Unverified50% confidenceFactExact time
DenseTwins is a multimodal self-supervised framework that learns dense representations by explicitly decoupling the dense feature space into shared components and modality-specific components.
1
Sources
50%
Confidence
Long-term
Relevance
11/4/2026
First Seen
Sources
Related Entities
Related Claims
Unverified乘积量化的核心思想是将高维向量空间分解为若干低维子空间，在每个子空间内独立进行码本学习与编码66% similarUnverifiedFeature disentanglement is a representation learning technique in deep learning, aimed at separating different attributes of input data into independent latent dimensions.65% similarUnverifiedKernel fusion is a common optimization used by deep learning compilers and inference engines, merging multiple small operators into one large kernel to reduce launch overhead and memory round trips.65% similarUnverifiedTowards Monosemanticity研究在单层MLP上训练SAE，从约512维激活空间中提炼出超过4000个可解释特征64% similarUnverified多头分类器的架构思路源自深度学习中的多任务学习，其核心是共享一个主干编码器生成向量表示，再通过多个独立分类头并行输出不同维度预测63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1133858API
curl https://kongchang.com/api/v1/knowledge/claims/1133858MCP
get_claim(id=1133858)