⬅ configs/datasets.py source

1 dataset_defaults = {
2 'ogb-molpcba': {
3 'num_domains': 120084,
4 'split_scheme': 'official',
  • E501 Line too long (89 > 79 characters)
5 'model_kwargs': {'three_wl_in_dim': 186, 'dropout': 0.5, 'dataset_group': 'mol'},
6 'default_frac': 1.0,
7 'loss_function': 'multitask_bce',
8 'groupby_fields': ['scaffold', ],
9 'val_metric': 'ap',
10 'val_metric_decreasing': False,
11 'optimizer': 'Adam',
12 'batch_size': 128,
13 'lr': 1e-03,
14 'weight_decay': 0.,
15 'n_epochs': 250,
16 'n_groups_per_batch': 4,
17 'no_group_logging': True,
18 'process_outputs_function': None,
19 'algo_log_metric': 'multitask_binary_accuracy',
20 'gsn_id_type': 'cycle_graph',
21 'gsn_k': 6,
22 'model': 'gin_virtual'
23 },
24 'ogb-molhiv': {
25 'num_domains': 19089,
26 'split_scheme': 'official',
  • E501 Line too long (89 > 79 characters)
27 'model_kwargs': {'three_wl_in_dim': 186, 'dropout': 0.5, 'dataset_group': 'mol'},
28 'default_frac': 1.0,
29 'loss_function': 'BCEWithLogitsLoss',
30 'groupby_fields': ['scaffold', ],
31 'val_metric': 'rocauc',
32 'val_metric_decreasing': False,
33 'optimizer': 'Adam',
34 'batch_size': 128,
35 'lr': 1e-03,
36 'weight_decay': 0.,
37 'n_epochs': 200,
38 'n_groups_per_batch': 4,
39 'no_group_logging': True,
40 'process_outputs_function': None,
41 'algo_log_metric': 'binary_accuracy',
42 'gsn_id_type': 'cycle_graph',
43 'gsn_k': 6,
44 'model': 'gin_virtual',
45 'flag_step_size': 1e-3,
46 'gcl_aug_ratio': 0.3
47 },
48 'ogbg-ppa': {
49 'num_domains': 1581, # not mapped
50 'split_scheme': 'official',
  • E501 Line too long (87 > 79 characters)
51 'model_kwargs': {'three_wl_in_dim': 7, 'dropout': 0.5, 'dataset_group': 'ppa'},
52 'default_frac': 1.0,
53 'loss_function': 'cross_entropy',
54 'groupby_fields': ['species', ],
55 'val_metric': 'acc',
56 'val_metric_decreasing': False,
57 'optimizer': 'Adam',
58 'batch_size': 32,
59 'lr': 1e-03,
60 'weight_decay': 0.,
61 'n_epochs': 150,
62 'n_groups_per_batch': 4,
63 'no_group_logging': True,
64 'process_outputs_function': None,
65 'algo_log_metric': 'multiclass_accuracy',
66 'gsn_id_type': 'cycle_graph',
67 'gsn_k': 6,
68 'model': 'gin_virtual',
69 'flag_step_size': 1e-3
70 },
71 'RotatedMNIST': {
72 'num_domains': 6,
73 'num_train_domains': 5,
74 'split_scheme': 'official',
  • E501 Line too long (96 > 79 characters)
75 'model_kwargs': {'three_wl_in_dim': 1, 'dropout': 0.5, 'dataset_group': 'RotatedMNIST'},
76 'default_frac': 1.0,
77 'loss_function': 'cross_entropy',
78 'groupby_fields': ['angle', ],
79 'val_metric': 'acc',
80 'val_metric_decreasing': False,
81 'optimizer': 'Adam',
82 'batch_size': 128,
83 'lr': 1e-03,
84 'weight_decay': 0.,
85 'n_epochs': 150,
86 'n_groups_per_batch': 4,
87 'no_group_logging': True,
88 'process_outputs_function': None,
89 'algo_log_metric': 'multiclass_accuracy',
90 'gsn_id_type': 'cycle_graph',
91 'gsn_k': 6,
92 'model': 'cheb'
93 },
94 'ColoredMNIST': {
95 'num_domains': 3,
  • E231 Missing whitespace after ':'
96 'num_train_domains':2,
97 'split_scheme': 'official',
  • E231 Missing whitespace after ':'
  • E501 Line too long (95 > 79 characters)
98 'model_kwargs': {'three_wl_in_dim':2, 'dropout': 0.5, 'dataset_group': 'ColoredMNIST'},
99 'default_frac': 1.0,
100 'loss_function': 'BCEWithLogitsLoss',
101 'groupby_fields': ['color', ],
102 'val_metric': 'rocauc',
103 'val_metric_decreasing': False,
104 'optimizer': 'Adam',
105 'batch_size': 128,
106 'lr': 1e-03,
107 'weight_decay': 0.,
108 'n_epochs': 100,
109 'n_groups_per_batch': 2,
110 'no_group_logging': True,
111 'process_outputs_function': None,
112 'algo_log_metric': 'binary_accuracy',
113 'model': 'cheb',
114 'irm_lambda': 100.
115 },
116 'SBM-Environment': {
117 'num_domains': 4,
  • E231 Missing whitespace after ':'
118 'num_train_domains':3,
119 'split_scheme': 'official',
  • E501 Line too long (87 > 79 characters)
120 'model_kwargs': {'three_wl_in_dim': 8, 'dropout': 0.5, 'dataset_group': 'SBM'},
121 'default_frac': 1.0,
122 'loss_function': 'cross_entropy',
123 'groupby_fields': ['composition', ],
124 'val_metric': 'acc',
125 'val_metric_decreasing': False,
126 'optimizer': 'Adam',
127 'batch_size': 128,
128 'lr': 1e-03,
129 'weight_decay': 0.,
130 'n_epochs': 200,
131 'n_groups_per_batch': 2,
132 'no_group_logging': True,
133 'process_outputs_function': None,
134 'algo_log_metric': 'multiclass_accuracy',
135 'gsn_id_type': 'cycle_graph',
136 'gsn_k': 6,
137 'model': 'gin_virtual',
138 'dann_lambda': 1,
139 'mldg_beta': 10,
140 'gcl_aug_ratio': 0.2
141 },
142 'SBM-Isolation': {
143 'num_domains': 10,
  • E231 Missing whitespace after ':'
144 'num_train_domains':8,
145 'split_scheme': 'official',
  • E501 Line too long (87 > 79 characters)
146 'model_kwargs': {'three_wl_in_dim': 8, 'dropout': 0.5, 'dataset_group': 'SBM'},
147 'default_frac': 1.0,
148 'loss_function': 'cross_entropy',
149 'groupby_fields': ['composition', ],
150 'val_metric': 'acc',
151 'val_metric_decreasing': False,
152 'optimizer': 'Adam',
153 'batch_size': 128,
154 'lr': 1e-03,
155 'weight_decay': 0.,
156 'n_epochs': 150,
157 'n_groups_per_batch': 4,
158 'no_group_logging': True,
159 'process_outputs_function': None,
160 'algo_log_metric': 'multiclass_accuracy',
161 'gsn_id_type': 'cycle_graph',
162 'gsn_k': 6,
163 'model': 'cheb',
164 'dann_lambda': 1,
165 'mldg_beta': 1,
166 'gcl_aug_ratio': 0.3
167 },
168 'UPFD': {
169 'num_domains': 10,
  • E231 Missing whitespace after ':'
170 'num_train_domains':8,
171 'split_scheme': 'official',
  • E501 Line too long (88 > 79 characters)
172 'model_kwargs': {'three_wl_in_dim': 8, 'dropout': 0.5, 'dataset_group': 'UPFD'},
173 'default_frac': 1.0,
174 'loss_function': 'BCEWithLogitsLoss',
175 'groupby_fields': ['size', ],
176 'val_metric': 'rocauc',
177 'val_metric_decreasing': False,
178 'optimizer': 'Adam',
179 'batch_size': 128,
180 'lr': 1e-03,
181 'weight_decay': 0.,
182 'n_epochs': 150,
183 'n_groups_per_batch': 4,
184 'no_group_logging': True,
185 'process_outputs_function': None,
186 'algo_log_metric': 'binary_accuracy',
187 'gsn_id_type': 'cycle_graph',
188 'gsn_k': 6,
189 'model': 'cheb'
190 },
191 }