Expired50% confidenceEventExact time
The laboratory proactively suspended reinforcement learning (RL) training on its latest deployment-targeted model for two weeks.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
8/31/2026
First Seen
Valid until: 9/14/2026(expired)
Sources
Related Entities
Related Claims
Unverified预训练阶段在海量无标注语料上通过自监督学习完成，计算成本极高、周期通常以月计，而指令微调、RLHF对齐等工程阶段可在数周甚至数天内完成58% similarVerifiedOpenAI's self-training loop mechanism has compressed the model iteration cycle from months to weeks.55% similarUnverifiedThe training data cutoff date for AI models typically lags reality by 6 to 12 months.54% similarUnverifiedMeta在OPT-175B训练日志中披露，175天的训练过程中经历了超过100次硬件相关的中断事件，平均每1-2天出现一次需要人工干预的故障54% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/827366API
curl https://kongchang.com/api/v1/knowledge/claims/827366MCP
get_claim(id=827366)