Matthew Hausknecht

AFAIK it's still an issue. Albeit a very hard one to trigger. Also not clear how to fix it short of hacking Stella internals. On Fri, Feb 6, 2015 at...

Hi, the train.sh script is designed to start run jobs on my local computational cluster. Specifically, the cluster command refers to this repo: https://github.com/LARG/cluster and is not the same cluster...

Hi, I've pushed an update to the repo that adds a --verbose flag which unsupresses output from the game server. Can you try the code again using the verbose flag...

Also, thanks for the suggestion to include Caffe Definitions - I've added it to the makefile!

Great! monitor-condor-job is a script that's specific to the computational cluster I'm using. If you wanted to try the experiments in the train.sh file, just comment out the monitor-condor-job line.

Can you try running some of the example C++ HFO agents (located in HFO/example) to determine if the problem is specific to HFO?

Hi Hongjie, you are correct - the features in HFO are designed to be overcomplete, because neural networks are able to handle redundant information gracefully. I have not tried using...

Sorry to hear that the approach is giving you trouble. In general high gradients like this are not a good indication. On Tue, Mar 20, 2018 at 11:54 PM, Arpit...

There was an issue with a version a couple of weeks ago where the CMake would have permissions errors grabbing the sub-libraries. On Nov,11 I pushed a commit to fix...

Ahh, yes, there were some recent changes to HFO that weren't reflected yet in dqn_hfo. I've pushed a commit to dqn_hfo repo to (hopefully) fix the error you're getting above....