Unverified50% confidenceOpinionExact time
The extremely scarce samples of European faces in Gongbi painting training data cause models to face the challenge of training data bias.
1
Sources
50%
Confidence
Long-term
Relevance
9/12/2026
First Seen
Sources
Related Entities
Related Claims
Unverified长尾分布指训练数据中少数常见场景占据绝大多数样本，而罕见但重要的场景严重欠表示70% similarUnverifiedFor a small dataset of 5,600 images, training a standard DETR from scratch is almost certain to fail, and it is recommended to load COCO pretrained weights for fine-tuning.68% similarUnverifiedDeepMind's Chinchilla research pointed out that previous models were generally oversized and undertrained, meaning they had too many parameters but insufficient training data.67% similarUnverified过拟合是LoRA训练中最常见的失败模式，模型会死记硬背训练图像像素特征而非学习本质特征67% similarUnverified模型能力高度依赖训练数据的分布，法律文本充足而空间推理题的文本化表示相对稀少67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/915190API
curl https://kongchang.com/api/v1/knowledge/claims/915190MCP
get_claim(id=915190)