Unverified50% confidenceFactExact time
The paper proposes using Positive-Unlabeled Learning to reformulate negative guidance, giving the sampling process a well-defined target density and achieving asymptotic exactness in the particle limit.
1
Sources
50%
Confidence
Long-term
Relevance
10/15/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedSPLADE 是一种学习型稀疏检索模型，通过神经网络为文档生成稀疏向量表示，可在倒排索引上高效运行72% similarUnverifiedThe paper trains a bijection (typically implemented as a normalizing flow) so that the target distribution in the latent space approximates a reference density with simple structure.71% similarUnverifiedThe paper treats conditional samples as the positive class and the particles in the diffusion process as unlabeled data, training a lightweight classifier online to estimate the probability that each particle belongs to an undesirable condition.70% similarUnverifiedDistillation is the only compression technique in the study that yields fairness improvements, smoothing the treatment of different groups by having the small model learn the soft label distribution of the large model.69% similarUnverifiedFine-tuning embedding models with contrastive learning requires labeled data (positive samples of successful hires and negative samples of rejected candidates) to achieve significant performance gains.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1037375API
curl https://kongchang.com/api/v1/knowledge/claims/1037375MCP
get_claim(id=1037375)