Publications
AgentDiscover: Autonomous Discovery with Minimal Search Scaffolding
arXiv preprint, 2026
This paper introduces AgentDiscover, a framework that lets a coding agent plan its own search, run experiments, and build persistent memory to improve discovery across scientific and computational optimization tasks.
Reinforcement Learning for Diffusion LLMs with Entropy-Guided Step Selection and Stepwise Advantages
NeurIPS, 2026
This paper introduces EGSPO, an efficient reinforcement learning method for diffusion language models that uses entropy-guided denoising step selection and stepwise advantages to improve reasoning and code generation.
Diffusion Blend: Inference-Time Multi-Preference Alignment for Diffusion Models
ICLR, 2026
This paper proposes a method for inference-time alignment of diffusion models to multiple, possibly conflicting, user-specified preferences (rewards) without additional fine-tuning.
PowerMamba: A Deep State Space Model and Comprehensive Benchmark for Time Series Prediction in Electric Power Systems
IEEE Transaction on Power System, 2025
This paper introduces a deep state-space modelling approach and accompanying benchmark for multivariate time-series prediction in electric power systems, integrating high-resolution external forecasts and traditional dynamical structure.
Exploring the Capabilities and Limitations of Large Language Models in the Electric Energy Sector
Joule, 2024
This paper investigates how large language models (LLMs) perform in electric-energy systems tasks—identifying both their promise and their limitations for this safety-critical domain.