Text-to-Distribution Prediction with Quantile Tokens and Neighbor Context

TL;DR

This paper introduces Quantile Token Regression, a novel method for distributional regression with LLMs, using dedicated quantile tokens and neighbor retrieval to improve distribution predictions.

Contribution

It is the first to insert dedicated quantile tokens into input sequences and incorporate neighbor information, enhancing local grounding and prediction accuracy.

Findings

Quantile tokens with neighbors outperform baselines by ~4 points in MAPE.

The approach produces 2x narrower and more accurate prediction intervals.

Large gains are observed on smaller, challenging datasets.

Abstract

Many applications of LLM-based text regression require predicting a full conditional distribution rather than a single point value. We study distributional regression under empirical-quantile supervision, where each input is paired with multiple observed quantile outcomes, and the target distribution is represented by a dense grid of quantiles. We address two key limitations of current approaches: the lack of local grounding for distribution estimates, and the reliance on shared representations that create an indirect bottleneck between inputs and quantile outputs. In this paper, we introduce Quantile Token Regression, which, to our knowledge, is the first work to insert dedicated quantile tokens into the input sequence, enabling direct input-output pathways for each quantile through self-attention. We further augment these quantile tokens with retrieval, incorporating semantically similar neighbor instances and their empirical distributions to ground predictions with local evidence from similar instances. We also provide the first theoretical analysis of loss functions for quantile regression, clarifying which distributional objectives each optimizes. Experiments on the Inside Airbnb and StackSample benchmark datasets with LLMs ranging from 1.7B to 14B parameters show that quantile tokens with neighbors consistently outperform baselines (~4 points lower MAPE and 2x narrower prediction intervals), with especially large gains on smaller and more challenging datasets where quantile tokens produce substantially sharper and more accurate distributions.