OpenCodeReasoning Mini: 20,000 Python and C++ Examples for Fine-Tuning

OpenCodeReasoning-mini contains 20,000 code examples, with a balanced collection of 10,000 Python and 10,000 C++ examples. It is designed for supervised fine-tuning (SFT) of models for code generation tasks.

Use Cases

• Supervised fine-tuning of language models for Python code generation using the 10,000 Python examples.
• Supervised fine-tuning of language models for C++ code generation using the 10,000 C++ examples.
• Training a multi-language code generation model on the balanced collection of Python and C++ examples.
• Analyzing the stylistic or structural differences between the Python and C++ examples in the dataset.

Strengths

• Contains 20,000 total code examples.
• Balanced collection with exactly 10,000 examples each for Python and C++.
• Specifically designed for supervised fine-tuning (SFT).

Limitations

• No column definitions or sample data are provided, limiting understanding of data structure.
• The dataset is a small subset of a larger corpus, which may limit scope.
• The specific content and quality of the individual code examples are unknown.

Provenance

Source

OrionLLM via Hugging Face.

Collection Method

Subset of the OpenCodeReasoning dataset, curated for supervised fine-tuning.

Freshness

Last updated 2026-03-19.