Explainable Token-level Noise Filtering for LLM Fine-tuning Datasets

TL;DR

This paper introduces XTF, an explainable framework for token-level noise filtering in LLM fine-tuning datasets, improving performance by addressing token-level discrepancies.

Contribution

XTF decomposes token contributions into explicit attributes and masks noisy tokens, enhancing fine-tuning effectiveness across multiple tasks and models.

Findings

XTF improves downstream task performance by up to 13.7%.

XTF effectively assesses token importance, knowledge novelty, and task relevance.

Extensive experiments validate XTF's superiority over regular fine-tuning.

Abstract

Large Language Models (LLMs) have seen remarkable advancements, achieving state-of-the-art results in diverse applications. Fine-tuning, an important step for adapting LLMs to specific downstream tasks, typically involves further training on corresponding datasets. However, a fundamental discrepancy exists between current fine-tuning datasets and the token-level optimization mechanism of LLMs: most datasets are designed at the sentence-level, which introduces token-level noise, causing negative influence to final performance. In this paper, we propose XTF, an explainable token-level noise filtering framework. XTF decomposes the complex and subtle contributions of token-level data to the fine-tuning process into three distinct and explicit attributes (reasoning importance, knowledge novelty, and task relevance), which can be assessed using scoring methods, and then masks the gradients of selected noisy tokens accordingly to optimize the performance of fine-tuned LLMs. We conduct extensive experiments on three representative downstream tasks (math, code and medicine) across 7 mainstream LLMs. The results demonstrate that XTF can significantly improve downstream performance by up to 13.7% compared to regular fine-tuning. Our work highlights the importance of token-level dataset optimization, and demonstrates the potential of strategies based on attribute decomposition for explaining complex training mechanisms.