Unverified50% confidenceBenchmarkExact time
在同一细粒度汽车分类任务中，CLIP ViT-L的k-NN准确率约为59%
1
Sources
50%
Confidence
Long-term
Relevance
7/10/2026
First Seen
Sources
DINOv2 vs SigLIP: Why the Huge Gap in k-NN Classification? A Guide to Avoiding Vision Encoder Selection Pitfalls
redditr/MachineLearning7/8/2026
Related Claims
Unverified在细粒度汽车分类的k-NN实验中，SigLIP2 SO400M达到约92%准确率，CLIP ViT-L约59%，DINOv2 Giant约41%78% similarUnverified在同一细粒度汽车分类任务中，DINOv2 Giant的k-NN准确率约为41%，表现垫底78% similarUnverifiedK-Quant是llama.cpp生态中的混合精度量化方案，根据每层权重敏感度分配不同量化bit数，可在相同平均bit数下将精度损失降低40-60%61% similarUnverifiedTop-1% accuracy is a metric for evaluating the output consistency between a quantized model and the original full-precision model, focusing on whether the top 1% of tokens with the highest model confidence are consistent with the full-precision version.61% similarUnverifiedIn guided mode, the average end-to-end accuracy increased from 28.6% to 33.7%, and the ambiguity detection F1 score increased from 45.3% to 64.9%59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/456342API
curl https://kongchang.com/api/v1/knowledge/claims/456342MCP
get_claim(id=456342)