search

TeaPearce / Counter-Strike_Behavioural_Cloning

IEEE CoG & NeurIPS workshop paper 'Counter-Strike Deathmatch with Large-Scale Behavioural Cloning'
343 stars 45 forks source link

Record Crashes #7

Closed DuroDaCoder closed 2 years ago

DuroDaCoder commented 2 years ago

Hi, I have this issue: I open dm_record_data.py, and nothing happens, I press Q and nothing happens too. I set SHOW_IMAGE to True and window opens, that says resized that is crashing.

Video of error:

https://user-images.githubusercontent.com/48152410/152793200-c9b00f2f-64c8-449a-aae4-87947a06b05f.mp4 PS: I just requested for datasets, my email is r. h. 1234 @ gmail.com

TeaPearce commented 2 years ago

Hi, the script 'dm_record_data.py' is used for recording new data -- is this what you were trying to do? If so, you should also already have the game running (i.e. have joined a deathmatch) before running the script. I'm not exactly sure why it's crashing, would need more info.

Have you got the dm_run_agent.py script working?

DuroDaCoder commented 2 years ago

Hi, yes I am trying to record new data, and dm_run_agent.py works for me. I was in game, when trying to record. I have no idea, I tried to run screen_input.py and it works, but the script dm_record_data.py and dm_record_data_me_wasd.py doesn't work for me, I have no idea why.

TeaPearce commented 2 years ago

Hmm, I'd suggest debugging by gradually adding in the functionality in the dm_record_data.py script. e.g. run up to line 110 first, then add in the GSI stuff (I assume you installed the game state integration on your machine?).

Hopefully this will narrow down the problem a bit.

DuroDaCoder commented 2 years ago

Sorry, but I don't really know what you mean by that. I am new at python, and I have no idea how to do that.

DuroDaCoder commented 2 years ago

And when I have downloaded for example dataset_dm_expert_othermaps and want to use inferno 1-10, i can just jump to step 5 in the Code Overview ?

TeaPearce commented 2 years ago

Agreed, if you want to train on the existing datasets you can run dm_train_model.py directly (obviously still need to update paths and choose whether to fine tune an exisiting model etc).