Reliable Gradient-free and Likelihood-free Prompt Tuning

Maohao Shen, Soumya Ghosh, Prasanna Sattigeri, Subhro Das, Yuheng Bu, Gregory Wornell


Abstract
Due to privacy or commercial constraints, large pre-trained language models (PLMs) are often offered as black-box APIs. Fine-tuning such models to downstream tasks is challenging because one can neither access the model’s internal representations nor propagate gradients through it. This paper addresses these challenges by developing techniques for adapting PLMs with only API access. Building on recent work on soft prompt tuning, we develop methods to tune the soft prompts without requiring gradient computation. Further, we develop extensions that in addition to not requiring gradients also do not need to access any internal representation of the PLM beyond the input embeddings. Moreover, instead of learning a single prompt, our methods learn a distribution over prompts allowing us to quantify predictive uncertainty. Ours is the first work to consider uncertainty in prompts when only having API access to the PLM. Finally, through extensive experiments, we carefully vet the proposed methods and find them competitive with (and sometimes even improving on) gradient-based approaches with full access to the PLM.
Anthology ID:
2023.findings-eacl.183
Volume:
Findings of the Association for Computational Linguistics: EACL 2023
Month:
May
Year:
2023
Address:
Dubrovnik, Croatia
Editors:
Andreas Vlachos, Isabelle Augenstein
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
2416–2429
Language:
URL:
https://aclanthology.org/2023.findings-eacl.183
DOI:
10.18653/v1/2023.findings-eacl.183
Bibkey:
Cite (ACL):
Maohao Shen, Soumya Ghosh, Prasanna Sattigeri, Subhro Das, Yuheng Bu, and Gregory Wornell. 2023. Reliable Gradient-free and Likelihood-free Prompt Tuning. In Findings of the Association for Computational Linguistics: EACL 2023, pages 2416–2429, Dubrovnik, Croatia. Association for Computational Linguistics.
Cite (Informal):
Reliable Gradient-free and Likelihood-free Prompt Tuning (Shen et al., Findings 2023)
Copy Citation:
PDF:
https://aclanthology.org/2023.findings-eacl.183.pdf
Video:
 https://aclanthology.org/2023.findings-eacl.183.mp4