Prompts and Accuracy Scores for LLM Hazard Paper

9.6 MB of prompts and correctness scores for various large language model responses, compiled by Mary Cummings. The dataset is hosted on figshare and was last updated on 2026-05-26. It is released under a CC-BY-4.0 license.

Use Cases

• Benchmarking LLM performance based on correctness scores
• Analyzing prompt effectiveness for safety evaluations
• Comparing model outputs across different prompts

Strengths

• 9.6 MB of data provides a substantial collection of prompts and scores
• Explicitly includes correctness scores for model responses
• Has a clear author (Mary Cummings) and license (CC-BY-4.0)

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment
• Data is provided in PDF format, which may complicate automated extraction

Provenance

Source

Mary Cummings

Freshness

Last updated 2026-05-26 11:33:51