mattragoza/LiGAN

Environment.yml isnt complete and libmolgrid git command fails

Opened this issue · 3 comments

lark and gcc_linux-64==10.4.0 shoudl be added

also I dont think the public has access to clone this
git clone git@github.com:gnina/libmolgrid.git

I needed to use
git clone https://github.com/gnina/libmolgrid.git

Also the environment installs a torch that isnt gpu enabled.

ERROR tests/test_training.py::TestGenerativeSolver::test_solver_train_and_test[VAE2] - AssertionError: Torch not compiled with CUDA enabled
ERROR tests/test_training.py::TestGenerativeSolver::test_solver_train_and_test[CVAE2] - AssertionError: Torch not compiled with CUDA enabled
=================================================================== 21 failed, 126 passed, 11 warnings, 791 errors in 96.49s (0:01:36) ====================================================================

After fixing the environment there are still a few errors:

========================================================================================= short test summary info =========================================================================================
FAILED tests/test_data.py::TestMolDataset::test_benchmark[False] - ValueError: Could not read data/crossdock2020/1A02_HUMAN_25_199_pep_0/1t1z_A_rec.pdb
FAILED tests/test_data.py::TestMolDataset::test_benchmark[True] - AssertionError: file does not exist
FAILED tests/test_models.py::TestGridGenerator::test_gen_benchmark[AE-0-c-0] - AssertionError: 27.7M params 1.10s / batch
FAILED tests/test_models.py::TestGridGenerator::test_gen_benchmark[CE-0-c-0] - AssertionError: 30.1M params 1.25s / batch
FAILED tests/test_models.py::TestGridGenerator::test_gen_benchmark[VAE-0-c-0] - AssertionError: 34.8M params 1.11s / batch
FAILED tests/test_models.py::TestGridGenerator::test_gen_benchmark[CVAE-0-c-0] - AssertionError: 57.2M params 1.76s / batch
FAILED tests/test_models.py::TestGridGenerator::test_gen_benchmark[CVAE-1-c-0] - AssertionError: 57.1M params 1.12s / batch
FAILED tests/test_models.py::TestGridGenerator::test_gen_benchmark[CGAN-0-c-0] - AssertionError: 37.2M params 1.25s / batch
FAILED tests/test_models.py::TestGridGenerator::test_gen_benchmark[VAE2-0-c-0] - AssertionError: 34.8M params 1.11s / batch
FAILED tests/test_models.py::TestGridGenerator::test_gen_benchmark[CVAE2-0-c-0] - torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 136.00 MiB. GPU 0 has a total capacty of 15.74 GiB of which 34.62 MiB is free. Including non-PyTorch memory, this process has 15.70...
FAILED tests/test_models.py::TestGridGenerator::test_gen_benchmark[CVAE2-1-c-0] - AssertionError: 57.1M params 1.12s / batch
======================================================================== 11 failed, 927 passed, 903 warnings in 1010.01s (0:16:50) ========================================================================

Most of these errors are time related (test_gen_benchmark)
E AssertionError: 37.2M params 1.25s / batch
E assert 1.2485175609588623 < 1

though it appears a file is missing

data/crossdock2020/1A02_HUMAN_25_199_pep_0/1t1z_A_rec.pdb

and one error because my gpu runs out of memory.