microsoft / SoM

Set-of-Mark Prompting for GPT-4V and LMMs
MIT License
1.2k stars 96 forks source link

Dependency Conflict Error When Installing SEEM from GitHub Repository #14

Open ericqqqqq opened 12 months ago

ericqqqqq commented 12 months ago

had the below error when installing pip install git+https://github.com/UX-Decoder/Segment-Everything-Everywhere-All-At-Once.git@package

pip version: pip 23.3.1 os: macOS 14.1.1

ERROR: Cannot install SEEM and seem==1.0 because these package versions have conflicting dependencies.

The conflict is caused by:
    seem 1.0 depends on numpy==1.23.1
    accelerate 0.23.0 depends on numpy>=1.17
    cityscapesscripts 2.2.2 depends on numpy
    deepspeed 0.10.3 depends on numpy
    fvcore 0.1.5.post20221221 depends on numpy
    gradio 3.42.0 depends on numpy~=1.0
    mup 1.0.0 depends on numpy
    opencv-python 4.8.1.78 depends on numpy>=1.21.2; python_version >= "3.10"
    opencv-python 4.8.1.78 depends on numpy>=1.21.4; python_version >= "3.10" and platform_system == "Darwin"
    opencv-python 4.8.1.78 depends on numpy>=1.23.5; python_version >= "3.11"

To fix this you could try to:
1. loosen the range of package versions you've specified
2. remove package versions to allow pip attempt to solve the dependency conflict
yunyiliu commented 11 months ago

Same issue here

pbarker commented 11 months ago

Same here with poetry

francisrafal commented 11 months ago

same here pip: 23.3.2 OS: macOS 14.2.1

francisrafal commented 11 months ago

Okay, I solved this by installing Python 3.10.11 and creating a virtual environment. But now I can't run sh make.sh from the installation instructions because it says that the environment variable CUDA_HOME is not set. It seems to me that it's not possible to run this on a Mac without an NVIDIA graphics card?

cuizheng0520 commented 10 months ago

Okay, I solved this by installing Python 3.10.11 and creating a virtual environment. But now I can't run sh make.sh from the installation instructions because it says that the environment variable CUDA_HOME is not set. It seems to me that it's not possible to run this on a Mac without an NVIDIA graphics card?

same,very interesting, but it is definitely not legal, and there are other similar project?

ericqqqqq commented 10 months ago

We managed to run it smoothly on PC(windows operating system and Nvidia GPU)

abrichr commented 9 months ago

CUDA is indeed required, therefore local installation on MacOS is not currently available.

For now, you can deploy to EC2 on AWS via https://github.com/microsoft/SoM/pull/19. Using the configured g4dn.xlarge instance type costs $0.526/hr.