TorchCraft / TorchCraftAI

A platform that lets you build agents to learn to play StarCraft: Brood War.

Home Page:https://torchcraft.github.io/TorchCraftAI

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Run micro_tutorial get error

JeremyLee268 opened this issue · comments

I have bulit TorchCraftAI and CherryPi on unbuntu 18.04.

But I got this error when I input this cmd
./build/tutorials/micro/micro_tutorial -vmodule state=-1,micro_tutorial=1 -scenario vu_zl

I 6788/XXXXX 04/24 13:02:39 [micro_tutorial.cpp:246] Scenario: vu_zl
I 6788/XXXXX 04/24 13:02:39 [micro_tutorial.cpp:247] Model: PF
I 6788/XXXXX 04/24 13:02:39 [micro_tutorial.cpp:248] Resume: false
I 6788/XXXXX 04/24 13:02:39 [micro_tutorial.cpp:249] Evaluate: false
I 6788/XXXXX 04/24 13:02:39 [micro_tutorial.cpp:256] resultsJSON: ./metrics-rank-0.json
I 6788/XXXXX 04/24 13:02:39 [micro_tutorial.cpp:257] resultsCheckpoint: ./trainer_latest.bin
I 6788/XXXXX 04/24 13:02:39 [micro_tutorial.cpp:284] Model has 221568 total parameters
I 6788/XXXXX 04/24 13:02:39 [micro_tutorial.cpp:290] Begin training!

F0424 13:02:39.818872 6796 forkserver.cpp:575] You must call ForkServer::startForkServer! Call it as early as possible (in main, after parsing command line flags, but before initializing gloo/mpi/anything else)!

*** Check failure stack trace: ***
F0424 13:02:39.818872 6797 forkserver.cpp:575] You must call ForkServer::startForkServer! Call it as early as possible (in main, after parsing command line flags, but before initializing gloo/mpi/anything else)!F0424 13:02:39.819079 6798 forkserver.cpp:575] You must call ForkServer::startForkServer! Call it as early as possible (in main, after parsing command line flags, but before initializing gloo/mpi/anything else)!F0424 13:02:39.819214 6799 forkserver.cpp:575] You must call ForkServer::startForkServer! Call it as early as possible (in main, after parsing command line flags, but before initializing gloo/mpi/anything else)!F0424 13:02:39.819252 6800 forkserver.cpp:575] You must call ForkServer::startForkServer! Call it as early as possible (in main, after parsing command line flags, but before initializing gloo/mpi/anything else)!

Sorry Jeremy, I just saw this. The tutorial fell a little out of date. Try adding this line prior to cherrypi::init() in micro_tutorial.cpp: https://github.com/TorchCraft/TorchCraftAI/blob/develop/scripts/botplay/botplay.cpp#L36