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

The recent popularity of large language models (LLMs) has brought asignificant impact to boundless fields, particularly through their open-endedecosystem such as the APIs, open-sourced models, and plugins. However, withtheir widespread deployment, there is a general lack of research thatthoroughly discusses and analyzes the potential risks concealed. In that case,we intend to conduct a preliminary but pioneering study covering therobustness, consistency, and credibility of LLMs systems. With most of therelated literature in the era of LLM uncharted, we propose an automatedworkflow that copes with an upscaled number of queries/responses. Overall, weconduct over a million queries to the mainstream LLMs including ChatGPT, LLaMA,and OPT. Core to our workflow consists of a data primitive, followed by anautomated interpreter that evaluates these LLMs under different adversarialmetrical systems. As a result, we draw several, and perhaps unfortunate,conclusions that are quite uncommon from this trendy community. Briefly, theyare: (i)-the minor but inevitable error occurrence in the user-generated queryinput may, by chance, cause the LLM to respond unexpectedly; (ii)-LLMs possesspoor consistency when processing semantically similar query input. In addition,as a side finding, we find that ChatGPT is still capable to yield the correctanswer even when the input is polluted at an extreme level. While thisphenomenon demonstrates the powerful memorization of the LLMs, it raisesserious concerns about using such data for LLM-involved evaluation in academicdevelopment. To deal with it, we propose a novel index associated with adataset that roughly decides the feasibility of using such data forLLM-involved evaluation. Extensive empirical studies are tagged to support theaforementioned claims.

Assessing Hidden Risks of LLMs: An Empirical Study on Robustness, Consistency, and Credibility