Form Over Content In Gradient-Based Data Attribution Methods

2026-09-16

AI Summary

研究解决了梯度相似性在数据归因方法中的争议，通过独立改变任务和答案格式，发现梯度对齐主要受答案格式影响而非任务语义。

Abstract

Data attribution methods using gradient similarity are widely used to analyze and select training data for large language models, but what gradient similarity actually measures is debated. Some interpret it as identifying task-relevant skills, while other work reports that surface form is the main factor. We resolve this debate for supervised fine-tuning examples by varying task and answer format independently. Specifically, we render benchmarks in different answer formats, such that datasets can share a task without a format or a format without a task. We find that gradient alignment follows the answer format, as benchmark pairs sharing an answer format align strongly (disattenuated cosine near 0.4), while same benchmarks rendered with different answer format classes show no alignment (near 0.0). We demonstrate that this ordering holds from the earliest pretraining checkpoints through post-training, and across model scales and families. We then analyze the released selections of LESS, a gradient-based data selection method for instruction tuning, and find that each target's selections over-represent the target's own answer format. Hence, we demonstrate that gradient-based attribution methods track format similarity more than task semantics, meaning that such methods, as well as the semantic interpretation of the gradient, should be tested on data where answer format and task vary independently for greater robustness and reliability.

Problem

• Gradient-based Data Attribution
• Task-relevant Skills
• Surface Form
• Answer Format
• Semantic Interpretation

Innovation

• Gradient-based Data Attribution
• Answer Format
• Task Semantics
• Gradient Alignment