Turkish Math Problems for Multi-Step Reasoning Training

Phase 2 of a curriculum learning pipeline contains moderately difficult math problems requiring multiple reasoning steps in Turkish. The dataset was created by author 'erayalp' and was last updated in April 2025. It is designed to bridge the gap between basic arithmetic and complex problem-solving for language models.

Use Cases

• Supervised fine-tuning of language models on Turkish two-part arithmetic problems.
• Training models to perform logical reasoning and comparisons as described in the problem statements.
• Evaluating model performance on multi-step reasoning tasks in a non-English language.
• Curriculum learning by progressing from this medium-difficulty dataset to more complex GSM8K-style problems.

Strengths

• Part of a structured curriculum learning pipeline (Phase 2).
• Focuses on multi-step reasoning problems, a distinct challenge.
• Created specifically for the Turkish language, addressing a less common language in this domain.

Limitations

• The exact number of problem rows and samples is unknown.
• Specific problem types and their distribution are not detailed.
• Lacks sample data or column definitions to assess structure and complexity.

Provenance

Source

huggingface

Collection Method

null

Time Range

null

Freshness

Last updated 2025-04-22.

Geography

null