April 4, 2024, 4:48 a.m. | Yifu Qiu, Varun Embar, Shay B. Cohen, Benjamin Han

cs.CL updates on arXiv.org arxiv.org

arXiv:2311.09467v2 Announce Type: replace
Abstract: Knowledge-to-text generators often struggle to faithfully generate descriptions for the input facts: they may produce hallucinations that contradict the input, or describe facts not present in the input. To reduce hallucinations, we propose a decoding-only method, TWEAK (Think While Effectively Articulating Knowledge), which can be integrated with any generator without retraining. TWEAK treats the generated sequences at each decoding step and its future sequences as hypotheses, and ranks each generation candidate based on the extent …

abstract arxiv cs.ai cs.cl decoding facts generate generators hallucinations hypothesis knowledge reduce struggle text text generation think type verification

Data Scientist (m/f/x/d)

@ Symanto Research GmbH & Co. KG | Spain, Germany

Senior DevOps/MLOps

@ Global Relay | Vancouver, British Columbia, Canada

Senior Statistical Programmer for Clinical Development

@ Novo Nordisk | Aalborg, North Denmark Region, DK

Associate, Data Analysis

@ JLL | USA-CLIENT Boulder CO-Google

AI Compiler Engineer, Model Optimization, Quantization & Framework

@ Renesas Electronics | Duesseldorf, Germany

Lead AI Security Researcher

@ Grammarly | United States; Hybrid