Learn to select tools based on user needs and complete bookings with end-to-end pet service booking trajectories.

Train and evaluate RAG systems to retrieve controlling, context-complete evidence for conflicting clauses in lease agreements.

Evaluate how well models move from understanding a word problem to structuring its quantities, relationships, and answer target.

Train models to answer specialized humanities questions with context grounded in digital textbook sources and traceable evidence.

Train models to answer common SAT score questions with focused coverage of score components and report interpretation.

Train evidence-grounded legal AI on life sciences licensing contracts and their core commercial and rights provisions.

Train models to verify affordable housing policy claims against program rules and paired source evidence.

This dataset is specifically constructed to enhance English speech recognition systems, covering general daily contexts.

Train and evaluate domain models with answered and explained practice questions on relay protection principles and power engineering applications.

Train and evaluate legal question answering with structured links between case facts, legal rules, reasoning, and conclusions.

Train and evaluate ship weight and center-of-gravity calculation checkers with JSON samples containing itemized weights, reference results, and correctness labels.

Train models to answer confidentiality questions in M&A due diligence using agreement text and supporting clauses.

Explore computer and tablet service trajectories covering warranty checks, repair options, service locations, and appointment booking.

Train models to verify buy now, pay later product claims and assess source authority using provider terms and quoted evidence.

The reception communication process speech dialogue dataset provides high-quality training data for intelligent speech analysis and human-computer interaction development.

Extract tasks, deadlines, and collaboration relationships from Korean workplace documents for business language learning and model training.

Train models to interpret 5G SA QoS constraints, detect policy conflicts, and derive appropriate handling decisions from scenario-based QA.

Built around key technology commercialization license terms, this dataset supports precise retrieval, structured extraction, and evidence traceability.

Train models to reason about patients' potential costs by linking health plan terms to care scenarios.

The spatial guidance sign image dataset is used to improve the accuracy and efficiency of navigation systems.