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 Plus annual

Presents the access of premium content as premium feature

Premium Content

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 Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

We introduce Inference-Time Intervention (ITI), a technique designed toenhance the "truthfulness" of large language models (LLMs). ITI operates byshifting model activations during inference, following a set of directionsacross a limited number of attention heads. This intervention significantlyimproves the performance of LLaMA models on the TruthfulQA benchmark. On aninstruction-finetuned LLaMA called Alpaca, ITI improves its truthfulness from32.5% to 65.1%. We identify a tradeoff between truthfulness and helpfulness anddemonstrate how to balance it by tuning the intervention strength. ITI isminimally invasive and computationally inexpensive. Moreover, the technique isdata efficient: while approaches like RLHF require extensive annotations, ITIlocates truthful directions using only few hundred examples. Our findingssuggest that LLMs may have an internal representation of the likelihood ofsomething being true, even as they produce falsehoods on the surface.

Inference-Time Intervention: Eliciting Truthful Answers from a Language Model