The Stack Smol Xl: 870,000 Code Samples Across 87 Programming Languages

Bigcode provides a multilingual source code dataset containing 10,000 random samples for each of 87 programming languages. The dataset is a subset of the larger 'the-stack' collection and was last updated on February 10, 2023. It likely contains raw code snippets intended for analysis and model training.

Use Cases

• Training multilingual code generation models based on the 87 programming languages mentioned.
• Analyzing coding patterns and language-specific syntax across the provided code samples.
• Benchmarking model performance on code-related tasks across different programming languages.
• Creating synthetic training data for specialized or low-resource programming languages.

Strengths

• Contains a broad scope of 87 distinct programming languages.
• Provides a standardized 10,000 samples per language for consistent sampling.
• Derived from the established 'the-stack' dataset, suggesting a known provenance.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last updated 2023-02-10 17:22:38; freshness should be verified.

Provenance

Source

bigcode on Hugging Face

Collection Method

Random sampling from the original 'the-stack' dataset.

Time Range

null

Freshness

2023-02-10 17:22:38

Geography

null