DRLC: Reinforcement Learning with Dense Rewards from LLM Critic

Reinforcement learning (RL) can align language models with non-differentiable reward signals, such as human preferences. However, a major challenge arises from the sparsity of these reward signals - typically, there is only one reward for the entire generation. This sparsity of rewards can lead to inefficient and unstable learning. In this paper, we introduce a novel framework leveraging the critique ability of LLMs to produce dense rewards throughout the learning process. Our approach incorporates a critic language model alongside the policy model. This critic is prompted with the task description, question, policy model's output, and environment's reward signal as input, and provides token or span-level dense rewards that reflect the quality of each segment of the output. We assess our approach on three text generation tasks: sentiment control, language model detoxification, and summarization. Experimental results show that incorporating artificial dense rewards in training yields consistent performance gains over the PPO baseline with holistic rewards. Furthermore, in a setting where the same model serves as both policy and critic, we demonstrate that "self-critique" rewards also boost learning efficiency.

Plug-in Diffusion Model for Sequential Recommendation

Pioneering efforts have verified the effectiveness of the diffusion models in exploring the informative uncertainty for recommendation. Considering the difference between recommendation and image synthesis tasks, existing methods have undertaken tailored refinements to the diffusion and reverse process. However, these approaches typically use the highest-score item in corpus for user interest prediction, leading to the ignorance of the user's generalized preference contained within other items, thereby remaining constrained by the data sparsity issue. To address this issue, this paper presents a novel Plug-in Diffusion Model for Recommendation (PDRec) framework, which employs the diffusion model as a flexible plugin to jointly take full advantage of the diffusion-generating user preferences on all items. Specifically, PDRec first infers the users' dynamic preferences on all items via a time-interval diffusion model and proposes a Historical Behavior Reweighting (HBR) mechanism to identify the high-quality behaviors and suppress noisy behaviors. In addition to the observed items, PDRec proposes a Diffusion-based Positive Augmentation (DPA) strategy to leverage the top-ranked unobserved items as the potential positive samples, bringing in informative and diverse soft signals to alleviate data sparsity. To alleviate the false negative sampling issue, PDRec employs Noise-free Negative Sampling (NNS) to select stable negative samples for ensuring effective model optimization. Extensive experiments and analyses on four datasets have verified the superiority of the proposed PDRec over the state-of-the-art baselines and showcased the universality of PDRec as a flexible plugin for commonly-used sequential encoders in different recommendation scenarios. The code is available in https://github.com/hulkima/PDRec.

Fusion-Eval: Integrating Evaluators with LLMs

Evaluating Large Language Models (LLMs) is a complex task, especially considering the intricacies of natural language understanding and the expectations for high-level reasoning. Traditional evaluations typically lean on human-based, model-based, or automatic-metrics-based paradigms, each with its own advantages and shortcomings. We introduce "Fusion-Eval", a system that employs LLMs not solely for direct evaluations, but to skillfully integrate insights from diverse evaluators. This gives Fusion-Eval flexibility, enabling it to work effectively across diverse tasks and make optimal use of multiple references. In testing on the SummEval dataset, Fusion-Eval achieved a Spearman correlation of 0.96, outperforming other evaluators. The success of Fusion-Eval underscores the potential of LLMs to produce evaluations that closely align human perspectives, setting a new standard in the field of LLM evaluation.

SiRA: Sparse Mixture of Low Rank Adaptation

Parameter Efficient Tuning has been an prominent approach to adapt the Large Language Model to downstream tasks. Most previous works considers adding the dense trainable parameters, where all parameters are used to adapt certain task. We found this less effective empirically using the example of LoRA that introducing more trainable parameters does not help. Motivated by this we investigate the importance of leveraging "sparse" computation and propose SiRA: sparse mixture of low rank adaption. SiRA leverages the Sparse Mixture of Expert(SMoE) to boost the performance of LoRA. Specifically it enforces the top $k$ experts routing with a capacity limit restricting the maximum number of tokens each expert can process. We propose a novel and simple expert dropout on top of gating network to reduce the over-fitting issue. Through extensive experiments, we verify SiRA performs better than LoRA and other mixture of expert approaches across different single tasks and multitask settings.

Critique Ability of Large Language Models

Critical thinking is essential for rational decision-making and problem-solving. This skill hinges on the ability to provide precise and reasoned critiques and is a hallmark of human intelligence. In the era of large language models (LLMs), this study explores the ability of LLMs to deliver accurate critiques across various tasks. We are interested in this topic as a capable critic model could not only serve as a reliable evaluator, but also as a source of supervised signals for model tuning. Particularly, if a model can self-critique, it has the potential for autonomous self-improvement. To examine this, we introduce a unified evaluation framework for assessing the critique abilities of LLMs. We develop a benchmark called CriticBench, which comprises 3K high-quality natural language queries and corresponding model responses; and annotate the correctness of these responses. The benchmark cover tasks such as math problem-solving, code completion, and question answering. We evaluate multiple LLMs on the collected dataset and our analysis reveals several noteworthy insights: (1) Critique is generally challenging for most LLMs, and this capability often emerges only when models are sufficiently large. (2) In particular, self-critique is especially difficult. Even top-performing LLMs struggle to achieve satisfactory performance. (3) Models tend to have lower critique accuracy on problems where they are most uncertain. To this end, we introduce a simple yet effective baseline named self-check, which leverages self-critique to improve task performance for various models. We hope this study serves as an initial exploration into understanding the critique abilities of LLMs, and aims to inform future research, including the development of more proficient critic models and the application of critiques across diverse tasks.

Towards an On-device Agent for Text Rewriting

Large Language Models (LLMs) have demonstrated impressive capabilities for text rewriting. Nonetheless, the large sizes of these models make them impractical for on-device inference, which would otherwise allow for enhanced privacy and economical inference. Creating a smaller yet potent language model for text rewriting presents a formidable challenge because it requires balancing the need for a small size with the need to retain the emergent capabilities of the LLM, that requires costly data collection. To address the above challenge, we introduce a new instruction tuning approach for building a mobile-centric text rewriting model. Our strategies enable the generation of high quality training data without any human labeling. In addition, we propose a heuristic reinforcement learning framework which substantially enhances performance without requiring preference data. To further bridge the performance gap with the larger server-side model, we propose an effective approach that combines the mobile rewrite agent with the server model using a cascade. To tailor the text rewriting tasks to mobile scenarios, we introduce MessageRewriteEval, a benchmark that focuses on text rewriting for messages through natural language instructions. Through empirical experiments, we demonstrate that our on-device model surpasses the current state-of-the-art LLMs in text rewriting while maintaining a significantly reduced model size. Notably, we show that our proposed cascading approach improves model performance.

Class-level Structural Relation Modelling and Smoothing for Visual Representation Learning

Representation learning for images has been advanced by recent progress in more complex neural models such as the Vision Transformers and new learning theories such as the structural causal models. However, these models mainly rely on the classification loss to implicitly regularize the class-level data distributions, and they may face difficulties when handling classes with diverse visual patterns. We argue that the incorporation of the structural information between data samples may improve this situation. To achieve this goal, this paper presents a framework termed \textbf{C}lass-level Structural Relation Modeling and Smoothing for Visual Representation Learning (CSRMS), which includes the Class-level Relation Modelling, Class-aware Graph Sampling, and Relational Graph-Guided Representation Learning modules to model a relational graph of the entire dataset and perform class-aware smoothing and regularization operations to alleviate the issue of intra-class visual diversity and inter-class similarity. Specifically, the Class-level Relation Modelling module uses a clustering algorithm to learn the data distributions in the feature space and identify three types of class-level sample relations for the training set; Class-aware Graph Sampling module extends typical training batch construction process with three strategies to sample dataset-level sub-graphs; and Relational Graph-Guided Representation Learning module employs a graph convolution network with knowledge-guided smoothing operations to ease the projection from different visual patterns to the same class. Experiments demonstrate the effectiveness of structured knowledge modelling for enhanced representation learning and show that CSRMS can be incorporated with any state-of-the-art visual representation learning models for performance gains. The source codes and demos have been released at https://github.com/czt117/CSRMS.

Cross-Silo Prototypical Calibration for Federated Learning with Non-IID Data

Federated Learning aims to learn a global model on the server side that generalizes to all clients in a privacy-preserving manner, by leveraging the local models from different clients. Existing solutions focus on either regularizing the objective functions among clients or improving the aggregation mechanism for the improved model generalization capability. However, their performance is typically limited by the dataset biases, such as the heterogeneous data distributions and the missing classes. To address this issue, this paper presents a cross-silo prototypical calibration method (FedCSPC), which takes additional prototype information from the clients to learn a unified feature space on the server side. Specifically, FedCSPC first employs the Data Prototypical Modeling (DPM) module to learn data patterns via clustering to aid calibration. Subsequently, the cross-silo prototypical calibration (CSPC) module develops an augmented contrastive learning method to improve the robustness of the calibration, which can effectively project cross-source features into a consistent space while maintaining clear decision boundaries. Moreover, the CSPC module's ease of implementation and plug-and-play characteristics make it even more remarkable. Experiments were conducted on four datasets in terms of performance comparison, ablation study, in-depth analysis and case study, and the results verified that FedCSPC is capable of learning the consistent features across different data sources of the same class under the guidance of calibrated model, which leads to better performance than the state-of-the-art methods. The source codes have been released at https://github.com/qizhuang-qz/FedCSPC.

RewriteLM: An Instruction-Tuned Large Language Model for Text Rewriting

Large Language Models (LLMs) have demonstrated impressive zero-shot capabilities in long-form text generation tasks expressed through natural language instructions. However, user expectations for long-form text rewriting is high, and unintended rewrites (''hallucinations'') produced by the model can negatively impact its overall performance. Existing evaluation benchmarks primarily focus on limited rewriting styles and sentence-level rewriting rather than long-form open-ended rewriting.We introduce OpenRewriteEval, a novel benchmark that covers a wide variety of rewriting types expressed through natural language instructions. It is specifically designed to facilitate the evaluation of open-ended rewriting of long-form texts. In addition, we propose a strong baseline model, RewriteLM, an instruction-tuned large language model for long-form text rewriting. We develop new strategies that facilitate the generation of diverse instructions and preference data with minimal human intervention. We conduct empirical experiments and demonstrate that our model outperforms the current state-of-the-art LLMs in text rewriting. Specifically, it excels in preserving the essential content and meaning of the source text, minimizing the generation of ''hallucinated'' content, while showcasing the ability to generate rewrites with diverse wording and structures.

Triple Sequence Learning for Cross-domain Recommendation

Cross-domain recommendation (CDR) aims to leverage the users' behaviors in both source and target domains to improve the target domain's performance. Conventional CDR methods typically explore the dual relations between the source and target domains' behavior sequences. However, they ignore modeling the third sequence of mixed behaviors that naturally reflects the user's global preference. To address this issue, we present a novel and model-agnostic Triple sequence learning for cross-domain recommendation (Tri-CDR) framework to jointly model the source, target, and mixed behavior sequences in CDR. Specifically, Tri-CDR independently models the hidden user representations for the source, target, and mixed behavior sequences, and proposes a triple cross-domain attention (TCA) to emphasize the informative knowledge related to both user's target-domain preference and global interests in three sequences. To comprehensively learn the triple correlations, we design a novel triple contrastive learning (TCL) that jointly considers coarse-grained similarities and fine-grained distinctions among three sequences, ensuring the alignment while preserving the information diversity in multi-domain. We conduct extensive experiments and analyses on two real-world datasets with four domains. The significant improvements of Tri-CDR with different sequential encoders on all datasets verify the effectiveness and universality. The source code will be released in the future.