Unverified50% confidenceFactExact time
The computation of the GW distance is usually transformed into a quadratic assignment problem, whose objective function is non-convex with respect to the coupling matrix.
1
Sources
50%
Confidence
Long-term
Relevance
11/8/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedDue to the non-convexity of GW distance optimization, different initializations may converge to different local optima, causing the same pair of inputs to yield different GW distance values.75% similarUnverifiedPlain gradient descent-ascent (GDA) may diverge in the case of monotone operators, whereas the extragradient method can theoretically guarantee convergence for monotone operators.66% similarUnverifiedQuantized GGUF weights are stored in blocks, each block containing a scale factor and several low-bit integers, and inference requires dequantization before participating in matrix multiplication.66% similarUnverifiedWhen ε ≈ 1, the gradient second-moment term in the denominator is diluted, adaptive scaling has almost no effect, and the optimizer behaves close to pure SGD.66% 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% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1148779API
curl https://kongchang.com/api/v1/knowledge/claims/1148779MCP
get_claim(id=1148779)