Open axin1301 opened 1 month ago
Hello! Could you try creating an environment with these steps?
I encountered the same issue, which was due to numpy 2.0. Was fixed by setting numpy==1.26.4
in requirements.txt
.
@piperwolters would you be open to pinning all the dependencies in the repo? One way of doing it could be by using pip-compile
: https://github.com/jazzband/pip-tools?tab=readme-ov-file#requirements-from-requirementsin
If you make a brief adjustment by writing the dependencies in a requirements.in
file like this:
basicsr==1.4.2
lpips
numpy==1.26.4
scikit-image
Pillow
opencv_python
kornia
wandb
git+https://github.com/openai/CLIP.git
open_clip_torch
torch==2.1.0
torchvision==0.16.0
--find-links=https://girder.github.io/large_image_wheels
GDAL
Then you can run pip install pip-tools
and then pip-compile requirements.in
, and you should get a requirements.txt
file similar to the one below. That would secure the versioning of all the required packages, and also add gdal
to requirements.txt
, so it's automatically installed when running pip install -r requirements.txt
. I usually then add both the requirements.in
and requirements.txt
files to the repo, such that they're available for reference.
#
# This file is autogenerated by pip-compile with Python 3.9
# by the following command:
#
# pip-compile requirements.in
#
--find-links https://girder.github.io/large_image_wheels
absl-py==2.1.0
# via tb-nightly
addict==2.4.0
# via basicsr
basicsr==1.4.2
# via -r requirements.in
certifi==2024.7.4
# via
# requests
# sentry-sdk
charset-normalizer==3.3.2
# via requests
click==8.1.7
# via wandb
clip @ git+https://github.com/openai/CLIP.git
# via -r requirements.in
docker-pycreds==0.4.0
# via wandb
filelock==3.15.4
# via
# huggingface-hub
# torch
# triton
fsspec==2024.6.1
# via
# huggingface-hub
# torch
ftfy==6.2.0
# via
# clip
# open-clip-torch
future==1.0.0
# via basicsr
gdal==3.10.0
# via -r requirements.in
gitdb==4.0.11
# via gitpython
gitpython==3.1.43
# via wandb
grpcio==1.65.1
# via tb-nightly
huggingface-hub==0.24.2
# via
# open-clip-torch
# timm
idna==3.7
# via requests
imageio==2.34.2
# via scikit-image
importlib-metadata==8.2.0
# via
# markdown
# yapf
jinja2==3.1.4
# via torch
kornia==0.7.3
# via -r requirements.in
kornia-rs==0.1.5
# via kornia
lazy-loader==0.4
# via scikit-image
lmdb==1.5.1
# via basicsr
lpips==0.1.4
# via -r requirements.in
markdown==3.6
# via tb-nightly
markupsafe==2.1.5
# via
# jinja2
# werkzeug
mpmath==1.3.0
# via sympy
networkx==3.2.1
# via
# scikit-image
# torch
numpy==1.26.4
# via
# -r requirements.in
# basicsr
# imageio
# lpips
# opencv-python
# scikit-image
# scipy
# tb-nightly
# tifffile
# torchvision
nvidia-cublas-cu12==12.1.3.1
# via
# nvidia-cudnn-cu12
# nvidia-cusolver-cu12
# torch
nvidia-cuda-cupti-cu12==12.1.105
# via torch
nvidia-cuda-nvrtc-cu12==12.1.105
# via torch
nvidia-cuda-runtime-cu12==12.1.105
# via torch
nvidia-cudnn-cu12==8.9.2.26
# via torch
nvidia-cufft-cu12==11.0.2.54
# via torch
nvidia-curand-cu12==10.3.2.106
# via torch
nvidia-cusolver-cu12==11.4.5.107
# via torch
nvidia-cusparse-cu12==12.1.0.106
# via
# nvidia-cusolver-cu12
# torch
nvidia-nccl-cu12==2.18.1
# via torch
nvidia-nvjitlink-cu12==12.5.82
# via
# nvidia-cusolver-cu12
# nvidia-cusparse-cu12
nvidia-nvtx-cu12==12.1.105
# via torch
open-clip-torch==2.26.1
# via -r requirements.in
opencv-python==4.10.0.84
# via
# -r requirements.in
# basicsr
packaging==24.1
# via
# clip
# huggingface-hub
# kornia
# lazy-loader
# scikit-image
pillow==10.4.0
# via
# -r requirements.in
# basicsr
# imageio
# scikit-image
# torchvision
platformdirs==4.2.2
# via
# wandb
# yapf
protobuf==4.25.4
# via
# tb-nightly
# wandb
psutil==6.0.0
# via wandb
pyyaml==6.0.1
# via
# basicsr
# huggingface-hub
# timm
# wandb
regex==2024.7.24
# via
# clip
# open-clip-torch
requests==2.32.3
# via
# basicsr
# huggingface-hub
# torchvision
# wandb
safetensors==0.4.3
# via timm
scikit-image==0.24.0
# via
# -r requirements.in
# basicsr
scipy==1.13.1
# via
# basicsr
# lpips
# scikit-image
sentry-sdk==2.11.0
# via wandb
setproctitle==1.3.3
# via wandb
six==1.16.0
# via
# docker-pycreds
# tb-nightly
smmap==5.0.1
# via gitdb
sympy==1.13.1
# via torch
tb-nightly==2.18.0a20240729
# via basicsr
tensorboard-data-server==0.7.2
# via tb-nightly
tifffile==2024.7.24
# via scikit-image
timm==1.0.8
# via open-clip-torch
tomli==2.0.1
# via yapf
torch==2.1.0
# via
# -r requirements.in
# basicsr
# clip
# kornia
# lpips
# open-clip-torch
# timm
# torchvision
torchvision==0.16.0
# via
# -r requirements.in
# basicsr
# clip
# lpips
# open-clip-torch
# timm
tqdm==4.66.4
# via
# basicsr
# clip
# huggingface-hub
# lpips
# open-clip-torch
triton==2.1.0
# via torch
typing-extensions==4.12.2
# via
# huggingface-hub
# torch
# wandb
urllib3==2.2.2
# via
# requests
# sentry-sdk
wandb==0.17.5
# via -r requirements.in
wcwidth==0.2.13
# via ftfy
werkzeug==3.0.3
# via tb-nightly
yapf==0.40.2
# via basicsr
zipp==3.19.2
# via importlib-metadata
# The following packages are considered to be unsafe in a requirements file:
# setuptools
Hi Jacob, thanks for the suggestion. I have pinned numpy in requirements.txt for now, and will look into pinning the rest of them (maybe using pip-compile).
Hi, thank you for your contribution to our community. When I reproduce the results, I got stuck by "ImportError: numpy._core.multiarray failed to import". Could you please show me your python package versions so that I can run the inference with no errors? Thank you.