Wei Wei | Tharindu Cyril Weerasooriya

HUMANLM: Simulating Users with State Alignment Beats Response Imitation

Thu, 01 Jan 2026 00:00:00 +0000

MultiGA: Leveraging Multi-Source Seeding in Genetic Algorithms

Wed, 01 Jan 2025 00:00:00 +0000

Different LLMs (like GPT-4, Claude, DeepSeek) are good at different things. Instead of picking one model and hoping for the best, MultiGA uses genetic algorithms—inspired by natural evolution—to combine outputs from multiple LLMs. It starts by generating diverse answers from several models, evaluates which ones are best, then “breeds” improved solutions by mixing and mutating them over multiple generations. We tested this on text-to-SQL generation, trip planning, and grad-level science questions, and found it matches or beats the best single model. Think of it as getting a second (and third, and fourth) opinion from different AI experts, then synthesizing their best ideas.

ProRefine: Inference-Time Prompt Refinement with Textual Feedback

Wed, 01 Jan 2025 00:00:00 +0000

When multiple AI agents work together (like one agent planning and another executing), they rely heavily on good prompts. But writing perfect prompts is hard, and even small mistakes can cascade through the system, causing failures. ProRefine solves this by creating a feedback loop: one AI agent evaluates how well the prompt worked and suggests improvements, then another agent refines the prompt based on that feedback—all happening automatically at inference time without any training.

We tested ProRefine on mathematical reasoning tasks and it improved accuracy by 3-37% compared to standard methods. Even better, it lets smaller, cheaper AI models perform nearly as well as much larger, expensive ones. This makes powerful AI more accessible and cost-effective, especially for complex multi-agent systems that are becoming critical in commercial applications.

Harnessing Business and Media Insights with Large Language Models

Mon, 01 Jan 2024 00:00:00 +0000