For-Value: Efficient Forward-Only Data Valuation for finetuning LLMs and VLMs

For-Value estimates training-data influence from a single forward pass, avoiding gradient computations for scalable LLM and VLM data valuation.

Textual Equilibrium Propagation for Deep Compound AI Systems

TEP uses local prompt optimization to mitigate exploding and vanishing textual gradients in deep, long-horizon compound AI systems.

Enhancing Clinical Multiple-Choice Questions Benchmarks with Knowledge Graph Guided Distractor Generation

KGGDG generates clinically plausible but incorrect answer choices from medical knowledge graphs to make medical QA benchmarks more challenging.