From Haystack to Needle: Label Space Reduction for Zero-shot Classification
This addresses the challenge of inefficient label space handling in zero-shot classification for LLM users, offering a method to enhance accuracy with computational efficiency, though it is incremental as it builds on existing zero-shot approaches.
The paper tackles the problem of improving zero-shot classification performance in Large Language Models by introducing Label Space Reduction (LSR), which iteratively refines the label space to focus on relevant options, resulting in average macro-F1 score improvements of 7.0% with Llama-3.1-70B and 3.3% with Claude-3.5-Sonnet across seven benchmarks.
We present Label Space Reduction (LSR), a novel method for improving zero-shot classification performance of Large Language Models (LLMs). LSR iteratively refines the classification label space by systematically ranking and reducing candidate classes, enabling the model to concentrate on the most relevant options. By leveraging unlabeled data with the statistical learning capabilities of data-driven models, LSR dynamically optimizes the label space representation at test time. Our experiments across seven benchmarks demonstrate that LSR improves macro-F1 scores by an average of 7.0% (up to 14.2%) with Llama-3.1-70B and 3.3% (up to 11.1%) with Claude-3.5-Sonnet compared to standard zero-shot classification baselines. To reduce the computational overhead of LSR, which requires an additional LLM call at each iteration, we propose distilling the model into a probabilistic classifier, allowing for efficient inference.