How to run mmaction2 on Mac M1 chip?

[datascientistlin]

Hi Team,

May I know how I could set the mmaction2 config such that it can be run on a mac M1 chip? I had tried to instantiate my model with mps device like below
model = init_recognizer(config, checkpoint, device='mps')
However, when I do an inference with this model (inference_recognizer(model, video)), it returned me the following error:

RuntimeError: Input type (torch.FloatTensor) and weight type (MPSFloatType) should be the same or input should be a MKLDNN tensor and weight is a dense tensor

My packages installed:

absl-py=1.4.0=pypi_0
addict=2.4.0=pypi_0
blas=1.0=mkl
brotlipy=0.7.0=py38h9ed2024_1003
bzip2=1.0.8=h1de35cc_0
ca-certificates=2022.12.7=h033912b_0
cachetools=5.3.0=pypi_0
certifi=2022.12.7=pyhd8ed1ab_0
cffi=1.15.1=py38h6c40b1e_3
charset-normalizer=2.0.4=pyhd3eb1b0_0
click=8.1.3=pypi_0
colorama=0.4.6=pypi_0
contourpy=1.0.7=pypi_0
cryptography=38.0.4=py38hf6deb26_0
cudatoolkit=9.0=h41a26b3_0
cycler=0.11.0=pypi_0
cython=0.29.33=pypi_0
decorator=4.4.2=pypi_0
decord=0.6.0=pypi_0
einops=0.6.0=pypi_0
ffmpeg=4.3=h0a44026_0
flit-core=3.6.0=pyhd3eb1b0_0
fonttools=4.38.0=pypi_0
freetype=2.12.1=hd8bbffd_0
fvcore=0.1.5.post20221221=pypi_0
gettext=0.21.0=h7535e17_0
giflib=5.2.1=h6c40b1e_1
gmp=6.2.1=he9d5cce_3
gnutls=3.6.15=hed9c0bf_0
google-auth=2.16.1=pypi_0
google-auth-oauthlib=0.4.6=pypi_0
grpcio=1.51.1=pypi_0
icu=58.2=h0a44026_3
idna=3.4=py38hecd8cb5_0
imageio=2.25.1=pypi_0
imageio-ffmpeg=0.4.8=pypi_0
importlib-metadata=6.0.0=pypi_0
importlib-resources=5.10.2=pypi_0
intel-openmp=2021.4.0=hecd8cb5_3538
iopath=0.1.10=pypi_0
joblib=1.2.0=pypi_0
jpeg=9e=hca72f7f_0
kiwisolver=1.4.4=pypi_0
lame=3.100=h1de35cc_0
lcms2=2.12=hf1fd2bf_0
lerc=3.0=he9d5cce_0
libcxx=14.0.6=h9765a3e_0
libdeflate=1.8=h9ed2024_5
libffi=3.4.2=hecd8cb5_6
libiconv=1.16=hca72f7f_2
libidn2=2.3.2=h9ed2024_0
libpng=1.6.37=ha441bb4_0
libtasn1=4.16.0=h9ed2024_0
libtiff=4.5.0=hcec6c5f_1
libunistring=0.9.10=h9ed2024_0
libwebp=1.2.4=h56c3ce4_0
libwebp-base=1.2.4=hca72f7f_0
libxml2=2.9.14=hbf8cd5e_0
llvm-openmp=14.0.6=h0dcd299_0
lz4-c=1.9.4=hcec6c5f_0
markdown=3.4.1=pypi_0
markdown-it-py=2.1.0=pypi_0
markupsafe=2.1.2=pypi_0
matplotlib=3.7.0=pypi_0
mdurl=0.1.2=pypi_0
mkl=2021.4.0=hecd8cb5_637
mkl-service=2.4.0=py38h9ed2024_0
mkl_fft=1.3.1=py38h4ab4a9b_0
mkl_random=1.2.2=py38hb2f4e1b_0
mmaction2=0.24.1=dev_0
mmcv-full=1.7.1=pypi_0
mmdet=2.28.1=dev_0
model-index=0.1.11=pypi_0
moviepy=1.0.3=pypi_0
ncurses=6.4=hcec6c5f_0
nettle=3.7.3=h230ac6f_1
numpy=1.23.5=py38he696674_0
numpy-base=1.23.5=py38h9cd3388_0
oauthlib=3.2.2=pypi_0
opencv-contrib-python=4.7.0.68=pypi_0
opencv-python=4.7.0.68=pypi_0
openh264=2.1.1=h8346a28_0
openmim=0.3.6=pypi_0
openssl=1.1.1t=hfd90126_0
ordered-set=4.1.0=pypi_0
packaging=23.0=pypi_0
pandas=1.5.3=pypi_0
pillow=9.3.0=py38hcec6c5f_2
pip=22.3.1=py38hecd8cb5_0
portalocker=2.7.0=pypi_0
proglog=0.1.10=pypi_0
protobuf=4.22.0=pypi_0
psutil=5.9.4=pypi_0
pyasn1=0.4.8=pypi_0
pyasn1-modules=0.2.8=pypi_0
pycocotools=2.0.6=pypi_0
pycparser=2.21=pyhd3eb1b0_0
pygments=2.14.0=pypi_0
pyopenssl=22.0.0=pyhd3eb1b0_0
pyparsing=3.0.9=pypi_0
pysocks=1.7.1=py38_1
python=3.8.16=h218abb5_2
python-dateutil=2.8.2=pypi_0
pytorch=1.13.1=py3.8_0
pytz=2022.7.1=pypi_0
pyyaml=6.0=pypi_0
readline=8.2=hca72f7f_0
requests=2.28.1=py38hecd8cb5_0
requests-oauthlib=1.3.1=pypi_0
rich=13.3.1=pypi_0
rsa=4.9=pypi_0
scipy=1.10.0=pypi_0
setuptools=65.6.3=py38hecd8cb5_0
six=1.16.0=pyhd3eb1b0_1
sqlite=3.40.1=h880c91c_0
tabulate=0.9.0=pypi_0
tensorboard=2.12.0=pypi_0
tensorboard-data-server=0.7.0=pypi_0
tensorboard-plugin-wit=1.8.1=pypi_0
termcolor=2.2.0=pypi_0
terminaltables=3.1.10=pypi_0
tk=8.6.12=h5d9f67b_0
torchaudio=0.13.1=py38_cpu
torchvision=0.14.1=py38_cpu
tqdm=4.64.1=pypi_0
typing_extensions=4.4.0=py38hecd8cb5_0
urllib3=1.26.14=py38hecd8cb5_0
webcolors=1.12=pypi_0
werkzeug=2.2.3=pypi_0
wheel=0.37.1=pyhd3eb1b0_0
xz=5.2.10=h6c40b1e_1
yacs=0.1.8=pypi_0
yapf=0.32.0=pypi_0
zipp=3.13.0=pypi_0
zlib=1.2.13=h4dc903c_0
zstd=1.5.2=hcb37349_0

Thanks in advance for any help.