Wenjin Zhang

Yiyou Sun, Xinyang Han, Weichen Zhang et al.

Recent AI systems have achieved strong results on a wide range of benchmarks, yet these gains have not translated into economically meaningful deployment across many professional domains. We argue that this gap is largely an evaluation problem: widely used benchmarks lack sustained performance measurement on real and economically valuable workflows. This paper introduces Agents' Last Exam (ALE), a benchmark designed to evaluate AI agents on long-horizon, economically valuable, real-world tasks with verifiable outcomes. Developed in collaboration with 250+ industry experts, ALE covers non-physical industries defined with reference to O*NET / SOC 2018 (the U.S. federal occupational taxonomy). It is organized around a task taxonomy with 55 subfields grouped into 13 industry clusters covering 1K+ tasks. Current results show that the hardest tier remains far from saturated: across mainstream harness and backbone configurations, the average full pass rate is 2.6%. ALE is designed as a living benchmark: its task pool grows continuously as new workflows and industries are onboarded. More broadly, ALE is intended not merely as another leaderboard, but as an instrument for closing the gap between benchmark success and GDP-relevant impact.

Long Phan, Alice Gatti, Ziwen Han et al. · amazon-science, apple-ml

Benchmarks are important tools for tracking the rapid advancements in large language model (LLM) capabilities. However, benchmarks are not keeping pace in difficulty: LLMs now achieve over 90\% accuracy on popular benchmarks like MMLU, limiting informed measurement of state-of-the-art LLM capabilities. In response, we introduce Humanity's Last Exam (HLE), a multi-modal benchmark at the frontier of human knowledge, designed to be the final closed-ended academic benchmark of its kind with broad subject coverage. HLE consists of 2,500 questions across dozens of subjects, including mathematics, humanities, and the natural sciences. HLE is developed globally by subject-matter experts and consists of multiple-choice and short-answer questions suitable for automated grading. Each question has a known solution that is unambiguous and easily verifiable, but cannot be quickly answered via internet retrieval. State-of-the-art LLMs demonstrate low accuracy and calibration on HLE, highlighting a significant gap between current LLM capabilities and the expert human frontier on closed-ended academic questions. To inform research and policymaking upon a clear understanding of model capabilities, we publicly release HLE at https://lastexam.ai.

Wenjin Zhang, Keyi Li, Sen Yang et al.

Conventional methods in semi-supervised learning (SSL) often face challenges related to limited data utilization, mainly due to their reliance on threshold-based techniques for selecting high-confidence unlabeled data during training. Various efforts (e.g., FreeMatch) have been made to enhance data utilization by tweaking the thresholds, yet none have managed to use 100% of the available data. To overcome this limitation and improve SSL performance, we introduce \algo, a novel algorithm that fully utilizes unlabeled data to boost semi-supervised learning. \algo integrates a self-supervised learning strategy, i.e., Masked Autoencoder (MAE), that uses all available data to enforce the visual representation learning. This enables the SSL algorithm to leverage all available data, including samples typically filtered out by traditional methods. In addition, we propose a synthetic data training approach to further increase data utilization and improve generalization. These innovations lead \algo to achieve state-of-the-art results on challenging datasets. For instance, on CIFAR-100 with 2 labels per class, STL-10 with 4 labels per class, and Euro-SAT with 2 labels per class, \algo achieves low error rates of 18.71%, 9.47%, and 3.07%, respectively. The code will be made publicly available.

Zhangjian Ji, Wenjin Zhang, Shaotong Qiao et al.

Human pose estimation has been widely applied in the human-centric understanding and generation, but most existing state-of-the-art human pose estimation methods require heavy computational resources for accurate predictions. In order to obtain an accurate, robust yet lightweight human pose estimator, one feasible way is to transfer pose knowledge from a powerful teacher model to a less-parameterized student model by knowledge distillation. However, the traditional knowledge distillation framework does not fully explore the contextual information among human joints. Thus, in this paper, we propose a novel coarse-to-fine two-stage knowledge distillation framework for human pose estimation. In the first-stage distillation, we introduce the human joints structure loss to mine the structural information among human joints so as to transfer high-level semantic knowledge from the teacher model to the student model. In the second-stage distillation, we utilize an Image-Guided Progressive Graph Convolutional Network (IGP-GCN) to refine the initial human pose obtained from the first-stage distillation and supervise the training of the IGP-GCN in the progressive way by the final output pose of teacher model. The extensive experiments on the benchmark dataset: COCO keypoint and CrowdPose datasets, show that our proposed method performs favorably against lots of the existing state-of-the-art human pose estimation methods, especially for the more complex CrowdPose dataset, the performance improvement of our model is more significant.

Nikolai Warner, Wenjin Zhang, Irfan Essa et al. · stanford

Lifting-based methods for 3D Human Pose Estimation (HPE), which predict 3D poses from detected 2D keypoints, often generalize poorly to new datasets and real-world settings. To address this, we propose \emph{AugLift}, a simple yet effective reformulation of the standard lifting pipeline that significantly improves generalization performance without requiring additional data collection or sensors. AugLift sparsely enriches the standard input -- the 2D keypoint coordinates $(x, y)$ -- by augmenting it with a keypoint detection confidence score $c$ and a corresponding depth estimate $d$. These additional signals are computed from the image using off-the-shelf, pre-trained models (e.g., for monocular depth estimation), thereby inheriting their strong generalization capabilities. Importantly, AugLift serves as a modular add-on and can be readily integrated into existing lifting architectures. Our extensive experiments across four datasets demonstrate that AugLift boosts cross-dataset performance on unseen datasets by an average of $10.1\%$, while also improving in-distribution performance by $4.0\%$. These gains are consistent across various lifting architectures, highlighting the robustness of our method. Our analysis suggests that these sparse, keypoint-aligned cues provide robust frame-level context, offering a practical way to significantly improve the generalization of any lifting-based pose estimation model. Code will be made publicly available.