IRWOZ 2.0 Improves Industrial Robot Dialogue State Tracking
September 4, 2026
IRWOZ 2.0 uses Mistral and Claude-3.5 to refine an industrial human-robot interaction dataset across four domains. The updated 390-dialogue set increased GPT-2's BLEU-4 score from 0.1651 to 0.5604 on dialogue state tracking benchmarks.
HOW THIS AFFECTS YOU
●
builderThis provides a cleaner dataset for developing more robust industrial robot conversational interfaces.
●
researcherYou can use this improved dataset to test HRI dialogue state tracking accuracy.