WithinUsAI's TRM_Genesis_RealWorld_5k dataset contains 5000 examples of fact-based question-and-answer pairs. Each answer includes citations (fact_sources) and a verification tag, and prompts are designed to be short and optimized for tiny contexts. The dataset was last updated on 2026-01-04.
Use Cases
- Training or evaluating small language models on factual accuracy based on the dataset's fact-based Q&A structure.
- Developing citation-aware AI systems based on the included fact_sources field.
- Benchmarking model performance in constrained contexts based on the short, constraint-aware prompts.
- Researching automated fact verification based on the verification tags attached to answers.
Strengths
- Contains 5000 examples, providing a substantial corpus for training.
- Each answer includes citation sources (fact_sources) and a verification tag, supporting factual integrity.
- Prompts are explicitly designed to be short and optimized for tiny contexts, aligning with specific model requirements.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is known, but the specific JSON schema beyond the described fields is unknown.
- Last updated 2026-01-04 00:02:42; freshness should be verified.
Provenance
- Source
- WithinUsAI
- Freshness
- 2026-01-04 00:02:42