all AI news
Deductive Closure Training of Language Models for Coherence, Accuracy, and Updatability
June 28, 2024, 4:42 a.m. | Afra Feyza Aky\"urek, Ekin Aky\"urek, Leshem Choshen, Derry Wijaya, Jacob Andreas
cs.CL updates on arXiv.org arxiv.org
Abstract: While language models (LMs) can sometimes generate factually correct text and estimate truth values of individual claims, these generally do not reflect a globally coherent, manipulable model of the world. As a consequence, current LMs also generate incorrect or nonsensical content, and are difficult to edit and bring up to date. We present a method called Deductive Closure Training (DCT) that uses LMs themselves to identify implications of (and contradictions within) the text that they …
abstract accuracy arxiv cs.cl current generate language language models lms replace text training truth type values while world
More from arxiv.org / cs.CL updates on arXiv.org
ReFT: Reasoning with Reinforced Fine-Tuning
2 days, 10 hours ago |
arxiv.org
Exploring Defeasibility in Causal Reasoning
2 days, 10 hours ago |
arxiv.org
A Large Language Model Approach to Educational Survey Feedback Analysis
2 days, 10 hours ago |
arxiv.org
Jobs in AI, ML, Big Data
Data Scientist
@ Ford Motor Company | Chennai, Tamil Nadu, India
Systems Software Engineer, Graphics
@ Parallelz | Vancouver, British Columbia, Canada - Remote
Engineering Manager - Geo Engineering Team (F/H/X)
@ AVIV Group | Paris, France
Data Analyst
@ Microsoft | San Antonio, Texas, United States
Azure Data Engineer
@ TechVedika | Hyderabad, India
Senior Data & AI Threat Detection Researcher (Cortex)
@ Palo Alto Networks | Tel Aviv-Yafo, Israel