WARNING:tensorflow:From /Users/domore/Documents/deeplearning/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/base.py:198: retry (from tensorflow.contrib.learn.python.learn.datasets.base) is deprecated and will be removed in a future version.
Instructions for updating:
Use the retry module or similar alternatives.
algorithm : a2c
timesteps : 2000000
exploration_fraction : 0.5
prioritized : True
dueling : True
num_agents : 4
lr : 0.0005
random lr : 0.0005
Traceback (most recent call last):
File "/Users/domore/starcraft/pysc2-examples/train_mineral_shards.py", line 302, in
main()
File "/Users/domore/starcraft/pysc2-examples/train_mineral_shards.py", line 168, in main
env = SubprocVecEnv(FLAGS.num_agents + FLAGS.num_scripts, FLAGS.map)

TypeError: Can't instantiate abstract class SubprocVecEnv with abstract methods step_async, step_wait

Hi there!
Thanks for adding this useful project for pysc2 .

After cloning the repo and solve some requirement and installation issues, It appears the following error when training the map, in both deepq and acktr.

Can you point what to do here?
Many thanks !

I'm trying to get replay data from .SC2Replay files by running replay_minerals.py. The program runs successfully but outputs empty arrays and values. I've tried several different replay files. Here is a sample output:

============================================= Summary 10 secs =============================================
Replays: 0, Steps total: 0

Camera move: 0, Select pt: 0, Select rect: 0, Control group: 0

Maps: 0
{}

Races: 0
{}

Unit ids: 0
{}

Valid abilities: 0
{}

Made abilities: 0
{}

Valid actions: 0
{}

Made actions: 0
{}

Crashing replays: 0
[]

Invalid replays: 0
[]
---------------------------------------------- Process stats ----------------------------------------------
[ 0] replay: , replays: 0, steps: 0, game loops: 0, last: launch, 9 s ago

The program then finishes with 0 exit code.

Any ideas of what I should try to get this file to work properly?

I get this error when running any of the scripts. I cant seem to get this fixed. Maybe its a problem with pysc2 itself?

I ran the script last night. It started with ~11 mean reward, and ended with ~15.5 mean reward.
I tried to play this mini-game myself, and I could get ~100 score or more.
Deepmind reached ~100 score in their video.

Just like this video
https://www.youtube.com/watch?v=DXUOWXidcY0

When i run
python train_mineral_shards.py --algorithm=a2c

it says:
ValueError: You must specify the list of players.

does anyone know what happen?

I'm using Python 3.5.3 in macOS with following packages:

git+https://github.com/openai/baselines.git@699919f1cf2527b184f4445a3758a773f333a1ba#egg=baselines
PySC2==1.1

When I run the command:

$ python train_mineral_shards.py 

Error raises:

Traceback (most recent call last):
  File "train_mineral_shards.py", line 51, in <module>
    main()
  File "train_mineral_shards.py", line 24, in main
    visualize=True) as env:
  File "/Users/haje01/.pyenv/versions/pysc2/lib/python3.5/site-packages/pysc2/env/sc2_env.py", line 128, in __init__
    raise ValueError("All arguments must be passed as keyword arguments.")
ValueError: All arguments must be passed as keyword arguments.

Any ideas?

I am not sure how to best address this, but it looks like train_defeat_zerglings (and potentially the Q learning version of train_mineral_shards) are not operational with TF 1.14.

The error is thrown in line 1024 of layers.py

    if conv_dims is not None and conv_dims + 2 != input_rank:
      raise ValueError('Convolution expects input with rank %d, got %d' %
                       (conv_dims + 2, input_rank))

Hello,

I have an issue after installing the project and its prerequisites through Git Shell.

When I run train_mineral_shards.py, I get a bunch of stack traces, and the culprit is an ImportError:
ImportError: No module named '_pywrap_tensorflow_internal'

I 've found a workaround on this and currently working on it, but is there another way to resolve this?

I'm getting the following error when I run enjoy_mineral_shards. I didn't exactly have a seamless baselines installation, but I think baselines is installed correctly now and not the problem.

C:\Users\Rick\Desktop\pysc2-examples>python enjoy_mineral_shards.py
Traceback (most recent call last):
File "enjoy_mineral_shards.py", line 162, in
main()
File "enjoy_mineral_shards.py", line 55, in main
"mineral_shards.pkl", act_params=act_params)
File "C:\Users\Rick\Desktop\pysc2-examples\deepq_mineral_shards.py", line 93, in load
return ActWrapper.load(path, num_cpu=num_cpu, act_params=act_params)
File "C:\Users\Rick\Desktop\pysc2-examples\deepq_mineral_shards.py", line 44, in load
act = deepq.build_act(**act_params)
File "c:\users\rick\desktop\baselines\baselines\deepq\build_graph.py", line 146, in build_act
observations_ph = make_obs_ph("observation")
File "enjoy_mineral_shards.py", line 46, in make_obs_ph
return U.BatchInput((64, 64), name=name)
AttributeError: module 'baselines.common.tf_util' has no attribute 'BatchInput'
W0204 12:41:43.175422 10984 sc_process.py:183] Killing the process.

After cleanups done in baselines repo I am getting new errors while running train_mineral_shards (with enjoy_mineral_shards everything works just fine)

Traceback (most recent call last):
  File "C:/Users/Artur/Documents/PySC2/train/train_mineral_shards.py", line 14, in <module>
    from a2c.policies import CnnPolicy
  File "C:\Users\Artur\Documents\PySC2\train\a2c\policies.py", line 3, in <module>
    from baselines.acktr.utils import conv, fc, dense, conv_to_fc, sample, kl_div
ImportError: cannot import name 'conv'

Process finished with exit code 1

Any help?

show an error when i run train_mineral_shards

File "C:\Anaconda3\lib\site-packages\pysc2\run_configs_init_.py", line 39, in get
if FLAGS.sc2_run_config is None: # Find the highest priority as default.
File "C:\Anaconda3\lib\site-packages\absl\flags_flagvalues.py", line 491, in getattr
raise _exceptions.UnparsedFlagAccessError(error_message)
absl.flags._exceptions.UnparsedFlagAccessError: Trying to access flag --sc2_run_config before flags were parsed.

Hi, this is somewhat related to Issue #13 .
As of 3/11/2019, openaI/baselines does not support BatchInput in baselines.deepq.utils anymore.

The change was made in commit 24fe3d6576dd8f4cdd5f017805be689d6fa6be8c May 21, 2018.
It seems like BatchInput has been changed to ObvservationInput.

openai/baselines issue #524 also states this issue, but there were no replies made yet.

Since ObvservationInput only accepts gym spaces, I'm not really sure how to modify enjoy_mineral_shards.py to run on the latest baselines since the code below will not work anymore.

def make_obs_ph(name):
      return U.BatchInput((64, 64), name=name)

I think I should be using a code something like the below, but if I do so I would get another error saying ValueError: Convolution expects input with rank 4, got 2 while building the act.

def make_obs_ph(name):
      U.ObservationInput(Box(low=-32.0, high=32.0, shape=(2,)), name=name)

Thanks in advance!

I can't get started train_minteral_shards.py example. Getting this error:

$ python train_mineral_shards.py
Traceback (most recent call last):
File "train_mineral_shards.py", line 14, in
from common.vec_env.subproc_vec_env import SubprocVecEnv
File "D:\devel\repositories\pysc2-examples\common_init_.py", line 1, in
from baselines.common.console_util import *
ImportError: No module named 'baselines.common.console_util'

btw, I am using windows, can't it be problem? I've tried build baselines from repository, but I've failed on atari.py dependency (not very compatible with windows). baselines0.1.4 or lower libraries can be installed using pip, but seem to be not sufficient for example. Please help how to proceed.

Hallo , when I run command "python enjoy_mineral_shards.py" in terminal ,the error is:

Traceback (most recent call last):
File "enjoy_mineral_shards.py", line 163, in
main()
File "enjoy_mineral_shards.py", line 39, in main
game_steps_per_episode=steps * step_mul) as env:
File "/home/m/.local/lib/python3.5/site-packages/pysc2/env/sc2_env.py", line 248, in init
raise ValueError("Please specify agent_interface_format.")
ValueError: Please specify agent_interface_format.

Can anyone else can solve it ??? Thanks

I get the following error:

pygame 1.9.4
Hello from the pygame community. https://www.pygame.org/contribute.html
Error
Traceback (most recent call last):
File "/anaconda3/envs/conda/lib/python3.6/unittest/case.py", line 59, in testPartExecutor
yield
File "/anaconda3/envs/conda/lib/python3.6/unittest/case.py", line 605, in run
testMethod()
File "/Users/donggeun/Downloads/pysc2-examples-master/tests/scripted_test.py", line 33, in test_defeat_zerglings
game_steps_per_episode=self.steps * self.step_mul) as env:
File "/anaconda3/envs/conda/lib/python3.6/site-packages/pysc2/env/sc2_env.py", line 248, in init
raise ValueError("Please specify agent_interface_format.")
ValueError: Please specify agent_interface_format.

Hi! I tried to run the train_mineral_shards.py using pycharm on the windows 7, the error is :
Traceback (most recent call last):
File "D:\Anaconda3\lib\multiprocessing\process.py", line 258, in _bootstrap
self.run()
File "D:\Anaconda3\lib\multiprocessing\process.py", line 93, in run
self._target(*self._args, **self._kwargs)
File "C:\Users\Administrator.SC-201608140925\Desktop\pysc2-examples-master\common\vec_env\subproc_vec_env.py", line 20, in worker
minimap_size_px=(32, 32)) as env:
File "D:\Anaconda3\lib\site-packages\pysc2\env\sc2_env.py", line 182, in init
"screen_size_px and minimap_size_px are deprecated. Use the feature "
DeprecationWarning: screen_size_px and minimap_size_px are deprecated. Use the feature or rgb variants instead. Make sure to check your observations too since they also switched from screen/minimap to feature and rgb variants.
I hope someone can answer my question, thank you！

Hi,
I was wondering if you have or know of any examples that implemented supervised learning using Starcraft II game data (implimented in section 5 of this paper : StarCraft II: A New Challenge for
Reinforcement Learning). Methods for how the features were generated from game replays or anything of that sort is always appreciated.

Thank you,

Hi there,

This is more of a question, I'm using your code to train some agents and I would like to generate graphs of their performance rating over time. Is there already a way to do that already or I would need to export the scores at the end of each iteration?

Thanks

Trying to run train_defeat_zerglings.py using pycharm on Mac in a virtual env.

I get this. By chance would he api of baselines have changed? Does anyone know if this code is up to date with the libraries?

thanks

In pysc2, How can I use auto attack? Now it seems disable.

$ python train_mineral_shards.py --algorithm=a2c
/home/guoyoutian/anaconda3/lib/python3.6/site-packages/h5py/init.py:36: FutureWarning: Conversion of the second argument of issubdtype from float to np.floating is deprecated. In future, it will be treated as np.float64 == np.dtype(float).type.
from ._conv import register_converters as _register_converters
pygame 1.9.4
Hello from the pygame community. https://www.pygame.org/contribute.html
Traceback (most recent call last):
File "train_mineral_shards.py", line 15, in
from a2c import a2c
File "/home/guoyoutian/PycharmProjects/pysc2-examples-master/a2c/a2c.py", line 12, in
from baselines.a2c.utils import cat_entropy, mse
ImportError: cannot import name 'mse'

there ara an error:
NotFoundError (see above for traceback): Key deepq_x/q_func/state_value/fully_connected/biases not found in checkpoint

when i run
python enjoy_mineral_shards.py

does anyone know how to solve it?

Hi! I tried to run it on the windows 7. But got the problem that I can't install git+https://github.com/openai/baselines (it fails during installing atari-py). Did someone win it?

What's the baselines' version in this repo?
Baselines have pr so many times, some functions can't work well.

We have setup a Discord Server @ see below dedicated for PySC2 collaboration.

• I should add this is not endorsed by this project and is open to anyone

Hello, I am trying to get the example work in Windows 10. Here is an error which I encountered.

Traceback (most recent call last):
  File "train_mineral_shards.py", line 302, in <module>
    main()
  File "train_mineral_shards.py", line 168, in main
    env = SubprocVecEnv(FLAGS.num_agents + FLAGS.num_scripts, FLAGS.map)
TypeError: __init__() missing 1 required positional argument: 'map_name'