Incompatibility with pydantic

[jyoung105]

Hello,
I know engineers in @replicate work with this issue in Pull requests.
But want to know how I can push the image with uninstalling and installing new version of packages in cog.yaml, too

Thank you!

Validating model schema...

Traceback (most recent call last):
  File "/root/.pyenv/versions/3.10.14/lib/python3.10/runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/root/.pyenv/versions/3.10.14/lib/python3.10/runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "/root/.pyenv/versions/3.10.14/lib/python3.10/site-packages/cog/command/openapi_schema.py", line 25, in <module>
    raise CogError(app.state.setup_result.logs)
cog.errors.CogError: Error while loading predictor:

Traceback (most recent call last):
  File "/root/.pyenv/versions/3.10.14/lib/python3.10/site-packages/cog/server/http.py", line 137, in create_app
    predictor = load_predictor_from_ref(predictor_ref)
  File "/root/.pyenv/versions/3.10.14/lib/python3.10/site-packages/cog/predictor.py", line 187, in load_predictor_from_ref
    spec.loader.exec_module(module)
  File "<frozen importlib._bootstrap_external>", line 883, in exec_module
  File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed
  File "/src/predict.py", line 10, in <module>
    from insightface.app import FaceAnalysis
  File "/root/.pyenv/versions/3.10.14/lib/python3.10/site-packages/insightface/__init__.py", line 18, in <module>
    from . import app
  File "/root/.pyenv/versions/3.10.14/lib/python3.10/site-packages/insightface/app/__init__.py", line 2, in <module>
    from .mask_renderer import *
  File "/root/.pyenv/versions/3.10.14/lib/python3.10/site-packages/insightface/app/mask_renderer.py", line 4, in <module>
    import albumentations as A
  File "/root/.pyenv/versions/3.10.14/lib/python3.10/site-packages/albumentations/__init__.py", line 3, in <module>
    from .augmentations import *
  File "/root/.pyenv/versions/3.10.14/lib/python3.10/site-packages/albumentations/augmentations/__init__.py", line 1, in <module>
    from .blur.functional import *
  File "/root/.pyenv/versions/3.10.14/lib/python3.10/site-packages/albumentations/augmentations/blur/__init__.py", line 1, in <module>
    from .functional import *
  File "/root/.pyenv/versions/3.10.14/lib/python3.10/site-packages/albumentations/augmentations/blur/functional.py", line 9, in <module>
    from albumentations.augmentations.geometric.functional import scale
  File "/root/.pyenv/versions/3.10.14/lib/python3.10/site-packages/albumentations/augmentations/geometric/__init__.py", line 2, in <module>
    from .resize import *
  File "/root/.pyenv/versions/3.10.14/lib/python3.10/site-packages/albumentations/augmentations/geometric/resize.py", line 6, in <module>
    from pydantic import Field, ValidationInfo, field_validator
ImportError: cannot import name 'ValidationInfo' from 'pydantic' (/root/.pyenv/versions/3.10.14/lib/python3.10/site-packages/pydantic/__init__.cpython-310-x86_64-linux-gnu.so)

[cryptexis]

I have the same issue, fighting against it 12 hours already

[cryptexis]

@jyoung105 found the problem it is actually because of algumentations : specify the version 1.4.3, two days ago they release 1.4.4 and it breaks everything

[fofr]

albumentations-team/albumentations#1624

[jyoung105]

Thanks @cryptexis @fofr

After arriving at office, I will take a look.
I didn't check albumentation, just pydantic and insightface.

Thanks once again 😀

[ElyasMoshirpanahi]

@jyoung105 After dealing with this issue I found out that in my case it was due to a dependency mismatch , instant id is currently compatible with this library with this version : albumentations==1.4.3 and if you check albumentations is not pinned in the yaml file therefore newest version (1.4.4) would be installed and raises a pydantic error , My yaml file looks like this and it works:

build:
  # set to true if your model requires a GPU
  gpu: true
  # cuda: "12.1"
  
  # a list of ubuntu apt packages to install
  system_packages:
    - "libgl1-mesa-glx"
    - "libglib2.0-0"

  # python version in the form '3.11' or '3.11.4'
  python_version: "3.11"

  # a list of packages in the format <package-name>==<version>
  python_packages:
    - "albumentations==1.4.3"
    - "opencv-python==4.9.0.80"
    - "transformers==4.37.0"
    - "accelerate==0.26.1"
    - "insightface==0.7.3"
    - "diffusers==0.25.1"
    - "onnxruntime==1.16.3"
    - "omegaconf==2.3.0"
    - "pandas==2.0.3"
  # commands run after the environment is setup
  run:
    - curl -o /usr/local/bin/pget -L "https://github.com/replicate/pget/releases/download/v0.6.0/pget_linux_x86_64" && chmod +x /usr/local/bin/pget
    - pip freeze
# predict.py defines how predictions are run on your model
predict: "cog/predict.py:Predictor"

Hope that it fixes your case as well 😊