bark-simulator/bark-ml

logging in ```iqn_agent.py```

tinmodeHuang opened this issue · 3 comments

self.writer.add_scalar('stats/mean_Q', mean_q, 4 * self.steps)
hello! authors.
I wonder if self.steps times 4 in order to just scale horizontal axis.

Could you please provide a link to the source code? I am not familiar with this code and @juloberno might be able to help you out here. Seems like it scales the axis, or that the steps have to be scaled.

well,

self.writer.add_scalar('stats/mean_Q', mean_q, 4 * self.steps)

I guess, this comment could give a clue about that?

# We log evaluation results along with training frames = 4 * steps.