About
Research & Development
On-the-Fly LLM Improvement with Textual Self-Attention Networks
TLDR: The Textual Self-Attention Network (TSAN) is a novel method for enhancing Large Language Model (LLM) outputs at inference time, without requiring any model...
Research & Development
New Method Overcomes S-Wave Amplitude Suppression in Seismic Picking
TLDR: Deep learning models often fail to detect S-waves accurately due to "amplitude suppression," where predictions remain too low. This paper diagnoses this as...
Research & Development
Boosting LLM Performance with Implicit Federated In-Context Learning
TLDR: Implicit Federated In-Context Learning (IFed-ICL) is a novel framework that enhances Large Language Model (LLM) performance on specific tasks by leveraging private data...
Research & Development
Improving LLM Accuracy and Efficiency in Telecommunications Mathematics with Data Trajectory Alignment
TLDR: Data Trajectory Alignment (DTA) is a two-phase framework that adapts large language models (LLMs) for specialized domains like telecommunications mathematics. It synthesizes diverse...
Research & Development
AgentSUMO: Simplifying Traffic Simulation for Urban Planners
TLDR: AgentSUMO is a novel framework that leverages large language models (LLMs) to make complex urban traffic simulations, typically performed with SUMO, accessible to...
Research & Development
Advancing Online Reinforcement Learning with Trajectory-Level Flow Matching
TLDR: CtrlFlow is a new online reinforcement learning method that uses conditional flow matching to directly generate entire high-return trajectories, avoiding the cumulative errors...