in sparse_autoencoder/kernels.py [0:0]
def triton_sparse_dense_matmul_kernel(
sparse_indices_ptr,
sparse_values_ptr,
dense_ptr,
out_ptr,
stride_dn,
stride_db,
A,
B,
N,
K,
BLOCK_SIZE_K: tl.constexpr,
BLOCK_SIZE_B: tl.constexpr,