Yuxuan Li

Yuxuan Li, James L. McClelland

Transformer networks have seen great success in natural language processing and machine vision, where task objectives such as next word prediction and image classification benefit from nuanced context sensitivity across high-dimensional inputs. However, there is an ongoing debate about how and when transformers can acquire highly structured behavior and achieve systematic generalization. Here, we explore how well a causal transformer can perform a set of algorithmic tasks, including copying, sorting, and hierarchical compositions of these operations. We demonstrate strong generalization to sequences longer than those used in training by replacing the standard positional encoding typically used in transformers with labels arbitrarily paired with items in the sequence. We search for the layer and head configuration sufficient to solve these tasks, then probe for signs of systematic processing in latent representations and attention patterns. We show that two-layer transformers learn reliable solutions to multi-level problems, develop signs of task decomposition, and encode input items in a way that encourages the exploitation of shared computation across related tasks. These results provide key insights into how attention layers support structured computation both within a task and across multiple tasks.

Ren Wang, Yuxuan Li, Can Chen et al.

Adversarial robustness is a critical measure of a neural network's ability to withstand adversarial attacks at inference time. While robust training techniques have improved defenses against individual $\ell_p$-norm attacks (e.g., $\ell_2$ or $\ell_\infty$), models remain vulnerable to diversified $\ell_p$ perturbations. To address this challenge, we propose a novel Robust Mode Connectivity (RMC)-oriented adversarial defense framework comprising two population-based learning phases. In Phase I, RMC searches the parameter space between two pre-trained models to construct a continuous path containing models with high robustness against multiple $\ell_p$ attacks. To improve efficiency, we introduce a Self-Robust Mode Connectivity (SRMC) module that accelerates endpoint generation in RMC. Building on RMC, Phase II presents RMC-based optimization, where RMC modules are composed to further enhance diversified robustness. To increase Phase II efficiency, we propose Efficient Robust Mode Connectivity (ERMC), which leverages $\ell_1$- and $\ell_\infty$-adversarially trained models to achieve robustness across a broad range of $p$-norms. An ensemble strategy is employed to further boost ERMC's performance. Extensive experiments across diverse datasets and architectures demonstrate that our methods significantly improve robustness against $\ell_\infty$, $\ell_2$, $\ell_1$, and hybrid attacks. Code is available at https://github.com/wangren09/MCGR.

Qun Dai, Chunyang Yuan, Yimian Dai et al.

Land Surface Temperature (LST) is a critical parameter for environmental studies, but directly obtaining high spatial resolution LST data remains challenging due to the spatio-temporal trade-off in satellite remote sensing. Guided LST downscaling has emerged as an alternative solution to overcome these limitations, but current methods often neglect spatial non-stationarity, and there is a lack of an open-source ecosystem for deep learning methods. In this paper, we propose the Modality-Conditional Large Selective Kernel (MoCoLSK) Network, a novel architecture that dynamically fuses multi-modal data through modality-conditioned projections. MoCoLSK achieves a confluence of dynamic receptive field adjustment and multi-modal feature fusion, leading to enhanced LST prediction accuracy. Furthermore, we establish the GrokLST project, a comprehensive open-source ecosystem featuring the GrokLST dataset, a high-resolution benchmark, and the GrokLST toolkit, an open-source PyTorch-based toolkit encapsulating MoCoLSK alongside 40+ state-of-the-art approaches. Extensive experimental results validate MoCoLSK's effectiveness in capturing complex dependencies and subtle variations within multispectral data, outperforming existing methods in LST downscaling. Our code, dataset, and toolkit are available at https://github.com/GrokCV/GrokLST.

Jianling Li, Shangzhan Li, Zhenye Gao et al.

Triton, a high-level Python-like language designed for building efficient GPU kernels, is widely adopted in deep learning frameworks due to its portability, flexibility, and accessibility. However, programming and parallel optimization still require considerable trial and error from Triton developers. Despite advances in large language models (LLMs) for conventional code generation, these models struggle to generate accurate, performance-optimized Triton code, as they lack awareness of its specifications and the complexities of GPU programming. More critically, there is an urgent need for systematic evaluations tailored to Triton. In this work, we introduce TritonBench, the first comprehensive benchmark for Triton operator generation. TritonBench features two evaluation channels: a curated set of 184 real-world operators from GitHub and a collection of operators aligned with PyTorch interfaces. Unlike conventional code benchmarks prioritizing functional correctness, TritonBench also profiles efficiency performance on widely deployed GPUs aligned with industry applications. Our study reveals that current state-of-the-art code LLMs struggle to generate efficient Triton operators, highlighting a significant gap in high-performance code generation. TritonBench will be available at https://github.com/thunlp/TritonBench.

Yuxuan Li, Ying Lin, Chenang Liu

Investigating the causal relationships between characteristics and expressions plays a critical role in healthcare analytics. Effective synthesis for expressions using given characteristics can make great contributions to health risk management and medical decision-making. For example, predicting the resulting physiological symptoms on patients from given treatment characteristics is helpful for the disease prevention and personalized treatment strategy design. Therefore, the objective of this study is to effectively synthesize the expressions based on given characteristics. However, the mapping from characteristics to expressions is usually from a relatively low dimension space to a high dimension space, but most of the existing methods such as regression models could not effectively handle such mapping. Besides, the relationship between characteristics and expressions may contain not only deterministic patterns, but also stochastic patterns. To address these challenges, this paper proposed a novel selective ensemble characteristic-to-expression synthesis (SE-CTES) approach inspired by generative adversarial network (GAN). The novelty of the proposed method can be summarized into three aspects: (1) GAN-based architecture for deep neural networks are incorporated to learn the relatively low dimensional mapping to high dimensional mapping containing both deterministic and stochastic patterns; (2) the weights of the two mismatching errors in the GAN-based architecture are proposed to be different to reduce the learning bias in the training process; and (3) a selective ensemble learning framework is proposed to reduce the prediction bias and improve the synthesis stability. To validate the effectiveness of the proposed approach, extensive numerical simulation studies and a real-world healthcare case study were applied and the results demonstrated that the proposed method is very promising.

Yuxuan Li, Luca Weihs · allen-ai

Pretrained representations from large-scale vision models have boosted the performance of downstream embodied policy learning. We look to understand whether additional self-supervised pretraining on exploration trajectories can build on these general-purpose visual representations to better support embodied planning in realistic environments. We evaluated four common auxiliary losses in embodied AI, two hindsight-based losses, and a standard imitation learning loss, by pretraining the agent's visual compression module and state belief representations with each objective and using CLIP as a representative visual backbone. The learned representations are then frozen for downstream multi-step evaluation on two goal-directed tasks. Surprisingly, we find that imitation learning on these exploration trajectories out-performs all other auxiliary losses even despite the exploration trajectories being dissimilar from the downstream tasks. This suggests that imitation of exploration may be ''all you need'' for building powerful planning representations. Additionally, we find that popular auxiliary losses can benefit from simple modifications to improve their support for downstream planning ability.

Nan Lin, Yuxuan Li, Yujun Zhu et al.

Traditionally, reinforcement learning methods predict the next action based on the current state. However, in many situations, directly applying actions to control systems or robots is dangerous and may lead to unexpected behaviors because action is rather low-level. In this paper, we propose a novel hierarchical reinforcement learning framework without explicit action. Our meta policy tries to manipulate the next optimal state and actual action is produced by the inverse dynamics model. To stabilize the training process, we integrate adversarial learning and information bottleneck into our framework. Under our framework, widely available state-only demonstrations can be exploited effectively for imitation learning. Also, prior knowledge and constraints can be applied to meta policy. We test our algorithm in simulation tasks and its combination with imitation learning. The experimental results show the reliability and robustness of our algorithms.

Yu-Xuan Li, Jin-Yuan Liu, Liang Li et al.

Recurrent neural networks have been widely used in sequence learning tasks. In previous studies, the performance of the model has always been improved by either wider or deeper structures. However, the former becomes more prone to overfitting, while the latter is difficult to optimize. In this paper, we propose a simple new model named Thick-Net, by expanding the network from another dimension: thickness. Multiple parallel values are obtained via more sets of parameters in each hidden state, and the maximum value is selected as the final output among parallel intermediate outputs. Notably, Thick-Net can efficiently avoid overfitting, and is easier to optimize than the vanilla structures due to the large dropout affiliated with it. Our model is evaluated on four sequential tasks including adding problem, permuted sequential MNIST, text classification and language modeling. The results of these tasks demonstrate that our model can not only improve accuracy with faster convergence but also facilitate a better generalization ability.