Qiang Fu

Jiate Liu, Yiqin Zhu, Kaiwen Xiao et al.

The goal of program synthesis, or code generation, is to generate executable code based on given descriptions. Recently, there has been an increasing number of studies employing reinforcement learning (RL) to improve the performance of large language models (LLMs) for code. However, current representative works either rely solely on offline frameworks, limiting the exploration of new sample spaces, or fall short in the utilization of unit test signals, not accounting for specific error locations within the code. To address these issues, we propose RLTF, i.e., Reinforcement Learning from Unit Test Feedback, a novel online RL framework with unit test feedback of multi-granularity for refining code LLMs. Our approach generates data in real-time during training and simultaneously utilizes fine-grained feedback signals to guide the model towards producing higher-quality code. Extensive experiments show that RLTF achieves state-of-the-art performance on the APPS and the MBPP benchmarks. Our code is available at: https://github.com/Zyq-scut/RLTF.

Yifei Li, Zeqi Lin, Shizhuo Zhang et al.

Few-shot learning is a challenging task that requires language models to generalize from limited examples. Large language models like GPT-3 and PaLM have made impressive progress in this area, but they still face difficulties in reasoning tasks such as GSM8K, a benchmark for arithmetic problems. To improve their reasoning skills, previous work has proposed to guide the language model with prompts that elicit a series of reasoning steps before giving the final answer, achieving a significant improvement on GSM8K from 17.9% to 58.1% in problem-solving rate. In this paper, we present DIVERSE (Diverse Verifier on Reasoning Step), a novel approach that further enhances the reasoning capability of language models. DIVERSE has three main components: first, it generates diverse prompts to explore different reasoning paths for the same question; second, it uses a verifier to filter out incorrect answers based on a weighted voting scheme; and third, it verifies each reasoning step individually instead of the whole chain. We evaluate DIVERSE on the latest language model code-davinci-002 and show that it achieves new state-of-the-art results on six of eight reasoning benchmarks (e.g., GSM8K 74.4% to 83.2%).

Hua Wei, Jingxiao Chen, Xiyang Ji et al.

This paper introduces Honor of Kings Arena, a reinforcement learning (RL) environment based on Honor of Kings, one of the world's most popular games at present. Compared to other environments studied in most previous work, ours presents new generalization challenges for competitive reinforcement learning. It is a multi-agent problem with one agent competing against its opponent; and it requires the generalization ability as it has diverse targets to control and diverse opponents to compete with. We describe the observation, action, and reward specifications for the Honor of Kings domain and provide an open-source Python-based interface for communicating with the game engine. We provide twenty target heroes with a variety of tasks in Honor of Kings Arena and present initial baseline results for RL-based methods with feasible computing resources. Finally, we showcase the generalization challenges imposed by Honor of Kings Arena and possible remedies to the challenges. All of the software, including the environment-class, are publicly available at https://github.com/tencent-ailab/hok_env . The documentation is available at https://aiarena.tencent.com/hok/doc/ .

Zichuan Lin, Xiapeng Wu, Mingfei Sun et al.

Recent success in Deep Reinforcement Learning (DRL) methods has shown that policy optimization with respect to an off-policy distribution via importance sampling is effective for sample reuse. In this paper, we show that the use of importance sampling could introduce high variance in the objective estimate. Specifically, we show in a principled way that the variance of importance sampling estimate grows quadratically with importance ratios and the large ratios could consequently jeopardize the effectiveness of surrogate objective optimization. We then propose a technique called sample dropout to bound the estimation variance by dropping out samples when their ratio deviation is too high. We instantiate this sample dropout technique on representative policy optimization algorithms, including TRPO, PPO, and ESPO, and demonstrate that it consistently boosts the performance of those DRL algorithms on both continuous and discrete action controls, including MuJoCo, DMControl and Atari video games. Our code is open-sourced at \url{https://github.com/LinZichuan/sdpo.git}.

Jian Yao, Weiming Liu, Haobo Fu et al.

Policy-Space Response Oracles (PSRO) is an influential algorithm framework for approximating a Nash Equilibrium (NE) in multi-agent non-transitive games. Many previous studies have been trying to promote policy diversity in PSRO. A major weakness in existing diversity metrics is that a more diverse (according to their diversity metrics) population does not necessarily mean (as we proved in the paper) a better approximation to a NE. To alleviate this problem, we propose a new diversity metric, the improvement of which guarantees a better approximation to a NE. Meanwhile, we develop a practical and well-justified method to optimize our diversity metric using only state-action samples. By incorporating our diversity regularization into the best response solving in PSRO, we obtain a new PSRO variant, Policy Space Diversity PSRO (PSD-PSRO). We present the convergence property of PSD-PSRO. Empirically, extensive experiments on various games demonstrate that PSD-PSRO is more effective in producing significantly less exploitable policies than state-of-the-art PSRO variants.

Wendong Bi, Lun Du, Qiang Fu et al.

Graph Neural Networks (GNNs) have shown expressive performance on graph representation learning by aggregating information from neighbors. Recently, some studies have discussed the importance of modeling neighborhood distribution on the graph. However, most existing GNNs aggregate neighbors' features through single statistic (e.g., mean, max, sum), which loses the information related to neighbor's feature distribution and therefore degrades the model performance. In this paper, inspired by the method of moment in statistical theory, we propose to model neighbor's feature distribution with multi-order moments. We design a novel GNN model, namely Mix-Moment Graph Neural Network (MM-GNN), which includes a Multi-order Moment Embedding (MME) module and an Element-wise Attention-based Moment Adaptor module. MM-GNN first calculates the multi-order moments of the neighbors for each node as signatures, and then use an Element-wise Attention-based Moment Adaptor to assign larger weights to important moments for each node and update node representations. We conduct extensive experiments on 15 real-world graphs (including social networks, citation networks and web-page networks etc.) to evaluate our model, and the results demonstrate the superiority of MM-GNN over existing state-of-the-art models.

Tiantian Zhang, Zichuan Lin, Yuxing Wang et al.

A key challenge of continual reinforcement learning (CRL) in dynamic environments is to promptly adapt the RL agent's behavior as the environment changes over its lifetime, while minimizing the catastrophic forgetting of the learned information. To address this challenge, in this article, we propose DaCoRL, i.e., dynamics-adaptive continual RL. DaCoRL learns a context-conditioned policy using progressive contextualization, which incrementally clusters a stream of stationary tasks in the dynamic environment into a series of contexts and opts for an expandable multihead neural network to approximate the policy. Specifically, we define a set of tasks with similar dynamics as an environmental context and formalize context inference as a procedure of online Bayesian infinite Gaussian mixture clustering on environment features, resorting to online Bayesian inference to infer the posterior distribution over contexts. Under the assumption of a Chinese restaurant process prior, this technique can accurately classify the current task as a previously seen context or instantiate a new context as needed without relying on any external indicator to signal environmental changes in advance. Furthermore, we employ an expandable multihead neural network whose output layer is synchronously expanded with the newly instantiated context, and a knowledge distillation regularization term for retaining the performance on learned tasks. As a general framework that can be coupled with various deep RL algorithms, DaCoRL features consistent superiority over existing methods in terms of the stability, overall performance and generalization ability, as verified by extensive experiments on several robot navigation and MuJoCo locomotion tasks.

Shengnan An, Zeqi Lin, Bei Chen et al.

Abstraction is a desirable capability for deep learning models, which means to induce abstract concepts from concrete instances and flexibly apply them beyond the learning context. At the same time, there is a lack of clear understanding about both the presence and further characteristics of this capability in deep learning models. In this paper, we introduce a systematic probing framework to explore the abstraction capability of deep learning models from a transferability perspective. A set of controlled experiments are conducted based on this framework, providing strong evidence that two probed pre-trained language models (PLMs), T5 and GPT2, have the abstraction capability. We also conduct in-depth analysis, thus shedding further light: (1) the whole training phase exhibits a "memorize-then-abstract" two-stage process; (2) the learned abstract concepts are gathered in a few middle-layer attention heads, rather than being evenly distributed throughout the model; (3) the probed abstraction capabilities exhibit robustness against concept mutations, and are more robust to low-level/source-side mutations than high-level/target-side ones; (4) generic pre-training is critical to the emergence of abstraction capability, and PLMs exhibit better abstraction with larger model sizes and data scales.

Haoxuan Pan, Deheng Ye, Xiaoming Duan et al.

We revisit the estimation bias in policy gradients for the discounted episodic Markov decision process (MDP) from Deep Reinforcement Learning (DRL) perspective. The objective is formulated theoretically as the expected returns discounted over the time horizon. One of the major policy gradient biases is the state distribution shift: the state distribution used to estimate the gradients differs from the theoretical formulation in that it does not take into account the discount factor. Existing discussion of the influence of this bias was limited to the tabular and softmax cases in the literature. Therefore, in this paper, we extend it to the DRL setting where the policy is parameterized and demonstrate how this bias can lead to suboptimal policies theoretically. We then discuss why the empirically inaccurate implementations with shifted state distribution can still be effective. We show that, despite such state distribution shift, the policy gradient estimation bias can be reduced in the following three ways: 1) a small learning rate; 2) an adaptive-learning-rate-based optimizer; and 3) KL regularization. Specifically, we show that a smaller learning rate, or, an adaptive learning rate, such as that used by Adam and RSMProp optimizers, makes the policy optimization robust to the bias. We further draw connections between optimizers and the optimization regularization to show that both the KL and the reverse KL regularization can significantly rectify this bias. Moreover, we provide extensive experiments on continuous control tasks to support our analysis. Our paper sheds light on how successful PG algorithms optimize policies in the DRL setting, and contributes insights into the practical issues in DRL.

Ren-Jian Wang, Ke Xue, Yutong Wang et al.

Diversity plays a significant role in many problems, such as ensemble learning, reinforcement learning, and combinatorial optimization. How to define the diversity measure is a longstanding problem. Many methods rely on expert experience to define a proper behavior space and then obtain the diversity measure, which is, however, challenging in many scenarios. In this paper, we propose the problem of learning a behavior space from human feedback and present a general method called Diversity from Human Feedback (DivHF) to solve it. DivHF learns a behavior descriptor consistent with human preference by querying human feedback. The learned behavior descriptor can be combined with any distance measure to define a diversity measure. We demonstrate the effectiveness of DivHF by integrating it with the Quality-Diversity optimization algorithm MAP-Elites and conducting experiments on the QDax suite. The results show that the behavior learned by DivHF is much more consistent with human requirements than the one learned by direct data-driven approaches without human feedback, and makes the final solutions more diverse under human preference. Our contributions include formulating the problem, proposing the DivHF method, and demonstrating its effectiveness through experiments.

Hang Xu, Kai Li, Haobo Fu et al.

Counterfactual regret minimization (CFR) is a family of algorithms for effectively solving imperfect-information games. To enhance CFR's applicability in large games, researchers use neural networks to approximate its behavior. However, existing methods are mainly based on vanilla CFR and struggle to effectively integrate more advanced CFR variants. In this work, we propose an efficient model-free neural CFR algorithm, overcoming the limitations of existing methods in approximating advanced CFR variants. At each iteration, it collects variance-reduced sampled advantages based on a value network, fits cumulative advantages by bootstrapping, and applies discounting and clipping operations to simulate the update mechanisms of advanced CFR variants. Experimental results show that, compared with model-free neural algorithms, it exhibits faster convergence in typical imperfect-information games and demonstrates stronger adversarial performance in a large poker game.

Hang Xu, Kai Li, Bingyun Liu et al.

Counterfactual regret minimization (CFR) is a family of algorithms for effectively solving imperfect-information games. It decomposes the total regret into counterfactual regrets, utilizing local regret minimization algorithms, such as Regret Matching (RM) or RM+, to minimize them. Recent research establishes a connection between Online Mirror Descent (OMD) and RM+, paving the way for an optimistic variant PRM+ and its extension PCFR+. However, PCFR+ assigns uniform weights for each iteration when determining regrets, leading to substantial regrets when facing dominated actions. This work explores minimizing weighted counterfactual regret with optimistic OMD, resulting in a novel CFR variant PDCFR+. It integrates PCFR+ and Discounted CFR (DCFR) in a principled manner, swiftly mitigating negative effects of dominated actions and consistently leveraging predictions to accelerate convergence. Theoretical analyses prove that PDCFR+ converges to a Nash equilibrium, particularly under distinct weighting schemes for regrets and average strategies. Experimental results demonstrate PDCFR+'s fast convergence in common imperfect-information games. The code is available at https://github.com/rpSebastian/PDCFRPlus.

Yizhou Zhang, Lun Du, Defu Cao et al.

Foundation models, such as Large language Models (LLMs), have attracted significant amount of interest due to their large number of applications. However, when handling tasks involving repetitive sub-tasks and/or deceptive contents, such as arithmetic calculation and article-level fake news detection, simple instructional prompts suffer from inaccurate responses. Existing works show that more complicated prompting strategies, such as Chain-of-Thoughts and Least-to-Most, can unlock LLM's powerful capacity in diverse areas. Recent researches reveal that simple divide-and-conquer prompting strategy, i.e. simply dividing the input sequence to multiple sub-inputs, can also substantially improve LLM's performance in some specific tasks such as misinformation detection. In this paper, we aim at examining the utility of divide-and-conquer prompting strategy and answer on which kind of tasks this strategy gets advantages. Specifically, we provide a theoretic analysis to divide-and-conquer prompting strategy and help us identify the specific tasks where DaC prompting can bring performance boost with theoretic guarantee. We then present two cases (large integer arithmetic and fact verification) where experimental results aligns with our theoretic analysis.

Guangyu Zhao, Kewei Lian, Haowei Lin et al. · pku

A glowing body of work has emerged focusing on instruction-following policies for open-world agents, aiming to better align the agent's behavior with human intentions. However, the performance of these policies is highly susceptible to the initial prompt, which leads to extra efforts in selecting the best instructions. We propose a framework named Preference Goal Tuning (PGT). PGT allows an instruction following policy to interact with the environment to collect several trajectories, which will be categorized into positive and negative samples based on preference. Then we use preference learning to fine-tune the initial goal latent representation with the categorized trajectories while keeping the policy backbone frozen. The experiment result shows that with minimal data and training, PGT achieves an average relative improvement of 72.0% and 81.6% over 17 tasks in 2 different foundation policies respectively, and outperforms the best human-selected instructions. Moreover, PGT surpasses full fine-tuning in the out-of-distribution (OOD) task-execution environments by 13.4%, indicating that our approach retains strong generalization capabilities. Since our approach stores a single latent representation for each task independently, it can be viewed as an efficient method for continual learning, without the risk of catastrophic forgetting or task interference. In short, PGT enhances the performance of agents across nearly all tasks in the Minecraft Skillforge benchmark and demonstrates robustness to the execution environment.

Lun Du, Xiaozhou Shi, Qiang Fu et al.

Graph Neural Networks (GNNs) are widely used on a variety of graph-based machine learning tasks. For node-level tasks, GNNs have strong power to model the homophily property of graphs (i.e., connected nodes are more similar) while their ability to capture the heterophily property is often doubtful. This is partially caused by the design of the feature transformation with the same kernel for the nodes in the same hop and the followed aggregation operator. One kernel cannot model the similarity and the dissimilarity (i.e., the positive and negative correlation) between node features simultaneously even though we use attention mechanisms like Graph Attention Network (GAT), since the weight calculated by attention is always a positive value. In this paper, we propose a novel GNN model based on a bi-kernel feature transformation and a selection gate. Two kernels capture homophily and heterophily information respectively, and the gate is introduced to select which kernel we should use for the given node pairs. We conduct extensive experiments on various datasets with different homophily-heterophily properties. The experimental results show consistent and significant improvements against state-of-the-art GNN methods.

Ziteng Wang, Yueyue Na, Biao Tian et al.

Acoustic echo cancellation (AEC), noise suppression (NS) and automatic gain control (AGC) are three often required modules for real-time communications (RTC). This paper proposes a neural network supported algorithm for RTC, namely NN3A, which incorporates an adaptive filter and a multi-task model for residual echo suppression, noise reduction and near-end speech activity detection. The proposed algorithm is shown to outperform both a method using separate models and an end-to-end alternative. It is further shown that there exists a trade-off in the model between residual suppression and near-end speech distortion, which could be balanced by a novel loss weighting function. Several practical aspects of training the joint model are also investigated to push its performance to limit.

Qiang Fu, Francis Grady, Bjoern Flemming Broberg et al.

We need ways to improve the code quality. Programmers have different level of tenure and experience. Standard and programming languages change and we are forced to re-use legacy code with minimum revision. Programmers develop their habits and can be slow to incorporate new technologies to simplify the code or improve the performance. We rolled out our customized code review and pair programming process to address these issues. The paper discusses the about the improvement of mandatory code review and pair programming practiced in the commercial software development, and also proposes effective approaches to customize the code review and pair programming to avoid the pitfalls and keep the benefits.