In this contribution, four methods proposed in JVET-AC0056 have been cross-checked. They are float and integerized models and their models with or without RDOQ. The experiment was conducted using Kodak 24 test sequences and 7 lambda parameters. However, the integerized model with lambda equal to 0.08 is retrained using quantization-aware training (QAT). As experimental results, the results of two models without RDOQ are identical to that of JVET-AC0056. The results of the two models with RDOQ are almost identical to them; however, they are not matched with them precisely due to parallel optimization.