85,000 Python functions paired with short natural-language instructions derived from repository docstrings. The dataset was created by NickIBrody and last updated on April 20, 2026. It includes deterministic train, validation, and test splits.
Use Cases
- Training instruction-following models for code generation based on natural language prompts.
- Evaluating the alignment between code and its documentation based on paired instructions.
- Benchmarking models on the task of generating Python code from short descriptions.
- Studying the relationship between function implementation and its docstring summary.
Strengths
- Contains 85,000 examples, providing a substantial corpus for training.
- Exact duplicate rows were removed using normalized instruction and output hashing.
- Includes deterministic splits for training, validation, and testing.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
- Freshness should be verified as the last update date is in the future (2026).
Provenance
- Source
- NickIBrody
- Collection Method
- Derived from repository docstrings.
- Freshness
- Last updated 2026-04-20 18:15:54