shihzy
commented
5 years ago hi @rt2017bk - can you provide a bit more detail on how you are training. Can you also provide the python mlagents-learn command you are using?
Closed rt2017bk closed 5 years ago
shihzy
commented
5 years ago hi @rt2017bk - can you provide a bit more detail on how you are training. Can you also provide the python mlagents-learn command you are using?
shihzy
commented
5 years ago Thanks for reaching out to us. We are closing this due to inactivity, but if you need additional assistance, feel free to reopen the issue.
Lucci93
commented
5 years ago Hi, I reopen the issue because I have the same problem. Could you help me? I used the ML-Agent library for a project in October 2018 with Unity 2018.2.2, Python 3.6.6 and the 0.4.0b version of the library with my custom game. Today I have tried to train the AI of the project with the same configuration of October 2018 but I have got the following error:
escapist 16 ~/Desktop/TOGEXP/run $ python3 ../python/learn.py --run-id=testMC --train /Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:517: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'. _np_qint8 = np.dtype([("qint8", np.int8, 1)]) /Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:518: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'. _np_quint8 = np.dtype([("quint8", np.uint8, 1)]) /Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:519: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'. _np_qint16 = np.dtype([("qint16", np.int16, 1)]) /Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:520: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'. _np_quint16 = np.dtype([("quint16", np.uint16, 1)]) /Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:521: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'. _np_qint32 = np.dtype([("qint32", np.int32, 1)]) /Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:526: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'. np_resource = np.dtype([("resource", np.ubyte, 1)])
UNITY LOGO HERE
INFO:unityagents:{'--curriculum': 'None',
'--docker-target-name': 'Empty',
'--help': False,
'--keep-checkpoints': '5',
'--lesson': '0',
'--load': False,
'--no-graphics': False,
'--run-id': 'testMC',
'--save-freq': '50000',
'--seed': '-1',
'--slow': False,
'--train': True,
'--worker-id': '0',
'
Unity brain name: SurvivalBrainComp
Number of Visual Observations (per agent): 0
Vector Observation space type: continuous
Vector Observation space size (per agent): 108
Number of stacked Vector Observation: 25
Vector Action space type: continuous
Vector Action space size (per agent): 5
Vector Action descriptions: Forward, Back, Fire, Left, Right
/Users/dario/Desktop/TOGEXP/python/unitytrainers/trainer_controller.py:194: YAMLLoadWarning: calling yaml.load() without Loader=... is deprecated, as the default Loader is unsafe. Please read https://msg.pyyaml.org/load for full details.
trainer_config = yaml.load(data_file)
2019-10-30 11:13:15.188220: I tensorflow/core/platform/cpu_feature_guard.cc:140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
INFO:unityagents:Hyperparameters for the PPO Trainer of brain SurvivalBrainComp:
batch_size: 2048
beta: 0.001
buffer_size: 20480
epsilon: 0.2
gamma: 0.9
hidden_units: 128
lambd: 0.93
learning_rate: 0.0001
max_steps: 5.0e5
normalize: True
num_epoch: 5
num_layers: 2
time_horizon: 256
sequence_length: 64
summary_freq: 2000
use_recurrent: True
graph_scope:
summary_path: ./summaries/testMC
memory_size: 256
use_curiosity: False
curiosity_strength: 0.01
curiosity_enc_size: 128
Traceback (most recent call last):
File "../python/learn.py", line 71, in
shihzy
commented
5 years ago hi @Lucci93 - we are no longer supporting that version of ML-Agents. Can you try using the latest version to see if you are still getting these errors?
Lucci93
commented
5 years ago I will try, but I think to reproduce my experiments I need to re-run the training with the 0.4.0b version of the code. Is it possible that some library is no longer backwards compatible with my project?
shihzy
commented
5 years ago hi @Lucci93 - i think without knowing what exactly was changed in your setup, it would be difficult to debug. If nothing was touched or updated, it would run as expected.
Lucci93
commented
5 years ago Hi @unityjeffrey. In October 2018 I ran the requirement.txt file with the actual version of the libraries written inside. Using the libraries written in the file today is probably the reason why my code doesn't work. Is the only thing I have re-imported in my project. Unity, python and Ml-Agent versions are the same. In support of my thesis even the examples inside the ML-Agent folder doesn't work with a clear installation following the specifications of the guide written for the version 0.4.0b. Do you have any idea?
NFO:unityagents:Hyperparameters for the PPO Trainer of brain Ball3DBrain:
batch_size: 64
beta: 0.001
buffer_size: 12000
epsilon: 0.2
gamma: 0.995
hidden_units: 128
lambd: 0.99
learning_rate: 0.0003
max_steps: 5.0e4
normalize: True
num_epoch: 3
num_layers: 2
time_horizon: 1000
sequence_length: 64
summary_freq: 1000
use_recurrent: False
graph_scope:
summary_path: ./summaries//Users/danielepiergigli/Desktop/test
memory_size: 256
use_curiosity: False
curiosity_strength: 0.01
curiosity_enc_size: 128
INFO:unityagents: Ball3DBrain: Step: 1000. Mean Reward: 1.255. Std of Reward: 0.736.
INFO:unityagents: Ball3DBrain: Step: 2000. Mean Reward: 1.389. Std of Reward: 0.751.
INFO:unityagents: Ball3DBrain: Step: 3000. Mean Reward: 1.677. Std of Reward: 1.003.
INFO:unityagents: Ball3DBrain: Step: 4000. Mean Reward: 2.363. Std of Reward: 1.699.
INFO:unityagents: Ball3DBrain: Step: 5000. Mean Reward: 3.512. Std of Reward: 2.837.
INFO:unityagents: Ball3DBrain: Step: 6000. Mean Reward: 5.752. Std of Reward: 5.195.
INFO:unityagents: Ball3DBrain: Step: 7000. Mean Reward: 11.211. Std of Reward: 10.930.
INFO:unityagents: Ball3DBrain: Step: 8000. Mean Reward: 18.083. Std of Reward: 19.186.
INFO:unityagents: Ball3DBrain: Step: 9000. Mean Reward: 23.474. Std of Reward: 22.393.
Traceback (most recent call last):
File "learn.py", line 84, in
github-actions[bot]
commented
3 years ago This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.
my training stop and I get the following error, can someone help me ???
INFO:unityagents: Ball3DBrain: Step: 4000. Mean Reward: 5.405. Std of Reward: 7.848. INFO:unityagents: Ball3DBrain: Step: 5000. Mean Reward: 17.955. Std of Reward: 17.415. Traceback (most recent call last): File "python/learn.py", line 62, in
tc.start_learning()
File "C:\Users\pc\Desktop \ml-agents-0.3.1a\python\unitytrainers\trainer_controller.py", line 259, in start_learning trainer.update_model()
File " C:\Users\pc\Desktop \ml-agents-0.3.1a\python\unitytrainers\ppo\trainer.py", line 360, in update_model
self.model.returns_holder: np.array(_buffer['discounted_returns'][start:end]).reshape( ValueError: could not broadcast input array from shape (1001) into shape (1)