Clip-Tuning: Towards Derivative-free Prompt Learning with a Mixture of Rewards
Yekun Chai, Shuohuan Wang, Yu Ming Sun, Hao Tian, Hua Wu, Haifeng Wang · 2022
Derivative-free prompt learning has emerged as a lightweight alternative to prompt tuning, which only requires model inference to optimize the prompts.However, existing work did not take full advantage of the over-parameterized characteristics of large pretrained language models (PLMs).In this paper, we propose Clip-Tuning, a simple yet effective method that adopts diverse frozen "thinned" networks of PLMs to obtain a mixture of rewards and thus advance the derivative-free prompt learning.The thinned networks consist of all the hidden units that survive a stationary dropout strategy, whose inference predictions reflect an ensemble of partial views over prompted training samples.Our method outperforms previous gradient-free prompt learning methods and achieves parity with gradientbased counterparts on seven language understanding benchmarks under few-shot settings.