This contribution reports the cross-check of the contribution JVET-AM0172: a combination of EE2-3.3 and EE2-3.4. In JVET-AM0172, EE2-3.5 combines EE2-3.3 and EE2-3.4c. In EE2-3.3, a third transform set selection method for the neural network-based intra prediction mode is proposed. EE2-3.4c increases the amount of output non-zero coefficients for NSPT for block sizes 4x8, 8x4, 4x16, 16x4, 8x32, and 32x8, and retrains all NSPT kernels. Moreover, the number of NSPT kernels is reduced as, for some block sizes, retrained NSPT kernels are shared between intra prediction modes.