update for keras 3.0

kgcnn/literature/GAT/_make.py

@@ -76,11 +76,13 @@ def make_model(inputs: list = None,
     Model inputs:
     Model uses the list template of inputs and standard output template.
     The supported inputs are  :obj:`[nodes, edges, edge_indices, ...]`
-    with '...' indicating mask or id tensors following the template below:
+    with '...' indicating mask or ID tensors following the template below:
+
     %s
 
     Model outputs:
     The standard output template:
+
     %s
 
 

training/results/MatProjectJdft2dDataset/PAiNN_make_crystal_model/PAiNN_MatProjectJdft2dDataset_score.yaml

@@ -0,0 +1,154 @@
+OS: nt_win32
+backend: tensorflow
+cuda_available: 'False'
+data_unit: meV/atom
+date_time: '2023-11-15 15:22:58'
+device_id: '[LogicalDevice(name=''/device:CPU:0'', device_type=''CPU'')]'
+device_memory: '[]'
+device_name: '[{}]'
+epochs:
+- 800
+- 800
+- 800
+- 800
+- 800
+execute_folds: null
+kgcnn_version: 4.0.0
+learning_rate:
+- 2.1409230612334795e-05
+- 2.1409230612334795e-05
+- 2.1409230612334795e-05
+- 2.1409230612334795e-05
+- 2.1409230612334795e-05
+loss:
+- 0.01342229824513197
+- 0.01582755520939827
+- 0.015029882080852985
+- 0.016817400231957436
+- 0.014265685342252254
+max_learning_rate:
+- 9.999999747378752e-05
+- 9.999999747378752e-05
+- 9.999999747378752e-05
+- 9.999999747378752e-05
+- 9.999999747378752e-05
+max_loss:
+- 0.5201069712638855
+- 0.5315175652503967
+- 0.451824426651001
+- 0.5210411548614502
+- 0.4851284623146057
+max_scaled_mean_absolute_error:
+- 75.4897232055664
+- 74.12128448486328
+- 61.5352897644043
+- 67.2265625
+- 57.95024871826172
+max_scaled_root_mean_squared_error:
+- 149.78330993652344
+- 143.6664276123047
+- 140.2177734375
+- 131.99143981933594
+- 122.83416748046875
+max_val_loss:
+- 0.2821546494960785
+- 0.3245273530483246
+- 0.3746875822544098
+- 0.5050886273384094
+- 0.618653416633606
+max_val_scaled_mean_absolute_error:
+- 40.90802001953125
+- 45.22543716430664
+- 51.08058166503906
+- 65.11273956298828
+- 73.94288635253906
+max_val_scaled_root_mean_squared_error:
+- 84.96401977539062
+- 111.83113861083984
+- 129.0309600830078
+- 159.9274444580078
+- 187.96551513671875
+min_learning_rate:
+- 9.99999993922529e-09
+- 9.99999993922529e-09
+- 9.99999993922529e-09
+- 9.99999993922529e-09
+- 9.99999993922529e-09
+min_loss:
+- 0.013281172141432762
+- 0.01350078172981739
+- 0.01428682915866375
+- 0.013030022382736206
+- 0.011631934903562069
+min_scaled_mean_absolute_error:
+- 1.9360378980636597
+- 1.8882485628128052
+- 1.9466769695281982
+- 1.6665455102920532
+- 1.3933351039886475
+min_scaled_root_mean_squared_error:
+- 9.1520357131958
+- 7.898106575012207
+- 9.539583206176758
+- 9.564427375793457
+- 6.542611122131348
+min_val_loss:
+- 0.21616201102733612
+- 0.2801247239112854
+- 0.3179992735385895
+- 0.44068655371665955
+- 0.5248962044715881
+min_val_scaled_mean_absolute_error:
+- 31.34011459350586
+- 39.01124572753906
+- 43.35005569458008
+- 56.815731048583984
+- 62.80753707885742
+min_val_scaled_root_mean_squared_error:
+- 68.39794158935547
+- 100.04585266113281
+- 115.25602722167969
+- 143.70919799804688
+- 162.13404846191406
+model_class: make_crystal_model
+model_name: PAiNN
+model_version: '2023-10-04'
+multi_target_indices: null
+number_histories: 5
+scaled_mean_absolute_error:
+- 1.9560686349868774
+- 2.2146451473236084
+- 2.046118974685669
+- 2.1549291610717773
+- 1.7081328630447388
+scaled_root_mean_squared_error:
+- 9.223356246948242
+- 8.082352638244629
+- 9.63218879699707
+- 9.677685737609863
+- 6.588728427886963
+seed: 42
+time_list:
+- '0:31:53.683697'
+- '0:33:27.269425'
+- '1:00:06.044732'
+- '1:13:54.304826'
+- '1:49:35.146251'
+val_loss:
+- 0.2428491860628128
+- 0.29850465059280396
+- 0.3226911723613739
+- 0.4874231219291687
+- 0.5294967293739319
+val_scaled_mean_absolute_error:
+- 35.20934295654297
+- 41.602272033691406
+- 43.98591613769531
+- 62.783592224121094
+- 63.3631706237793
+val_scaled_root_mean_squared_error:
+- 80.62264251708984
+- 101.75128936767578
+- 115.29129791259766
+- 147.98414611816406
+- 162.89393615722656

training/results/MatProjectJdft2dDataset/PAiNN_make_crystal_model/PAiNN_hyper.json

@@ -0,0 +1 @@
+{"model": {"module_name": "kgcnn.literature.PAiNN", "class_name": "make_crystal_model", "config": {"name": "PAiNN", "inputs": [{"shape": [null], "name": "node_number", "dtype": "int64", "ragged": true}, {"shape": [null, 3], "name": "node_coordinates", "dtype": "float32", "ragged": true}, {"shape": [null, 2], "name": "range_indices", "dtype": "int64", "ragged": true}, {"shape": [null, 3], "name": "range_image", "dtype": "int64", "ragged": true}, {"shape": [3, 3], "name": "graph_lattice", "dtype": "float32", "ragged": false}], "input_tensor_type": "ragged", "cast_disjoint_kwargs": {}, "input_node_embedding": {"input_dim": 95, "output_dim": 128}, "equiv_initialize_kwargs": {"dim": 3, "method": "eye"}, "bessel_basis": {"num_radial": 20, "cutoff": 5.0, "envelope_exponent": 5}, "pooling_args": {"pooling_method": "scatter_mean"}, "conv_args": {"units": 128, "cutoff": null, "conv_pool": "scatter_sum"}, "update_args": {"units": 128}, "depth": 2, "verbose": 10, "equiv_normalization": false, "node_normalization": false, "output_embedding": "graph", "output_mlp": {"use_bias": [true, true], "units": [128, 1], "activation": ["swish", "linear"]}}}, "training": {"cross_validation": {"class_name": "KFold", "config": {"n_splits": 5, "random_state": 42, "shuffle": true}}, "fit": {"batch_size": 32, "epochs": 800, "validation_freq": 10, "verbose": 2, "callbacks": [{"class_name": "kgcnn>LinearWarmupLinearLearningRateScheduler", "config": {"learning_rate_start": 0.0001, "learning_rate_stop": 1e-06, "epo_warmup": 25, "epo": 1000, "verbose": 0}}]}, "compile": {"optimizer": {"class_name": "Adam", "config": {"learning_rate": 0.0001}}, "loss": "mean_absolute_error"}, "scaler": {"class_name": "StandardLabelScaler", "module_name": "kgcnn.data.transform.scaler.standard", "config": {"with_std": true, "with_mean": true, "copy": true}}, "multi_target_indices": null}, "data": {"data_unit": "meV/atom"}, "info": {"postfix": "", "postfix_file": "", "kgcnn_version": "4.0.0"}, "dataset": {"class_name": "MatProjectJdft2dDataset", "module_name": "kgcnn.data.datasets.MatProjectJdft2dDataset", "config": {}, "methods": [{"map_list": {"method": "set_range_periodic", "max_distance": 5.0}}, {"map_list": {"method": "count_nodes_and_edges", "total_edges": "total_ranges", "count_edges": "range_indices", "count_nodes": "node_number", "total_nodes": "total_nodes"}}]}}

training/results/README.md

@@ -138,6 +138,7 @@ Materials Project dataset from Matbench with 636 crystal structures and their co
 
 | model                     | kgcnn   |   epochs | MAE [meV/atom]           | RMSE [meV/atom]           |
 |:--------------------------|:--------|---------:|:-------------------------|:--------------------------|
+| PAiNN.make_crystal_model  | 4.0.0   |      800 | 49.3889 ± 11.5376     | 121.7087 ± 30.0472     |
 | Schnet.make_crystal_model | 4.0.0   |      800 | **45.2412 ± 11.6395** | **115.6890 ± 39.0929** |
 
 #### MatProjectLogGVRHDataset