search

sb-ai-lab / Py-Boost

Python based GBDT implementation on GPU. Efficient multioutput (multiclass/multilabel/multitask) training
Apache License 2.0
163 stars 13 forks source link

error install #17

Open TDL77 opened 10 months ago

TDL77 commented 10 months ago

image image

btbpanda commented 10 months ago

@TDL77 According to what you provided, this is unobvious why the error occured, but it is definetly the environment issue. You can check the following:

1) Ensure that notebook kernel is set to the correct python path. Compare command line which python output and Notebook output of


import sys
print(sys.executable)

2) Ensure that you are running on GPU available machine. Check the nvidia-smi output

3) Ensure that there is single cupy installed inside the environment. Check the pip freeze | grep cupy output. Since, for example, conda cupy and pypi cupy-cuda110 or cupy-cuda11x are all the different packages! And import conflicts can occur. If cupy was already installed inside conda/mamba env, just pip uninstall cupy-cuda110 and try again

4) Check that you installed the correct cupy for your cuda. cupy-cuda110 is the package for CUDA11.0. If you are running under another CUDA just uninstall cupy-cuda110 and install the version you need

If nonthing helps, just let me know again

TDL77 commented 10 months ago

все переустановил , но таже ошибка

image