Unverified50% confidenceFactExact time
ReWY uses algebraic restructuring to compress the seven local matrix multiplications related to the WY matrix into three, and eliminates the inverse adjoint.
1
Sources
50%
Confidence
Long-term
Relevance
11/3/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedReWY replaces pairwise vector gating with centralized matrix contraction.74% similarUnverifiedThe forward pass of a fully connected layer is essentially matrix multiplication plus a bias (y=Wx+b), and convolution operations can be reformulated as multiplication by a Toeplitz matrix63% similarUnverifiedLow-rank parameterization decomposes the full covariance matrix into the form Σ = WW^T + D, reducing the parameter count from O(d²) to O(dr).63% similarUnverifiedIn the Transformer, the Q, K and V matrices are obtained by multiplying the hidden representations of the input sequence by three learnable weight matrices W_Q, W_K and W_V respectively.63% similarUnverifiedSimpleSpread supports incremental recalc, recalculating only affected cells rather than recalculating the entire sheet.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1129579API
curl https://kongchang.com/api/v1/knowledge/claims/1129579MCP
get_claim(id=1129579)