notebook.community

Edit and run

Baselines





In [0]:



    


!pip install git+https://github.com/openai/baselines >/dev/null



    











In [2]:



    


!rm -r ~/logs



    


















    






rm: cannot remove '/root/logs': No such file or directory

















In [3]:



    


# MsPacmanNoFrameskip-v4: Atari has no entropy source, so these are deterministic environments 

!OPENAI_LOGDIR=$HOME/logs/atari-ppo OPENAI_LOG_FORMAT=csv python -m baselines.run --alg=ppo2 --env=MsPacmanNoFrameskip-v4 --num_timesteps=30000 --nsteps=128



    


















    






env_type: atari
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
Training ppo2 on atari:MsPacmanNoFrameskip-v4 with arguments 
{'nsteps': 128, 'nminibatches': 4, 'lam': 0.95, 'gamma': 0.99, 'noptepochs': 4, 'log_interval': 1, 'ent_coef': 0.01, 'lr': <function atari.<locals>.<lambda> at 0x7f25a6b84d08>, 'cliprange': <function atari.<locals>.<lambda> at 0x7f25944920d0>, 'network': 'cnn'}
2018-12-20 10:57:04.182978: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:964] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2018-12-20 10:57:04.183467: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1432] Found device 0 with properties: 
name: Tesla K80 major: 3 minor: 7 memoryClockRate(GHz): 0.8235
pciBusID: 0000:00:04.0
totalMemory: 11.17GiB freeMemory: 11.10GiB
2018-12-20 10:57:04.183513: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1511] Adding visible gpu devices: 0
2018-12-20 10:57:04.641697: I tensorflow/core/common_runtime/gpu/gpu_device.cc:982] Device interconnect StreamExecutor with strength 1 edge matrix:
2018-12-20 10:57:04.641779: I tensorflow/core/common_runtime/gpu/gpu_device.cc:988]      0 
2018-12-20 10:57:04.641806: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1001] 0:   N 
2018-12-20 10:57:04.642164: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1115] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 10758 MB memory) -> physical GPU (device: 0, name: Tesla K80, pci bus id: 0000:00:04.0, compute capability: 3.7)
[629229239f8e:02256] *** Process received signal ***
[629229239f8e:02256] Signal: Segmentation fault (11)
[629229239f8e:02256] Signal code: Address not mapped (1)
[629229239f8e:02256] Failing at address: 0x7f7bcb7d320d
[629229239f8e:02256] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x12890)[0x7f7bce883890]
[629229239f8e:02256] [ 1] /lib/x86_64-linux-gnu/libc.so.6(getenv+0xa5)[0x7f7bce4c2785]
[629229239f8e:02256] [ 2] /usr/lib/x86_64-linux-gnu/libtcmalloc.so.4(_ZN13TCMallocGuardD1Ev+0x34)[0x7f7bced2de44]
[629229239f8e:02256] [ 3] /lib/x86_64-linux-gnu/libc.so.6(__cxa_finalize+0xf5)[0x7f7bce4c3615]
[629229239f8e:02256] [ 4] /usr/lib/x86_64-linux-gnu/libtcmalloc.so.4(+0x13cb3)[0x7f7bced2bcb3]
[629229239f8e:02256] *** End of error message ***

















In [4]:



    


!ls -l ~/logs



    


















    






total 4
drwxr-xr-x 2 root root 4096 Dec 20 10:57 atari-ppo

















In [5]:



    


from baselines.common import plot_util as pu
results = pu.load_results('~/logs/atari-ppo')



    


















    






Logging to /tmp/openai-2018-12-20-10-59-23-325901










    






/usr/local/lib/python3.6/dist-packages/baselines/bench/monitor.py:164: UserWarning: Pandas doesn't allow columns to be created via a new attribute name - see https://pandas.pydata.org/pandas-docs/stable/indexing.html#attribute-access
  df.headers = headers # HACK to preserve backwards compatibility

















In [6]:



    


import matplotlib.pyplot as plt
import numpy as np
r = results[0]
plt.plot(r.progress.total_timesteps, r.progress.eprewmean)



    


















    
Out[6]:








[<matplotlib.lines.Line2D at 0x7f220c6764e0>]










    
























In [0]:



    






    











In [7]:



    


!for seed in $(seq 0 5); do OPENAI_LOG_FORMAT=csv OPENAI_LOGDIR=$HOME/logs/atari-ppo-hp/b128-$seed python -m baselines.run --alg=ppo2 --env=MsPacmanNoFrameskip-v4 --num_timesteps=30000 --seed=$seed --nsteps=128; done



    


















    






env_type: atari
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
Training ppo2 on atari:MsPacmanNoFrameskip-v4 with arguments 
{'nsteps': 128, 'nminibatches': 4, 'lam': 0.95, 'gamma': 0.99, 'noptepochs': 4, 'log_interval': 1, 'ent_coef': 0.01, 'lr': <function atari.<locals>.<lambda> at 0x7f8671d7fd08>, 'cliprange': <function atari.<locals>.<lambda> at 0x7f865f68d0d0>, 'network': 'cnn'}
2018-12-20 10:59:34.016061: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:964] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2018-12-20 10:59:34.016570: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1432] Found device 0 with properties: 
name: Tesla K80 major: 3 minor: 7 memoryClockRate(GHz): 0.8235
pciBusID: 0000:00:04.0
totalMemory: 11.17GiB freeMemory: 11.10GiB
2018-12-20 10:59:34.016648: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1511] Adding visible gpu devices: 0
2018-12-20 10:59:34.451052: I tensorflow/core/common_runtime/gpu/gpu_device.cc:982] Device interconnect StreamExecutor with strength 1 edge matrix:
2018-12-20 10:59:34.451138: I tensorflow/core/common_runtime/gpu/gpu_device.cc:988]      0 
2018-12-20 10:59:34.451168: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1001] 0:   N 
2018-12-20 10:59:34.451592: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1115] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 10758 MB memory) -> physical GPU (device: 0, name: Tesla K80, pci bus id: 0000:00:04.0, compute capability: 3.7)
[629229239f8e:02523] *** Process received signal ***
[629229239f8e:02523] Signal: Segmentation fault (11)
[629229239f8e:02523] Signal code: Address not mapped (1)
[629229239f8e:02523] Failing at address: 0x7f602b57820d
[629229239f8e:02523] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x12890)[0x7f602e628890]
[629229239f8e:02523] [ 1] /lib/x86_64-linux-gnu/libc.so.6(getenv+0xa5)[0x7f602e267785]
[629229239f8e:02523] [ 2] /usr/lib/x86_64-linux-gnu/libtcmalloc.so.4(_ZN13TCMallocGuardD1Ev+0x34)[0x7f602ead2e44]
[629229239f8e:02523] [ 3] /lib/x86_64-linux-gnu/libc.so.6(__cxa_finalize+0xf5)[0x7f602e268615]
[629229239f8e:02523] [ 4] /usr/lib/x86_64-linux-gnu/libtcmalloc.so.4(+0x13cb3)[0x7f602ead0cb3]
[629229239f8e:02523] *** End of error message ***
env_type: atari
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
Training ppo2 on atari:MsPacmanNoFrameskip-v4 with arguments 
{'nsteps': 128, 'nminibatches': 4, 'lam': 0.95, 'gamma': 0.99, 'noptepochs': 4, 'log_interval': 1, 'ent_coef': 0.01, 'lr': <function atari.<locals>.<lambda> at 0x7f67275cad08>, 'cliprange': <function atari.<locals>.<lambda> at 0x7f6714ed80d0>, 'network': 'cnn'}
2018-12-20 11:02:00.085028: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:964] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2018-12-20 11:02:00.085530: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1432] Found device 0 with properties: 
name: Tesla K80 major: 3 minor: 7 memoryClockRate(GHz): 0.8235
pciBusID: 0000:00:04.0
totalMemory: 11.17GiB freeMemory: 11.10GiB
2018-12-20 11:02:00.085576: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1511] Adding visible gpu devices: 0
2018-12-20 11:02:00.521632: I tensorflow/core/common_runtime/gpu/gpu_device.cc:982] Device interconnect StreamExecutor with strength 1 edge matrix:
2018-12-20 11:02:00.521695: I tensorflow/core/common_runtime/gpu/gpu_device.cc:988]      0 
2018-12-20 11:02:00.521725: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1001] 0:   N 
2018-12-20 11:02:00.522124: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1115] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 10758 MB memory) -> physical GPU (device: 0, name: Tesla K80, pci bus id: 0000:00:04.0, compute capability: 3.7)
[629229239f8e:02786] *** Process received signal ***
[629229239f8e:02786] Signal: Segmentation fault (11)
[629229239f8e:02786] Signal code: Address not mapped (1)
[629229239f8e:02786] Failing at address: 0x7fd071c5520d
[629229239f8e:02786] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x12890)[0x7fd074d05890]
[629229239f8e:02786] [ 1] /lib/x86_64-linux-gnu/libc.so.6(getenv+0xa5)[0x7fd074944785]
[629229239f8e:02786] [ 2] /usr/lib/x86_64-linux-gnu/libtcmalloc.so.4(_ZN13TCMallocGuardD1Ev+0x34)[0x7fd0751afe44]
[629229239f8e:02786] [ 3] /lib/x86_64-linux-gnu/libc.so.6(__cxa_finalize+0xf5)[0x7fd074945615]
[629229239f8e:02786] [ 4] /usr/lib/x86_64-linux-gnu/libtcmalloc.so.4(+0x13cb3)[0x7fd0751adcb3]
[629229239f8e:02786] *** End of error message ***
env_type: atari
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
Training ppo2 on atari:MsPacmanNoFrameskip-v4 with arguments 
{'nsteps': 128, 'nminibatches': 4, 'lam': 0.95, 'gamma': 0.99, 'noptepochs': 4, 'log_interval': 1, 'ent_coef': 0.01, 'lr': <function atari.<locals>.<lambda> at 0x7f380dfc0d08>, 'cliprange': <function atari.<locals>.<lambda> at 0x7f37fb8ce0d0>, 'network': 'cnn'}
2018-12-20 11:04:25.462601: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:964] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2018-12-20 11:04:25.463086: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1432] Found device 0 with properties: 
name: Tesla K80 major: 3 minor: 7 memoryClockRate(GHz): 0.8235
pciBusID: 0000:00:04.0
totalMemory: 11.17GiB freeMemory: 11.10GiB
2018-12-20 11:04:25.463132: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1511] Adding visible gpu devices: 0
2018-12-20 11:04:25.904357: I tensorflow/core/common_runtime/gpu/gpu_device.cc:982] Device interconnect StreamExecutor with strength 1 edge matrix:
2018-12-20 11:04:25.904430: I tensorflow/core/common_runtime/gpu/gpu_device.cc:988]      0 
2018-12-20 11:04:25.904460: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1001] 0:   N 
2018-12-20 11:04:25.904821: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1115] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 10758 MB memory) -> physical GPU (device: 0, name: Tesla K80, pci bus id: 0000:00:04.0, compute capability: 3.7)
[629229239f8e:03046] *** Process received signal ***
[629229239f8e:03046] Signal: Segmentation fault (11)
[629229239f8e:03046] Signal code: Address not mapped (1)
[629229239f8e:03046] Failing at address: 0x7fb3cef3a20d
[629229239f8e:03046] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x12890)[0x7fb3d1fea890]
[629229239f8e:03046] [ 1] /lib/x86_64-linux-gnu/libc.so.6(getenv+0xa5)[0x7fb3d1c29785]
[629229239f8e:03046] [ 2] /usr/lib/x86_64-linux-gnu/libtcmalloc.so.4(_ZN13TCMallocGuardD1Ev+0x34)[0x7fb3d2494e44]
[629229239f8e:03046] [ 3] /lib/x86_64-linux-gnu/libc.so.6(__cxa_finalize+0xf5)[0x7fb3d1c2a615]
[629229239f8e:03046] [ 4] /usr/lib/x86_64-linux-gnu/libtcmalloc.so.4(+0x13cb3)[0x7fb3d2492cb3]
[629229239f8e:03046] *** End of error message ***
env_type: atari
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
Training ppo2 on atari:MsPacmanNoFrameskip-v4 with arguments 
{'nsteps': 128, 'nminibatches': 4, 'lam': 0.95, 'gamma': 0.99, 'noptepochs': 4, 'log_interval': 1, 'ent_coef': 0.01, 'lr': <function atari.<locals>.<lambda> at 0x7f8dc4509d08>, 'cliprange': <function atari.<locals>.<lambda> at 0x7f8db1e170d0>, 'network': 'cnn'}
2018-12-20 11:06:54.072034: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:964] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2018-12-20 11:06:54.072578: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1432] Found device 0 with properties: 
name: Tesla K80 major: 3 minor: 7 memoryClockRate(GHz): 0.8235
pciBusID: 0000:00:04.0
totalMemory: 11.17GiB freeMemory: 11.10GiB
2018-12-20 11:06:54.072661: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1511] Adding visible gpu devices: 0
2018-12-20 11:06:54.514341: I tensorflow/core/common_runtime/gpu/gpu_device.cc:982] Device interconnect StreamExecutor with strength 1 edge matrix:
2018-12-20 11:06:54.514409: I tensorflow/core/common_runtime/gpu/gpu_device.cc:988]      0 
2018-12-20 11:06:54.514438: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1001] 0:   N 
2018-12-20 11:06:54.514800: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1115] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 10758 MB memory) -> physical GPU (device: 0, name: Tesla K80, pci bus id: 0000:00:04.0, compute capability: 3.7)
[629229239f8e:03306] *** Process received signal ***
[629229239f8e:03306] Signal: Segmentation fault (11)
[629229239f8e:03306] Signal code: Address not mapped (1)
[629229239f8e:03306] Failing at address: 0x7f88d9e6620d
[629229239f8e:03306] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x12890)[0x7f88dcf16890]
[629229239f8e:03306] [ 1] /lib/x86_64-linux-gnu/libc.so.6(getenv+0xa5)[0x7f88dcb55785]
[629229239f8e:03306] [ 2] /usr/lib/x86_64-linux-gnu/libtcmalloc.so.4(_ZN13TCMallocGuardD1Ev+0x34)[0x7f88dd3c0e44]
[629229239f8e:03306] [ 3] /lib/x86_64-linux-gnu/libc.so.6(__cxa_finalize+0xf5)[0x7f88dcb56615]
[629229239f8e:03306] [ 4] /usr/lib/x86_64-linux-gnu/libtcmalloc.so.4(+0x13cb3)[0x7f88dd3becb3]
[629229239f8e:03306] *** End of error message ***
env_type: atari
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
Training ppo2 on atari:MsPacmanNoFrameskip-v4 with arguments 
{'nsteps': 128, 'nminibatches': 4, 'lam': 0.95, 'gamma': 0.99, 'noptepochs': 4, 'log_interval': 1, 'ent_coef': 0.01, 'lr': <function atari.<locals>.<lambda> at 0x7fbfe3bd6d08>, 'cliprange': <function atari.<locals>.<lambda> at 0x7fbfd14e40d0>, 'network': 'cnn'}
2018-12-20 11:09:20.595344: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:964] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2018-12-20 11:09:20.595869: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1432] Found device 0 with properties: 
name: Tesla K80 major: 3 minor: 7 memoryClockRate(GHz): 0.8235
pciBusID: 0000:00:04.0
totalMemory: 11.17GiB freeMemory: 11.10GiB
2018-12-20 11:09:20.595916: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1511] Adding visible gpu devices: 0
2018-12-20 11:09:21.041229: I tensorflow/core/common_runtime/gpu/gpu_device.cc:982] Device interconnect StreamExecutor with strength 1 edge matrix:
2018-12-20 11:09:21.041293: I tensorflow/core/common_runtime/gpu/gpu_device.cc:988]      0 
2018-12-20 11:09:21.041328: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1001] 0:   N 
2018-12-20 11:09:21.041710: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1115] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 10758 MB memory) -> physical GPU (device: 0, name: Tesla K80, pci bus id: 0000:00:04.0, compute capability: 3.7)
[629229239f8e:03569] *** Process received signal ***
[629229239f8e:03569] Signal: Segmentation fault (11)
[629229239f8e:03569] Signal code: Address not mapped (1)
[629229239f8e:03569] Failing at address: 0x7f029a93820d
[629229239f8e:03569] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x12890)[0x7f029d9e8890]
[629229239f8e:03569] [ 1] /lib/x86_64-linux-gnu/libc.so.6(getenv+0xa5)[0x7f029d627785]
[629229239f8e:03569] [ 2] /usr/lib/x86_64-linux-gnu/libtcmalloc.so.4(_ZN13TCMallocGuardD1Ev+0x34)[0x7f029de92e44]
[629229239f8e:03569] [ 3] /lib/x86_64-linux-gnu/libc.so.6(__cxa_finalize+0xf5)[0x7f029d628615]
[629229239f8e:03569] [ 4] /usr/lib/x86_64-linux-gnu/libtcmalloc.so.4(+0x13cb3)[0x7f029de90cb3]
[629229239f8e:03569] *** End of error message ***
env_type: atari
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
/usr/local/lib/python3.6/dist-packages/gym/envs/registration.py:14: PkgResourcesDeprecationWarning: Parameters to load are deprecated.  Call .resolve and .require separately.
  result = entry_point.load(False)
Training ppo2 on atari:MsPacmanNoFrameskip-v4 with arguments 
{'nsteps': 128, 'nminibatches': 4, 'lam': 0.95, 'gamma': 0.99, 'noptepochs': 4, 'log_interval': 1, 'ent_coef': 0.01, 'lr': <function atari.<locals>.<lambda> at 0x7f5988a16d08>, 'cliprange': <function atari.<locals>.<lambda> at 0x7f59763240d0>, 'network': 'cnn'}
2018-12-20 11:11:44.767812: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:964] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2018-12-20 11:11:44.768275: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1432] Found device 0 with properties: 
name: Tesla K80 major: 3 minor: 7 memoryClockRate(GHz): 0.8235
pciBusID: 0000:00:04.0
totalMemory: 11.17GiB freeMemory: 11.10GiB
2018-12-20 11:11:44.768332: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1511] Adding visible gpu devices: 0
2018-12-20 11:11:45.211531: I tensorflow/core/common_runtime/gpu/gpu_device.cc:982] Device interconnect StreamExecutor with strength 1 edge matrix:
2018-12-20 11:11:45.211648: I tensorflow/core/common_runtime/gpu/gpu_device.cc:988]      0 
2018-12-20 11:11:45.211686: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1001] 0:   N 
2018-12-20 11:11:45.212043: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1115] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 10758 MB memory) -> physical GPU (device: 0, name: Tesla K80, pci bus id: 0000:00:04.0, compute capability: 3.7)
[629229239f8e:03829] *** Process received signal ***
[629229239f8e:03829] Signal: Segmentation fault (11)
[629229239f8e:03829] Signal code: Address not mapped (1)
[629229239f8e:03829] Failing at address: 0x7f34912c320d
[629229239f8e:03829] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x12890)[0x7f3494373890]
[629229239f8e:03829] [ 1] /lib/x86_64-linux-gnu/libc.so.6(getenv+0xa5)[0x7f3493fb2785]
[629229239f8e:03829] [ 2] /usr/lib/x86_64-linux-gnu/libtcmalloc.so.4(_ZN13TCMallocGuardD1Ev+0x34)[0x7f349481de44]
[629229239f8e:03829] [ 3] /lib/x86_64-linux-gnu/libc.so.6(__cxa_finalize+0xf5)[0x7f3493fb3615]
[629229239f8e:03829] [ 4] /usr/lib/x86_64-linux-gnu/libtcmalloc.so.4(+0x13cb3)[0x7f349481bcb3]
[629229239f8e:03829] *** End of error message ***

















In [8]:



    


!ls -l ~/logs/atari-ppo-hp



    


















    






total 24
drwxr-xr-x 2 root root 4096 Dec 20 10:59 b128-0
drwxr-xr-x 2 root root 4096 Dec 20 11:01 b128-1
drwxr-xr-x 2 root root 4096 Dec 20 11:04 b128-2
drwxr-xr-x 2 root root 4096 Dec 20 11:06 b128-3
drwxr-xr-x 2 root root 4096 Dec 20 11:09 b128-4
drwxr-xr-x 2 root root 4096 Dec 20 11:11 b128-5

















In [9]:



    


results = pu.load_results('~/logs/atari-ppo-hp')



    


















    






/usr/local/lib/python3.6/dist-packages/baselines/bench/monitor.py:164: UserWarning: Pandas doesn't allow columns to be created via a new attribute name - see https://pandas.pydata.org/pandas-docs/stable/indexing.html#attribute-access
  df.headers = headers # HACK to preserve backwards compatibility

















In [15]:



    


pu.plot_results(results)
# pu.plot_results(results, average_group=True)



    


















    
Out[15]:








(<matplotlib.figure.Figure at 0x7f22099e0630>,
 array([[<matplotlib.axes._subplots.AxesSubplot object at 0x7f2209a467b8>]],
       dtype=object))










    
























In [0]:

Content source: DJCordhose/ai

Similar notebooks:

notebook.community | gallery | about