arXiv subjects

Donghan Yang

Publications and source records attributed to Donghan Yang.

3 recordsLinked to original sources

arXiv2025

MedAgentGym: A Scalable Agentic Training Environment for Code-Centric Reasoning in Biomedical Data Science

We introduce MedAgentGym, a scalable and interactive training environment designed to enhance coding-based biomedical reasoning capabilities in large language model (LLM) agents. MedAgentGym comprises 72,413 task instances across 129 categories derived from 12 authentic real-world biomedical scenarios. Tasks are encapsulated within executable sandbox environments, each featuring detailed task specifications, interactive feedback mechanisms, verifiable ground truth annotations, and scalable training trajectory generation. Extensive benchmarking of 29 LLMs reveals substantial performance disparities in biomedical data science between commercial and open-source LLMs. Leveraging efficient multi-threaded and multi-turn trajectory sampling in MedAgentGym, Med-Copilot achieves performance gains of +43.02% and +45.28% from offline and online reinforcement learning, respectively, demonstrating MedAgentGym as an effective training ground while establishing itself as a cost-effective, privacy-preserving alternative competitive with proprietary LLMs (gpt-4o). By offering a unified execution environment with a comprehensive benchmark and accessible, extensible training resources, MedAgentGym delivers an integrated platform to develop LLM-based coding assistants for advanced biomedical data science.

cs.CL↗

arXiv2022

SGM-Net: Semantic Guided Matting Net

Human matting refers to extracting human parts from natural images with high quality, including human detail information such as hair, glasses, hat, etc. This technology plays an essential role in image synthesis and visual effects in the film industry. When the green screen is not available, the existing human matting methods need the help of additional inputs (such as trimap, background image, etc.), or the model with high computational cost and complex network structure, which brings great difficulties to the application of human matting in practice. To alleviate such problems, most existing methods (such as MODNet) use multi-branches to pave the way for matting through segmentation, but these methods do not make full use of the image features and only utilize the prediction results of the network as guidance information. Therefore, we propose a module to generate foreground probability map and add it to MODNet to obtain Semantic Guided Matting Net (SGM-Net). Under the condition of only one image, we can realize the human matting task. We verify our method on the P3M-10k dataset. Compared with the benchmark, our method has significantly improved in various evaluation indicators.

cs.CV↗

arXiv2016

23Na and 1H NMR Relaxometry of Shale at High Magnetic Field

Formation evaluation of unconventional reservoirs is challenging due to the coexistence of different phases such as kerogen, bitumen, movable and bound light hydrocarbon and water. Current low-frequency (0.05 T) nuclear magnetic resonance (NMR) laboratory and logging methods are incapable of quantitatively separating the different phases. We demonstrate the utility of high-field (9 T) NMR 2D T1-T2 measurements for separating hydrocarbon and the clay-interacting aqueous phases in shale based on the difference in the frequency dependence of the spin-lattice relaxation time. Furthermore, we demonstrate 23Na NMR as a promising complementary technique to conventional 1H NMR for shale fluid typing, taking advantage of the fact that sodium ions are only present in the aqueous phase. We validate high-field (9 T) 23Na-1H NMR relaxometry for assessing brine-filled porosity and brine salinity in various porous materials, including porous glass, conventional rocks, clays, and shale, and apply it for differentiating hydrocarbon versus aqueous components and also the clay-associated versus free water in Eagle Ford shale cores. This work lays the groundwork for developing future downhole 23Na-1H NMR logging techniques.

physics.geo-ph↗