CMATH: 1.7k Chinese Elementary School Math Word Problems

CMATH is a dataset of 1.7k elementary school-level math word problems with detailed annotations, sourced from actual Chinese workbooks and exams. It was created by weitianwen and last updated on Hugging Face in October 2023. The dataset aims to benchmark the mathematical reasoning abilities of large language models against specific grade levels.

Use Cases

• Benchmarking LLM performance on grade-specific math problems based on the annotated difficulty levels.
• Analyzing the types of reasoning required for Chinese elementary math based on the word problem structure.
• Training or fine-tuning models for educational question-answering tasks using the provided problems and annotations.

Strengths

• Contains 1.7k math word problems, providing a substantial corpus for evaluation.
• Problems are sourced from real Chinese educational materials (workbooks and exams), suggesting authentic content.
• Includes detailed annotations intended for benchmarking LLM abilities against specific grade levels.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last updated 2023-10-14 07:27:22; freshness should be verified.

Provenance

Source

Sourced from actual Chinese elementary school workbooks and exams.

Freshness

Last updated 2023-10-14 07:27:22.

Geography

China