About
Research & Development
Unlocking LLM Potential: How PromptPilot Improves Prompt Engineering
TLDR: PromptPilot is an interactive AI assistant designed to help users, especially non-experts, create more effective prompts for large language models (LLMs). A study...
Research & Development
Enhancing Hierarchical Classification with Feature-Aware Contrastive Learning
TLDR: This paper introduces two novel hierarchical contrastive learning methods, G-HMLC and A-HMLC, that explicitly identify and leverage relevant features for each level of...
Research & Development
World-Language Models: Integrating Dynamic Priors for Enhanced Visual Understanding
TLDR: The research paper introduces World-Language Models (WorldLMs), exemplified by DynamicVisionAligner (DyVA), which integrate generative world models (like Stable Video Diffusion) as 'generative encoders'...
Research & Development
RiskPO: Enhancing LLM Reasoning by Tackling Challenging Problems with Risk-Based Optimization
TLDR: RiskPO is a novel post-training method for Large Language Models (LLMs) that addresses the limitations of existing mean-based reinforcement learning techniques, such as...
Research & Development
Drones Learn to Land Safely in Emergencies Using Advanced AI
TLDR: A new research paper introduces a hybrid AI pipeline enabling autonomous drones to make sudden, safe landing decisions in emergencies. It combines traditional...
Research & Development
DualTune: Enhancing On-Device AI Agents for Better Tool Calling
TLDR: DualTune is a novel methodology and inference framework designed to significantly improve the tool-calling capabilities of local, on-device Large Language Models (LLMs). It...