Train models to diagnose catfish pond hypoxia through causal-chain analysis and explicit reasoning about evidence limits.

Train models to make evidence-based electrical safety judgments with explicit premises and reasoning chains for floating solar PV.

Analyze requests, responses, objections, and disputes in commercial litigation discovery documents to evaluate legal document understanding.

Make SAT answer choice decisions more retrievable and explainable with evidence that supports correct answers and refutes distractors.

Train models to assess chromatography compatibility by reasoning across method parameters, sample conditions, and instrument specifications.

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

Train and evaluate Korean document understanding with structured examples of ambiguity and meaning that is implied rather than stated directly.

A structured QA resource on hamza and Arabic orthography details, designed to help language models answer specialized spelling questions accurately.

Turn vocational digital textbook content into workflow and safety question-answer examples with source excerpts to support domain-specific model training.

Combine diverse rescue records with failure analyses to help models produce evidence-based, professional incident reviews.

Train models to reason about feeding changes, growth outcomes, and confounding factors in trout farming scenarios.

Train models to diagnose FTTB faults more reliably by grounding conclusions in network dependencies and operational evidence.

Train models to combine evidence across religious text passages and produce grounded, traceable answers within a defined corpus.

Train multi-step reasoning about self-consumption, grid exports, and curtailment under commercial solar operating constraints.

Train and evaluate pharmacy-focused models with practice questions and explanations on prescriptions, drug risks, and medication safety.

Train models to turn advertising briefs into structured social media creative that fits directly into advertising workflows.

Answer questions about trademark licensing and quality control clauses with traceable support from contract text.

Pairing exam syllabus pages with evidence-grounded candidate Q&A, this dataset helps models answer questions about tested content and exam rules.

Train models to interpret fault conditions and protection responses in battery management firmware through code snippets and structured explanations.

An audio dataset covering front desk service dialogues, supporting the training of intent classification models.