Frontier DataText

Chinese College Entrance Exam STEM Reasoning Dataset

Type
CoT Reasoning Dataset
Domain
Science and Engineering
Language
English

Overview

This reasoning dataset pairs Chinese college entrance exam questions in mathematics, physics, and chemistry with their key concepts, final answers, and step-by-step solutions.

Potential Use Cases

  • Supervised Fine-Tuning for STEM Reasoning: Worked solutions expose the derivation path rather than the answer alone, training models to select a governing principle and carry it through to a result.
  • Process Supervision & Reward Modeling: Step-separated solutions give step-level training signal for process reward models and verifiers that locate where a derivation goes wrong.
  • Subject-Resolved Evaluation: Because each item carries its subject, key concept, and final answer as separate fields, accuracy can be scored per subject and separately from solution quality.
Flitto Curation Data Flitto Curation Data