Add minibatch_size hyperparameter

downloads-generation/models_class1/hyperparameters.yaml

@@ -8,9 +8,10 @@
 # OPTIMIZATION
 ##########################################
 "max_epochs": 500,
-"patience": 10,
+"patience": 20,
 "early_stopping": true,
 "validation_split": 0.2,
+"minibatch_size": 128,
 
 ##########################################
 # RANDOM NEGATIVE PEPTIDES

mhcflurry/class1_affinity_prediction/class1_neural_network.py

@@ -50,11 +50,6 @@ class Class1NeuralNetwork(object):
         batch_normalization=False,
         embedding_init_method="glorot_uniform",
         locally_connected_layers=[
-            {
-                "filters": 8,
-                "activation": "tanh",
-                "kernel_size": 3
-            },
             {
                 "filters": 8,
                 "activation": "tanh",
@@ -77,6 +72,7 @@ class Class1NeuralNetwork(object):
         take_best_epoch=False,  # currently unused
         validation_split=0.2,
         early_stopping=True,
+        minibatch_size=128,
         random_negative_rate=0.0,
         random_negative_constant=25,
         random_negative_affinity_min=20000.0,
@@ -447,6 +443,7 @@ class Class1NeuralNetwork(object):
                 x_dict_with_random_negatives,
                 y_dict_with_random_negatives,
                 shuffle=True,
+                batch_size=self.hyperparameters['minibatch_size'],
                 verbose=verbose,
                 epochs=1,
                 validation_split=self.hyperparameters['validation_split'],