Performance lower than reported

[chenz97]

Hi, thank you for your great work and code!

I ran your training code of node classification by make test_phase=1 save_dir=save, but have found that the performance on the validation set is relatively low at accuracy : 0.4146, hit ratio : 0.5090, macro f1 : 0.2762, expected return : -0.0087, which seems to be much lower than the figures reported in the paper.

Is there any configurations that should be modified to reproduce your results? Thank you very much!

[280777510]

Hi, thank you for your great work and code!

I ran your training code of node classification by make test_phase=1 save_dir=save, but have found that the performance on the validation set is relatively low at accuracy : 0.4146, hit ratio : 0.5090, macro f1 : 0.2762, expected return : -0.0087, which seems to be much lower than the figures reported in the paper.

Is there any configurations that should be modified to reproduce your results? Thank you very much!
Hello, I cannot download the stock data and Wikipedia relationship data provided in the original paper, could you please give me a copy?