def model_fn(features, labels, mode):
model = plx.models.Regressor(mode,
graph_fn=graph_fn,
loss=MeanSquaredErrorConfig(),
optimizer=SGDConfig(learning_rate=0.001),
summaries='all')
return model(features, labels)
def model_fn(features, labels, mode):
model = plx.models.Regressor(
mode,
graph_fn=graph_fn,
loss=AbsoluteDifferenceConfig(),
optimizer=SGDConfig(learning_rate=0.5, decay_type='exponential_decay', decay_steps=10),
summaries='all',
name='xor')
return model(features, labels)
def model_fn(features, labels, mode):
model = plx.models.DDQNModel(
mode,
graph_fn=graph_fn,
loss=HuberLossConfig(),
num_states=env.num_states,
num_actions=env.num_actions,
optimizer=SGDConfig(learning_rate=0.01),
exploration_config=DecayExplorationConfig(),
target_update_frequency=10,
summaries='all')
return model(features, labels)
def test_sgd_config(self):
config_dict = {
'learning_rate': 0.001,
'decay_type': "",
'decay_rate': 0.,
'decay_steps': 100,
'start_decay_at': 0,
'stop_decay_at': 1e10,
'min_learning_rate': 1e-12,
'staircase': False,
'global_step': None,
'use_locking': False,
'name': 'optimizer'
}
config = SGDConfig.from_dict(config_dict)
assert_equal_dict(config.to_dict(), config_dict)