Math_Archon_25k consists of 25,000 mathematical examples developed by WithinUsAI to train models in graduate-level techniques and research-grade exposition. Updated in January 2026, the collection provides structured prompt-answer pairs across various mathematical domains.
Use Cases
- Fine-tuning language models for advanced mathematical reasoning using the 'prompt' and 'answer' columns
- Domain-specific performance benchmarking using the 'domain' field
- Training models in research-grade mathematical exposition via the 'answer' text
Strengths
- 25,000 structured mathematical records
- Includes 'domain' labels for topical filtering
- Focuses on graduate-level and research-grade complexity
Limitations
- Unknown provenance regarding whether content is synthetic or human-curated
- Relatively small sample size of 25,000 for broad LLM pre-training
Provenance
- Source
- WithinUsAI
- Freshness
- Last updated January 2026.