Unverified50% confidenceFactExact time
Existing orthogonal transform methods fall roughly into two categories: data-independent fixed transforms (such as the Hadamard rotation) and rotation matrices learned during the calibration stage (such as SpinQuant and DartQuant).
1
Sources
50%
Confidence
Long-term
Relevance
10/15/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedLearned rotation methods (such as SpinQuant and DartQuant) require repeated iterative optimization of the rotation matrix before quantization, resulting in high memory usage and long runtime.79% similarUnverified哈达玛变换作为正交变换可对激活值和权重进行旋转，将离群值打散并平滑化，使量化后的分布更均匀，减小量化误差68% similarUnverifiedSpinQuant通过Cayley优化在正交矩阵空间中搜索最优旋转，使用少量校准数据微调旋转矩阵，以获得比固定哈达玛矩阵更好的量化效果66% similarUnverifiedQuaRot采用固定的哈达玛矩阵作为旋转矩阵，无需额外训练即可应用，通过将旋转吸收到相邻权重矩阵中而不改变模型输出66% similarUnverifiedInverse kinematics solving is typically accomplished through Jacobian matrix iteration, using the pseudo-inverse method or damped least squares method to convert 3D coordinates into multi-degree-of-freedom joint angles within a few milliseconds.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1032994API
curl https://kongchang.com/api/v1/knowledge/claims/1032994MCP
get_claim(id=1032994)