Henry Charlesworth
Henry Charlesworth
Thanks for the reply. I guess my thought was it'd be better to broadcast onto the height dimension - so you could one-hot encode the actions and end up with...
Hi! Yes initially when I trained it (to get the parameters which are included in the repository) this never happened - this was using 150 million training steps. But when...
Yeah that's very peculiar. For me, I can train it on my GPU (1070) for more updates than that before getting NaNs, but they still come eventually. I really don't...
OK well I appreciate you spending the time trying to figure it out, let me know if you're able to fix the issue! On the second question, I don't think...
For me it was because I was using Python 3.5, once I updated to 3.6 it was working (I was getting the error: pytorch-a2c-ppo-acktr-gail/a2c_ppo_acktr/envs.py", line 146 assert len(op) == 3,...
I think so - using 3.17.5. I tried a number of earlier versions and this didn't seem to help.
Same issue for me. Everything had been working but now suddenly getting this.
Hi Marvin, thanks for getting back to me! I appreciate that you're both very busy people and so it's easy to miss an issue on here! Haha well I guess...
So I ran the example in the release branch. It certainly runs OK but the final results don't seem right at all. After training the model, the next stage leads...
Yeah, so these are the results I got (here I changed the number of iterations of 10 to 50, but the other parameters were the default, and it didn't do...