The contribution reports the results of the training cross-check of EE1-1.2 described in JVET-AK0150. For the LOP4 model, the results from the training cross-check match those from the proponents, despite some differences in the luma/chroma balancing. For the VLOP3 model, the training of the cross-checked model did not converge as well as the training of the proponent model. The inference results using the VLOP3 model from the training cross-check are worse than the inference results using the VLOP3 model from the proponents.