This contribution reports the crosscheck results of EE1-1.6 described in JVET-Z2023 [1], which evaluates the effectiveness of the deep in-loop filter presented in JVET-Z0113 [2]. Compared with the simulation results in JVET-Z0113, the retrained models bring a BD-rate difference of -0.10%, -0.71%, -0.96% under AI configuration, and 0.40%, -0.17%, -0.73% under RA configuration. The simulation results reveal that the training process of JVET-Z0113 is reproducible. As for EE1.6.2, the BD-rate performance exactly matches the results provided by the proponents under the fixed-point calculation. While the results with the float-point calculation is very close to the proponent’s results.