TAL-SCQ5K: 5,000 Math Competition Questions in English and Chinese

TAL-SCQ5K-EN and TAL-SCQ5K-CN are high-quality mathematical competition datasets created by TAL Education Group, each consisting of 5,000 questions. The datasets contain 3,000 training and 2,000 testing questions in multiple-choice format covering primary, junior high, and high school levels. Detailed solution steps are provided to facilitate chain-of-thought training.

Use Cases

• Training language models for mathematical reasoning based on multiple-choice questions.
• Benchmarking model performance on bilingual (English/Chinese) educational tasks.
• Developing chain-of-thought reasoning models based on the provided detailed solution steps.
• Analyzing question difficulty and topic coverage across primary, junior high, and high school levels.

Strengths

• Contains 5,000 questions per language version, providing substantial scale.
• Includes detailed solution steps for each question, supporting chain-of-thought training.
• Covers mathematical topics across three educational levels: primary, junior high, and high school.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Last updated 2023-09-15 06:37:10; freshness should be verified.
• Data may reflect bias inherent to the specific educational context of the TAL Education Group.

Provenance

Source

TAL Education Group, created by author 'math-eval'.

Collection Method

Likely curated from educational materials and competitions.

Freshness

Last updated 2023-09-15 06:37:10.