Let's Verify Step by Step
by OpenAI · free · Last verified 2026-03-17
Demonstrated that process-based reward models (PRMs), which provide feedback on each reasoning step, substantially outperform outcome-based reward models (ORMs) for training LLMs to solve mathematical reasoning problems. The paper also introduced PRM800K, a dataset of 800K step-level human feedback labels on MATH solutions.
https://arxiv.org/abs/2305.20050 ↗B+
B+—Good
Adoption: AQuality: A+Freshness: ACitations: AEngagement: F
Specifications
- License
- Open Access
- Pricing
- free
- Capabilities
- step-level-feedback, math-reasoning, reward-modeling, process-supervision
- Integrations
- Use Cases
- mathematical-reasoning, process-supervision, rlhf-training
- API Available
- No
- Tags
- process-reward-models, reasoning, rlhf, math, step-by-step
- Added
- 2026-03-17
- Completeness
- 100%
Index Score
71.6Adoption
82
Quality
94
Freshness
81
Citations
80
Engagement
0
Put AI to work for your business
Deploy this paper alongside autonomous AaaS agents that handle tasks end-to-end — no babysitting required.