Search

Graph Neural Networks (GNNs) are powerful deep learning models designed for graph-structured data, demonstrating effectiveness across a wide range of applications.The softmax function is the most commonly used classifier for semi-supervised node classification. However, the softmax function lacks spatial information of the graph structure. In this paper, we propose a graph similarity regularized softmax for GNNs in semi-supervised node classification. By incorporating non-local total variation (TV) regularization into the softmax activation function, we can more effectively capture the spatial information inherent in graphs. The weights in the non-local gradient and divergence operators are determined based on the graph's adjacency matrix. We apply the proposed method into the architecture of GCN and GraphSAGE, testing them on citation and webpage linking datasets, respectively. Numerical experiments demonstrate its good performance in node classification and generalization capabilities. These results indicate that the graph similarity regularized softmax is effective on both assortative and disassortative graphs.

The optimal training configurations of large language models (LLMs) with respect to model sizes and compute budgets have been extensively studied. But how to optimally configure LLMs during inference has not been explored in sufficient depth. We study compute-optimal inference: designing models and inference strategies that optimally trade off additional inference-time compute for improved performance. As a first step towards understanding and designing compute-optimal inference methods, we assessed the effectiveness and computational efficiency of multiple inference strategies such as Greedy Search, Majority Voting, Best-of-N, Weighted Voting, and their variants on two different Tree Search algorithms, involving different model sizes and computational budgets. We found that a smaller language model with a novel tree search algorithm typically achieves a Pareto-optimal trade-off. These results highlight the potential benefits of deploying smaller models equipped with more sophisticated decoding algorithms in budget-constrained scenarios, e.g., on end-devices, to enhance problem-solving accuracy. For instance, we show that the Llemma-7B model can achieve competitive accuracy to a Llemma-34B model on MATH500 while using $2\times$ less FLOPs. Our findings could potentially apply to any generation task with a well-defined measure of success.

Traditional language model-based theorem proving assumes that by training on a sufficient amount of formal proof data, a model will learn to prove theorems. Our key observation is that a wealth of informal information that is not present in formal proofs can be useful for learning to prove theorems. For instance, humans think through steps of a proof, but this thought process is not visible in the resulting code. We present Lean-STaR, a framework for training language models to produce informal thoughts prior to each step of a proof, thereby boosting the model's theorem-proving capabilities. Lean-STaR uses retrospective ground-truth tactics to generate synthetic thoughts for training the language model. At inference time, the trained model directly generates the thoughts prior to the prediction of the tactics in each proof step. Building on the self-taught reasoner framework, we then apply expert iteration to further fine-tune the model on the correct proofs it samples and verifies using the Lean solver. Lean-STaR achieves state-of-the-art results on the miniF2F-test benchmark within the Lean theorem proving environment, significantly outperforming base models ($\boldsymbol{43.4\% \rightarrow 46.3\%,}$ Pass@64). We also analyze the impact of the augmented thoughts on various aspects of the theorem proving process, providing insights into their effectiveness.

An open challenge in recent machine learning is about how to improve the reasoning capability of large language models (LLMs) in a black-box setting, i.e., without access to detailed information such as output token probabilities. Existing approaches either rely on accessibility (which is often unrealistic) or involve significantly increased train- and inference-time costs. This paper addresses those limitations or shortcomings by proposing a novel approach, namely CoBB (Correct for improving QA reasoning of Black-Box LLMs). It uses a trained adaptation model to perform a seq2seq mapping from the often-imperfect reasonings of the original black-box LLM to the correct or improved reasonings. Specifically, the adaptation model is initialized with a relatively small open-source LLM and adapted over a collection of sub-sampled training pairs. To select the representative pairs of correct and incorrect reasonings, we formulated the dataset construction as an optimization problem that minimizes the statistical divergence between the sampled subset and the entire collection, and solved it via a genetic algorithm. We then train the adaptation model over the sampled pairs by contrasting the likelihoods of correct and incorrect reasonings. Our experimental results demonstrate that CoBB significantly improves reasoning accuracy across various QA benchmarks, compared to the best-performing adaptation baselines., Comment: preprint, 18 pages

As the diversity of users increases, the capability of providing personalized responses by large language models (LLMs) has become increasingly important. Existing approaches have only limited successes in LLM personalization, due to the absence of personalized learning or the reliance on shared personal data. This paper proposes a new approach for a few-shot personalization of LLMs with their mis-aligned responses (Fermi). Our key idea is to learn a set of personalized prompts for each user by progressively improving the prompts using LLMs, based on user profile (e.g., demographic information) and a few examples of previous opinions. During an iterative process of prompt improvement, we incorporate the contexts of mis-aligned responses by LLMs, which are especially crucial for the effective personalization of LLMs. In addition, we develop an effective inference method to further leverage the context of the test query and the personalized prompts. Our experimental results demonstrate that Fermi significantly improves performance across various benchmarks, compared to the best-performing baselines., Comment: preprint, 30 pages

Axion-like particles (ALPs) with lepton flavour violating (LFV) interactions are predicted within a wide range of flavoured ALP models. The proposed $\mu$TRISTAN high-energy $e^-\mu^+$ and $\mu^+\mu^+$ collider will provide a good opportunity to explore flavour physics in the charged lepton sector. In this work, based on a model-independent effective Lagrangian describing the ALP leptonic interactions, we investigate the potential of $\mu$TRISTAN to probe ALP LFV couplings. We analyse the testability of selected ALP production channels with potential sensitivity at $\mu$TRISTAN, considering different beams and collision energies, including $e^- \mu^+ \to a \gamma$, $e^- \mu^+ \to e^- \tau^+ a$, $\mu^+ \mu^+ \to \mu^+ \tau^+ a$, and $e^- \mu^+ \to \tau^- \mu^+ a$. The produced ALP $a$ is either long-lived or can promptly decay to flavour violating or conserving charged lepton final states. In particular, combining the above LFV ALP production modes with a suitable LFV decay mode, one can identify signatures that are virtually free of Standard Model background. We show the resulting sensitivity of $\mu$TRISTAN to LFV ALP couplings and compare it with multiple low-energy leptonic constraints and the future improvements thereof. We find that $\mu$TRISTAN can be generally complementary to searches for low-energy LFV processes and measurements of the leptonic magnetic dipole moments and has the capability to explore unconstrained parameter space for ALP masses in the $\mathcal{O}(1)$ to $\mathcal{O}(100)$~GeV range. In the light ALP regime, however, the parameter space that $\mu$TRISTAN is sensitive to, has been already excluded by low-energy searches for LFV decays., Comment: 30 pages, 10 figures, 8 tables

Traditional reinforcement learning from human feedback (RLHF) approaches relying on parametric models like the Bradley-Terry model fall short in capturing the intransitivity and irrationality in human preferences. Recent advancements suggest that directly working with preference probabilities can yield a more accurate reflection of human preferences, enabling more flexible and accurate language model alignment. In this paper, we propose a self-play-based method for language model alignment, which treats the problem as a constant-sum two-player game aimed at identifying the Nash equilibrium policy. Our approach, dubbed Self-Play Preference Optimization (SPPO), approximates the Nash equilibrium through iterative policy updates and enjoys a theoretical convergence guarantee. Our method can effectively increase the log-likelihood of the chosen response and decrease that of the rejected response, which cannot be trivially achieved by symmetric pairwise loss such as Direct Preference Optimization (DPO) and Identity Preference Optimization (IPO). In our experiments, using only 60k prompts (without responses) from the UltraFeedback dataset and without any prompt augmentation, by leveraging a pre-trained preference model PairRM with only 0.4B parameters, SPPO can obtain a model from fine-tuning Mistral-7B-Instruct-v0.2 that achieves the state-of-the-art length-controlled win-rate of 28.53% against GPT-4-Turbo on AlpacaEval 2.0. It also outperforms the (iterative) DPO and IPO on MT-Bench and the Open LLM Leaderboard. Starting from a stronger base model Llama-3-8B-Instruct, we are able to achieve a length-controlled win rate of 38.77%. Notably, the strong performance of SPPO is achieved without additional external supervision (e.g., responses, preferences, etc.) from GPT-4 or other stronger language models. Codes are available at https://github.com/uclaml/SPPO., Comment: 27 pages, 4 figures, 5 tables

Reconfigurable Intelligent Surface (RIS) is attracting more and more research interest because of its ability to reprogram the radio environment. Designing and implementing the RIS, however, is challenging because of limitations of printed circuit board (PCB) technology related to manufacturing of large sizes as well as the cost of switches. Thus, a low-cost manufacturing process suitable for large size and volume of devices, such as screen-printing is necessary. In this paper, for the first time, a fully screen-printed reconfigurable intelligent surface (RIS) with vanadium dioxide (VO2) switches for 5G and beyond communications is proposed. A VO2 ink has been prepared and batches of switches have been printed and integrated with the resonator elements. These switches are a fraction of the cost of commercial switches. Furthermore, the printing of these switches directly on metal patterns negates the need of any minute soldering of the switches. To avoid the complications of multilayer printing and realizing the RIS without vias, the resonators and the biasing lines are realized on a single layer. However, this introduces the challenge of interference between the biasing lines and the resonators, which is tackled in this work by designing the bias lines as part of the resonator. By adjusting the unit cell periodicity and the dimension of the H-shaped resonator, we achieve a 220 to 170{\deg} phase shift from 23.5 GHz to 29.5 GHz covering both n257 and n258 bands. Inside the wide bandwidth, the maximum ON reflection magnitude is 74%, and the maximum OFF magnitude is 94%. The RIS array comprises 20x20 unit cells (4.54x4.54{\lambda}^2 at 29.5 GHz). Each column of unit cells is serially connected to a current biasing circuit. To validate the array's performance, we conduct full-wave simulations as well as near-field and far-field measurements.

Preference modeling techniques, such as direct preference optimization (DPO), has shown effective in enhancing the generalization abilities of large language model (LLM). However, in tasks involving video instruction-following, providing informative feedback, especially for detecting hallucinations in generated responses, remains a significant challenge. Previous studies have explored using large large multimodal models (LMMs) as reward models to guide preference modeling, but their ability to accurately assess the factuality of generated responses compared to corresponding videos has not been conclusively established. This paper introduces a novel framework that utilizes detailed video captions as a proxy of video content, enabling language models to incorporate this information as supporting evidence for scoring video Question Answering (QA) predictions. Our approach demonstrates robust alignment with OpenAI GPT-4V model's reward mechanism, which directly takes video frames as input. Furthermore, we show that applying this tailored reward through DPO significantly improves the performance of video LMMs on video QA tasks.

Current AI alignment methodologies rely on human-provided demonstrations or judgments, and the learned capabilities of AI systems would be upper-bounded by human capabilities as a result. This raises a challenging research question: How can we keep improving the systems when their capabilities have surpassed the levels of humans? This paper answers this question in the context of tackling hard reasoning tasks (e.g., level 4-5 MATH problems) via learning from human annotations on easier tasks (e.g., level 1-3 MATH problems), which we term as \textit{easy-to-hard generalization}. Our key insight is that an evaluator (reward model) trained on supervisions for easier tasks can be effectively used for scoring candidate solutions of harder tasks and hence facilitating easy-to-hard generalization over different levels of tasks. Based on this insight, we propose a novel approach to scalable alignment, which firstly trains the process-supervised reward models on easy problems (e.g., level 1-3), and then uses them to evaluate the performance of policy models on hard problems. We show that such \textit{easy-to-hard generalization from evaluators} can enable \textit{easy-to-hard generalizations in generators} either through re-ranking or reinforcement learning (RL). Notably, our process-supervised 7b RL model achieves an accuracy of 34.0\% on MATH500, despite only using human supervision on easy problems. Our approach suggests a promising path toward AI systems that advance beyond the frontier of human supervision.

Hallucinations pose a significant challenge to the reliability of large language models (LLMs) in critical domains. Recent benchmarks designed to assess LLM hallucinations within conventional NLP tasks, such as knowledge-intensive question answering (QA) and summarization, are insufficient for capturing the complexities of user-LLM interactions in dynamic, real-world settings. To address this gap, we introduce HaluEval-Wild, the first benchmark specifically designed to evaluate LLM hallucinations in the wild. We meticulously collect challenging (adversarially filtered by Alpaca) user queries from ShareGPT, an existing real-world user-LLM interaction datasets, to evaluate the hallucination rates of various LLMs. Upon analyzing the collected queries, we categorize them into five distinct types, which enables a fine-grained analysis of the types of hallucinations LLMs exhibit, and synthesize the reference answers with the powerful GPT-4 model and retrieval-augmented generation (RAG). Our benchmark offers a novel approach towards enhancing our comprehension of and improving LLM reliability in scenarios reflective of real-world interactions. Our benchmark is available at https://github.com/HaluEval-Wild/HaluEval-Wild.

In-context learning (ICL, also known as few-shot prompting) has been the standard method of adapting LLMs to downstream tasks, by learning from a few input-output examples. Nonetheless, all ICL-based approaches only learn from correct input-output pairs. In this paper, we revisit this paradigm, by learning more from the few given input-output examples. We introduce Learning Principles (LEAP): First, we intentionally induce the model to make mistakes on these few examples; then we reflect on these mistakes, and learn explicit task-specific "principles" from them, which help solve similar problems and avoid common mistakes; finally, we prompt the model to answer unseen test questions using the original few-shot examples and these learned general principles. We evaluate LEAP on a wide range of benchmarks, including multi-hop question answering (Hotpot QA), textual QA (DROP), Big-Bench Hard reasoning, and math problems (GSM8K and MATH); in all these benchmarks, LEAP improves the strongest available LLMs such as GPT-3.5-turbo, GPT-4, GPT-4 turbo and Claude-2.1. For example, LEAP improves over the standard few-shot prompting using GPT-4 by 7.5% in DROP, and by 3.3% in HotpotQA. Importantly, LEAP does not require any more input or examples than the standard few-shot prompting settings.

The potential of Vision-Language Models (VLMs) often remains underutilized in handling complex text-based problems, particularly when these problems could benefit from visual representation. Resonating with humans' ability to solve complex text-based problems by (1) creating a visual diagram from the problem and (2) deducing what steps they need to take to solve it, we propose Self-Imagine. We leverage a single Vision-Language Model (VLM) to generate a structured representation of the question using HTML, then render the HTML as an image, and finally use the same VLM to answer the question using both the question and the image. Our approach does not require any additional training data or training. We evaluate our approach on three mathematics tasks and nine general-purpose reasoning tasks using state-of-the-art (LLAVA-1.5 and GEMINI PRO) VLMs. Our approach boosts the performance of LLAVA-1.5 and GEMINI PRO on all math tasks (on average GSM8K: +3.1%; ASDIV: +3.2%; SVAMP: +6.9%) and the majority of the general-purpose reasoning tasks by 3.2% to 6.0% on average., Comment: 18 pages, 9 figures, 12 tables

This paper investigates how representation learning can enable optimal control in unknown and complex dynamics, such as chaotic and non-linear systems, without relying on prior domain knowledge of the dynamics. The core idea is to establish an equivariant geometry that is diffeomorphic to the manifold defined by a dynamical system and to perform optimal control within this corresponding geometry, which is a non-trivial task. To address this challenge, Koopman Embed to Equivariant Control (KEEC) is proposed for model learning and control. Inspired by Lie theory, KEEC begins by learning a non-linear dynamical system defined on a manifold and embedding trajectories into a Lie group. Subsequently, KEEC formulates an equivariant value function equation in reinforcement learning on the equivariant geometry, ensuring an invariant effect as the value function on the original manifold. By deriving analytical-form optimal actions on the equivariant value function, KEEC theoretically achieves quadratic convergence for the optimal equivariant value function by leveraging the differential information on the equivariant geometry. The effectiveness of KEEC is demonstrated in challenging dynamical systems, including chaotic ones like Lorenz-63. Notably, our results show that isometric functions, which maintain the compactness and completeness of geometry while preserving metric and differential information, consistently outperform loss functions lacking these characteristics.

Large Language Models (LLMs), despite their great power in language generation, often encounter challenges when dealing with intricate and knowledge-demanding queries in specific domains. This paper introduces a novel approach to enhance LLMs by effectively extracting the relevant knowledge from domain-specific textual sources, and the adaptive training of a chatbot with domain-specific inquiries. Our two-step approach starts from training a knowledge miner, namely LLMiner, which autonomously extracts Question-Answer pairs from relevant documents through a chain-of-thought reasoning process. Subsequently, we blend the mined QA pairs with a conversational dataset to fine-tune the LLM as a chatbot, thereby enriching its domain-specific expertise and conversational capabilities. We also developed a new evaluation benchmark which comprises four domain-specific text corpora and associated human-crafted QA pairs for testing. Our model shows remarkable performance improvement over generally aligned LLM and surpasses domain-adapted models directly fine-tuned on domain corpus. In particular, LLMiner achieves this with minimal human intervention, requiring only 600 seed instances, thereby providing a pathway towards self-improvement of LLMs through model-synthesized training data., Comment: Work in progress

Large language models (LLMs) are now available from cloud API providers in various sizes and configurations. While this diversity offers a broad spectrum of choices, effectively leveraging the options to optimize computational cost and performance remains challenging. In this work, we present Automix, an approach that strategically routes queries to larger LMs, based on the approximate correctness of outputs from a smaller LM. Central to Automix are two key technical contributions. First, it has a few-shot self-verification mechanism, which estimates the reliability of its own outputs without requiring extensive training. Second, given that self-verification can be noisy, it employs a POMDP based router that can effectively select an appropriately sized model, based on answer confidence. Experiments across five language models and five challenging datasets show that Automix consistently surpasses strong baselines, reducing computational cost by over 50% for comparable performance., Comment: The first two authors contributed equally. Work started and partly done during Aman's internship at Google. This version adds results on additional models and datasets

Supervised Fine-Tuning (SFT) on response demonstrations combined with Reinforcement Learning from Human Feedback (RLHF) constitutes a powerful paradigm for aligning LLM-based AI agents. However, a significant limitation of such an approach is its dependency on high-quality human annotations, making its application to intricate tasks challenging due to difficulties in obtaining consistent response demonstrations and in-distribution response preferences. This paper presents a novel approach, namely SALMON, to align base language models with minimal human supervision, using only a small set of human-defined principles, yet achieving superior performance. Central to our approach is an instructable reward model. Trained on synthetic preference data, this model can generate reward scores based on arbitrary human-defined principles. By merely adjusting these principles during the RL training phase, we gain full control over the preferences with the instructable reward model, subsequently influencing the behavior of the RL-trained policy models, and reducing the reliance on the collection of online human preferences. Applying our method to the LLaMA-2-70b base language model, we developed an AI assistant named Dromedary-2. With only 6 exemplars for in-context learning and 31 human-defined principles, Dromedary-2 significantly surpasses the performance of several state-of-the-art AI systems, including LLaMA-2-Chat-70b, on various benchmark datasets. We have open-sourced the code and model weights to encourage further research into aligning LLM-based AI agents with enhanced supervision efficiency, improved controllability, and scalable oversight., Comment: Previous Title: SALMON: Self-Alignment with Principle-Following Reward Models. Accepted to ICLR 2024. Project page: https://github.com/IBM/SALMON

Preventing the performance decay of Transformers on inputs longer than those used for training has been an important challenge in extending the context length of these models. Though the Transformer architecture has fundamentally no limits on the input sequence lengths it can process, the choice of position encoding used during training can limit the performance of these models on longer inputs. We propose a novel functional relative position encoding with progressive interpolation, FIRE, to improve Transformer generalization to longer contexts. We theoretically prove that this can represent some of the popular relative position encodings, such as T5's RPE, Alibi, and Kerple. We next empirically show that FIRE models have better generalization to longer contexts on both zero-shot language modeling and long text benchmarks., Comment: 26 pages; ICLR 2024 camera ready version

Large Multimodal Models (LMM) are built across modalities and the misalignment between two modalities can result in "hallucination", generating textual outputs that are not grounded by the multimodal information in context. To address the multimodal misalignment issue, we adapt the Reinforcement Learning from Human Feedback (RLHF) from the text domain to the task of vision-language alignment, where human annotators are asked to compare two responses and pinpoint the more hallucinated one, and the vision-language model is trained to maximize the simulated human rewards. We propose a new alignment algorithm called Factually Augmented RLHF that augments the reward model with additional factual information such as image captions and ground-truth multi-choice options, which alleviates the reward hacking phenomenon in RLHF and further improves the performance. We also enhance the GPT-4-generated training data (for vision instruction tuning) with previously available human-written image-text pairs to improve the general capabilities of our model. To evaluate the proposed approach in real-world scenarios, we develop a new evaluation benchmark MMHAL-BENCH with a special focus on penalizing hallucinations. As the first LMM trained with RLHF, our approach achieves remarkable improvement on the LLaVA-Bench dataset with the 94% performance level of the text-only GPT-4 (while previous best methods can only achieve the 87% level), and an improvement by 60% on MMHAL-BENCH over other baselines. We opensource our code, model, data at https://llava-rlhf.github.io., Comment: Preprint

Graph-based diffusion models have shown promising results in terms of generating high-quality solutions to NP-complete (NPC) combinatorial optimization (CO) problems. However, those models are often inefficient in inference, due to the iterative evaluation nature of the denoising diffusion process. This paper proposes to use progressive distillation to speed up the inference by taking fewer steps (e.g., forecasting two steps ahead within a single step) during the denoising process. Our experimental results show that the progressively distilled model can perform inference 16 times faster with only 0.019% degradation in performance on the TSP-50 dataset., Comment: Published at ICML 2023, Sampling and Optimization in Discrete Space Workshop. The implementation is at https://github.com/jwrh/Accelerating-Diffusion-based-Combinatorial-Optimization-Solvers-by-Progressive-Distillation

Temporal Sentence Grounding in Videos (TSGV) aims to detect the event timestamps described by the natural language query from untrimmed videos. This paper discusses the challenge of achieving efficient computation in TSGV models while maintaining high performance. Most existing approaches exquisitely design complex architectures to improve accuracy with extra layers and loss, suffering from inefficiency and heaviness. Although some works have noticed that, they only make an issue of feature fusion layers, which can hardly enjoy the highspeed merit in the whole clunky network. To tackle this problem, we propose a novel efficient multi-teacher model (EMTM) based on knowledge distillation to transfer diverse knowledge from both heterogeneous and isomorphic networks. Specifically, We first unify different outputs of the heterogeneous models into one single form. Next, a Knowledge Aggregation Unit (KAU) is built to acquire high-quality integrated soft labels from multiple teachers. After that, the KAU module leverages the multi-scale video and global query information to adaptively determine the weights of different teachers. A Shared Encoder strategy is then proposed to solve the problem that the student shallow layers hardly benefit from teachers, in which an isomorphic teacher is collaboratively trained with the student to align their hidden states. Extensive experimental results on three popular TSGV benchmarks demonstrate that our method is both effective and efficient without bells and whistles.

The mini-batch versions of StochAstic Recursive grAdient algoritHm and Semi-Stochastic Gradient Descent method, employed the random Barzilai-Borwein step sizes (shorted as MB-SARAH-RBB and mS2GD-RBB), have surged into prominence through timely step size sequence. Inspired by modern adaptors and variance reduction techniques, we propose two new variant rules in the paper, referred to as RHBB and RHBB+, thereby leading to four algorithms MB-SARAH-RHBB, MB-SARAH-RHBB+, mS2GD-RHBB and mS2GD-RHBB+ respectively. RHBB+ is an enhanced version that additionally incorporates the importance sampling technique. They are aggressive in updates, robust in performance and self-adaptive along iterative periods. We analyze the flexible convergence structures and the corresponding complexity bounds in strongly convex cases. Comprehensive tuning guidance is theoretically provided for reference in practical implementations. Experiments show that the proposed methods consistently outperform the original and various state-of-the-art methods on frequently tested data sets. In particular, tests on the RHBB+ verify the efficacy of applying the importance sampling technique to the step size level. Numerous explorations display the promising scalability of our iterative adaptors., Comment: 44 pages, 33 figures

Goal-Conditioned Hierarchical Reinforcement Learning (GCHRL) is a promising paradigm to address the exploration-exploitation dilemma in reinforcement learning. It decomposes the source task into subgoal conditional subtasks and conducts exploration and exploitation in the subgoal space. The effectiveness of GCHRL heavily relies on subgoal representation functions and subgoal selection strategy. However, existing works often overlook the temporal coherence in GCHRL when learning latent subgoal representations and lack an efficient subgoal selection strategy that balances exploration and exploitation. This paper proposes HIerarchical reinforcement learning via dynamically building Latent Landmark graphs (HILL) to overcome these limitations. HILL learns latent subgoal representations that satisfy temporal coherence using a contrastive representation learning objective. Based on these representations, HILL dynamically builds latent landmark graphs and employs a novelty measure on nodes and a utility measure on edges. Finally, HILL develops a subgoal selection strategy that balances exploration and exploitation by jointly considering both measures. Experimental results demonstrate that HILL outperforms state-of-the-art baselines on continuous control tasks with sparse rewards in sample efficiency and asymptotic performance. Our code is available at https://github.com/papercode2022/HILL., Comment: Accepted by the conference of International Joint Conference on Neural Networks (IJCNN) 2023

We present PESCO, a novel contrastive learning framework that substantially improves the performance of zero-shot text classification. We formulate text classification as a neural text matching problem where each document is treated as a query, and the system learns the mapping from each query to the relevant class labels by (1) adding prompts to enhance label matching, and (2) using retrieved labels to enrich the training set in a self-training loop of contrastive learning. PESCO achieves state-of-the-art performance on four benchmark text classification datasets. On DBpedia, we achieve 98.5\% accuracy without any labeled data, which is close to the fully-supervised result. Extensive experiments and analyses show all the components of PESCO are necessary for improving the performance of zero-shot text classification., Comment: accepted by ACL 2023

Popular reinforcement learning (RL) algorithms tend to produce a unimodal policy distribution, which weakens the expressiveness of complicated policy and decays the ability of exploration. The diffusion probability model is powerful to learn complicated multimodal distributions, which has shown promising and potential applications to RL. In this paper, we formally build a theoretical foundation of policy representation via the diffusion probability model and provide practical implementations of diffusion policy for online model-free RL. Concretely, we character diffusion policy as a stochastic process, which is a new approach to representing a policy. Then we present a convergence guarantee for diffusion policy, which provides a theory to understand the multimodality of diffusion policy. Furthermore, we propose the DIPO which is an implementation for model-free online RL with DIffusion POlicy. To the best of our knowledge, DIPO is the first algorithm to solve model-free online RL problems with the diffusion model. Finally, extensive empirical results show the effectiveness and superiority of DIPO on the standard continuous control Mujoco benchmark.

A popular approach for improving the correctness of output from large language models (LLMs) is Self-Consistency - poll the LLM multiple times and output the most frequent solution. Existing Self-Consistency techniques always generate a constant number of samples per question, where a better approach will be to non-uniformly distribute the available budget based on the amount of agreement in the samples generated so far. In response, we introduce Adaptive-Consistency, a cost-efficient, model-agnostic technique that dynamically adjusts the number of samples per question using a lightweight stopping criterion. Our experiments over 17 reasoning and code generation datasets and three LLMs demonstrate that Adaptive-Consistency reduces sample budget by up to 7.9 times with an average accuracy drop of less than 0.1%. Our code and data are available at https://www.sample-step-by-step.info, Comment: Published at EMNLP 2023

Despite the remarkable ability of large language models (LMs) to comprehend and generate language, they have a tendency to hallucinate and create factually inaccurate output. Augmenting LMs by retrieving information from external knowledge resources is one promising solution. Most existing retrieval augmented LMs employ a retrieve-and-generate setup that only retrieves information once based on the input. This is limiting, however, in more general scenarios involving generation of long texts, where continually gathering information throughout generation is essential. In this work, we provide a generalized view of active retrieval augmented generation, methods that actively decide when and what to retrieve across the course of the generation. We propose Forward-Looking Active REtrieval augmented generation (FLARE), a generic method which iteratively uses a prediction of the upcoming sentence to anticipate future content, which is then utilized as a query to retrieve relevant documents to regenerate the sentence if it contains low-confidence tokens. We test FLARE along with baselines comprehensively over 4 long-form knowledge-intensive generation tasks/datasets. FLARE achieves superior or competitive performance on all tasks, demonstrating the effectiveness of our method. Code and datasets are available at https://github.com/jzbjyb/FLARE., Comment: EMNLP 2023

Recent AI-assistant agents, such as ChatGPT, predominantly rely on supervised fine-tuning (SFT) with human annotations and reinforcement learning from human feedback (RLHF) to align the output of large language models (LLMs) with human intentions, ensuring they are helpful, ethical, and reliable. However, this dependence can significantly constrain the true potential of AI-assistant agents due to the high cost of obtaining human supervision and the related issues on quality, reliability, diversity, self-consistency, and undesirable biases. To address these challenges, we propose a novel approach called SELF-ALIGN, which combines principle-driven reasoning and the generative power of LLMs for the self-alignment of AI agents with minimal human supervision. Our approach encompasses four stages: first, we use an LLM to generate synthetic prompts, and a topic-guided method to augment the prompt diversity; second, we use a small set of human-written principles for AI models to follow, and guide the LLM through in-context learning from demonstrations (of principles application) to produce helpful, ethical, and reliable responses to user's queries; third, we fine-tune the original LLM with the high-quality self-aligned responses so that the resulting model can generate desirable responses for each query directly without the principle set and the demonstrations anymore; and finally, we offer a refinement step to address the issues of overly-brief or indirect responses. Applying SELF-ALIGN to the LLaMA-65b base language model, we develop an AI assistant named Dromedary. With fewer than 300 lines of human annotations (including < 200 seed prompts, 16 generic principles, and 5 exemplars for in-context learning). Dromedary significantly surpasses the performance of several state-of-the-art AI systems, including Text-Davinci-003 and Alpaca, on benchmark datasets with various settings., Comment: Accepted at NeurIPS 2023 (Spotlight). Project page: https://github.com/IBM/Dromedary

The remarkable performance of large language models (LLMs) in zero-shot language understanding has garnered significant attention. However, employing LLMs for large-scale inference or domain-specific fine-tuning requires immense computational resources due to their substantial model size. To overcome these limitations, we introduce a novel method, namely GenCo, which leverages the strong generative power of LLMs to assist in training a smaller and more adaptable language model. In our method, an LLM plays an important role in the self-training loop of a smaller model in two important ways. Firstly, the LLM is used to augment each input instance with a variety of possible continuations, enriching its semantic context for better understanding. Secondly, it helps crafting additional high-quality training pairs, by rewriting input texts conditioned on predicted labels. This ensures the generated texts are highly relevant to the predicted labels, alleviating the prediction error during pseudo-labeling, while reducing the dependency on large volumes of unlabeled text. In our experiments, GenCo outperforms previous state-of-the-art methods when only limited ($<5\%$ of original) in-domain text data is available. Notably, our approach surpasses the performance of Alpaca-7B with human prompts, highlighting the potential of leveraging LLM for self-training.

Despite the growing interest in reconfigurable intelligent surfaces (RISs) for millimeter-wave (mm-wave) bands, and the considerable theoretical work reported by the communication community, there is a limited number of published works demonstrating practical implementations and experimental results. To the authors' knowledge, no published literature has reported experimental results for RISs covering the n257 and n258 mm-wave bands. In this work, we propose a novel wideband RIS design that covers the entire mm-wave 5G n257 and n258 bands. In simulations, the unit cell can maintain a phase difference of 180{\deg} +- 20{\deg} and a reflection magnitude greater than -2.8 dB within 22.7 to 30.5 GHz (29.3% bandwidth) using one-bit PIN switches. The proposed unit cell design with four circular cutouts and long vias could realize wideband performance by exciting two adjacent high-order resonances (2.5f and 3.5f). The periodic unit cells can maintain an angular stability of 30{\deg}. Based on the proposed unit cell, a 20 by 20 RIS array is designed and fabricated with a size of 7.1{\lambda} by 7.1{\lambda}. The measurement results demonstrate that the proposed RIS could maintain a 3 dB peak gain variation bandwidth among various array configurations within 22.5 to 29.5 GHz (26.9%) and with a beam scanning capability of 50{\deg}, making this design a good candidate for 5G mm-wave applications.

Like humans, large language models (LLMs) do not always generate the best output on their first try. Motivated by how humans refine their written text, we introduce Self-Refine, an approach for improving initial outputs from LLMs through iterative feedback and refinement. The main idea is to generate an initial output using an LLMs; then, the same LLMs provides feedback for its output and uses it to refine itself, iteratively. Self-Refine does not require any supervised training data, additional training, or reinforcement learning, and instead uses a single LLM as the generator, refiner, and feedback provider. We evaluate Self-Refine across 7 diverse tasks, ranging from dialog response generation to mathematical reasoning, using state-of-the-art (GPT-3.5, ChatGPT, and GPT-4) LLMs. Across all evaluated tasks, outputs generated with Self-Refine are preferred by humans and automatic metrics over those generated with the same LLM using conventional one-step generation, improving by ~20% absolute on average in task performance. Our work demonstrates that even state-of-the-art LLMs like GPT-4 can be further improved at test time using our simple, standalone approach., Comment: Code, data, and demo at https://selfrefine.info/

Neural network-based Combinatorial Optimization (CO) methods have shown promising results in solving various NP-complete (NPC) problems without relying on hand-crafted domain knowledge. This paper broadens the current scope of neural solvers for NPC problems by introducing a new graph-based diffusion framework, namely DIFUSCO. Our framework casts NPC problems as discrete {0, 1}-vector optimization problems and leverages graph-based denoising diffusion models to generate high-quality solutions. We investigate two types of diffusion models with Gaussian and Bernoulli noise, respectively, and devise an effective inference schedule to enhance the solution quality. We evaluate our methods on two well-studied NPC combinatorial optimization problems: Traveling Salesman Problem (TSP) and Maximal Independent Set (MIS). Experimental results show that DIFUSCO strongly outperforms the previous state-of-the-art neural solvers, improving the performance gap between ground-truth and neural solvers from 1.76% to 0.46% on TSP-500, from 2.46% to 1.17% on TSP-1000, and from 3.19% to 2.58% on TSP10000. For the MIS problem, DIFUSCO outperforms the previous state-of-the-art neural solver on the challenging SATLIB benchmark., Comment: Accepted at NeurIPS 2023 (Spotlight)

Numerical simulation of non-linear partial differential equations plays a crucial role in modeling physical science and engineering phenomena, such as weather, climate, and aerodynamics. Recent Machine Learning (ML) models trained on low-resolution spatio-temporal signals have shown new promises in capturing important dynamics in high-resolution signals, under the condition that the models can effectively recover the missing details. However, this study shows that significant information is often lost in the low-resolution down-sampled features. To address such issues, we propose a new approach, namely Temporal Stencil Modeling (TSM), which combines the strengths of advanced time-series sequence modeling (with the HiPPO features) and state-of-the-art neural PDE solvers (with learnable stencil modeling). TSM aims to recover the lost information from the PDE trajectories and can be regarded as a temporal generalization of classic finite volume methods such as WENO. Our experimental results show that TSM achieves the new state-of-the-art simulation accuracy for 2-D incompressible Navier-Stokes turbulent flows: it significantly outperforms the previously reported best results by 19.9% in terms of the highly-correlated duration time and reduces the inference latency into 80%. We also show a strong generalization ability of the proposed method to various out-of-distribution turbulent flow settings. Our code is available at "https://github.com/Edward-Sun/TSM-PDE".

In this paper, we propose a via-less fully screen-printed reconfigurable intelligent surface which can establish a second line-of-sight communication from 23.5GHz to 29.5GHz. By serially connecting the H shaped resonator along the H field of the incident wave, we minimize the effect of the biasing lines and make a via-less design, which reduces the fabrication difficulty and cost. The unit-cell simulation of the array with screen-printed VO2 switches shows a 215{\deg} to 160{\deg} phase shift difference between the ON and OFF states within bandwidth. During the field testing of the ideal arrays, we verify that the array can redirect the 45{\deg} incident wave to 0{\deg} reflection with a signal enhancement of at least 10 dB as compared to the array which has all unit cells in the OFF condition., Comment: 2 pages, 3 figures, submitted to 2023 IEEE International Symposium on Antennas and Propagation

We propose a new class of linear Transformers called FourierLearner-Transformers (FLTs), which incorporate a wide range of relative positional encoding mechanisms (RPEs). These include regular RPE techniques applied for sequential data, as well as novel RPEs operating on geometric data embedded in higher-dimensional Euclidean spaces. FLTs construct the optimal RPE mechanism implicitly by learning its spectral representation. As opposed to other architectures combining efficient low-rank linear attention with RPEs, FLTs remain practical in terms of their memory usage and do not require additional assumptions about the structure of the RPE mask. Besides, FLTs allow for applying certain structural inductive bias techniques to specify masking strategies, e.g. they provide a way to learn the so-called local RPEs introduced in this paper and give accuracy gains as compared with several other linear Transformers for language modeling. We also thoroughly test FLTs on other data modalities and tasks, such as image classification, 3D molecular modeling, and learnable optimizers. To the best of our knowledge, for 3D molecular data, FLTs are the first Transformer architectures providing linear attention and incorporating RPE masking., Comment: AISTATS 2024

We study the possibility of probing the existence of a light, invisible, axion-like particle (ALP) $a$ in leptonic decays of the $Z$ boson at the proposed high-energy $e^+e^-$ colliders, CEPC and FCC-ee. Both projects plan to run at the $Z$ pole, collecting $10^{12}-10^{13}$ visible $Z$ decays. We show that, searching for the emission of an invisible ALP from leptons in leptonic $Z$ decays, this enormous statistics could allow to constrain the ALP couplings to leptons at an unprecedented level for laboratory experiments. In particular, within a Monte Carlo simulation framework, we estimate that CEPC/FCC-ee can be sensitive to the coupling of an invisible ALP to muons up to $f_a/C^A_{\mu\mu} \approx 1$ TeV - where $f_a$ is the ALP decay constant - corresponding to ${\rm BR}(Z \to \mu^+\mu^-\,a) \approx 3\times 10^{-11}$., Comment: 16 pages, 5 figures; v2, improved analysis and discussion, conclusions unchanged, version to appear in PRD; v3, typos corrected

Solid-liquid phase transitions are fundamental physical processes, but atomically-resolved microscopy has yet to capture both the solid and liquid dynamics for such a transition. We have developed a new technique for controlling the melting and freezing of 2D molecular layers on a graphene field-effect transistor (FET) that allows us to image phase transition dynamics via atomically-resolved scanning tunneling microscopy. Back-gate voltages applied to a F4TCNQ-decorated graphene FET induce reversible transitions between a charge-neutral solid phase and a negatively charged liquid phase. Nonequilibrium molecular melting dynamics are visualized by rapidly heating the graphene surface with electrical current and imaging the resulting evolution toward new equilibrium states. An analytical model has been developed that explains the observed equilibrium mixed-state phases based on spectroscopic measurement of both solid and liquid molecular energy levels. Observed non-equilibrium melting dynamics are consistent with Monte Carlo simulations.