Reinforcement Learning from Human Feedback (RLHF) is commonly utilized to improve the alignment of Large Language Models (LLMs) with human preferences. Given the evolving nature of human preferences, continual alignment becomes more crucial and practical in comparison to traditional static alignment. Nevertheless, making RLHF compatible with Continual Learning (CL) is challenging due to its complex process. Meanwhile, directly learning new human preferences may lead to Catastrophic Forgetting (CF) of historical preferences, resulting in helpless or harmful outputs. To overcome these challenges, we propose the Continual Optimal Policy Regularization (COPR) method, which draws inspiration from the optimal policy theory. COPR utilizes a sampling distribution as a demonstration and regularization constraints for CL. It adopts the Lagrangian Duality (LD) method to dynamically regularize the current policy based on the historically optimal policy, which prevents CF and avoids over-emphasizing unbalanced objectives. We also provide formal proof for 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. Furthermore, we validate the robustness of COPR under various CL settings, including different backbones, replay memory sizes, and learning orders.
In-context learning has become a popular paradigm in natural language processing. However, its performance can be significantly influenced by the order of in-context demonstration examples. In this paper, we found that causal language models (CausalLMs) are more sensitive to this order compared to prefix language models (PrefixLMs). We attribute this phenomenon to the auto-regressive attention masks within CausalLMs, which restrict each token from accessing information from subsequent tokens. This results in different receptive fields for samples at different positions, thereby leading to representation disparities across positions. To tackle this challenge, we introduce an unsupervised fine-tuning method, termed the Information-Augmented and Consistency-Enhanced approach. This approach utilizes contrastive learning to align representations of in-context examples across different positions and introduces a consistency loss to ensure similar representations for inputs with different permutations. This enhances the model's predictive consistency across permutations. Experimental results on four benchmarks suggest that our proposed method can reduce the sensitivity to the order of in-context examples and exhibit robust generalizability, particularly when demonstrations are sourced from a pool different from that used in the training phase, or when the number of in-context examples differs from what is used during training.
Counterfactual generation lies at the core of various machine learning tasks, including image translation and controllable text generation. This generation process usually requires the identification of the disentangled latent representations, such as content and style, that underlie the observed data. However, it becomes more challenging when faced with a scarcity of paired data and labeling information. Existing disentangled methods crucially rely on oversimplified assumptions, such as assuming independent content and style variables, to identify the latent variables, even though such assumptions may not hold for complex data distributions. For instance, food reviews tend to involve words like tasty, whereas movie reviews commonly contain words such as thrilling for the same positive sentiment. This problem is exacerbated when data are sampled from multiple domains since the dependence between content and style may vary significantly over domains. In this work, we tackle the domain-varying dependence between the content and the style variables inherent in the counterfactual generation task. We provide identification guarantees for such latent-variable models by leveraging the relative sparsity of the influences from different latent variables. Our theoretical insights enable the development of a doMain AdapTive counTerfactual gEneration model, called (MATTE). Our theoretically grounded framework achieves state-of-the-art performance in unsupervised style transfer tasks, where neither paired data nor style labels are utilized, across four large-scale datasets. Code is available at https://github.com/hanqi-qi/Matte.git
* Neurips23. Controllable generation in causal perspective with a case
study of ChatGPT, sheds light on theory-guaranteed alignment in language
While Large language models (LLMs) have the capability to iteratively reflect on their own outputs, recent studies have observed their struggles with knowledge-rich problems without access to external resources. In addition to the inefficiency of LLMs in self-assessment, we also observe that LLMs struggle to revisit their predictions despite receiving explicit negative feedback. Therefore, We propose Mirror, a Multiple-perspective self-reflection method for knowledge-rich reasoning, to avoid getting stuck at a particular reflection iteration. Mirror enables LLMs to reflect from multiple-perspective clues, achieved through a heuristic interaction between a Navigator and a Reasoner. It guides agents toward diverse yet plausibly reliable reasoning trajectory without access to ground truth by encouraging (1) diversity of directions generated by Navigator and (2) agreement among strategically induced perturbations in responses generated by the Reasoner. The experiments on five reasoning datasets demonstrate that Mirror's superiority over several contemporary self-reflection approaches. Additionally, the ablation study studies clearly indicate that our strategies alleviate the aforementioned challenges.
Task embedding, a meta-learning technique that captures task-specific information, has become prevalent, especially in areas such as multi-task learning, model editing, and interpretability. However, it faces challenges with the emergence of prompt-guided Large Language Models (LLMs) operating in a gradientfree manner. Existing task embedding methods rely on fine-tuned, task-specific language models, which hinders the adaptability of task embeddings across diverse models, especially prompt-based LLMs. To unleash the power of task embedding in the era of LLMs, we propose a framework for unified task embeddings (FUTE), harmonizing task embeddings from various models, including smaller language models and LLMs with varied prompts, within a single vector space. Such uniformity enables the comparison and analysis of similarities amongst different models, extending the scope and utility of existing task embedding methods in addressing multi-model scenarios, whilst maintaining their performance to be comparable to architecture-specific methods.
Stance detection is a challenging task that aims to identify public opinion from social media platforms with respect to specific targets. Previous work on stance detection largely focused on pure texts. In this paper, we study multi-modal stance detection for tweets consisting of texts and images, which are prevalent in today's fast-growing social media platforms where people often post multi-modal messages. To this end, we create five new multi-modal stance detection datasets of different domains based on Twitter, in which each example consists of a text and an image. In addition, we propose a simple yet effective Targeted Multi-modal Prompt Tuning framework (TMPT), where target information is leveraged to learn multi-modal stance features from textual and visual modalities. Experimental results on our three benchmark datasets show that the proposed TMPT achieves state-of-the-art performance in multi-modal stance detection.
Large language models (LLMs) have achieved remarkable progress in many natural language processing tasks. However, our experiment reveals that, in stance detection tasks, LLMs may generate biased stances due to spurious sentiment-stance correlation and preference towards certain individuals and topics, thus harming their performance. Therefore, in this paper, we propose to Mitigate Biases of LLMs in stance detection with Calibration (MB-Cal). In which, a novel gated calibration network is devised to mitigate the biases on the stance reasoning results from LLMs. Further, to make the calibration more accurate and generalizable, we construct counterfactual augmented data to rectify stance biases. Experimental results on in-target and zero-shot stance detection tasks show that the proposed MB-Cal can effectively mitigate biases of LLMs, achieving state-of-the-art results.
Existing datasets for narrative understanding often fail to represent the complexity and uncertainty of relationships in real-life social scenarios. To address this gap, we introduce a new benchmark, Conan, designed for extracting and analysing intricate character relation graphs from detective narratives. Specifically, we designed hierarchical relationship categories and manually extracted and annotated role-oriented relationships from the perspectives of various characters, incorporating both public relationships known to most characters and secret ones known to only a few. Our experiments with advanced Large Language Models (LLMs) like GPT-3.5, GPT-4, and Llama2 reveal their limitations in inferencing complex relationships and handling longer narratives. The combination of the Conan dataset and our pipeline strategy is geared towards understanding the ability of LLMs to comprehend nuanced relational dynamics in narrative contexts.
The performance of wireless communication systems is fundamentally constrained by random and uncontrollable wireless channels. Recently, reconfigurable intelligent surfaces (RIS) has emerged as a promising solution to enhance wireless network performance by smartly reconfiguring the radio propagation environment. While significant research has been conducted on RIS-assisted wireless systems, this paper focuses specifically on the deployment of RIS in a wideband millimeter wave (mmWave) multiple-input-multiple-output (MIMO) system to achieve maximum sum-rate. First, we derive the average user rate as well as the lower bound rate when the covariance of the channel follows the Wishart distribution. Based on the lower bound of users' rate, we propose a heuristic method that transforms the problem of optimizing the RIS's orientation into maximizing the number of users served by the RIS. Simulation results show that the proposed RIS deployment strategy can effectively improve the sum-rate. Furthermore, the performance of the proposed RIS deployment algorithm is only approximately 7.6\% lower on average than that of the exhaustive search algorithm.