Unlocking the Power of LLM Uncertainty for Active In-Context Example Selection

2024-08-17

AI Summary

This work addresses the lack of uncertainty awareness and active example selection mechanisms in large language models (LLMs) for in-context learning. We propose Unc-TTP, a novel paradigm that constructs a tripartite uncertainty taxonomy grounded in output inconsistency, via three rounds of sampling under label perturbation, thereby explicitly quantifying the model’s intrinsic uncertainty. Crucially, Unc-TTP is the first method to directly leverage uncertainty—rather than conventional confidence scores or entropy—for active selection of in-context examples, overcoming inherent limitations of existing uncertainty proxies. Empirically, on multi-task few-shot benchmarks, Unc-TTP consistently outperforms random sampling and state-of-the-art active learning baselines, yielding substantial performance gains. Moreover, it is model-agnostic, demonstrating compatibility with both open-source and proprietary LLMs.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Uncertainty Representations
• Multiagent Systems: Multiagent Systems under Uncertainty

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

Large Language Models (LLMs) have shown remarkable performance across a wide range of downstream tasks. However, it is challenging for users to discern whether the responses of LLM are generated with certainty or are fabricated to meet user expectations. In this paper, we introduce Uncertainty Tripartite Testing Paradigm (Unc-TTP), a novel method for classifying LLM uncertainty by leveraging output inconsistency. Specifically, Unc-TTP performs three rounds of sampling under varying label injection interference, enumerating all possible outcomes, and uses the degree of output inconsistency as the indicator of the LLM's intrinsic uncertainty. To validate the effectiveness of this inconsistency-defined uncertainty, we draw inspiration from Active Learning, comparing the informativeness of actively selected in-context examples. Our experiments show that uncertainty examples selected via Unc-TTP are more informative than certainty examples. Furthermore, the Unc-TTP-guided uncertainty-based active example selection strategy outperforms existing methods, highlighting its effectiveness in classifying LLM uncertainty and enhancing in-context learning. This work not only underscores the potential of inconsistency-based uncertainty classification for both open- and closed-source LLMs but also presents a practical approach for leveraging uncertainty to improve LLM performance in real-world tasks.

Problem

• Uncertainty Estimation
• Large Language Models
• Selective Sampling

Innovation

• Uncertainty Estimation
• Unc-TTP Methodology
• Active Learning Enhancement