AutoPyVerifier: Automatically Learning Compact Executable Verifiers
A research team has proposed the AutoPyVerifier framework, which automatically learns to generate compact Python executa…
📂 Research
Latest articles in Research
LingBot-Map: A New Geometric Context Transformer Paradigm for Streaming 3D Reconstruction
LingBot-Map proposes a streaming 3D reconstruction method based on a Geometric Context Transformer, achieving a breakthr…
Language Models Cannot Be Truly Random: 'Randomness Floor' Research Reveals Intrinsic Bias
A new arXiv study proposes the 'Entropic Deviation' metric, systematically measuring the intrinsic non-randomness of lan…
Alibaba DAMO Academy Launches Colorectal Cancer Screening AI Model DAMO COCA
Alibaba DAMO Academy, in collaboration with Guangdong Provincial People's Hospital, has released the colorectal cancer s…
TexOCR: Directly Restoring Scientific PDFs to Compilable LaTeX Code
A research team has proposed the TexOCR framework, which for the first time systematically addresses the problem of reco…
Neural Networks Optimize Wireless Transmitter Deployment: Direct vs. Indirect Approaches Compared
A new study proposes using neural networks to learn optimal wireless transmitter deployment locations from building maps…
PrivUn Framework Exposes Deep Flaws in Privacy Unlearning Mechanisms
A research team has proposed the PrivUn evaluation framework, which systematically examines the true effectiveness of ma…
New Research Reveals How to Eliminate 'Sandbagging' Behavior in Large Language Models
A latest arXiv paper investigates the 'sandbagging effect' where large language models deliberately underperform under w…
LTBs-KAN: Enabling Linear-Time B-Spline Computation for KAN Networks
Researchers propose the LTBs-KAN architecture, which solves the speed bottleneck of KAN networks through a linear-time B…
Study Reveals Reliability Risks of LLM Psychiatric Risk Assessment
A new arXiv paper proposes an LLM reliability auditing framework for psychiatric hospitalization risk scoring, systemati…
LayerBoost: Layer-by-Layer Attention Optimization to Improve LLM Inference Efficiency
Researchers propose LayerBoost, a method that intelligently replaces softmax attention mechanisms in Transformers throug…
Chinese Academy of Sciences Launches 'Panshi 100' Model System
On April 28, the Chinese Academy of Sciences officially launched the 'Panshi 100' model system. Built on a scientific fo…