Unverified50% confidenceFactExact time
2022年Hoffman等人的Chinchilla论文重新校准了算力最优的参数-数据比例，指出此前许多大模型训练不充分
1
Sources
50%
Confidence
Long-term
Relevance
7/15/2026
First Seen
Sources
Related Claims
UnverifiedMultiple independent studies in 2023 showed that some top models scored extremely high on standard mathematical reasoning tests but performed significantly worse on equivalent rephrased versions, indicating that the models learned test patterns rather than genuine reasoning ability.75% similarUnverifiedIn 2023, researchers demonstrated that contaminating less than 1% of training data is sufficient to alter model behavior under specific trigger conditions71% similarUnverifiedScale AI在2024年的分析显示，部分模型在MMLU上的分数膨胀可能高达5-10个百分点，因测试集已泄露到训练数据中70% similarUnverifiedAnthropic's research shows that even mild alignment training can cause a 2–8 percentage point decline in model performance on certain reasoning-intensive tasks.69% similarUnverifiedDeepMind于2022年的Chinchilla论文修正了Scaling比例，发现相同计算预算下应使用更小模型但喂入更多数据68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/523906API
curl https://kongchang.com/api/v1/knowledge/claims/523906MCP
get_claim(id=523906)