nodepred_cora_gat.yaml 1.15 KB
Newer Older
Jinjing Zhou's avatar
Jinjing Zhou committed
1
# Accuracy across 10 runs: 0.8208 ± 0.00663
2
version: 0.0.2
Jinjing Zhou's avatar
Jinjing Zhou committed
3
pipeline_name: nodepred
4
pipeline_mode: train
Jinjing Zhou's avatar
Jinjing Zhou committed
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
device: cuda:0
data:
  name: cora
model:
  name: gat
  embed_size: -1              # The dimension of created embedding table. -1 means using original node embedding
  num_layers: 2               # Number of layers.
  hidden_size: 8              # Hidden size.
  heads:
  - 8
  - 1
  activation: elu             # Activation function.
  feat_drop: 0.6              # Dropout rate for features.
  attn_drop: 0.6              # Dropout rate for attentions.
  negative_slope: 0.2
  residual: false             # If true, the GATConv will use residule connection
general_pipeline:
  early_stop:
    patience: 100             # Steps before early stop
    checkpoint_path: checkpoint.pth # Early stop checkpoint model file path
  num_epochs: 200             # Number of training epochs
  eval_period: 5              # Interval epochs between evaluations
  optimizer:
    name: Adam
    lr: 0.005
    weight_decay: 0.0005
  loss: CrossEntropyLoss
32
  save_path: "results"        # Directory to save the experiment results
Jinjing Zhou's avatar
Jinjing Zhou committed
33
  num_runs: 10                # Number of experiments to run