Sign in to view source links and access this dataset
Description
Arithmetic Word Problem Compendium Dataset (AWPCD) is a collection of mathematical word problems spanning multiple domains with natural language variations. The dataset, created by HelloCephalopod, contains a sample of 1,000 problems, each requiring 1 to 5 steps of mathematical operations. It was last updated on the Hugging Face platform on February 15, 2025.
Use Cases
Train language models for multi-step arithmetic reasoning based on the described 1-5 step problems.
Benchmark model performance on maintaining decimal precision throughout calculations as highlighted in the description.
Develop educational tools that encourage showing work, a key feature of the dataset's design.
Strengths
Contains a sample of 1,000 problems, providing a substantial base for experimentation.
Problems are designed with 1 to 5 mathematical steps, offering a range of complexity.
Includes rich metadata and natural language variations for each problem.
Limitations
Row count is unknown, which may limit suitability assessment.
Column-level documentation is absent; field semantics must be inferred after download.
Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
Source
HelloCephalopod on Hugging Face.
Freshness
Last updated 2025-02-15 17:57:00; freshness should be verified.
The available data is described as a sample; the full description is on the dataset page. License information is unknown.