Failed to reproducing result of RNG_GFN.ipynb

[Mr-Peach0301]

I directly run the code in RNG_GFN.ipynb, but I couldn't get the displayed result. Instead, I got:

Does anyone know how to solve this?

[MJ10]

Hi @Mr-Peach0301, could you please provide some more details? I just ran the notebook in clean environment and it seems to reproduce the results in the paper.

[Mr-Peach0301]

Thanks for your reply @MJ10 , I will try it again.
I noticed that the length of these numbers is only one token. Have you tried fitting distributions containing more tokens? Do you think this can succeed?

[Mr-Peach0301]

Additionally, can you tell me how many epochs were used for this result?

[MJ10]

I noticed that the length of these numbers is only one token. Have you tried fitting distributions containing more tokens? Do you think this can succeed?

The numbers are not all single token - there can be multiple tokens.

Additionally, can you tell me how many epochs were used for this result?

This is using the default config - which uses 10 epochs.