1000 examples of fact-based question-and-answer pairs, created by WithinUsAI and updated on January 4, 2026. Each example includes citations and a verification tag. The prompts are designed to be short and optimized for tiny contexts.
Use Cases
- Training tiny models for factual question answering based on the described Q&A structure.
- Benchmarking model performance on cited, verifiable answers based on the fact_sources field.
- Developing constraint-aware prompts for limited-context environments based on the TRM-native prompt design.
Strengths
- Contains 1000 examples, providing a substantial training corpus.
- Each answer includes cited fact_sources and a verification tag, supporting factual accuracy checks.
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.
Provenance
- Source
- WithinUsAI
- Freshness
- Last updated 2026-01-04 00:00:58; freshness should be verified.