Efficient Quantized Sparse Matrix Operations on Tensor Cores