Unverified50% confidenceFactExact time
A sparse autoencoder is an unsupervised learning architecture that compresses high-dimensional model activation vectors into an overcomplete sparse latent space and then reconstructs the original activations, forcing each latent variable to be sensitive to only a few input patterns.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
Unverified从表征到有限词汇的映射过程本质上是一种有损编码（Lossy Encoding）77% similarUnverifiedArithmetic coding is a lossless compression method that approaches the theoretical entropy limit, encoding the entire message sequence as a decimal within an interval.75% similarUnverifiedLatent-Space Distillation uses the latent representations produced by a larger, more powerful encoder as supervision signals to guide a smaller, lighter student encoder in learning, aligning in the representation space read by the language model.74% similarUnverified程序化生成的纹理能被极端压缩是因为它们具有低柯尔莫哥洛夫复杂度，即存在简短程序可完整描述它们74% similarUnverified文本编码器将提示词编码为固定维度嵌入向量，abliteration针对的是token生成过程中的拒绝行为，与嵌入向量输出情况不同74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1023413API
curl https://kongchang.com/api/v1/knowledge/claims/1023413MCP
get_claim(id=1023413)