula, Rogério Abreu Dognin, Pierre Farchi, Eitan Ghosh, Soumya Hind, Michael Horesh, Raya Kour, George Lee, Ja Young Madaan, Nishtha Mehta, Sameep Miehling, Erik Murugesan, Keerthiram Nagireddy, Manish Padhi, Inkit Piorkowski, David Rawat, Ambrish Raz, Orna Sattigeri, Prasanna Strobelt, Hendrik Swaminathan, Sarathkrishna Tillmann, Christoph Trivedi, Aashka Varshney, Kush R. Wei, Dennis Witherspooon, Shalisha Zalmanovici, Marcel |
| contents | Large language models (LLMs) are susceptible to a variety of risks, from non-faithful output to biased and toxic generations. Due to several limiting factors surrounding LLMs (training cost, API access, data availability, etc.), it may not always be feasible to impose direct safety constraints on a deployed model. Therefore, an efficient and reliable alternative is required. To this end, we present our ongoing efforts to create and deploy a library of detectors: compact and easy-to-build classification models that provide labels for various harms. In addition to the detectors themselves, we discuss a wide range of uses for these detector models - from acting as guardrails to enabling effective AI governance. We also deep dive into inherent challenges in their development and discuss future work aimed at making the detectors more reliable and broadening their scope. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_06009 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Detectors for Safe and Reliable LLMs: Implementations, Uses, and Limitations Achintalwar, Swapnaja Garcia, Adriana Alvarado Anaby-Tavor, Ateret Baldini, Ioana Berger, Sara E. Bhattacharjee, Bishwaranjan Bouneffouf, Djallel Chaudhury, Subhajit Chen, Pin-Yu Chiazor, Lamogha Daly, Elizabeth M. DB, Kirushikesh de Paula, Rogério Abreu Dognin, Pierre Farchi, Eitan Ghosh, Soumya Hind, Michael Horesh, Raya Kour, George Lee, Ja Young Madaan, Nishtha Mehta, Sameep Miehling, Erik Murugesan, Keerthiram Nagireddy, Manish Padhi, Inkit Piorkowski, David Rawat, Ambrish Raz, Orna Sattigeri, Prasanna Strobelt, Hendrik Swaminathan, Sarathkrishna Tillmann, Christoph Trivedi, Aashka Varshney, Kush R. Wei, Dennis Witherspooon, Shalisha Zalmanovici, Marcel Machine Learning Large language models (LLMs) are susceptible to a variety of risks, from non-faithful output to biased and toxic generations. Due to several limiting factors surrounding LLMs (training cost, API access, data availability, etc.), it may not always be feasible to impose direct safety constraints on a deployed model. Therefore, an efficient and reliable alternative is required. To this end, we present our ongoing efforts to create and deploy a library of detectors: compact and easy-to-build classification models that provide labels for various harms. In addition to the detectors themselves, we discuss a wide range of uses for these detector models - from acting as guardrails to enabling effective AI governance. We also deep dive into inherent challenges in their development and discuss future work aimed at making the detectors more reliable and broadening their scope. |
| title | Detectors for Safe and Reliable LLMs: Implementations, Uses, and Limitations |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2403.06009 |