Unverified50% confidenceTradeoffExact time
The advantage of the learned nominal distribution does not hold unconditionally: as the record size grows, or when the distribution shift deviates from the structure learned by the model, the gain gradually diminishes.
1
Sources
50%
Confidence
Long-term
Relevance
10/30/2026
First Seen
Sources
Related Entities
Related Claims
Unverified数据规模的增长并不等同于有效学习信号的增长，在模型能力提升到一定阈值后简单堆砌样本的收益会急剧下降78% similarUnverifiedOff-policy reinforcement learning reuses samples generated by older policies, which brings efficiency advantages but introduces distribution shift and high-variance gradient estimation problems.77% similarUnverifiedThe median improvement of 5.3 percentage points indicates that current models are mostly doing 'local patching' and have not yet formed cross-level knowledge accumulation, which is the key difference between continual learning and single-round self-correction.73% similarUnverifiedAblation experiments show that increasing example usage does not yield consistent gains, but instead presents a trade-off relationship between dynamics and reward.73% similarUnverified当生成模型反复在自身生成的数据上训练时，输出分布会逐渐退化，低概率事件被系统性遗忘73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1115661API
curl https://kongchang.com/api/v1/knowledge/claims/1115661MCP
get_claim(id=1115661)