4 papers
Spike Hijacking in Late-Interaction Retrieval
Karthik Suresh, Tushar Vatsa, Tracy King +2
Late-interaction retrieval models rely on hard maximum similarity (MaxSim) to aggregate token-level similarities. Although effective, this winner-take-all pooling rule may structur…
FUSE : Failure-aware Usage of Subagent Evidence for MultiModal Search and Recommendation
Tushar Vatsa, Vibha Belavadi, Priya Shanmugasundaram +2
Multimodal creative assistants decompose user goals and route tasks to subagents for layout, styling, retrieval, and generation. Retrieval quality is pivotal, yet failures can aris…
RouteNator: A Router-Based Multi-Modal Architecture for Generating Synthetic Training Data for Function Calling LLMs
Vibha Belavadi, Tushar Vatsa, Dewang Sultania +5
This paper addresses fine-tuning Large Language Models (LLMs) for function calling tasks when real user interaction data is unavailable. In digital content creation tools, where us…
Domain-specific Question Answering with Hybrid Search
Dewang Sultania, Zhaoyu Lu, Twisha Naik +11
Domain specific question answering is an evolving field that requires specialized solutions to address unique challenges. In this paper, we show that a hybrid approach combining a…