Reference library
DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
Paper summary
DeepSeekMath combines mathematical-data selection, continued pretraining, and Group Relative Policy Optimization. Experiments evaluate mathematical reasoning and compare training and sampling choices.
Relevance to the project
Helps separate contributions of data, update procedure, and evaluation. Its explicit correctness rewards are a comparator for externally imposed behavior selection.
Math benchmarks do not establish general autonomous learning.
Source record
- Reference ID
- deepseekmath
- Archived version
- 2402.03300v3
- Summary provenance
- reference-library/papers/deepseekmath/metadata.json · summary
- Review status
- initial orientation summary; not a critical full-paper review
- Reading scope
- Selected abstract/opening and available closing discussion/conclusion passages in the preserved text extract; broader project notes provide context. No reproduction of experiments, exhaustive methods review, or equation/figure audit.