Indonesian Math Problems With Chain-of-Thought Reasoning

50,000 synthetic samples of math problems written in Bahasa Indonesia. The dataset includes explicit chain-of-thought reasoning traces for each problem, designed to train language models on arithmetic problem solving. It was created by Sandroeth and last updated in April 2026.

Use Cases

• Fine-tune language models to generate the internal reasoning trace (reason) from a given math problem statement.
• Train models to solve arithmetic problems by predicting the final answer based on the provided chain-of-thought reasoning.
• Evaluate model performance on Indonesian-language math problems across basic to advanced arithmetic operations.
• Study the structure and patterns of synthetic chain-of-thought reasoning for educational AI applications.

Strengths

• Contains 50,000 samples, providing a substantial volume for model training.
• Includes explicit chain-of-thought reasoning (reason) for each problem, a key feature for instruction tuning.
• Covers a range of arithmetic operations from basic to advanced, adding task diversity.

Limitations

• The dataset is synthetic, which may limit the realism and linguistic variation compared to human-written problems.
• Specific sample size details, column definitions, and data distribution (e.g., class balance across operation types) are not provided.
• The geographic and demographic scope is undefined, as the data is synthetic and not tied to a specific population.

Provenance

Source

Synthetic dataset created by Sandroeth, hosted on Hugging Face.

Collection Method

Synthetically generated, designed for training language models on math reasoning with explicit reasoning traces.

Freshness

Last updated on 2026-04-04, indicating recent maintenance.