Posts for Netbrian

Posted: 10/14/2018 5:01 PM
Post subject: Re: Apply machine learning in Bizhawk
Quote
Netbrian
Experienced Forum User
Joined: 8/9/2009
Posts: 5
HeyRay wrote:
I'm using Bizhawk on Windows and want to apply some ml methods to games using python or lua. However, I always fail to require torch in lua, so I want to get data into python. Is there any great way to make it? Or how to install torch into Bizhawk lua enviornment? I tried a lot. Thanks in advance!
The most reliable way I've found to get this working is use sockets to have LUA act as sort of an API for Python. I was able to successfully do this using ZMQ, and knew nothing about socket programming (or LUA) when I tried. It'd be nice to have a standardized way to do this, but it isn't a huge obstacle.
Posted: 3/25/2018 11:55 PM
Quote
Netbrian
Experienced Forum User
Joined: 8/9/2009
Posts: 5
feos wrote:
Does this help? https://github.com/Ashafix/PyEmuHawk
Yes, that's perfect! Thank you very much! :)
Posted: 3/25/2018 4:41 PM
Quote
Netbrian
Experienced Forum User
Joined: 8/9/2009
Posts: 5
I'm experimenting a bit with the new communication library (have you thought about bundling zmq?) The unit test says "Please run TestCommunication_All.lua with the supplied Python server for a more comprehensive test", but I'm not able to find the Python server in question. Can anyone point me in the right direction?
Posted: 7/1/2011 3:04 AM
Quote
Netbrian
Experienced Forum User
Joined: 8/9/2009
Posts: 5
I'd appreciate seeing an Angband TAS, though it'd probably be somewhat miserable to put together.
Posted: 8/27/2010 4:31 AM
Quote
Netbrian
Experienced Forum User
Joined: 8/9/2009
Posts: 5
I'm trying to use the symbolic debugging feature and having problems -- the program doesn't seem to load my name files. For instance, I created a name file for Solar Wars (a public domain ROM) called SolarWars2001.NES.1.nl, with the comment $E43E#ChangeX# (newline) However, the debugger doesn't seem to have noticed the file. Are there any known issues, or am I missing something? Thank you!