About
Research & Development
Improving LLM Consensus: New Algorithms Outperform Majority Voting by Analyzing Model Strengths and Correlations
TLDR: This research paper introduces two novel algorithms, Optimal Weight (OW) and Inverse Surprising Popularity (ISP), for aggregating responses from multiple large language models...
Research & Development
Enhancing Search Autocomplete: A Data-Centric Approach to Reducing Bias
TLDR: This paper introduces a novel data-centric method to reduce presentation bias in real-time neural query autocomplete systems. By generating synthetic prefixes from unbiased...
Research & Development
Next-Generation Recommendations: Adapting LLMs to Evolving Preferences and Rich Media
TLDR: The paper introduces DynMM-Explain-LLMRec, a framework that enhances Large Language Model-based recommendation systems. It addresses limitations of existing systems by incorporating dynamic adaptation...
Research & Development
REFEREE: Learning Continuously Across Domains Without Relying on Source Data
TLDR: The research paper introduces Source-Free Cross-Domain Continual Learning (SFCDCL), a new problem where models must learn continuously across different data domains without access...
Research & Development
Investigating Function Composition in Large Language Models
TLDR: Large language models (LLMs) solve complex, multi-step tasks using two distinct internal mechanisms: a 'compositional' approach that explicitly computes intermediate steps, and a...
Research & Development
Understanding Scene Geometry: Introducing InstaFormer for Holistic Order Prediction
TLDR: InstaFormer is a new AI model that can quickly and accurately determine how objects in a natural image are positioned relative to each...