Merge branch 'dev' into 152-make-tensor-flow-optional

nasa · Aug 7, 2024 · c236657 · c236657
2 parents 93f467e + 1bc324f
commit c236657
Showing 1 changed file with 2 additions and 11 deletions.
diff --git a/src/progpy/data_models/lstm_model.py b/src/progpy/data_models/lstm_model.py
@@ -449,8 +449,6 @@ def from_data(cls, inputs, outputs, event_states=None, t_met=None, **kwargs):
                 If early stopping is desired. Default is True
             early_stop.cfg (dict):
                 Configuration to pass into early stopping callback (if enabled). See keras documentation (https://keras.io/api/callbacks/early_stopping) for options. E.g., {'patience': 5}
-            workers (int):
-                Number of workers to use when training. One worker indicates no multiprocessing
 
         Returns:
             LSTMStateTransitionModel: Generated Model
@@ -470,7 +468,6 @@ def from_data(cls, inputs, outputs, event_states=None, t_met=None, **kwargs):
             'normalize': True,
             'early_stop': True,
             'early_stop.cfg': {'patience': 3, 'monitor': 'loss'},
-            'workers': 1
         }.copy()  # Copy is needed to avoid updating default
 
         params.update(LSTMStateTransitionModel.default_params)
@@ -508,10 +505,6 @@ def from_data(cls, inputs, outputs, event_states=None, t_met=None, **kwargs):
             raise TypeError(f"epochs must be an integer greater than 0, not {type(params['epochs'])}")
         if params['epochs'] < 1:
             raise ValueError(f"epochs must be greater than 0, got {params['epochs']}")
-        if not isinstance(params['workers'], int):
-            raise TypeError(f"workers must be positive integer, got {type(params['workers'])}")
-        if params['workers'] < 1:
-            raise ValueError(f"workers must be positive integer, got {params['workers']}")
         if np.isscalar(inputs):  # Is scalar (e.g., SimResult)
             inputs = [inputs]
         if np.isscalar(outputs):
@@ -592,17 +585,15 @@ def from_data(cls, inputs, outputs, event_states=None, t_met=None, **kwargs):
             output_data.append(t_all)
 
         model = keras.Model(inputs, outputs)
-        model.compile(optimizer="rmsprop", loss="mse", metrics=["mae"])
+        model.compile(optimizer="rmsprop", loss="mse", metrics=["mae"]*len(outputs))
 
         # Train model
         history = model.fit(
             u_all,
             output_data,
             epochs=params['epochs'],
             callbacks=callbacks,
-            validation_split=params['validation_split'],
-            workers=params['workers'],
-            use_multiprocessing=(params['workers'] > 1))
+            validation_split=params['validation_split'])
 
         # Split model into separate models
         n_state_layers = params['layers'] + 1 + (params['dropout'] > 0) + (params['normalize'])