Issues with Python 3.10 compatibility

[lucachitayat]

@gguzzy

I first downgraded decorator to decorator<5.0,>=4.0.2 to resolve issues with moviepy and another package I can't recall.

I run into this error during a sample training followed from this article: https://www.datature.io/blog/how-to-perform-action-recogition-on-keypoints-with-stgcnn

I tried downgrading numpy, but a new issue arises in this scenario. I am wondering if there is any additional changes needed for the 3.10 compatibility. Currently trying pyskl.yaml instead to see if I can config this way, which so far seems to be running as intended. I believe there is a few package incompatibilities remaining.

See detailed information below.

##Thank you!

Environment

• Python: 3.10.9
• Conda environment: pyskl_310
• Relevant packages:
  • torchvision: 0.13.1
  • mmcv: (version not specified in traceback)
  • yapf: 0.32.0

Steps to Reproduce

1. Activate the conda environment: conda activate pyskl_310
2. Run the training script: python tools/train.py

Expected Behavior

The training script should run without errors.

Actual Behavior

The script fails with a TypeError related to the FormatCode function in mmcv/utils/config.py.

Additional Context

The error seems to be caused by an unexpected keyword argument verify in the FormatCode function call.

Suggested Fix

Please provide guidance on how to resolve this issue. Should I update or modify any specific package or function?

Thank you!

(pyskl_310) lchit@LUCA-PC:~/action-recog/pyskl$ bash tools/dist_train.sh configs/j.py 1 --validate --test-last --test-best

• CONFIG=configs/j.py

• GPUS=1
  ++ dirname tools/dist_train.sh

• MKL_SERVICE_FORCE_INTEL=1
  ++ dirname tools/dist_train.sh

• PYTHONPATH=tools/..:

• python -m torch.distributed.launch --nproc_per_node=1 --master_port=21820 tools/train.py configs/j.py --launcher pytorch --validate --test-last --test-best
  /home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/site-packages/torch/distributed/launch.py:178: FutureWarning: The module torch.distributed.launch is deprecated
  and will be removed in future. Use torchrun.
  Note that --use_env is set by default in torchrun.
  If your script expects --local_rank argument to be set, please
  change it to read from os.environ['LOCAL_RANK'] instead. See
  https://pytorch.org/docs/stable/distributed.html#launch-utility for
  further instructions

  warnings.warn(
  Traceback (most recent call last):
  File "/home/lchit/action-recog/pyskl/tools/train.py", line 16, in
  from pyskl.apis import init_random_seed, train_model
  File "/home/lchit/action-recog/pyskl/pyskl/apis/init.py", line 4, in
  from .inference import inference_recognizer, init_recognizer
  File "/home/lchit/action-recog/pyskl/pyskl/apis/inference.py", line 14, in
  from pyskl.datasets.pipelines import Compose
  File "/home/lchit/action-recog/pyskl/pyskl/datasets/init.py", line 2, in
  from .base import BaseDataset
  File "/home/lchit/action-recog/pyskl/pyskl/datasets/base.py", line 16, in
  from .pipelines import Compose
  File "/home/lchit/action-recog/pyskl/pyskl/datasets/pipelines/init.py", line 7, in
  from .multi_modality import * # noqa: F401, F403
  File "/home/lchit/action-recog/pyskl/pyskl/datasets/pipelines/multi_modality.py", line 6, in
  from .pose_related import PoseDecode
  File "/home/lchit/action-recog/pyskl/pyskl/datasets/pipelines/pose_related.py", line 3, in
  from scipy.stats import mode as get_mode
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/site-packages/scipy/stats/init.py", line 624, in
  from ._stats_py import *
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/site-packages/scipy/stats/_stats_py.py", line 39, in
  from scipy.spatial import distance_matrix
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/site-packages/scipy/spatial/init.py", line 116, in
  from ._geometric_slerp import geometric_slerp
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/site-packages/scipy/spatial/_geometric_slerp.py", line 7, in
  from scipy.spatial.distance import euclidean
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/site-packages/scipy/spatial/distance.py", line 121, in
  from ..special import rel_entr
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/site-packages/scipy/special/init.py", line 826, in
  from . import _basic
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/site-packages/scipy/special/_basic.py", line 22, in
  from ._multiufuncs import (assoc_legendre_p_all,
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/site-packages/scipy/special/_multiufuncs.py", line 142, in
  sph_legendre_p = MultiUFunc(
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/site-packages/scipy/special/_multiufuncs.py", line 41, in init
  raise ValueError("All ufuncs must have type numpy.ufunc."
  ValueError: All ufuncs must have type numpy.ufunc. Received (<ufunc 'sph_legendre_p'>, <ufunc 'sph_legendre_p'>, <ufunc 'sph_legendre_p'>)
  ERROR:torch.distributed.elastic.multiprocessing.api:failed (exitcode: 1) local_rank: 0 (pid: 37345) of binary: /home/lchit/miniconda3/envs/pyskl_310/bin/python
  Traceback (most recent call last):
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/runpy.py", line 196, in _run_module_as_main
  return _run_code(code, main_globals, None,
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/runpy.py", line 86, in _run_code
  exec(code, run_globals)
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/site-packages/torch/distributed/launch.py", line 193, in
  main()
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/site-packages/torch/distributed/launch.py", line 189, in main
  launch(args)
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/site-packages/torch/distributed/launch.py", line 174, in launch
  run(args)
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/site-packages/torch/distributed/run.py", line 752, in run
  elastic_launch(
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 131, in call
  return launch_agent(self._config, self._entrypoint, list(args))
  File "/home/lchit/miniconda3/envs/pyskl_310/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 245, in launch_agent
  raise ChildFailedError(
  torch.distributed.elastic.multiprocessing.errors.ChildFailedError:
  ============================================================
  tools/train.py FAILED

---

Failures:
<NO_OTHER_FAILURES>

[gguzzy]

My suggestion, if you using a conda environment, is to install the base environment as described, and insert the new required packages in this form 'pip install / conda install', using --no-deps, in such a way you don't create conflicts on packages.

Try to do this, install the remaining and if other errors pop up, please give me the updated error, or not working compatible library.

Thanks.

[lucachitayat]

@gguzzy Will do. If it does work, it should probably be in the install instructions or the correctly installed versions updated in the yaml. I'm sure it will save others a headache. Thanks for the response!