Reinforcement Learning from Human Feedback (RLHF) is effective for aligning Large Language Models (LLMs) with human preferences. However, RLHF’s complex process limits its ability to continually learn human feedback, making it impractical for real-world applications where the deployed model continuously receives feedback from users. The non-RL-based method, such as Direct Preference Optimization (DPO), is not primitively favorable for Continual Learning (CL). We observe that when combined with Experiment Relay (ER) for CL, DPO tends to significantly widen the gap in the probability of human-preferred and dispreferred responses. Consequently, this diminishes the diversity in model generation, potentially leading to model collapse. To overcome the above challenges, we propose the Continual Optimal Policy Regularization (COPR), a novel non-RL offline method to convert the historical optimal policies into optimization constraints when continually learning new preferences. We first derive a moderate reward function from the pairwise ranking loss and then use the moderate reward to calculate a new sampling distribution to construct novel learning objectives and constraints. We also provide formal proof of the learnability of COPR. The experimental results show that COPR outperforms strong CL baselines on our proposed benchmark, in terms of reward-based, GPT-4 evaluations and human assessment.

The inevitable private information in legal data necessitates legal artificial intelligence to study privacy-preserving and decentralized learning methods. Federated learning (FL) has merged as a promising technique for multiple participants to collaboratively train a shared model while efficiently protecting the sensitive data of participants. However, to the best of our knowledge, there is no work on applying FL to legal NLP. To fill this gap, this paper presents the first real-world FL benchmark for legal NLP, coined FEDLEGAL, which comprises five legal NLP tasks and one privacy task based on the data from Chinese courts. Based on the extensive experiments on these datasets, our results show that FL faces new challenges in terms of real-world non-IID data. The benchmark also encourages researchers to investigate privacy protection using real-world data in the FL setting, as well as deploying models in resource-constrained scenarios. The code and datasets of FEDLEGAL are available here.

With increasing concerns about data privacy, there is an increasing necessity of fine-tuning pre-trained language models (PLMs) for adapting to downstream tasks located in end-user devices or local clients without transmitting data to the central server. This urgent necessity therefore calls the research of investigating federated learning (FL) for PLMs. However, large PLMs bring the curse of prohibitive communication overhead and local model adaptation costs for the FL system. To this end, we investigate the parameter-efficient tuning (PETuning) of PLMs and develop a corresponding federated benchmark for four representative PETuning methods, dubbed FedPETuning. Specifically, FedPETuning provides the first holistic empirical study of representative PLMs tuning methods in FL, covering privacy attacks, performance comparisons, and resource-constrained analysis. Intensive experimental results have indicated that FedPETuning can efficiently defend against privacy attacks and maintains acceptable performance with reducing heavy resource consumption. The open-source code and data are available at https://github.com/SMILELab-FL/FedPETuning.