continue keras core integration

aimat-lab · Sep 5, 2023 · 1bf70ae · 1bf70ae
1 parent 6cb727d
commit 1bf70ae
Show file tree

Hide file tree

Showing 4 changed files with 155 additions and 97 deletions.
diff --git a/kgcnn/backend/tensorflow.py b/kgcnn/backend/tensorflow.py
@@ -0,0 +1,12 @@
+import tensorflow as tf
+
+
+def scatter_min(indices, values, shape):
+    return tf.scatter_min(indices, values, shape)
+
+def scatter_max(indices, values, shape):
+    return tf.scatter_nd(indices, values, shape)
+
+def scatter_mean(indices, values, shape):
+
+    return tf.scatter_nd(indices, values, shape)
diff --git a/kgcnn/ops_core/scatter.py b/kgcnn/ops_core/scatter.py
@@ -0,0 +1,46 @@
+from kgcnn import backend
+from keras_core.backend import KerasTensor
+from keras_core.backend import any_symbolic_tensors
+from keras_core.ops.operation import Operation
+
+
+class ScatterMax(Operation):
+    def call(self, indices, values, shape):
+        return backend.scatter_max(indices, values, shape)
+
+    def compute_output_spec(self, indices, values, shape):
+        return KerasTensor(shape, dtype=values.dtype)
+
+
+def scatter_max(indices, values, shape):
+    if any_symbolic_tensors((indices, values, shape)):
+        return ScatterMax().symbolic_call(indices, values, shape)
+    return backend.scatter_max(indices, values, shape)
+
+
+class ScatterMin(Operation):
+    def call(self, indices, values, shape):
+        return backend.scatter_min(indices, values, shape)
+
+    def compute_output_spec(self, indices, values, shape):
+        return KerasTensor(shape, dtype=values.dtype)
+
+
+def scatter_min(indices, values, shape):
+    if any_symbolic_tensors((indices, values, shape)):
+        return ScatterMin().symbolic_call(indices, values, shape)
+    return backend.scatter_min(indices, values, shape)
+
+
+class ScatterMean(Operation):
+    def call(self, indices, values, shape):
+        return backend.scatter_mean(indices, values, shape)
+
+    def compute_output_spec(self, indices, values, shape):
+        return KerasTensor(shape, dtype=values.dtype)
+
+
+def scatter_mean(indices, values, shape):
+    if any_symbolic_tensors((indices, values, shape)):
+        return ScatterMean().symbolic_call(indices, values, shape)
+    return backend.scatter_mean(indices, values, shape)
diff --git a/training_core/results/ESOLDataset/GCN/GCN_ESOLDataset_score.yaml b/training_core/results/ESOLDataset/GCN/GCN_ESOLDataset_score.yaml
@@ -1,5 +1,5 @@
 data_unit: mol/L
-date_time: '2023-09-04 13:32:26'
+date_time: '2023-09-05 13:02:37'
 epochs:
 - 800
 - 800
@@ -15,134 +15,134 @@ learning_rate:
 - 5.1727271056734025e-05
 - 5.1727271056734025e-05
 loss:
-- 0.02065335027873516
-- 0.03046582266688347
-- 0.026986178010702133
-- 0.035748980939388275
-- 0.025066649541258812
+- 0.030111245810985565
+- 0.029905715957283974
+- 0.02666669711470604
+- 0.03320549055933952
+- 0.027655862271785736
 max_learning_rate:
 - 0.0010000000474974513
 - 0.0010000000474974513
 - 0.0010000000474974513
 - 0.0010000000474974513
 - 0.0010000000474974513
 max_loss:
-- 0.7785677909851074
-- 0.7740250825881958
-- 0.7712702751159668
-- 0.7745649814605713
-- 0.785307765007019
+- 0.7727673053741455
+- 0.766106128692627
+- 0.7710932493209839
+- 0.7851217985153198
+- 0.7929884791374207
 max_scaled_mean_absolute_error:
-- 1.630502700805664
-- 1.6439743041992188
-- 1.6447334289550781
-- 1.6330018043518066
-- 1.6777045726776123
+- 1.6122925281524658
+- 1.6240359544754028
+- 1.64657461643219
+- 1.6524299383163452
+- 1.694350242614746
 max_scaled_root_mean_squared_error:
-- 2.051668882369995
-- 2.079864501953125
-- 2.104031801223755
-- 2.0745174884796143
-- 2.1795969009399414
+- 2.045316457748413
+- 2.058136224746704
+- 2.1218948364257812
+- 2.086286783218384
+- 2.2071051597595215
 max_val_loss:
-- 0.38604891300201416
-- 0.4033467471599579
-- 0.43952175974845886
-- 0.4190915822982788
-- 0.4452379643917084
+- 0.4188742935657501
+- 0.3975107669830322
+- 0.36188578605651855
+- 0.4290943443775177
+- 0.3819858431816101
 max_val_scaled_mean_absolute_error:
-- 0.7995112538337708
-- 0.9048331379890442
-- 0.7838926911354065
-- 0.7904984951019287
-- 0.8072317838668823
+- 0.8083961009979248
+- 0.8527750968933105
+- 0.680066704750061
+- 0.8364616632461548
+- 0.7286641001701355
 max_val_scaled_root_mean_squared_error:
-- 1.0650511980056763
-- 1.1253228187561035
-- 1.0860916376113892
-- 1.0380667448043823
-- 1.0352442264556885
+- 1.0951517820358276
+- 1.0844537019729614
+- 0.9051862955093384
+- 1.0477832555770874
+- 0.9376411437988281
 min_learning_rate:
 - 5.1727271056734025e-05
 - 5.1727271056734025e-05
 - 5.1727271056734025e-05
 - 5.1727271056734025e-05
 - 5.1727271056734025e-05
 min_loss:
-- 0.020218532532453537
-- 0.025095166638493538
-- 0.026986178010702133
-- 0.027601951733231544
-- 0.025066649541258812
+- 0.022370334714651108
+- 0.02377462387084961
+- 0.026008524000644684
+- 0.02583014965057373
+- 0.027655862271785736
 min_scaled_mean_absolute_error:
-- 0.04289504885673523
-- 0.05423098802566528
-- 0.05875982344150543
-- 0.058775320649147034
-- 0.0540020689368248
+- 0.04759758338332176
+- 0.05133747309446335
+- 0.05656632035970688
+- 0.05493299663066864
+- 0.05945947393774986
 min_scaled_root_mean_squared_error:
-- 0.12030639499425888
-- 0.15153780579566956
-- 0.1605674773454666
-- 0.16291119158267975
-- 0.15338918566703796
+- 0.12559226155281067
+- 0.14683835208415985
+- 0.16013726592063904
+- 0.15730111300945282
+- 0.163068026304245
 min_val_loss:
-- 0.22336530685424805
-- 0.2090274691581726
-- 0.18527281284332275
-- 0.20915909111499786
-- 0.27025991678237915
+- 0.22113201022148132
+- 0.19192928075790405
+- 0.20304016768932343
+- 0.19797971844673157
+- 0.26365381479263306
 min_val_scaled_mean_absolute_error:
-- 0.46149882674217224
-- 0.4658408463001251
-- 0.4380309283733368
-- 0.4883240759372711
-- 0.456197589635849
+- 0.45169585943222046
+- 0.4429832398891449
+- 0.4455697536468506
+- 0.4632081687450409
+- 0.46592286229133606
 min_val_scaled_root_mean_squared_error:
-- 0.6655771136283875
-- 0.6615913510322571
-- 0.6196939945220947
-- 0.6811450123786926
-- 0.6317007541656494
+- 0.6493441462516785
+- 0.6086452007293701
+- 0.6509482264518738
+- 0.6738059520721436
+- 0.6359040141105652
 model_class: make_model
 model_name: GCN
 model_version: 2023.09.30
 multi_target_indices: null
 number_histories: 5
 scaled_mean_absolute_error:
-- 0.04376649111509323
-- 0.06548744440078735
-- 0.05875982344150543
-- 0.07592686265707016
-- 0.0540020689368248
+- 0.06371089071035385
+- 0.06444712728261948
+- 0.0579991340637207
+- 0.07066909223794937
+- 0.05945947393774986
 scaled_root_mean_squared_error:
-- 0.12186748534440994
-- 0.15425890684127808
-- 0.1605674773454666
-- 0.16617444157600403
-- 0.15425477921962738
+- 0.13027626276016235
+- 0.15026293694972992
+- 0.16108879446983337
+- 0.1621994823217392
+- 0.163068026304245
 seed: 42
 time_list:
-- '0:02:41.250184'
-- '0:02:48.877327'
-- '0:02:45.440704'
-- '0:02:51.494117'
-- '0:02:54.385403'
+- '0:06:16.883915'
+- '0:06:34.529023'
+- '0:07:15.199847'
+- '0:10:23.172025'
+- '0:10:06.016519'
 val_loss:
-- 0.22336530685424805
-- 0.2319173663854599
-- 0.22124597430229187
-- 0.21199733018875122
-- 0.30064311623573303
+- 0.23328134417533875
+- 0.21415188908576965
+- 0.21731525659561157
+- 0.20350980758666992
+- 0.2738124430179596
 val_scaled_mean_absolute_error:
-- 0.463508278131485
-- 0.5030187964439392
-- 0.4640654921531677
-- 0.4997897446155548
-- 0.48311617970466614
+- 0.46819519996643066
+- 0.4745296835899353
+- 0.4473111033439636
+- 0.47483858466148376
+- 0.5058296918869019
 val_scaled_root_mean_squared_error:
-- 0.666396975517273
-- 0.7374665141105652
-- 0.674218475818634
-- 0.694538414478302
-- 0.7118412852287292
+- 0.6607564687728882
+- 0.6795703172683716
+- 0.674023449420929
+- 0.6981791853904724
+- 0.7606579065322876
diff --git a/training_core/results/ESOLDataset/GCN/GCN_hyper.json b/training_core/results/ESOLDataset/GCN/GCN_hyper.json
@@ -1 +1 @@
-{"model": {"class_name": "make_model", "module_name": "kgcnn.literature_core.GCN", "config": {"name": "GCN", "inputs": [{"shape": [null, 41], "name": "node_attributes", "dtype": "float32"}, {"shape": [null, 1], "name": "edge_weights", "dtype": "float32"}, {"shape": [null, 2], "name": "edge_indices", "dtype": "int64"}, {"shape": [], "name": "graph_size", "dtype": "int64"}, {"shape": [], "name": "edge_count", "dtype": "int64"}], "input_node_embedding": {"input_dim": 95, "output_dim": 64}, "input_edge_embedding": {"input_dim": 25, "output_dim": 1}, "gcn_args": {"units": 140, "use_bias": true, "activation": "relu"}, "depth": 5, "verbose": 10, "output_embedding": "graph", "output_mlp": {"use_bias": [true, true, false], "units": [140, 70, 1], "activation": ["relu", "relu", "linear"]}}}, "training": {"fit": {"batch_size": 32, "epochs": 800, "validation_freq": 10, "verbose": 2, "callbacks": [{"class_name": "kgcnn>LinearLearningRateScheduler", "config": {"learning_rate_start": 0.001, "learning_rate_stop": 5e-05, "epo_min": 250, "epo": 800, "verbose": 0}}]}, "compile": {"optimizer": {"class_name": "Adam", "config": {"learning_rate": 0.001}}, "loss": "mean_absolute_error", "jit_compile": false}, "cross_validation": {"class_name": "KFold", "config": {"n_splits": 5, "random_state": 42, "shuffle": true}}, "scaler": {"class_name": "StandardScaler", "config": {"with_std": true, "with_mean": true, "copy": true}}}, "dataset": {"class_name": "ESOLDataset", "module_name": "kgcnn.data.datasets.ESOLDataset", "config": {}, "methods": [{"set_attributes": {}}, {"map_list": {"method": "normalize_edge_weights_sym"}}]}, "data": {"data_unit": "mol/L"}, "info": {"postfix": "", "postfix_file": "", "kgcnn_version": "2.0.3"}}
+{"model": {"class_name": "make_model", "module_name": "kgcnn.literature_core.GCN", "config": {"name": "GCN", "inputs": [{"shape": [null, 41], "name": "node_attributes", "dtype": "float32"}, {"shape": [null, 1], "name": "edge_weights", "dtype": "float32"}, {"shape": [null, 2], "name": "edge_indices", "dtype": "int64"}, {"shape": [], "name": "graph_size", "dtype": "int64"}, {"shape": [], "name": "edge_count", "dtype": "int64"}], "input_node_embedding": {"input_dim": 95, "output_dim": 64}, "input_edge_embedding": {"input_dim": 25, "output_dim": 1}, "gcn_args": {"units": 140, "use_bias": true, "activation": "relu"}, "depth": 5, "verbose": 10, "output_embedding": "graph", "output_mlp": {"use_bias": [true, true, false], "units": [140, 70, 1], "activation": ["relu", "relu", "linear"]}}}, "training": {"fit": {"batch_size": 32, "epochs": 800, "validation_freq": 10, "verbose": 2, "callbacks": [{"class_name": "kgcnn>LinearLearningRateScheduler", "config": {"learning_rate_start": 0.001, "learning_rate_stop": 5e-05, "epo_min": 250, "epo": 800, "verbose": 0}}]}, "compile": {"optimizer": {"class_name": "Adam", "config": {"learning_rate": 0.001}}, "loss": "mean_absolute_error", "jit_compile": true}, "cross_validation": {"class_name": "KFold", "config": {"n_splits": 5, "random_state": 42, "shuffle": true}}, "scaler": {"class_name": "StandardScaler", "config": {"with_std": true, "with_mean": true, "copy": true}}}, "dataset": {"class_name": "ESOLDataset", "module_name": "kgcnn.data.datasets.ESOLDataset", "config": {}, "methods": [{"set_attributes": {}}, {"map_list": {"method": "normalize_edge_weights_sym"}}]}, "data": {"data_unit": "mol/L"}, "info": {"postfix": "", "postfix_file": "", "kgcnn_version": "2.0.3"}}