Bo Shen

Fenia Christopoulou, Gerasimos Lampouras, Milan Gritta et al.

We present PanGu-Coder, a pretrained decoder-only language model adopting the PanGu-Alpha architecture for text-to-code generation, i.e. the synthesis of programming language solutions given a natural language problem description. We train PanGu-Coder using a two-stage strategy: the first stage employs Causal Language Modelling (CLM) to pre-train on raw programming language data, while the second stage uses a combination of Causal Language Modelling and Masked Language Modelling (MLM) training objectives that focus on the downstream task of text-to-code generation and train on loosely curated pairs of natural language program definitions and code functions. Finally, we discuss PanGu-Coder-FT, which is fine-tuned on a combination of competitive programming problems and code with continuous integration tests. We evaluate PanGu-Coder with a focus on whether it generates functionally correct programs and demonstrate that it achieves equivalent or better performance than similarly sized models, such as CodeX, while attending a smaller context window and training on less data.

Bo Shen, Weijun Xie, Zhenyu Kong

The objective of this study is to address the problem of background/foreground separation with missing pixels by combining the video acquisition, video recovery, background/foreground separation into a single framework. To achieve this, a smooth robust tensor completion (SRTC) model is proposed to recover the data and decompose it into the static background and smooth foreground, respectively. Specifically, the static background is modeled by the low-rank tucker decomposition and the smooth foreground (moving objects) is modeled by the spatiotemporal continuity, which is enforced by the total variation regularization. An efficient algorithm based on tensor proximal alternating minimization (tenPAM) is implemented to solve the proposed model with global convergence guarantee under very mild conditions. Extensive experiments on real data demonstrate that the proposed method significantly outperforms the state-of-the-art approaches for background/foreground separation with missing pixels.

Lifu Wang, Tianyu Wang, Shengwei Yi et al.

We study the learning ability of linear recurrent neural networks with Gradient Descent. We prove the first theoretical guarantee on linear RNNs to learn any stable linear dynamic system using any a large type of loss functions. For an arbitrary stable linear system with a parameter $ρ_C$ related to the transition matrix $C$, we show that despite the non-convexity of the parameter optimization loss if the width of the RNN is large enough (and the required width in hidden layers does not rely on the length of the input sequence), a linear RNN can provably learn any stable linear dynamic system with the sample and time complexity polynomial in $\frac{1}{1-ρ_C}$. Our results provide the first theoretical guarantee to learn a linear RNN and demonstrate how can the recurrent structure help to learn a dynamic system.

Daoguang Zan, Ailun Yu, Wei Liu et al.

The impressive performance of large language models (LLMs) on code-related tasks has shown the potential of fully automated software development. In light of this, we introduce a new software engineering task, namely Natural Language to code Repository (NL2Repo). This task aims to generate an entire code repository from its natural language requirements. To address this task, we propose a simple yet effective framework CodeS, which decomposes NL2Repo into multiple sub-tasks by a multi-layer sketch. Specifically, CodeS includes three modules: RepoSketcher, FileSketcher, and SketchFiller. RepoSketcher first generates a repository's directory structure for given requirements; FileSketcher then generates a file sketch for each file in the generated structure; SketchFiller finally fills in the details for each function in the generated file sketch. To rigorously assess CodeS on the NL2Repo task, we carry out evaluations through both automated benchmarking and manual feedback analysis. For benchmark-based evaluation, we craft a repository-oriented benchmark, SketchEval, and design an evaluation metric, SketchBLEU. For feedback-based evaluation, we develop a VSCode plugin for CodeS and engage 30 participants in conducting empirical studies. Extensive experiments prove the effectiveness and practicality of CodeS on the NL2Repo task.

Haiyan Zhao, Xuansheng Wu, Fan Yang et al.

Linear concept vectors effectively steer LLMs, but existing methods suffer from noisy features in diverse datasets that undermine steering robustness. We propose Sparse Autoencoder-Denoised Concept Vectors (SDCV), which selectively keep the most discriminative SAE latents while reconstructing hidden representations. Our key insight is that concept-relevant signals can be explicitly separated from dataset noise by scaling up activations of top-k latents that best differentiate positive and negative samples. Applied to linear probing and difference-in-mean, SDCV consistently improves steering success rates by 4-16\% across six challenging concepts, while maintaining topic relevance.

Lifu Wang, Bo Shen, Bo Hu et al.

Recurrent Neural Network (RNN) is a fundamental structure in deep learning. Recently, some works study the training process of over-parameterized neural networks, and show that over-parameterized networks can learn functions in some notable concept classes with a provable generalization error bound. In this paper, we analyze the training and generalization for RNNs with random initialization, and provide the following improvements over recent works: 1) For a RNN with input sequence $x=(X_1,X_2,...,X_L)$, previous works study to learn functions that are summation of $f(β^T_lX_l)$ and require normalized conditions that $||X_l||\leqε$ with some very small $ε$ depending on the complexity of $f$. In this paper, using detailed analysis about the neural tangent kernel matrix, we prove a generalization error bound to learn such functions without normalized conditions and show that some notable concept classes are learnable with the numbers of iterations and samples scaling almost-polynomially in the input length $L$. 2) Moreover, we prove a novel result to learn N-variables functions of input sequence with the form $f(β^T[X_{l_1},...,X_{l_N}])$, which do not belong to the "additive" concept class, i,e., the summation of function $f(X_l)$. And we show that when either $N$ or $l_0=\max(l_1,..,l_N)-\min(l_1,..,l_N)$ is small, $f(β^T[X_{l_1},...,X_{l_N}])$ will be learnable with the number iterations and samples scaling almost-polynomially in the input length $L$.

Lifu Wang, Bo Shen, Ning Zhao et al.

The residual network is now one of the most effective structures in deep learning, which utilizes the skip connections to ``guarantee" the performance will not get worse. However, the non-convexity of the neural network makes it unclear whether the skip connections do provably improve the learning ability since the nonlinearity may create many local minima. In some previous works \cite{freeman2016topology}, it is shown that despite the non-convexity, the loss landscape of the two-layer ReLU network has good properties when the number $m$ of hidden nodes is very large. In this paper, we follow this line to study the topology (sub-level sets) of the loss landscape of deep ReLU neural networks with a skip connection and theoretically prove that the skip connection network inherits the good properties of the two-layer network and skip connections can help to control the connectedness of the sub-level sets, such that any local minima worse than the global minima of some two-layer ReLU network will be very ``shallow". The ``depth" of these local minima are at most $O(m^{(η-1)/n})$, where $n$ is the input dimension, $η<1$. This provides a theoretical explanation for the effectiveness of the skip connection in deep learning.

Lifu Wang, Bo Shen, Ning Zhao

In machine learning, asynchronous parallel stochastic gradient descent (APSGD) is broadly used to speed up the training process through multi-workers. Meanwhile, the time delay of stale gradients in asynchronous algorithms is generally proportional to the total number of workers, which brings additional deviation from the accurate gradient due to using delayed gradients. This may have a negative influence on the convergence of the algorithm. One may ask: How many workers can we use at most to achieve a good convergence and the linear speedup? In this paper, we consider the second-order convergence of asynchronous algorithms in non-convex optimization. We investigate the behaviors of APSGD with consistent read near strictly saddle points and provide a theoretical guarantee that if the total number of workers is bounded by $\widetilde{O}(K^{1/3}M^{-1/3})$ ($K$ is the total steps and $M$ is the mini-batch size), APSGD will converge to good stationary points ($||\nabla f(x)||\leq ε, \nabla^2 f(x)\succeq -\sqrtε\bm{I}, ε^2\leq O(\sqrt{\frac{1}{MK}}) $) and the linear speedup is achieved. Our works give the first theoretical guarantee on the second-order convergence for asynchronous algorithms. The technique we provide can be generalized to analyze other types of asynchronous algorithms to understand the behaviors of asynchronous algorithms in distributed asynchronous parallel training.

Bo Shen, Wei Zhang, Haiyan Zhao et al.

The pictorial jigsaw (PJ) puzzle is a well-known leisure game for humans. Usually, a PJ puzzle game is played by one or several human players face-to-face in the physical space. In this paper, we focus on how to solve PJ puzzles in the cyberspace by a group of physically distributed human players. We propose an approach to solving PJ puzzle by stigmergy-inspired Internet-based human collective intelligence. The core of the approach is a continuously executing loop, named the EIF loop, which consists of three activities: exploration, integration, and feedback. In exploration, each player tries to solve the PJ puzzle alone, without direct interactions with other players. At any time, the result of a player's exploration is a partial solution to the PJ puzzle, and a set of rejected neighboring relation between pieces. The results of all players' exploration are integrated in real time through integration, with the output of a continuously updated collective opinion graph (COG). And through feedback, each player is provided with personalized feedback information based on the current COG and the player's exploration result, in order to accelerate his/her puzzle-solving process. Exploratory experiments show that: (1) supported by this approach, the time to solve PJ puzzle is nearly linear to the reciprocal of the number of players, and shows better scalability to puzzle size than that of face-to-face collaboration for 10-player groups; (2) for groups with 2 to 10 players, the puzzle-solving time decreases 31.36%-64.57% on average, compared with the best single players in the experiments.