quantlab.model.predefined.rnn.GRURegressor

class quantlab.model.predefined.rnn.GRURegressor(config)

Bases: _QlibRNNHead

Qlib’s GRU benchmark model on (bar, symbol) samples.

RNNNet with a GRU, trained like LSTMRegressor. Hyperparameters, with Qlib’s Alpha158 benchmark values (workflow_config_gru_Alpha158.yaml) as defaults: window_bars (20), hidden_size (64), num_layers (2), dropout (0.0), batch_size (800), lr (2e-4), weight_decay (0.0), epochs (200) and early_stop (10). The known differences are LSTMRegressor’s.

Parameters:

configModelConfig

Factors, labels, dates and hyperparameters.

Examples

>>> GRURegressor(config).collect().train().name
'GRURegressor_total.pth'

DEFAULTS = {'batch_size': 800, 'dropout': 0.0, 'early_stop': 10, 'epochs': 200, 'hidden_size': 64, 'lr': 0.0002, 'num_layers': 2, 'weight_decay': 0.0, 'window_bars': 20}

Every unset hyperparameter’s value; a head gives Qlib’s benchmark values.

rnn_type = 'GRU'

The recurrent network the head builds.