-
Notifications
You must be signed in to change notification settings - Fork 5
/
custom.yaml
75 lines (75 loc) · 1.83 KB
/
custom.yaml
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
features:
scaling: s
normalization_range: [0, 1]
node_features: [shape,vertex,facet,count,min,max,sum,first,second]
edge_features: [shape,vertex,facet,count,min,max,sum]
node_normalization_feature: null
edge_normalization_feature: null
graph:
num_hops: 4 # outer hops (unsupervised) == num_layers
additional_num_hops: 1 # inner hops (supervised), this is there for having a supervised connected component, if set to 0, only the last layer of num_hops is supervised in the form of unconnected nodes
clique_sizes: [-1]
self_loops: 0
inference:
dataset: reconbench
classes: null
shapes_per_conf_per_class: 1
files: null
scan_confs: -1
batch_size: 0
per_layer: 1
has_label: 0
model: best
graph_cut: true
fix_orientation: true
metrics: []
export: ["mesh"]
validation:
dataset: reconbench
shapes_per_conf_per_class: 20
scan_confs: 4
batch_size: 0
per_layer: 1
val_every: 2
graph_cut: true
fix_orientation: true
metrics: []
training:
dataset: modelnet
classes: null
data_percentage: 0.8
shapes_per_conf_per_class: 10
scan_confs: 4
files: null
batch_size: 2048
epochs: 100
loss: kl
learning_rate: 0.005
adjust_lr_every: 24
load_epoch: 0
print_every: 1000
val_every: 2000
export_every: 10000
model:
type: sage
encoder: null
convs: [64, 128, 128, 128]
edge_convs: 1
decoder: 2
concatenate: 0 # whether to concatenate (or add) target to source node embedding
edge_prediction: 0
normalization: b
paths:
data: path/to/data/input/
out: data/models/kf96 # also where the model is stored
regularization:
cell_type: vol
cell_norm: null # [null, sqrt, log]
edge_type: null # edge regularization
edge_epoch: null # edge regularization
edge_weight: 0.4 # edge regularization
graph_cut:
unary_weight: 10
unary_type: logits
binary_weight: 10
binary_type: beta