Unverified50% confidenceFactExact time
OpenAI's method decomposes the discrete Fourier transform (DFT) matrix into a product of several low-rank or sparse matrices, and uses random sampling or structured pruning to skip butterfly operations with extremely small contributions.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/13/2026
First Seen
Valid until: 1/11/2027
Sources
Related Entities
Related Claims
UnverifiedLow-rank parameterization decomposes the covariance matrix into a low-rank factor plus a diagonal matrix form, reducing the parameter count from O(d²) to O(dr).69% similarUnverifiedThe idea behind latent factor decomposition is to assume that weight matrices have redundancy and low-rank structure, and to reconstruct the original weights using a set of smaller latent factors along with indices or combination methods.67% similarUnverifiedGrouping several variables into a cluster node in a causal graph can reduce the size of the graph and lower the complexity of identification algorithms.67% similarUnverifiedLow-rank decomposition and quantization compression are orthogonally complementary dimensions; quantizing the smaller matrices A and B separately results in smaller errors, and the combined compression ratio can reach several times that of either method alone66% similarUnverifiedNyström sketching approximates the full kernel matrix by generating a low-rank matrix through randomly selecting m anchor points from n training samples, reducing the bottleneck from O(n^2) to O(nm).65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1013572API
curl https://kongchang.com/api/v1/knowledge/claims/1013572MCP
get_claim(id=1013572)