Yong Liu

Weiwen Liu, Xu Huang, Xingshan Zeng et al.

Function calling significantly extends the application boundary of large language models, where high-quality and diverse training data is critical for unlocking this capability. However, real function-calling data is quite challenging to collect and annotate, while synthetic data generated by existing pipelines tends to lack coverage and accuracy. In this paper, we present ToolACE, an automatic agentic pipeline designed to generate accurate, complex, and diverse tool-learning data. ToolACE leverages a novel self-evolution synthesis process to curate a comprehensive API pool of 26,507 diverse APIs. Dialogs are further generated through the interplay among multiple agents, guided by a formalized thinking process. To ensure data accuracy, we implement a dual-layer verification system combining rule-based and model-based checks. We demonstrate that models trained on our synthesized data, even with only 8B parameters, achieve state-of-the-art performance on the Berkeley Function-Calling Leaderboard, rivaling the latest GPT-4 models. Our model and a subset of the data are publicly available at https://huggingface.co/Team-ACE.

Anqi Chen, Yingda Cheng, Yong Liu et al.

We analyze the superconvergence properties of ultra-weak discontinuous Galerkin (UWDG) methods with various choices of flux parameters for one-dimensional linear Schrödinger equation. In our previous work [10], stability and optimal convergence rate are established for a large class of flux parameters. Depending on the flux choices and if the polynomial degree $k$ is even or odd, in this paper, we prove $2k$ or $(2k-1)$-th order superconvergence rate for cell averages and numerical flux of the function, as well as $(2k-1)$ or $(2k-2)$-th order for numerical flux of the derivative. In addition, we prove superconvergence of $(k+2)$ or $(k+3)$-th order of the DG solution towards a special projection. At a class of special points, the function values and the first and second order derivatives of the DG solution are superconvergent with order $k+2, k+1, k$, respectively. The proof relies on the correction function techniques initiated in [8], and applied to [6] for direct DG (DDG) methods for diffusion problems. Compared with [6], Schrödinger equation poses unique challenges for superconvergence proof because of the lack of the dissipation mechanism from the equation. One major highlight of our proof is that we introduce specially chosen test functions in the error equation and show the superconvergence of the second derivative and jump across the cell interfaces of the difference between numerical solution and projected exact solution. This technique was originally proposed in [12] and is essential to elevate the convergence order for our analysis. Finally, by negative norm estimates, we apply the post-processing technique and show that the accuracy of our scheme can be enhanced to order $2k.$ Theoretical results are verified by numerical experiments.

Manna Dai, Yang Jiang, Feng Yang et al.

Metasurfaces have widespread applications in fifth-generation (5G) microwave communication. Among the metasurface family, free-form metasurfaces excel in achieving intricate spectral responses compared to regular-shape counterparts. However, conventional numerical methods for free-form metasurfaces are time-consuming and demand specialized expertise. Alternatively, recent studies demonstrate that deep learning has great potential to accelerate and refine metasurface designs. Here, we present XGAN, an extended generative adversarial network (GAN) with a surrogate for high-quality free-form metasurface designs. The proposed surrogate provides a physical constraint to XGAN so that XGAN can accurately generate metasurfaces monolithically from input spectral responses. In comparative experiments involving 20000 free-form metasurface designs, XGAN achieves 0.9734 average accuracy and is 500 times faster than the conventional methodology. This method facilitates the metasurface library building for specific spectral responses and can be extended to various inverse design problems, including optical metamaterials, nanophotonic devices, and drug discovery.

Tianxin Huang, Qingyao Liu, Xiangrui Zhao et al.

As point clouds are 3D signals with permutation invariance, most existing works train their reconstruction networks by measuring shape differences with the average point-to-point distance between point clouds matched with predefined rules. However, the static matching rules may deviate from actual shape differences. Although some works propose dynamically-updated learnable structures to replace matching rules, they need more iterations to converge well. In this work, we propose a simple but effective reconstruction loss, named Learnable Chamfer Distance (LCD) by dynamically paying attention to matching distances with different weight distributions controlled with a group of learnable networks. By training with adversarial strategy, LCD learns to search defects in reconstructed results and overcomes the weaknesses of static matching rules, while the performances at low iterations can also be guaranteed by the basic matching algorithm. Experiments on multiple reconstruction networks confirm that LCD can help achieve better reconstruction performances and extract more representative representations with faster convergence and comparable training efficiency. The source codes are provided in https://github.com/Tianxinhuang/LCDNet.git.

Yong Liu, Lizhong Ding, Weiping Wang

In this paper, we study the statistical properties of kernel $k$-means and obtain a nearly optimal excess clustering risk bound, substantially improving the state-of-art bounds in the existing clustering risk analyses. We further analyze the statistical effect of computational approximations of the Nyström kernel $k$-means, and prove that it achieves the same statistical accuracy as the exact kernel $k$-means considering only $Ω(\sqrt{nk})$ Nyström landmark points. To the best of our knowledge, such sharp excess clustering risk bounds for kernel (or approximate kernel) $k$-means have never been proposed before.