Unverified50% confidenceFactExact time
DeepSeek R1 listed process reward models among its "unsuccessful attempts"; its reasoning training used only simple rule-based rewards based on final answers.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/30/2026
First Seen
Valid until: 1/28/2027
Sources
14 Core Concepts of Reasoning Models: From Chain of Thought to GRPO Explained
youtubeBrix AI10/29/2026
Related Entities
Related Claims
UnverifiedDeepSeek R1的强化学习奖励信号主要来自数学答案的正确性验证，而非工具调用格式的规范性，因此未针对Function Call专项对齐75% similarUnverifiedDeepSeek-R1's training report documented the practice of combining GRPO with a reward function: correct answers receive +1 point, while format errors or incorrect answers receive 0 or negative points.71% similarUnverifiedSome of DeepSeek's models restrict using their outputs to train competing models.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1116036API
curl https://kongchang.com/api/v1/knowledge/claims/1116036MCP
get_claim(id=1116036)