Tools for COMP6 benchmark (#196)

.gitmodules

+[submodule "dataset/COMP6"]
+	path = dataset/COMP6
+	url = https://github.com/isayev/COMP6.git

azure/runnable_submodules.yml

@@ -21,11 +21,11 @@ steps:
 - script: 'azure/install_dependencies.sh && pip install .'
   displayName: 'Install dependencies'
 
-- script: 'python -m torchani.neurochem.trainer --tqdm tests/test_data/inputtrain.ipt  dataset/ani_gdb_s01.h5 dataset/ani_gdb_s01.h5'
+- script: 'python -m torchani.neurochem.trainer --tqdm tests/test_data/inputtrain.ipt  dataset/ani1-up_to_gdb4/ani_gdb_s01.h5 dataset/ani1-up_to_gdb4/ani_gdb_s01.h5'
   displayName: NeuroChem Trainer
 
-- script: 'python -m torchani.neurochem.trainer --tqdm tests/test_data/inputtrain.yaml  dataset/ani_gdb_s01.h5 dataset/ani_gdb_s01.h5'
+- script: 'python -m torchani.neurochem.trainer --tqdm tests/test_data/inputtrain.yaml  dataset/ani1-up_to_gdb4/ani_gdb_s01.h5 dataset/ani1-up_to_gdb4/ani_gdb_s01.h5'
   displayName: NeuroChem Trainer YAML config
 
-- script: 'python -m torchani.data.cache_aev tmp dataset/ani_gdb_s01.h5 256'
+- script: 'python -m torchani.data.cache_aev tmp dataset/ani1-up_to_gdb4/ani_gdb_s01.h5 256'
   displayName: Cache AEV

azure/tests.yml

@@ -18,6 +18,9 @@ steps:
   inputs:
     versionSpec: '$(python.version)'
 
+- script: 'git submodule update --init'
+  displayName: 'Fetch submodules'
+
 - script: 'azure/install_dependencies.sh && pip install nose coverage codecov'
   displayName: 'Install dependencies'
 

azure/tools.yml

@@ -19,14 +19,20 @@ steps:
   inputs:
     versionSpec: '$(python.version)'
 
+- script: 'git submodule update --init'
+  displayName: 'Fetch submodules'
+
 - script: 'azure/install_dependencies.sh && pip install .'
   displayName: 'Install dependencies'
 
-- script: 'python tools/training-benchmark.py ./dataset/ani_gdb_s01.h5'
+- script: 'python tools/training-benchmark.py ./dataset/ani1-up_to_gdb4/ani_gdb_s01.h5'
   displayName: Training Benchmark
 
-- script: 'python tools/neurochem-test.py ./dataset/ani_gdb_s01.h5'
+- script: 'python tools/neurochem-test.py ./dataset/ani1-up_to_gdb4/ani_gdb_s01.h5'
   displayName: NeuroChem Test
 
-- script: 'python tools/inference-benchmark.py --tqdm ./xyz_files/CH4-5.xyz'
+- script: 'python tools/inference-benchmark.py --tqdm ./dataset/xyz_files/CH4-5.xyz'
   displayName: Inference Benchmark
+
+- script: 'python tools/comp6.py ./dataset/COMP6/COMP6v1/s66x8'
+  displayName: COMP6 Benchmark

COMP6 @ 79f41c15

+Subproject commit 79f41c156f8e19506fe951b5513b98e7c534a503

examples/cache_aev.py

@@ -26,8 +26,8 @@ try:
     path = os.path.dirname(os.path.realpath(__file__))
 except NameError:
     path = os.getcwd()
-training_path = os.path.join(path, '../dataset/ani_gdb_s01.h5')
-validation_path = os.path.join(path, '../dataset/ani_gdb_s01.h5')
+training_path = os.path.join(path, '../dataset/ani1-up_to_gdb4/ani_gdb_s01.h5')
+validation_path = os.path.join(path, '../dataset/ani1-up_to_gdb4/ani_gdb_s01.h5')  # noqa: E501
 
 # checkpoint file to save model when validation RMSE improves
 model_checkpoint = 'model.pt'

examples/neurochem_trainer.py

@@ -30,8 +30,8 @@ try:
 except NameError:
     path = os.getcwd()
 cfg_path = os.path.join(path, '../tests/test_data/inputtrain.ipt')
-training_path = os.path.join(path, '../dataset/ani_gdb_s01.h5')
-validation_path = os.path.join(path, '../dataset/ani_gdb_s01.h5')
+training_path = os.path.join(path, '../dataset/ani1-up_to_gdb4/ani_gdb_s01.h5')  # noqa: E501
+validation_path = os.path.join(path, '../dataset/ani1-up_to_gdb4/ani_gdb_s01.h5')  # noqa: E501
 
 ###############################################################################
 # We also need to set the device to run the training:

examples/nnp_training.py

@@ -34,8 +34,8 @@ try:
     path = os.path.dirname(os.path.realpath(__file__))
 except NameError:
     path = os.getcwd()
-training_path = os.path.join(path, '../dataset/ani_gdb_s01.h5')
-validation_path = os.path.join(path, '../dataset/ani_gdb_s01.h5')
+training_path = os.path.join(path, '../dataset/ani1-up_to_gdb4/ani_gdb_s01.h5')
+validation_path = os.path.join(path, '../dataset/ani1-up_to_gdb4/ani_gdb_s01.h5')  # noqa: E501
 
 # checkpoint file to save model when validation RMSE improves
 model_checkpoint = 'model.pt'

tests/test_data.py

@@ -5,8 +5,8 @@ import unittest
 from torchani.data.cache_aev import cache_aev
 
 path = os.path.dirname(os.path.realpath(__file__))
-dataset_path = os.path.join(path, '../dataset')
-dataset_path2 = os.path.join(path, '../dataset/ani_gdb_s01.h5')
+dataset_path = os.path.join(path, '../dataset/ani1-up_to_gdb4')
+dataset_path2 = os.path.join(path, '../dataset/ani1-up_to_gdb4/ani_gdb_s01.h5')
 batch_size = 256
 builtins = torchani.neurochem.Builtins()
 consts = builtins.consts

tests/test_ignite.py

@@ -8,7 +8,7 @@ import torchani
 import torchani.ignite
 
 path = os.path.dirname(os.path.realpath(__file__))
-path = os.path.join(path, '../dataset/ani_gdb_s01.h5')
+path = os.path.join(path, '../dataset/ani1-up_to_gdb4/ani_gdb_s01.h5')
 batchsize = 4
 threshold = 1e-5