Value Compass

Publications Hub

This page houses a collection of published papers covering various topics within the Value Compass project. It provides a detailed overview of our research progress and findings, focusing on interdisciplinary analysis, benchmarks, value evaluation and alignment techniques.

    Alignment

    Embedding an Ethical Mind - Aligning Text-to-Image Synthesis via Lightweight Value Optimization

    Analysis

    Elephant in the Room - Unveiling the Impact of Reward Model Quality in Alignment

    Evaluation

    Raising the Bar - Investigating the Values of Large Language Models via Generative Evolving Testing

    Dataset / Evaluation

    CLAVE - An Adaptive Framework for Evaluating Values of LLM Generated Responses

    Analysis / Evaluation

    Beyond Human Norms - Unveiling Unique Values of Large Language Models through Interdisciplinary Approaches

    Analysis

    On the Essence and Prospect - An Investigation of Alignment Approaches for Big Models

    Alignment

    Negating Negatives - Alignment without Human Positive Samples via Distributional Dispreference Optimization

    Dataset / Evaluation

    CDEval - A Benchmark for Measuring the Cultural Dimensions of Large Language Models

    Alignment / Dataset

    Value FULCRA - Mapping Large Lnguage Modeuls to the Multidimensional Spectrum of Basic Human Values

    Evaluation / Dataset

    DenEvil - Towards Deciphering and Navigating the Ethical Values of Large Language Models via Instruction Learning

    Analysis

    Unpacking the Ethical Value Alignment in Big Models

    Analysis

    From Instructions to Intrinsic Human Values —— A Survey of Alignment Goals for Big Models