Microsoft/malmo

Project Malmo is a platform for Artificial Intelligence experimentation and research built on top of Minecraft. We aim to inspire a new generation of research into challenging new problems presented by this unique environment.

SP Mohanty

@spMohanty

So theres a chance that this might be because of some issue in the actual submitted code.

iZuno-K

@iZuno-K

@spMohanty Thank you for your investigating.

I’ve succeeded the local test based on crowdAI’s instruction (https://github.com/crowdAI/marlo-multi-agent-starter-kit). I tested 1 game with 2 agents and 3 different games with 6 agents. Both tests succeed without errors.

How can I check my submitted code have any issue?

SP Mohanty

@spMohanty

@iZuno-K : Do you test it for multiple episodes locally ?

iZuno-K

@iZuno-K

@spMohanty I tested multiple episodes today. My code cause an error RuntimeError: Attempted to use a closed Session. The reason is I use tf.Session().close() at the end of the run_episode().
Then I modified run.py by moving the tf.Session().close() line to the end of the main function. After that I tested 15 episodes local test and succeeded.
I submitted the modified code but receive Unable to orchestrate evaluation. Core services failed to start....

ProQianXiao

@ProQianXiao

@spMohanty @AndKram I submitted my code but I got the following error ,

it seems that episode-1 and episode-2 are successful, but I do not know why the later episodes failed.

ProQianXiao

@ProQianXiao

@spMohanty @AndKram How to solve the problem "Unable to orchestrate evaluation. Core services failed to start...."? Is it a evaluation environment error or code error?

iZuno-K

@iZuno-K

@ProQianXiao @spMohanty @AndKram I solved the problem by removing apt.txt from my repository.

I removed these dependencies and then the evaluator works.
libopenmpi-dev, openmpi-bin (for mpi4py)
libgtk2.0-dev, libglib2.0 (for opencv)

ProQianXiao

@ProQianXiao

@iZuno-K Thanks for your reply, did it work for you ? I checked my environment.yml, in fact there are no these dependencies.

This is mine

iZuno-K

@iZuno-K

@ProQianXiao @spMohanty @AndKram Sorry, that is not the cause... I tried a new submission now with the same dependency with only modified trained parameters. But it failed...

ProQianXiao

@ProQianXiao

I always met this error, did you have same error?

iZuno-K

@iZuno-K

Yes. I met the error at the 17/18 submissions.

ProQianXiao

@ProQianXiao

That really makes me crazy!!! I have tested locally and everything is fine, I do not know why it failed in the server

iZuno-K

@iZuno-K

Me too. I'm in the same situation.

EC2EZ4RD

@EC2EZ4RD

I met the same problem.

Tejas Jogi

@tdjogi010

Having the same problem ... Unable to orchestrate submission, please contact Administrators.

ProQianXiao

@ProQianXiao

@spMohanty @AndKram It seems that many people have the same error, maybe it is caused by the evaluation environment? Are there any solutions about that?

Lennart

@biggzlar

Hi there, I was just pushing my latest tag and discovered that the challenge ended at 12:00 UTC - however in the text it states, that the challenge ends on 23:59 PST. What's up with that??

andrekramer

@andrekramer

https://xkcd.com/2092/ Happy New Year everyone!

EC2EZ4RD

@EC2EZ4RD

Hi, did anyone receive some email or get some information about how is the competition going?

d33ms

@matthew-egbert

Is there anyway to accelerate the passage of time in Malmo? Or to have the time of day change triggered by an agent touching something (or similar)?

SP Mohanty

@spMohanty

@EC2EZ4RD : We are running the post challenge evaluations, and will get back to you soon. We will pit all the submitted models against each other, and then will have the results soon. Hopefully by the end of this week. Thanks for your patience.

ProQianXiao

@ProQianXiao

@spMohanty Many people meet the error 'Unable to orchestrate submission, please contact Administrators.' when submitting, is this caused by the code or the evaluation server? If it is caused by the evaluation server, could you give us another chance to submit? Thanks very much.

Andre Kramer

@AndKram

@matthew-egbert You can speed up using `<ModSettings>

    <MsPerTick>50</MsPerTick>


</ModSettings>` in the mission XML.

d33ms

@matthew-egbert

Thanks Andre, but this speeds up the whole simulated world, not just the daytime/nighttime cycle. I see now my question wasn't as clear as it might have been.

I was curious about how to programmatically change from night to day in the middle of a simulation without resetting things (or having it cycle more rapidly between these times).

Andre Kramer

@AndKram

@matthew-egbert You can set the time for a mission or stop the cycle (https://github.com/Microsoft/malmo/blob/master/Schemas/Mission.xsd) but I don't think you can set the cycle duration in Malmo. Though I see there is a gamerule command for it.

Richsl

@Richsl

Hi Folks, excited to get Malmo up and running. I can't reconnect to the client when I quit to title in the client with a server still running. It still shows 'server Running' up in the corner even after I quit the server, and then I can't reconnect (still shows MCP 10000), I get "Error starting mission: Failed to find an available client for this mission - tried all the clients in the supplied client pool" Anyway I can just restart the client but it goes through a full forge build each time. Is there any way to restart the client quickly without the forge build each time? thx

Andre Kramer

@AndKram

@Richsl You should be able to call startMission again using the same AgentHost (assuming you are using the Malmo and not MarLo / open AI gym API). And, yes, running the Mod currently always requires starting a Froge build.

d33ms

@matthew-egbert

Similar to Richsl's issue, I often can't reconnect to the server. The launchClient.sh reports "REPLYING WITH: MALMOBUSY" and my clients (adapted from team_reward_test) repeated give the error "Calling startMission for role 0. Not enough available Minecraft instances running." I have two instances of Minecraft running, and things often work the first time (but not always), sometimes the second (but not always) and then never again until I reboot both instances of launchClient.sh

Any help would be much appreciated! Running ubuntu

iZuno-K

@iZuno-K

@spMohanty What day will you announce the evaluation result?

d33ms

@matthew-egbert

One last comment about the MALMOBUSY issue. It seems to me like the connection to Minecraft is not getting properly ended when my python script ends. Is that possible? Is there some explicit "I am done" message that needs to get sent?

Andre Kramer

@AndKram

@matthew-egbert There is an issue with closing the client (pull request #785) but exiting the Python script should disconnect everything. Could you open an issue on GitHub malmo and provide some more details on your script please?

d33ms

@matthew-egbert

I figured out that the issue was caused by my using a Cuboid decorator that had an invalid type specification. I guess the world couldn't be generated, but instead of getting an error to that effect, my python scripts were just reporting unable to connect as described above.

Andre Kramer

@AndKram

@matthew-egbert That's good .. but bad that the error was not obvious. The logs in Minecraft/run/logs are the place to be checking if there are any hangs or other issues with Malmo or MarLo api.

Linjie Xu

@egg-west

About the busy error or other error, I find it useful to simply restart the marlo server. Still, sometimes I got stuck during my training process. Wish to figure it out together and make this environment better. It is a good project indeed.

BCDav

@BCDav

Hi, I'm trying to get visual data from the agent. Is it possible to access the data acquired from ObservationFromRay or ColourMapProducer? If so, which directory would they be in? I'm using python by the way.

Andre Kramer

@AndKram

@BCDav Python_examples/build_test.py uses ObservationsFromRay (in the json observation data). ColorMapProducer is one of the video types returned as frames. The recording_dir is a string arg to build_test.py.

BCDav

@BCDav

So, just for confirmation, to access the frames from the video produces I’ll need to edit build_test.py? That is, store and use the frames again outside the mission, as done in https://www.youtube.com/watch?v=4Z-pXipl2K0

Andre Kramer

@AndKram

@BCDav There are several examples you could start from. sample_mission_loader.py which records MP4 for example (in Malmo/samples/Python_examples).

BCDav

@BCDav

Thanks, I'll give those a try

Andre Kramer

@AndKram

Hi, Here's a link to the official 2018 MARLÖ Competition announcement: https://www.facebook.com/MicrosoftResearchCam/posts/2457928897568422. Congratulations to the winners and also thanks to all those that took part!

Where communities thrive

People

Repo info

Activity