Popular repositories Loading

1. SRKL-Optimizer SRKL-Optimizer Public

   A specialized second-order autograd optimizer implementing dynamic kinetic braking and structural curvature damping to prevent gradient explosion with optimized memory allocation.

   Python

2. Geodesic-Rotational-Layer Geodesic-Rotational-Layer Public

   ​A non-Euclidean linear neural network layer that enforces parameter updates via skew-symmetric matrix exponentials to guarantee strict orthogonality and norm-preservation.

   Python

3. Dynamic-Kernel-Projection-Layer Dynamic-Kernel-Projection-Layer Public

   A learnable non-linear Hilbert space projection layer in PyTorch that dynamic-shapes spatial curvature and optimizes RBF kernel parameters via auto-differentiation.

   Python

4. Non-Euclidean-Deep-Learning-Framework Non-Euclidean-Deep-Learning-Framework Public

   Python