IDFU Code Negative Dataset: 100 Python Samples That Failed Execution

100 unique Python code samples that failed execution-based validation, curated by author 'namakoo'. This preview dataset is designed for training machine learning models and was last updated on 2026-04-26. The full description is available on the dataset page.

Use Cases

• Training reward models for reinforcement learning from human feedback (RLHF) based on the described failed code samples.
• Creating Direct Preference Optimization (DPO) rejected-side pairs based on the negative code examples.
• Training classifiers to detect code errors based on the collection of execution-invalidated samples.
• Benchmarking code generation models on their ability to avoid common errors represented across 19 CS domains.

Strengths

• Contains 100 unique Python samples, providing a defined scale for the preview.
• All samples are AST-validated, suggesting a basic level of syntactic integrity.
• Represents code from 19 computer science domains, including MCMC, FFT, and formal methods, indicating topic diversity.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the full dataset is unknown, which may limit suitability assessment for the paid version.

Provenance

Source

huggingface

Collection Method

Curated collection of Python code samples that failed execution.

Freshness

Last updated 2026-04-26 11:38:05; freshness should be verified.