Prompt-Based Bias Calibration for Better Zero/Few-Shot Learning of Language Models
This addresses performance degradation in zero/few-shot learning for users of language models, but it is incremental as it builds on existing bias calibration methods with a focus on efficiency.
The paper tackles the problem of intrinsic bias in pre-trained language models that degrades performance in prompt-based zero/few-shot learning by proposing a null-input prompting method for bias calibration, resulting in average improvements of 9% in zero-shot and 2% in few-shot learning across various datasets.
Prompt-based learning is susceptible to intrinsic bias present in pre-trained language models (LMs), leading to sub-optimal performance in prompt-based zero/few-shot settings. In this work, we propose a null-input prompting method to calibrate intrinsic bias encoded in pre-trained LMs. Different from prior efforts that address intrinsic bias primarily for social fairness and often involve excessive computational cost, our objective is to explore enhancing LMs' performance in downstream zero/few-shot learning while emphasizing the efficiency of intrinsic bias calibration. Specifically, we leverage a diverse set of auto-selected null-meaning inputs generated from GPT-4 to probe intrinsic bias of pre-trained LMs. Utilizing the bias-reflected probability distribution, we formulate a distribution disparity loss for bias calibration, where we exclusively update bias parameters ($0.1\%$ of total parameters) of LMs towards equal probability distribution. Experimental results show that the calibration promotes an equitable starting point for LMs while preserving language modeling abilities. Across a wide range of datasets, including sentiment analysis and topic classification, our method significantly improves zero/few-shot learning performance of LMs for both in-context learning and prompt-based fine-tuning (on average $9\%$ and $2\%$, respectively).