Is there a config file or dict-format settings to implement the rainbow-benchmark score in you paper?

[fmxFranky]

Thanks for the repo, i am now trying running the rainbow's example code in the CuLE on my DGX-1-Station. The CuLE's paper on arxiv given a table that using CuLE and distributed pytorch can implement Rainbow within 3h on DGX-1(shown in the follow):

However i dont know how to set the hyper-parameters to implement it.Can you provide me a config for those hyper-parameters for implementing the table's results above?
Or can you tell me how to modify the the items in 'cule/examples/dqn/benchmark.config'?
thanks a lot!~