TypeError when running examples

[VerwimpEli]

Hi,

Upon running the examples I run into a runtime TypeError:

in dataset_utils.py:
File "replay.py", line 31, in <module> from avalanche.benchmarks import nc_scenario
...
class SubsetWithTargets(Generic[T_co, TTargetType], Subset[T_co]): TypeError: 'type' object is not subscriptable

It's the generic Subset that's causing problems, I believe. I'm running in a clean conda environment, python 3.8.

Thanks!

[vlomonaco]

Thanks @VerwimpEli for reporting this. @lrzpellegrini will follow-up on this soon! In the meanwhile, can you provide a minimal script to reproduce the issue?

[GengDavid]

Hi, I also met this problem with python 3.7.7, pytorch version 1.5.1
Would there be any solution rather than updating the python version?

[vlomonaco]

mmh maybe you can try update the "typing_extensions" package? what's your version?

[GengDavid]

I have tried typing_extensions==3.7.4.3 / 3.10.0.0 / 3.10.0.1 / 3.10.0.2. The problem remains. 😢

[GengDavid]

Updating Python and PyTorch versions do solve the problem.

[SinHanYang]

Having the same issue when using python 3.7, torch 1.6.0, but fixed when using torch 1.12.0.