---
title: Weak-to-Strong Generalization
url: https://www.emergentmind.com/topics/weak-to-strong-generalization-w2sg
type: topic
---

# Weak-to-Strong Generalization

Searching arXiv for recent W2SG papers to ground the article.
arxiv_search.query({"search_query":"all:\"weak-to-strong generalization\" OR ti:\"weak-to-strong generalization\"","start":0,"max_results":10,"sort_by":"submittedDate","sort_order":"descending"})
Weak-to-strong generalization (W2SG) is the phenomenon where a stronger student model, trained only on labels or preferences produced by a weaker supervisor, nevertheless outperforms that supervisor on the true task. In current literature, W2SG is treated both as an empirical alignment

Source: https://www.emergentmind.com/topics/weak-to-strong-generalization-w2sg