English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Large Language Models (LLMs) have emerged as dominant tools for varioustasks, particularly when tailored for a specific target by prompt tuning.Nevertheless, concerns surrounding data privacy present obstacles due to thetuned prompts' dependency on sensitive private information. A practicalsolution is to host a local LLM and optimize a soft prompt privately usingdata. Yet, hosting a local model becomes problematic when model ownership isprotected. Alternative methods, like sending data to the model's provider fortraining, intensify these privacy issues facing an untrusted provider. In thispaper, we present a novel solution called Differentially-Private Offsite PromptTuning (DP-OPT) to address this challenge. Our approach involves tuning adiscrete prompt on the client side and then applying it to the desired cloudmodels. We demonstrate that prompts suggested by LLMs themselves can betransferred without compromising performance significantly. To ensure that theprompts do not leak private information, we introduce the first private promptgeneration mechanism, by a differentially-private (DP) ensemble of in-contextlearning with private demonstrations. With DP-OPT, generatingprivacy-preserving prompts by Vicuna-7b can yield competitive performancecompared to non-private in-context learning on GPT3.5 or local private prompttuning. Codes are available at https://github.com/VITA-Group/DP-OPT .

DP-OPT: Make Large Language Model Your Privacy-Preserving Prompt Engineer