Unverified60% confidenceFactExact time
OpenAI published research in 2023 showing that process reward models can significantly improve the reliability of mathematical reasoning compared to outcome supervision.
2
Sources
60%
Confidence
Long-term
Relevance
8/23/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedIn 2023, OpenAI published the paper "Let's Verify Step by Step," which systematically demonstrated the superiority of process reward models (PRM) in mathematical reasoning tasks.79% similarUnverifiedIn mathematics reasoning research, OpenAI found that models trained with process supervision significantly outperform models trained with only outcome supervision on complex reasoning tasks.79% similarUnverified2021年的Chain-of-Thought Prompting论文证明，通过在提示词中加入推理步骤示例，模型的数学推理准确率可提升数十个百分点73% similarVerifiedIn 2020, Kaplan et al. at OpenAI revealed that model performance follows predictable power-law relationships with parameter count, training data volume, and compute budget.73% similarVerifiedResearch from OpenAI and Anthropic shows that using models matched to tasks of different complexity levels not only reduces inference costs but also improves overall output quality.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/789849API
curl https://kongchang.com/api/v1/knowledge/claims/789849MCP
get_claim(id=789849)