StreamKL introduces a fused GPU primitive that eliminates quadratic memory usage in attention distillation by streaming query-key tiles through on-chip SRAM. It achieves up to 43x speedup in forward and 14x in backward passes, reducing extra HBM footprint from O(N_QN_K) to O(1), enabling long-context distillation on a single GPU.
StreamKL: Fast and Memory-Efficient KL Divergence for Attention Distillation
from English