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

Language models (LMs) have yielded impressive results on many languagereasoning tasks, but their unexpected errors raise doubts about their reasoningabilities. In light of this, there is growing interest in finetuning/promptingLMs with both task instances and their associated free-text rationales (FTRs),which explain the correct reasoning process for predicting the correct taskoutput (i.e., how to be "right for the right reasons"). However, existingfinetuning methods fail to improve LM performance, while prompting needsprohibitively large (i.e., >50B) LMs to work well. We propose KNIFE, whichshows that reasoning knowledge can be effectively distilled from FTRs into asmall (i.e., <1B) LM and improve the LM's performance. First, KNIFE finetunes ateacher LM (given task input and FTR) to predict the task output, transferringreasoning knowledge from the FTRs to the teacher's hidden states. Second, KNIFEfinetunes a student LM (given task input only) such that its hidden states arealigned with the teacher's. Thus, the student is endowed with reasoningknowledge but can be used for inference without direct FTR input. On twoquestion-answering datasets, KNIFE outperforms various finetuning and promptingbaselines in fully-supervised and low-resource settings. Also, we observe thatFTR quality is crucial to KNIFE's performance.

KNIFE: Distilling Reasoning Knowledge From Free-Text Rationales