mistralai / mistral-inference

Official inference library for Mistral models
https://mistral.ai/
Apache License 2.0
9.16k stars 803 forks source link

I am unable to build the vLLM Container #142

Open AMGI-Pipeline opened 3 months ago

AMGI-Pipeline commented 3 months ago

I am unable to build the vLLM Container. I am getting these errors:

252.2 Building wheels for collected packages: vllm
252.2   Building wheel for vllm (pyproject.toml): started
313.8   Building wheel for vllm (pyproject.toml): still running...
551.9   Building wheel for vllm (pyproject.toml): still running...
774.4   Building wheel for vllm (pyproject.toml): still running...
936.1   Building wheel for vllm (pyproject.toml): still running...
983.2   Building wheel for vllm (pyproject.toml): finished with status 'error'
983.6   error: subprocess-exited-with-error
983.6
983.6   × Building wheel for vllm (pyproject.toml) did not run successfully.
983.6   │ exit code: 1
983.6   ╰─> [983 lines of output]
983.6       /tmp/pip-build-env-rt38cp4x/overlay/local/lib/python3.10/dist-packages/torch/nn/modules/transformer.py:20: UserWarning: Failed to initialize NumPy: No module named 'numpy' (Triggered internally at ../torch/csrc/utils/tensor_numpy.cpp:84.)
983.6         device: torch.device = torch.device(torch._C._get_default_device()),  # torch.device('cpu'),
.
.
.
983.6         File "/usr/lib/python3.10/subprocess.py", line 369, in check_call
983.6           raise CalledProcessError(retcode, cmd)
983.6       subprocess.CalledProcessError: Command '['cmake', '--build', '.', '--target', '_C', '-j', '32']' returned non-zero exit status 1.
983.6       [end of output]
983.6
983.6   note: This error originates from a subprocess, and is likely not a problem with pip.
983.6   ERROR: Failed building wheel for vllm
983.6 Failed to build vllm
983.6 ERROR: Could not build wheels for vllm, which is required to install pyproject.toml-based projects
------

Any ideas what is going on?

Any help is greatly apprecated!

vipulg13 commented 3 months ago

I am also getting the same error. Does anybody have any ideas what is causing this?

AMGI-Pipeline commented 3 months ago

Until this is fixed, I am pulling the pre-built container from the registry as documented here.

vipulg13 commented 3 months ago

@AMGI-Pipeline after downgrading the CUDA version from 12.3 to 12.1 the build is successful. However, I am dealing now with the following error while creating a container:

ImportError: cannot import name 'model_validator' from 'pydantic'

nathan-pugh-fsn commented 3 months ago

system level import file location is probably in the wrong place…check config files for CUDA