TeleGen: Benchmark for Automated Code Generation and Execution Evaluation

A 1.7 GB dataset released on figshare in May 2026 by Yujia Luo. It accompanies research on automated environment construction and execution-based evaluation for software and web development tasks. The dataset includes benchmark task metadata, structured input/output records, evaluation scripts, runtime observations, and experimental results.

Use Cases

• Analyze environment setup and configuration based on benchmark task metadata
• Study code execution and failure diagnosis based on runtime observations
• Evaluate observation-driven code refinement or repair based on structured input/output records
• Reproduce experiments on automated software engineering workflows based on evaluation scripts and results

Strengths

• Dataset size is 1.7 GB, indicating substantial content
• Includes multiple structured components: benchmark metadata, input/output records, scripts, observations, and results
• Released under CC-BY-4.0 license for reuse

Limitations

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

Provenance

Source

figshare

Collection Method

Likely generated as part of research on automated environment construction and execution-based evaluation

Freshness

Last updated 2026-05-25 17:34:48