train_dataset_speed.yaml

# 数据集
dataset:
  data_dir: data/dataset_s
  step_num_in: 12
  step_num_out: 12
  num_nodes: 307
  input_size: 1
  batch_size: 64
  device: cuda:0

AdapGLA:
  num_block: 2
  kernel_size: 5
  num_cheb_filter: 128
  num_time_filter: 128
  conv_type: cheb
  K: 3
  init_feature_num: 64
  adj_mx_path: data/dataset/adj_mx_distance.npy

AdapGLD:
  K: 3
  graph_num: 2
  rnn_hidden_size: 128
  rnn_layer_num: 1
  use_gc: True
  batch_first: True
  use_curriculum_learning: True
  cl_decay_steps: 2000
  activation: tanh
  adj_mx_path: data/dataset/adj_mx_distance.npy,data/dataset_s/adj_mx_corr.npy
  init_feature_num: 64

AdapGLT:
  K: 3
  rnn_hidden_size: 128
  rnn_layer_num: 1
  batch_first: True
  conv_type: cheb
  with_res: False
  num_block: 2
  num_cheb_filter: 128
  init_feature_num: 64
  adj_mx_path: data/dataset/adj_mx_distance.npy,data/dataset_s/adj_mx_corr.npy