# Dongxi NLP explains how reward models turn human choices into scores

> Original title: Reward model：Bradley-Terry 的倒影

- Source: Dongxi NLP
- Published: 2026-10-31T08:25:40.000Z
- Bad Influence: https://badinfluence.news/items/l0kc6kf2xf16yu07c720ix2ug
- Original: https://x.com/dongxi_nlp/status/2116446495033413960

## Summary

Dongxi NLP explains that a reward model learns from pairs of preferred and rejected answers rather than explicit scores. The post says the Bradley–Terry loss, L = −log σ(r(chosen) − r(rejected)), turns score differences into choice probabilities, and an illustrative two-weight example shows gradient descent raising the chosen answer's score. It adds that the reward model keeps the base LLM's body and replaces the LM head with a single-row score head that outputs one scalar.

## Post

https://x.com/i/article/2116301274433961984