This is your work, valued
QJL. QJL: 1-Bit Quantized JL transform for KV Cache Quantization with Zero Overhead
HyperAttention. Triton Implementation of HyperAttention Algorithm
NTK-CNTK-Sketch. Implementation of the NTK and CNTK Sketch, fast methods for accelerating the learning with NTK and CNTK kernels
