MinorBench: A Benchmark for Child-Safety in Large Language Models

MinorBench is a benchmark created by govtech to evaluate whether large language models respond appropriately to questions that may be unsuitable for children. It contains 299 prompts spanning various sensitive topics, designed to assess models in child-friendly assistant roles. The dataset was last updated on March 14, 2025.

Use Cases

• Benchmarking LLM safety filters based on the 299 sensitive prompts.
• Evaluating model refusal capabilities for child-inappropriate content as described.
• Training or fine-tuning models for educational assistant roles using the defined prompts.

Strengths

• Contains 299 specifically designed prompts for a focused evaluation.
• Targets a clear, defined application area: child safety in educational settings.
• Created by a named author (govtech) and has a specific last update date (2025-03-14).

Limitations

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

Provenance

Source

govtech

Collection Method

Likely manually curated or generated for benchmark purposes.

Time Range

Creation or update timeframe implied by the last updated date of 2025-03-14.

Freshness

Last updated 2025-03 14 07:40:09.

Geography

null