Abstract:Recent Language Models (LMs) have shown impressive capabilities in generating texts with the knowledge internalized in parameters. Yet, LMs often generate the factually incorrect responses to the given queries, since their knowledge may be inaccurate, incomplete, and outdated. To address this problem, previous works propose to augment LMs with the knowledge retrieved from an external knowledge source. However, such approaches often show suboptimal text generation performance due to two reasons: 1) the model may fail to retrieve the knowledge relevant to the given query, or 2) the model may not faithfully reflect the retrieved knowledge in the generated text. To overcome these, we propose to verify the output and the knowledge of the knowledge-augmented LMs with a separate verifier, which is a small LM that is trained to detect those two types of errors through instruction-finetuning. Then, when the verifier recognizes an error, we can rectify it by either retrieving new knowledge or generating new text. Further, we use an ensemble of the outputs from different instructions with a single verifier to enhance the reliability of the verification processes. We validate the effectiveness of the proposed verification steps on multiple question answering benchmarks, whose results show that the proposed verifier effectively identifies retrieval and generation errors, allowing LMs to provide more factually correct outputs. Our code is available at <a class="link-external link-https" href="https://github.com/JinheonBaek/KALMV" rel="external noopener nofollow">this https URL</a>.

A Novel Keyword Verification Algorithm

An utterance verification algorithm in keyword spotting system

Experimental Investigation into Alignment-based Acoustic Confidence Measures in Keyword Verification for Mandarin Speech

RESEARCH ON SPEECH KEYWORD VERIFICATION METHOD BASED ON SVM

A NOVEL TWO-LEVEL ARCHITECTURE PLUS CONFIDENCE MEASURES FOR A KEYWORD SPOTTING SYSTEM

Utterance verification based on dynamic garbage evaluation approach

An lstm-ctc based verification system for proxy-word based oov keyword search

Binary-Search Based Verification of Feature Models.

Utterance verification using modified segmental probability model

Knowledge-Augmented Language Model Verification

A Novel Discriminative Score Calibration Method for Keyword Search

A multimodel keyword spotting system based on lip movement and speech features

A novel speaker verification approach for certain noisy environment

A Novel Text-Independent Speaker Verification Method Based on the Global Speaker Model

Keyword Spotting Based on Restricting Model and Acoustic Confidence Measure

Model Shrinking for Embedded Keyword Spotting

ModelVerification.jl: a Comprehensive Toolbox for Formally Verifying Deep Neural Networks

A Keyword Spotting Method

Few-shot short utterance speaker verification using meta-learning

Feature Extraction and Test Algorithm for Speaker Verification

Speaker verification system based on M-vector and support vector machine