Reproducing results

[zerobayes]

Hi,

Thank you for open-sourcing your work! I have a question regarding reproduciblity.

I tried running the code on the Atari 100k benchmark and I did not manage to reproduce the results. Here are the results I am obtaining:

	Obtained	Reported in paper
Boxing	0.5	78
Breakout	10	20
CrazyClimber	3188	71820
Krull	3414	6349
Pong	-20	19

Is there a step I’m missing here? I installed the dependencies specified in the requirements file and ran the training command specified in the README file. The results I am reporting come from the eval/final_score_mean chart.

Thanks!

[yingchengyang]

Hi, I also have the same problem. I run Pong and the result is -21, are there any possible reasons? Thanks a lot!

[fancy1223]

Hi, where is the eval/final_score_mean chart?

[fancy1223]

Hi,

Thank you for open-sourcing your work! I have a question regarding reproduciblity.

I tried running the code on the Atari 100k benchmark and I did not manage to reproduce the results. Here are the results I am obtaining:

Obtained Reported in paper
Boxing 0.5 78
Breakout 10 20
CrazyClimber 3188 71820
Krull 3414 6349
Pong -20 19
Is there a step I’m missing here? I installed the dependencies specified in the requirements file and ran the training command specified in the README file. The results I am reporting come from the eval/final_score_mean chart.

Thanks!

Hi, where is the eval/final_score_mean chart?