search

tristandeleu / pytorch-meta

A collection of extensions and data-loaders for few-shot learning & meta-learning in PyTorch
https://tristandeleu.github.io/pytorch-meta/
MIT License
1.98k stars 256 forks source link

help installing torchmeta (for ppc64le IBM architecture) #95

Closed brando90 closed 4 years ago

brando90 commented 4 years ago

I got the following error:

ERROR: Could not find a version that satisfies the requirement torch<1.7.0,>=1.4.0 (from torchmeta) (from versions: 0.1.2, 0.1.2.post1, 0.1.2.post2)
ERROR: No matching distribution found for torch<1.7.0,>=1.4.0 (from torchmeta)

unfortunately, the cluster I use has only 1.3.1 and I believe it's due to hardware stuff. They don't use x86 basically.

what are solutions for me?

SO: https://stackoverflow.com/questions/64049603/how-does-one-install-torchmeta-for-a-ppc64le-architecture-in-pytorch SO: https://stackoverflow.com/questions/64033543/force-installing-torchvision-0-4-2-when-i-am-forced-to-use-pytorch-1-3-1-due-to ibm issue: https://github.com/IBM/powerai/issues/268

brando90 commented 4 years ago

when through the whole history of releases none worked: https://pypi.org/project/torchmeta/#history

with command `pip install torchmeta=x.x.x'

brando90 commented 4 years ago

My versions:

torchvision-base          0.4.2 
pytorch                   1.3.1            21932.g39c5d28    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
pytorch-base              1.3.1           gpu_py37_21932.g39c5d28    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
brando90 commented 4 years ago

any version that works with the above should be fine

tristandeleu commented 4 years ago

when through the whole history of releases none worked: https://pypi.org/project/torchmeta/#history

with command `pip install torchmeta=x.x.x'

Isn't torchmeta==1.3.1 compatible with PyTorch 1.3 & Torchvision 0.4? I haven't tried it yet, but the requirements in setup.py seem to match https://github.com/tristandeleu/pytorch-meta/blob/5bc8638f1f640724a6f03aafd3b7606595d308d8/setup.py#L33-L34

brando90 commented 4 years ago

https://pypi.org/project/torchmeta/#history

ok seems you suggest pip install torchmeta==1.3.1 should work. I will try it again in a bit and report what happened.

brando90 commented 4 years ago

when through the whole history of releases none worked: https://pypi.org/project/torchmeta/#history with command `pip install torchmeta=x.x.x'

Isn't torchmeta==1.3.1 compatible with PyTorch 1.3 & Torchvision 0.4? I haven't tried it yet, but the requirements in setup.py seem to match https://github.com/tristandeleu/pytorch-meta/blob/5bc8638f1f640724a6f03aafd3b7606595d308d8/setup.py#L33-L34

weird, I get this error:

$ pip install torchmeta==1.3.1
Defaulting to user installation because normal site-packages is not writeable
Collecting torchmeta==1.3.1
  Using cached torchmeta-1.3.1-py3-none-any.whl (144 kB)
Requirement already satisfied: tqdm>=4.0.0 in /opt/apps/anaconda3/lib/python3.8/site-packages (from torchmeta==1.3.1) (4.47.0)
Collecting h5py~=2.9.0
  Using cached h5py-2.9.0.tar.gz (287 kB)
ERROR: Could not find a version that satisfies the requirement torchvision<0.6.0,>=0.4.0 (from torchmeta==1.3.1) (from versions: 0.1.6, 0.1.7, 0.1.8, 0.1.9, 0.2.0, 0.2.1, 0.2.2, 0.2.2.post2, 0.2.2.post3)
ERROR: No matching distribution found for torchvision<0.6.0,>=0.4.0 (from torchmeta==1.3.1)

It says that torchmeta wants a version of torchvision less than 0.6.0 but more than 0.4.0.

After looking more into it seems I have (under conda list, nothing relevant under pip list)

torchvision 0.2.2 pypi_0 pypi

will see if I can update it.

brando90 commented 4 years ago

asking on forum how to force install the version of torchvision I need here: https://discuss.pytorch.org/t/force-installing-torchvision/97279/2

brando90 commented 4 years ago

@tristandeleu what do you think of installing torchmeta and ignoring the dependencies?

e.g.

pip install --no-deps torchmeta
brando90 commented 4 years ago

pip install --no-deps torchmeta

ok that didn't work without needed to modify torchmeta (to be expected): 

ModuleNotFoundError: No module named 'ordered_set'
brando90 commented 4 years ago

perhaps:

conda install torchvision==0.4.2 -c pytorch

or 

conda install torchvision -c pytorch

might work...it's running right now.

brando90 commented 4 years ago

about to try:

pip install torchmeta==1.3.1

seems torchvision installed...

brando90 commented 4 years ago

but that gave an error: 

pip install torchmeta==1.3.1
Collecting torchmeta==1.3.1
  Using cached https://files.pythonhosted.org/packages/a2/78/2260a3f2a4c4653ba4de5485b32995a5ee73be93b27c29f8dd235459127e/torchmeta-1.3.1-py3-none-any.whl
ERROR: Could not find a version that satisfies the requirement torchvision<0.6.0,>=0.4.0 (from torchmeta==1.3.1) (from versions: 0.1.6, 0.1.7, 0.1.8, 0.1.9, 0.2.0, 0.2.1, 0.2.2, 0.2.2.post2, 0.2.2.post3)
ERROR: No matching distribution found for torchvision<0.6.0,>=0.4.0 (from torchmeta==1.3.1)
brando90 commented 4 years ago

@tristandeleu what about this instead.

Is there a version of torchmeta that works with: 1) pytorch 1.3.1 + torchvision 0.2.2 2) pytorch 1.1.1 + torchvision 0.2.2

I am still struggling to get torchvision 0.4.2 so meanwhile those might work.

tristandeleu commented 4 years ago

Unfortunately the earliest version of Torchvision required seems to be 0.4 (even for the earliest version of Torchmeta). If you install the dependencies manually (in particular h5py & tqdm), then you might be able to install Torchmeta with pip install --no-deps torchmeta==1.3.1 (what you tried earlier, but using a specific version). The versions of Torchvision will not match the requirements from Torchmeta, so I can't guarantee that the library will work as expected, but this is probably your best option.

brando90 commented 4 years ago

Unfortunately the earliest version of Torchvision required seems to be 0.4 (even for the earliest version of Torchmeta). If you install the dependencies manually (in particular h5py & tqdm), then you might be able to install Torchmeta with pip install --no-deps torchmeta==1.3.1 (what you tried earlier, but using a specific version). The versions of Torchvision will not match the requirements from Torchmeta, so I can't guarantee that the library will work as expected, but this is probably your best option.

I did try that. Torchmeta started complaining about some orderdict not being there and more. I will try that (more deeply) path later (fork torchmeta and fix it) after I get help from the IBM people/sys admins to install the required version of pytorch + torchvision on their fancy ppc64le hardware architecture.

Thanks for the help. I will post the solution here once I solve it.

brando90 commented 4 years ago

For all details check (https://github.com/IBM/powerai/issues/268).

Make sure you have the right conda channel prepended:

conda config --prepend channels https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda/#/

then install the powerai wmlce that you want e.g. 1.7.0 (most recent as of this writing):

conda create -n my_new_env python=3.7 powerai=1.7.0
conda activate my_new_env

though I am still unable to install torchmeta. Now there is some h5py installation failure on my ppc64le architecture:

...
    building 'h5py.defs' extension
    creating build/temp.linux-ppc64le-3.7
    creating build/temp.linux-ppc64le-3.7/tmp
    creating build/temp.linux-ppc64le-3.7/tmp/pip-install-xw5w7nuz
    creating build/temp.linux-ppc64le-3.7/tmp/pip-install-xw5w7nuz/h5py
    creating build/temp.linux-ppc64le-3.7/tmp/pip-install-xw5w7nuz/h5py/h5py
    gcc -pthread -B /home/miranda9/.conda/envs/my_new_env/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -DH5_USE_16_API -I./h5py -I/tmp/pip-install-xw5w7nuz/h5py/lzf -I/opt/local/include -I/usr/local/include -I/home/miranda9/.conda/envs/my_new_env/lib/python3.7/site-packages/numpy/core/include -I/home/miranda9/.conda/envs/my_new_env/include/python3.7m -c /tmp/pip-install-xw5w7nuz/h5py/h5py/defs.c -o build/temp.linux-ppc64le-3.7/tmp/pip-install-xw5w7nuz/h5py/h5py/defs.o
    In file included from /home/miranda9/.conda/envs/my_new_env/lib/python3.7/site-packages/numpy/core/include/numpy/ndarraytypes.h:1830:0,
                     from /home/miranda9/.conda/envs/my_new_env/lib/python3.7/site-packages/numpy/core/include/numpy/ndarrayobject.h:12,
                     from /home/miranda9/.conda/envs/my_new_env/lib/python3.7/site-packages/numpy/core/include/numpy/arrayobject.h:4,
                     from /tmp/pip-install-xw5w7nuz/h5py/h5py/api_compat.h:26,
                     from /tmp/pip-install-xw5w7nuz/h5py/h5py/defs.c:654:
    /home/miranda9/.conda/envs/my_new_env/lib/python3.7/site-packages/numpy/core/include/numpy/npy_1_7_deprecated_api.h:17:2: warning: #warning "Using deprecated NumPy API, disable it with " "#define NPY_NO_DEPRECATED_API NPY_1_7_API_VERSION" [-Wcpp]
     #warning "Using deprecated NumPy API, disable it with " \
      ^
    In file included from /tmp/pip-install-xw5w7nuz/h5py/h5py/defs.c:654:0:
    /tmp/pip-install-xw5w7nuz/h5py/h5py/api_compat.h:27:18: fatal error: hdf5.h: No such file or directory
     #include "hdf5.h"
                      ^
    compilation terminated.
    error: command 'gcc' failed with exit status 1
    ----------------------------------------
  Rolling back uninstall of h5py
  Moving to /home/miranda9/.conda/envs/my_new_env/lib/python3.7/site-packages/h5py
   from /home/miranda9/.conda/envs/my_new_env/lib/python3.7/site-packages/~5py
  Moving to /home/miranda9/.conda/envs/my_new_env/lib/python3.7/site-packages/h5py-2.8.0-py3.7.egg-info
   from /home/miranda9/.conda/envs/my_new_env/lib/python3.7/site-packages/~5py-2.8.0-py3.7.egg-info
ERROR: Command errored out with exit status 1: /home/miranda9/.conda/envs/my_new_env/bin/python -u -c 'import sys, setuptools, tokenize; sys.argv[0] = '"'"'/tmp/pip-install-xw5w7nuz/h5py/setup.py'"'"'; __file__='"'"'/tmp/pip-install-xw5w7nuz/h5py/setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(__file__);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, __file__, '"'"'exec'"'"'))' install --record /tmp/pip-record-k4yqbpkb/install-record.txt --single-version-externally-managed --compile --install-headers /home/miranda9/.conda/envs/my_new_env/include/python3.7m/h5py Check the logs for full command output.
brando90 commented 4 years ago

@tristandeleu I am getting some issues with h5py that torchmeta requires. Is this really required? do you see a way around this?

tristandeleu commented 4 years ago

No unfortunately, h5py is a hard requirement because all the dataloaders are based on HDF5. You could probably still use some features of Torchmeta (such as the MetaModules and utility functions), but you would be missing out on all the dataloaders (which is one of the main features of the library).

brando90 commented 4 years ago

@tristandeleu

I'm not sure why pip doesn't let me install torchmeta. The setup.py says that any version except 2.9.0:

    install_requires=[
        'torch>=1.3.0,<1.5.0',
        'torchvision>=0.4.0,<0.6.0',
        'numpy>=1.14.0',
        'Pillow>=5.0.0,<7.0.0',
        'h5py~=2.9.0',
        'tqdm>=4.0.0',
        'requests' # Required by Torchvision
    ]

so I am not sure why it's not installing it.

I will try to install it with the no dependencies flag and see what happens since just from reading your list it should be fine.

pip install --no-deps torchmeta==1.3.1
brando90 commented 4 years ago

It seems I was wrong h5py really isn't there. That seems to be my only issue now:

$ conda list
# packages in environment at /home/miranda9/.conda/envs/dawei:
#
# Name                    Version                   Build  Channel
_libgcc_mutex             0.1                        main  
_pytorch_select           2.0             gpu_21940.g1a3a219    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
absl-py                   0.8.1                    py37_0  
apex                      0.1.0_1.7.0     py37_662.gb1ba068    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
atomicwrites              1.4.0                      py_0  
attrs                     20.2.0                     py_0  
blas                      1.0                    openblas  
brotlipy                  0.7.0           py37h7b6447c_1000  
bzip2                     1.0.8                h7b6447c_0  
c-ares                    1.16.1               h7b6447c_0  
ca-certificates           2020.7.22                     0  
cairo                     1.14.12              h8948797_3  
certifi                   2020.6.20                py37_0  
cffi                      1.12.3           py37h2e261b9_0  
chardet                   3.0.4                 py37_1003  
click                     7.0                      py37_0  
cloudpickle               1.6.0                      py_0  
coverage                  4.5.4            py37h7b6447c_0  
cryptography              3.1              py37h1ba5d50_0  
cudatoolkit               10.2.89            680.g0f7a43a    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
cudnn                     7.6.5_10.2         650.g338a052    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
cxxfilt                   0.2.0           py_622.gbc2955e    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
cycler                    0.10.0                   py37_0  
cytoolz                   0.10.1           py37h7b6447c_0  
dask-core                 2.27.0                     py_0  
ddl                       1.5.1           py37_1364.ged89432    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
decorator                 4.4.2                      py_0  
ffmpeg                    4.2.2                h20bf706_0  
fontconfig                2.13.0               h9420a91_0  
freeglut                  3.0.0                hf484d3e_5  
freetype                  2.9.1                h8a8886c_0  
future                    0.17.1                   py37_0  
giflib                    5.1.4                h14c3975_1  
glib                      2.63.1               h5a9c865_0  
gmp                       6.1.2                h7f7056e_2  
gnutls                    3.6.5             h71b1129_1002  
graphite2                 1.3.14               h23475e2_0  
graphsurgeon              0.4.1           py37_697.g9922bde    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
grpcio                    1.16.1           py37hf8bcb03_1  
harfbuzz                  1.8.8                hffaf4a1_0  
hdf5                      1.10.2               hba1933b_1  
hypothesis                3.59.1           py37h39e3cac_0  
icu                       58.2                 he6710b0_3  
idna                      2.8                      py37_0  
imageio                   2.9.0                      py_0  
importlib-metadata        1.7.0                    py37_0  
importlib_metadata        1.7.0                         0  
jasper                    2.0.14               h07fcdf6_1  
jpeg                      9b                   hcb7ba68_2  
kiwisolver                1.2.0            py37hfd86e86_0  
lame                      3.100                h7b6447c_0  
ld_impl_linux-ppc64le     2.33.1               h0f24833_7  
leveldb                   1.20                 hf484d3e_1  
libedit                   3.1.20191231         h14c3975_1  
libffi                    3.2.1                hf62a594_5  
libgcc-ng                 8.2.0                h822a55f_1  
libgfortran-ng            7.3.0                h822a55f_1  
libglu                    9.0.0                hf484d3e_1  
libopenblas               0.3.10               h5a2b251_0  
libopencv                 3.4.8           py37_784.g5a642ca    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
libopus                   1.3.1                h7b6447c_0  
libpng                    1.6.36               hbc83047_0  
libprotobuf               3.8.0              634.g08dc819    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
libstdcxx-ng              8.2.0                h822a55f_1  
libtiff                   4.1.0                h2733197_1  
libuuid                   1.0.3                h1bed415_2  
libvpx                    1.7.0                hf484d3e_0  
libwebp                   1.0.1                h8e7db2f_0  
libxcb                    1.14                 h7b6447c_0  
libxml2                   2.9.10               he19cac6_1  
llvmlite                  0.31.0           py37hd408876_0  
lmdb                      0.9.22               hf484d3e_1  
lz4-c                     1.9.2                he6710b0_1  
markdown                  3.1.1                    py37_0  
matplotlib                3.0.3            py37h5429711_0  
mock                      3.0.5                    py37_0  
more-itertools            8.5.0                      py_0  
nccl                      2.5.6              645.g51c2e94    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
ncurses                   6.2                  he6710b0_1  
nettle                    3.4.1                hbb512f6_0  
networkx                  2.3                        py_0  
ninja                     1.9.0            py37hfd86e86_0  
numactl                   2.0.12             628.gb5e1afd    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
numba                     0.47.0           py37h962f231_0  
numpy                     1.17.4           py37hd5be1e1_0  
numpy-base                1.17.4           py37h2f8d375_0  
olefile                   0.46                     py37_0  
onnx                      1.6.0           py37_671.g75d3229    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
opencv                    3.4.8           py37_784.g5a642ca    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
openh264                  2.1.0                hd408876_0  
openssl                   1.1.1h               h7b6447c_0  
pciutils                  3.6.2              627.g804ec60    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
pcre                      8.44                 he6710b0_0  
pillow                    7.1.2            py37haac5956_0  
pip                       20.2.2                   py37_0  
pixman                    0.40.0               h7b6447c_0  
pluggy                    0.13.1                   py37_0  
powerai-license           1.7.0              778.g91a0fd9    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
powerai-release           1.7.0              627.g1c389a2    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
powerai-tools             1.7.0              623.g843ad38    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
protobuf                  3.8.0           py37_642.gdc7b773    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
psutil                    5.6.7            py37h7b6447c_0  
py                        1.9.0                      py_0  
py-opencv                 3.4.8           py37_784.g5a642ca    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
pycparser                 2.20                       py_2  
pyopenssl                 19.1.0                     py_1  
pyparsing                 2.4.7                      py_0  
pysocks                   1.7.1                    py37_1  
pytest                    4.4.2                    py37_0  
python                    3.7.7           ha29dc6b_0_cpython  
python-dateutil           2.8.1                      py_0  
python-lmdb               0.98             py37he6710b0_0  
pytorch                   1.3.1            21940.g1a3a219    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
pytorch-base              1.3.1           gpu_py37_21940.g1a3a219    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
pytz                      2020.1                     py_0  
pywavelets                1.1.1            py37h7b6447c_0  
pyyaml                    5.3.1            py37h7b6447c_1  
readline                  8.0                  h7b6447c_0  
requests                  2.22.0                   py37_1  
scikit-image              0.15.0           py37he6710b0_0  
scipy                     1.3.1            py37he2b7bc3_0  
setuptools                49.6.0                   py37_0  
six                       1.13.0                   py37_0  
snappy                    1.1.8                he6710b0_0  
spectrum-mpi              10.03              678.ga72dafb    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
sqlite                    3.33.0               hbc83047_0  
tabulate                  0.8.2                    py37_0  
tbb                       2020.3               hfd86e86_0  
tensorboard               2.1.0           py37_3dc74fe_3959.g2372771    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
tensorrt                  7.0.0.11        py37_697.g9922bde    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
tk                        8.6.10               hbc83047_0  
toolz                     0.10.0                     py_0  
torchmeta                 1.3.1                    pypi_0    pypi
torchtext                 0.4.0           py37_633.g16a90d7    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
torchvision               0.4.2              661.g37d5d4d    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
torchvision-base          0.4.2           gpu_py37_661.g37d5d4d    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
tornado                   6.0.4            py37h7b6447c_1  
tqdm                      4.36.1                     py_0  
typing                    3.6.4                    py37_0  
uff                       0.6.5           py37_697.g9922bde    https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda
urllib3                   1.25.10                    py_0  
werkzeug                  0.16.0                     py_0  
wheel                     0.35.1                     py_0  
x264                      1!157.20191217       h7b6447c_0  
xz                        5.2.5                h7b6447c_0  
yaml                      0.2.5                h7b6447c_0  
zipp                      3.1.0                      py_0  
zlib                      1.2.11               h7b6447c_3  
zstd                      1.4.5                h9ceee32_0
brando90 commented 4 years ago

@tristandeleu

Sorry I wish I could use the most recent version...

I am having this error with my code:

    super().__init__(index=index, num_classes=None)  # Regression task
TypeError: __init__() got an unexpected keyword argument 'index'

is it ok if I remove that index that the newer version requires?

old:

class Task(Dataset):
    """Base class for a classification task.

    Parameters
    ----------
    num_classes : int
        Number of classes for the classification task.
    """
    def __init__(self, num_classes, transform=None, target_transform=None):
        super(Task, self).__init__(transform=transform,
            target_transform=target_transform)
        self.num_classes = num_classes

vs new:

class Task(Dataset):
    """Base class for a classification task.

    Parameters
    ----------
    num_classes : int
        Number of classes for the classification task.
    """
    def __init__(self, index, num_classes,
                 transform=None, target_transform=None):
        super(Task, self).__init__(index, transform=transform,
                                   target_transform=target_transform)
        self.num_classes = num_classes
tristandeleu commented 4 years ago

It should be ok to remove that, the only thing you'd be losing is some determinism (if you see the same task twice, then the datasets could contain different samples). Depending on the version you are rolling with, you might need to change the __hash__ method of Dataset and SubsetTask (you probably just generate a random number for the hash).

def __hash__(self):
    return random.randrange(1 << 32)
brando90 commented 4 years ago

to complement the close.

I eventually did get it to work by cloning the conda version of the appropriate binaries and doing everything from scratch. I don't think the commands I ran will be useful since they depend on my HPC's conda stuff but will paste them anyway:

module load wmlce/1.7.0-py3.7
module load wmlce/1.7.0-py3.7

create copy of wmlce env that HAL has
conda create --name=automl-meta-learning_wmlce-v1.7.0-py3.7 --clone=wmlce-v1.7.0-py3.7
conda activate automl-meta-learning_wmlce-v1.7.0-py3.7
conda remove --name xx --all

prepend the 1.7.0 code for wmlce
conda config --prepend channels https://public.dhe.ibm.com/ibmdl/export/pub/software/server/ibm-ai/conda/#/

pip install torchmeta==1.3.1
pip install --no-deps torchmeta==1.3.1

there is a way to get them using conda and the IBM channel but I forget how to do it now. Going through the other links above referenced here should eventually get them.