TypeError: Input must be namedtuple or namedarraytuple instance or class, got <class 'numpy.ndarray'>

[robjlyons]

I'm trying to run this in Windows.

I have tried several versions of Pytorch but with no luck.

Do I have a package version wrong?

Error and pip list pasted below.

Thanks,

Error output

Unable to import tensorboard SummaryWriter, proceeding without.
wandb: Currently logged in as: anony-mouse-162437 (use wandb login --relogin to force relogin)
Unable to import tensorboard SummaryWriter, proceeding without.
wandb: wandb version 0.12.14 is available! To upgrade, please run:
wandb: $ pip install wandb --upgrade
wandb: Tracking run with wandb version 0.12.13
wandb: Run data is saved locally in wandb\run-20220410_121943-2wa5jydb
wandb: Run wandb offline to turn off syncing.
wandb: Syncing run playful-spaceship-10
wandb: View project at https://wandb.ai/anony-mouse-162437/SPR?apiKey=8f90936030ab76c4ea4b4a01c63d86a1ba05bec7
wandb: View run at https://wandb.ai/anony-mouse-162437/SPR/runs/2wa5jydb?apiKey=8f90936030ab76c4ea4b4a01c63d86a1ba05bec7
wandb: WARNING Do NOT share these links with anyone. They can be used to claim your runs.
logger_context received log_dir outside of x:\ai plays pokemon\ai\spr\src\rlpyt\data: prepending by x:\ai plays pokemon\ai\spr\src\rlpyt\data/local//
2022-04-10 12:19:45.841568 | dqn_breakout_0 Runner master CPU affinity: [0, 1, 2, 3].
2022-04-10 12:19:45.842067 | dqn_breakout_0 Runner master Torch threads: 4.
←[32musing seed 0←[0m
Spatial latent size is torch.Size([64, 7, 7])
Initialized model with 3775151 parameters
Spatial latent size is torch.Size([64, 7, 7])
Initialized model with 3775151 parameters
2022-04-10 12:21:31.139408 | dqn_breakout_0 Sampler decorrelating envs, max steps: 0
2022-04-10 12:21:31.141906 | dqn_breakout_0 Agent at itr 0, sample eps 1.0 (min itr: 50, max_itr: 1000)
2022-04-10 12:21:31.142393 | dqn_breakout_0 Serial Sampler initialized.
2022-04-10 12:21:31.142393 | dqn_breakout_0 Running 100000 iterations of minibatch RL.
2022-04-10 12:21:33.068301 | dqn_breakout_0 Initialized agent model on device: cuda:0.
2022-04-10 12:21:33.087765 | dqn_breakout_0 From sampler batch size 1, training batch size 32, and replay ratio 64, computed 2 updates per iteration.
2022-04-10 12:21:33.088264 | dqn_breakout_0 Agent setting min/max epsilon itrs: 2000, 2001
2022-04-10 12:21:33.113717 | dqn_breakout_0 Frame-based buffer using 4-frame sequences.
Traceback (most recent call last):
File "x:\ai plays pokemon\ai\spr\src\rlpyt\rlpyt\utils\buffer.py", line 33, in buffer_from_example
buffer_type = namedarraytuple_like(example)
File "x:\ai plays pokemon\ai\spr\src\rlpyt\rlpyt\utils\collections.py", line 202, in namedarraytuple_like
raise TypeError("Input must be namedtuple or namedarraytuple instance"
TypeError: Input must be namedtuple or namedarraytuple instance or class, got <class 'numpy.ndarray'>.

Pip list

Package Version Location

---

ale-py 0.7.4
atari-py 0.2.9
certifi 2021.10.8
charset-normalizer 2.0.12
click 8.1.2
cloudpickle 1.6.0
colorama 0.4.4
cycler 0.11.0
docker-pycreds 0.4.0
fonttools 4.32.0
future 0.18.2
gitdb 4.0.9
GitPython 3.1.27
gym 0.17.3
gym-notices 0.0.6
idna 3.3
importlib-metadata 4.11.3
importlib-resources 5.6.0
joblib 1.1.0
kiwisolver 1.4.2
kornia 0.6.4
matplotlib 3.2.2
numpy 1.21.5
opencv-python 4.5.5.64
packaging 21.3
pathtools 0.1.2
Pillow 9.1.0
pip 21.2.4
promise 2.3
protobuf 3.20.0
psutil 5.9.0
pyglet 1.5.0
pyparsing 3.0.7
PyPrind 2.11.3
python-dateutil 2.8.2
PyYAML 6.0
recordclass 0.17.2
requests 2.27.1
rlpyt 0.1.1.dev0 x:\ai plays pokemon\ai\spr\src\rlpyt
scikit-learn 1.0.2
scipy 1.8.0
sentry-sdk 1.5.8
setproctitle 1.2.2
setuptools 58.0.4
shortuuid 1.0.8
six 1.16.0
smmap 5.0.0
threadpoolctl 3.1.0
torch 1.9.1+cu111
torchaudio 0.9.1
torchvision 0.10.1+cu111
typing_extensions 4.1.1
urllib3 1.26.9
wandb 0.12.13
wheel 0.37.1
wincertstore 0.2
zipp 3.8.0

[MaxASchwarzer]

What's 'example' in this stacktrace -- is it standard, or have you changed it to be from pokemon? If the latter, I think you need to process it so that it's a namedtuple of numpy arrays before you pass it in.

[robjlyons]

Hi, thanks for the reply.

I have not changed anything within buffer.py so example should still be example.

Thanks

[MSKBOX]

same question，have you solved it？