Unverified50% confidenceFactExact time
When the number of chunks is in the single digits, the entire "index" is just one matrix, and a full scan takes a single matrix multiplication at microsecond-level cost.
1
Sources
50%
Confidence
Long-term
Relevance
10/17/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe paper points out that the construction of the matrix mean confidence sequence depends on a specific matrix norm, and whether it can be generalized to more general norms is worth exploring.65% similarUnverifiedA positive definite matrix is a symmetric matrix with all strictly positive eigenvalues, satisfying x^TAx > 0 for any non-zero vector x65% similarUnverifiedOne-hot encoding assigns each word in the vocabulary a unique index, representing it with a vector that has a 1 only at that index position and 0s elsewhere; the dot product of any two different one-hot vectors is always 0 (orthogonal).65% similarUnverifiedIn self-attention, the input matrix X is multiplied by three learned weight matrices to obtain the Query, Key, and Value matrices.64% similarUnverifiedWhen there is no ordering relationship between the rows of matrix data (swapping rows does not affect the classification result), the data is essentially a Set rather than a Sequence.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1054544API
curl https://kongchang.com/api/v1/knowledge/claims/1054544MCP
get_claim(id=1054544)