自回归预训练中令牌角色的解耦

中文摘要 AI 辅助

自回归预训练日益依赖异构数据，这使得理解模型如何从单个令牌中学习变得重要。下一令牌预测目标自然地将令牌的贡献与其自身损失相关联。然而，每个令牌不仅是预测目标，也是后续内容的上下文。通过受控破坏，我们解耦了这两个角色，并发现了一个反转现象：使一个嘈杂的令牌更易于预测，会减少其作为目标的损害，但会增加其作为上下文的损害。同样的解耦有助于解释语言模型生成的文本：生成过程根据每个令牌与前缀的匹配度来选择它，而其作为上下文的角色从未针对独立确定的续文进行测试，因为该续文是为了适配它而生成的。在已知的破坏位置，通过上下文起作用可以减少移除令牌自身损失所无法减少的损害。因此，理解并控制模型从令牌中学习的内容需要解耦其角色。

英文摘要

Autoregressive pretraining increasingly draws on heterogeneous data, making it important to understand how a model learns from an individual token. The next-token prediction objective naturally identifies a token's contribution with its own loss. However, each token is not only a prediction target but also context for what follows. Using controlled corruption, we decouple these two roles and find a reversal: making a noisy token easier to predict reduces its damage as a target but increases it as context. The same decoupling helps explain text generated by language models: generation selects each token by its fit to the prefix, while its role as context is never tested against an independently determined continuation, because that continuation is generated to fit it. At known corrupted positions, acting through the context can reduce damage that removing the token's own loss does not. Understanding and controlling what a model learns from a token therefore requires decoupling its roles.

发表机构

• University of Sydney（悉尼大学）
• University of Oxford（牛津大学）
• Southeast University（东南大学）

机构由 AI 辅助整理，请以论文原文为准。