Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Evidence-based Distributional Alignment for Large Language Models

Viet-Thanh Pham, Lizhen Qu, Zhuang Li, Gholamreza Haffari

Abstract

Distributional alignment enables large language models (LLMs) to predict how a target population distributes its responses across answer options, rather than collapsing disagreement into a single consensus answer. However, existing LLM-based distribution prediction is often unstable and degrades under cultural and domain shift. Token score-based estimates can change with minor option wording or formatting, response sampling-based estimates are expensive and sensitive to prompts and decoding settings, and directly generated distributions are frequently miscalibrated. We propose Evi-DA, an evidence-based alignment technique that improves the fidelity and robustness of LLM-based distribution estimation under domain and cultural shift. Given a target country and a multiple-choice question, Evi-DA retrieves related World Values Survey items and their answer distributions, predicts a coarse Welzel value signature for each option, and infers the country-conditioned answer distribution in a structured format. We train the LLMs using a two-stage pipeline, where reinforcement learning optimizes survey-derived rewards that encourage accurate intermediate value predictions, faithful final distributions, well-formed structured outputs, and reduced cultural bias. Across in-domain and out-of-domain benchmarks and multiple open-source backbones, Evi-DA reduces Jensen-Shannon divergence between predicted and gold distributions relative to strong baselines, with average relative improvements of up to 44%.

Evidence-based Distributional Alignment for Large Language Models

Abstract

Distributional alignment enables large language models (LLMs) to predict how a target population distributes its responses across answer options, rather than collapsing disagreement into a single consensus answer. However, existing LLM-based distribution prediction is often unstable and degrades under cultural and domain shift. Token score-based estimates can change with minor option wording or formatting, response sampling-based estimates are expensive and sensitive to prompts and decoding settings, and directly generated distributions are frequently miscalibrated. We propose Evi-DA, an evidence-based alignment technique that improves the fidelity and robustness of LLM-based distribution estimation under domain and cultural shift. Given a target country and a multiple-choice question, Evi-DA retrieves related World Values Survey items and their answer distributions, predicts a coarse Welzel value signature for each option, and infers the country-conditioned answer distribution in a structured format. We train the LLMs using a two-stage pipeline, where reinforcement learning optimizes survey-derived rewards that encourage accurate intermediate value predictions, faithful final distributions, well-formed structured outputs, and reduced cultural bias. Across in-domain and out-of-domain benchmarks and multiple open-source backbones, Evi-DA reduces Jensen-Shannon divergence between predicted and gold distributions relative to strong baselines, with average relative improvements of up to 44%.
Paper Structure (46 sections, 9 equations, 3 figures, 5 tables)

This paper contains 46 sections, 9 equations, 3 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Method
  3. Overview
  4. Welzel values as a latent representation
  5. Evidence bank construction from the World Values Survey
  6. Retrieving evidence for input question
  7. Two-stage inference with structured intermediate representations
  8. Stage A: option value profiling.
  9. Stage B: distribution prediction conditioned on value evidence.
  10. Reinforcement learning with GRPO
  11. Experiments
  12. Evaluation Setup
  13. Benchmarks.
  14. Demographic Representation.
  15. Evaluation Metric.
  16. ...and 31 more sections

Figures (3)

  • Figure 1: Illustration of Evi-DA - our proposed method for distributional alignment with LLMs.
  • Figure 2: Illustration of Evi-DA with different LLM backbones on the out-of-domain benchmark when changing the value $K$ for number of retrieved samples.
  • Figure 3: Illustration of Evi-DA with different Embedding Models for Retrieval on the out-of-domain benchmark.