IDS-Bench: A Japanese LLM Evaluation Dataset for CSV-to-IDS Generation

Japan's GENIAC Project, promoted by the Ministry of Economy, Trade and Industry and NEDO, produced this dataset. It is an evaluation dataset for the CSV-to-IDS task, used to assess the Ishigaki-IDS model and evaluate whether an LLM can generate appropriate IDS from CSV. The dataset was authored by ONESTRUCTION and last updated on March 26, 2026.

Use Cases

• Benchmarking LLM performance on the CSV-to-IDS task based on the described evaluation purpose
• Training or fine-tuning models for structured data generation based on the CSV-to-IDS task
• Researching LLM capabilities in converting tabular data to IDS formats as mentioned in the description

Strengths

• Created under Japan's GENIAC Project, indicating institutional backing from METI and NEDO
• Specifically designed for evaluating the Ishigaki-IDS model, providing a clear benchmark purpose
• Last updated March 26, 2026, suggesting recent maintenance

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment
• The dataset's size and file formats are unknown, complicating download planning

Provenance

Source

ONESTRUCTION, as part of Japan's GENIAC Project promoted by METI and NEDO

Collection Method

Produced for evaluating the Ishigaki-IDS model and the CSV-to-IDS task

Time Range

Last updated March 2026

Freshness

Last updated 2026-03-26 03:15:01

Geography

Japan (inferred from project sponsors)