Adding support for Pytorch 2.x and Pytorch-Lightning 2.0.7

[YankoFelipe]

Hi!

Thanks for the cool pipeline. I was facing some crashes when I tried to run the code with PyTorch 2.x and PyTorch Lightning 2.0.7 so I fixed them. I used the nightly of 2.1 but it should be the same behaviour for the 2.0 release.
I was able to run the complete flow of Colab notebook and passed all the tests. The environment used was:

(deepethogram_39n) ➜  deepethogram pip freeze
aiohttp==3.8.5
aiosignal==1.3.1
antlr4-python3-runtime==4.9.3
async-timeout==4.0.3
attrs==23.1.0
brotlipy==0.7.0
certifi==2023.7.22
cffi @ file:///croot/cffi_1670423208954/work
chardet==3.0.4
charset-normalizer @ file:///tmp/build/80754af9/charset-normalizer_1630003229654/work
contourpy==1.1.0
cryptography @ file:///croot/cryptography_1677533068310/work
cycler==0.11.0
-e git+https://github.com/jbohnslav/deepethogram.git@a2565bb6a19f3f71ae4231dfd947a98ab9b3ae0f#egg=deepethogram
exceptiongroup==1.1.3
filelock @ file:///croot/filelock_1672387128942/work
fonttools==4.42.1
frozenlist==1.4.0
fsspec==2023.6.0
gmpy2 @ file:///tmp/build/80754af9/gmpy2_1645438755360/work
h5py==3.9.0
idna @ file:///croot/idna_1666125576474/work
importlib-resources==6.0.1
iniconfig==2.0.0
Jinja2 @ file:///croot/jinja2_1666908132255/work
joblib==1.3.2
kiwisolver==1.4.5
kornia==0.7.0
lightning-utilities==0.9.0
MarkupSafe @ file:///opt/conda/conda-bld/markupsafe_1654597864307/work
matplotlib==3.7.2
mkl-fft==1.3.6
mkl-random @ file:///work/mkl/mkl_random_1682950433854/work
mkl-service==2.4.0
mpmath @ file:///croot/mpmath_1690848262763/work
multidict==6.0.4
networkx @ file:///croot/networkx_1690561992265/work
numpy==1.22.4
omegaconf==2.3.0
opencv-python-headless==4.8.0.76
opencv-transforms==0.0.6
packaging==23.1
pandas==1.3.5
Pillow==9.4.0
pluggy==1.3.0
protobuf==4.24.2
pycparser @ file:///tmp/build/80754af9/pycparser_1636541352034/work
pyOpenSSL @ file:///croot/pyopenssl_1690223430423/work
pyparsing==3.0.9
PySide2==5.13.2
PySocks @ file:///tmp/build/80754af9/pysocks_1605305812635/work
pytest==7.4.1
python-dateutil @ file:///home/conda/feedstock_root/build_artifacts/python-dateutil_1626286286081/work
pytorch-lightning==2.0.7
pytz @ file:///home/conda/feedstock_root/build_artifacts/pytz_1680088766131/work
PyYAML @ file:///croot/pyyaml_1670514731622/work
requests @ file:///croot/requests_1690400202158/work
scikit-learn==1.0.2
scipy==1.7.3
shiboken2==5.13.2
six @ file:///home/conda/feedstock_root/build_artifacts/six_1620240208055/work
sympy @ file:///croot/sympy_1668202399572/work
tensorboardX==2.6.2.2
threadpoolctl==3.2.0
tomli==2.0.1
torch==2.1.0.dev20230831
torchaudio==2.2.0.dev20230901
torchmetrics==1.1.0
torchvision==0.16.0.dev20230831
tqdm==4.66.1
triton==2.1.0
typing_extensions @ file:///croot/typing_extensions_1690297465030/work
urllib3 @ file:///croot/urllib3_1686163155763/work
vidio==0.0.4
yarl==1.9.2
zipp==3.16.2

The Python version was 3.9.7 and here are the result of the tests:

(deepethogram_39n) ➜  deepethogram git:(pytorch2.0) ✗ pytest tests
============================================================================= test session starts =============================================================================
platform linux -- Python 3.9.7, pytest-7.4.1, pluggy-1.3.0
rootdir: /data/yanar/deepethogram
collected 18 items

tests/test_data.py .                                                                                                                                                    [  5%]
tests/test_flow_generator.py .                                                                                                                                          [ 11%]
tests/test_gui.py .                                                                                                                                                     [ 16%]
tests/test_models.py .                                                                                                                                                  [ 22%]
tests/test_projects.py ........                                                                                                                                         [ 66%]
tests/test_z_score.py .                                                                                                                                                 [ 72%]
tests/test_zz_commandline.py .....                                                                                                                                      [100%]

======================================================================= 18 passed in 412.60s (0:06:52) ========================================================================

I think it'll be complicated to merge this considering that there're no releases of DeepEthogram and that could mess the installation in environments with 1.x but it can be helpful for those working with PyTorch 2.x.