Zero-to-Strong Generalization: Eliciting Strong Capabilities of Large Language Models Iteratively without Gold Labels
This addresses the limitation of gold label availability in complex tasks for AI researchers and practitioners, though it is incremental as it builds on existing prompting and filtering techniques.
The study tackled the problem of eliciting strong capabilities from large language models without gold labels by proposing a zero-to-strong generalization paradigm that iteratively prompts models to annotate unlabeled data and filters high-quality labels, resulting in effective performance on classification and reasoning tasks across various model sizes.
Large Language Models (LLMs) have demonstrated remarkable performance through supervised fine-tuning or in-context learning using gold labels. However, this paradigm is limited by the availability of gold labels, while in certain scenarios, LLMs may need to perform tasks that are too complex for humans to provide such labels. To tackle this challenge, this study explores whether solely utilizing unlabeled data can elicit strong model capabilities. We propose a new paradigm termed zero-to-strong generalization. We iteratively prompt LLMs to annotate unlabeled data and retain high-quality labels by filtering. Surprisingly, we obverse that this iterative process gradually unlocks LLMs' potential on downstream tasks. Our experiments on extensive classification and reasoning tasks confirm the effectiveness of our proposed framework. Our analysis indicates that this paradigm is effective for both in-context learning and fine-tuning, and for various model sizes.