basic: dataset: PEMSD8 device: cuda:0 mode: train model: STFGNN seed: 2023 data: batch_size: 64 column_wise: false days_per_week: 7 horizon: 12 input_dim: 1 lag: 12 normalizer: std num_nodes: 170 steps_per_day: 288 test_ratio: 0.2 val_ratio: 0.2 model: activation: GLU first_layer_embedding_size: 64 hidden_dims: [[64, 64, 64], [64, 64, 64], [64, 64, 64]] horizon: 12 input_dim: 1 mask: None out_layer_dim: 128 output_dim: 1 spatial_emb: true temporal_emb: true window: 12 train: batch_size: 64 debug: false early_stop: true early_stop_patience: 15 epochs: 300 grad_norm: false log_step: 200 loss_func: mae lr_decay: false lr_decay_rate: 0.3 lr_decay_step: 5,20,40,70 lr_init: 0.003 mae_thresh: mape_thresh: 0.0 max_grad_norm: 5 output_dim: 1 plot: false real_value: true seed: 10 weight_decay: 0