Although, i have installed all the necessary dependencies, i still get the above error code which has prevented the training of the dataset.

[Gabbydgoat]

          Although, i have installed all the necessary dependencies, i still get the above error code which has prevented the training of the dataset.

Originally posted by @Gabbydgoat in #2 (comment)

[BohyungPaeng]

The error is related to the args.state_dim not corresponding with actual state vector generated in wrapper.py.
args.state_dim is correctly given as (10,62) by default configuration in config.py (when args.did=0 and args.F=10)
It means that wrapper.py is somewhat wrong, which generated 440 = 10*44 size state
Missing # of values per family is 62-44=18.. and hard to think obvious reason, however, some possibilities are as follows.

1. self.sopt in wrapper.py must remain as 'mat' (default argument), or 4*(# of job attributes)=16 can be omitted
2. args.use must remain as default value : [1,2,3,4,5,7]
   Try checking the above two items. The easist way to skip the error might be to force args.state_dim as [10,44]

I am sorry that you stuck in an error even with default configuration and right dependencies..
Thanks for reporting and Help me fix this error with yet unknown causes

[johnpapiscurious]

Hi there,
Has this issue been resolved yet? I am curious in testing the code out.
Thank you!
@BohyungPaeng

[Gabbydgoat]

The last time I checked, there were just too many bugs as a result of the conflicting versions of the tensorflow that was implemented in designing the project. I believe in our last correspondence, he did mention updating it to a more recent version of tensorflow. However, it still remained buggy when I did try it so I lost interest in the paper altogether and moved on. I believe you could still try it out if you are very much interested in the research. However, since he coded basically from first principles without using the standard available libraries, you can easily just implement your own using his logic. That would be a great learning curve. Wishing you the best of Luck

…

On Tue, 6 Feb 2024 at 18:56, John Pap ***@***.***> wrote: Hi there, Has this issue been resolved yet? I am curious in testing the code out. Thank you! — Reply to this email directly, view it on GitHub <#3 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AY3BRZRELU3DBIST7ERHE23YSJODXAVCNFSM6AAAAAAZ66RMNKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSMZQGM2TQMZSGM> . You are receiving this because you authored the thread.Message ID: ***@***.***>