hiv_config.yaml

task_name: train.hiv
dataset_name: ogbg-molhiv
metrics: mse
use_cuda: True

warm_start_from: null 
model_path_for_infer: null 

# new:
mgf_file: "./dataset/ogbg_molhiv/mgf_feat.npy"  # null

# data config
num_class: null

# deeper GCN
model_config:
    pt_model_type: DeeperGCNModel
    num_layers: 7
    embed_dim: 256
    hidden_size: 256
    dropout_rate: 0.2
    graph_pool_type: average
    norm: "no_layer_norm"  #no_layer_norm
    appnp: True
    alpha: 0.2
    GN: True

# runconfig
epochs: 100
batch_size: 32
lr: 0.001   # 0.00001 too small
num_workers: 1
shuffle: True
shuffle_size: 0
save_steps: 1620000000000000
log_steps: 200
max_ckpt: 8
skip_steps: 0  # 前n个steps不打log
eval_steps: 1020 # 636 #81
eval_max_steps: 10000

# logger
stdout: True
log_dir: ./logs
log_filename: log.txt
save_dir: ./checkpoints
output_dir: ./outputs
files2saved: ["*.yaml", "*.py", "./models", "./utils"]