renamed to have _keras2 for CANDLE

Author: brettinanl

examples/xform-smiles/sct_baseline_keras2.py

@@ -0,0 +1,75 @@
+# Setup
+
+import os
+import sys
+# import gzip
+
+# import math
+# import matplotlib
+# matplotlib.use('Agg')
+
+# import matplotlib.pyplot as plt
+
+from tensorflow.keras import backend as K
+from tensorflow.keras.optimizers import Adam  # RMSprop, SGD
+from tensorflow.keras.callbacks import ModelCheckpoint, CSVLogger, ReduceLROnPlateau, EarlyStopping
+
+file_path = os.path.dirname(os.path.realpath(__file__))
+lib_path = os.path.abspath(os.path.join(file_path, '..', '..', 'common'))
+sys.path.append(lib_path)
+
+import candle
+import smiles_transformer as st
+
+
+def initialize_parameters(default_model='class_default_model.txt'):
+
+    # Build benchmark object
+    sctBmk = st.BenchmarkST(st.file_path, default_model, 'keras',
+                            prog='p1b1_baseline',
+                            desc='Multi-task (DNN) for data extraction from clinical reports - Pilot 3 Benchmark 1')
+
+    # Initialize parameters
+    gParameters = candle.finalize_parameters(sctBmk)
+
+    return gParameters
+
+
+# Train and Evaluate
+
+def run(params):
+
+    x_train, y_train, x_val, y_val = st.load_data(params)
+
+    model = st.transformer_model(params)
+
+    model.compile(loss='sparse_categorical_crossentropy',
+                  optimizer=Adam(lr=0.000001),
+                  metrics=['accuracy'])
+
+    # set up a bunch of callbacks to do work during model training..
+
+    checkpointer = ModelCheckpoint(filepath='smile_class.autosave.model.h5', verbose=1, save_weights_only=True, save_best_only=True)
+    csv_logger = CSVLogger('smile_class.training.log')
+    reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.75, patience=20, verbose=1, mode='auto', epsilon=0.0001, cooldown=3, min_lr=0.000000001)
+    early_stop = EarlyStopping(monitor='val_loss', patience=100, verbose=1, mode='auto')
+
+    history = model.fit(x_train, y_train,
+                        batch_size=params['batch_size'],
+                        epochs=params['epochs'],
+                        verbose=1,
+                        validation_data=(x_val, y_val),
+                        callbacks=[checkpointer, csv_logger, reduce_lr, early_stop])
+
+    model.load_weights('smile_class.autosave.model.h5')
+
+
+def main():
+    params = initialize_parameters()
+    run(params)
+
+
+if __name__ == '__main__':
+    main()
+    if K.backend() == 'tensorflow':
+        K.clear_session()

examples/xform-smiles/srt_baseline_keras2.py

@@ -0,0 +1,98 @@
+# Setup
+
+import os
+import sys
+# import gzip
+
+# import math
+# import matplotlib
+# matplotlib.use('Agg')
+
+# import matplotlib.pyplot as plt
+
+from tensorflow.keras import backend as K
+import tensorflow.keras.optimizers as optimizers
+from tensorflow.keras.callbacks import ModelCheckpoint, CSVLogger, ReduceLROnPlateau, EarlyStopping
+
+file_path = os.path.dirname(os.path.realpath(__file__))
+lib_path = os.path.abspath(os.path.join(file_path, '..', '..', 'common'))
+sys.path.append(lib_path)
+
+import candle
+import smiles_transformer as st
+
+import tensorflow.config.experimental
+gpus = tensorflow.config.experimental.list_physical_devices('GPU')
+try:
+    for gpu in gpus:
+        print("setting memory growth")
+        tensorflow.config.experimental.set_memory_growth(gpu, True)
+except RuntimeError as e:
+    print(e)
+
+
+def initialize_parameters(default_model='regress_default_model.txt'):
+
+    # Build benchmark object
+    sctBmk = st.BenchmarkST(st.file_path, default_model, 'keras',
+                            prog='p1b1_baseline',
+                            desc='Multi-task (DNN) for data extraction from clinical reports - Pilot 3 Benchmark 1')
+
+    # Initialize parameters
+    gParameters = candle.finalize_parameters(sctBmk)
+
+    return gParameters
+
+
+# Train and Evaluate
+
+def run(params):
+
+    x_train, y_train, x_val, y_val = st.load_data(params)
+
+    model = st.transformer_model(params)
+
+    optimizer = optimizers.deserialize({'class_name': params['optimizer'], 'config': {}})
+
+    # I don't know why we set base_lr. It doesn't appear to be used.
+    if 'base_lr' in params and params['base_lr'] > 0:
+        base_lr = params['base_lr']
+    else:
+        base_lr = K.get_value(optimizer.lr)
+
+    if 'learning_rate' in params and params['learning_rate'] > 0:
+        K.set_value(optimizer.lr, params['learning_rate'])
+        print('Done setting optimizer {} learning rate to {}'.format(
+            params['optimizer'],params['learning_rate']))
+
+    model.compile(loss='mean_squared_error',
+                  optimizer=optimizer,
+                  metrics=['mae', st.r2])
+
+    # set up a bunch of callbacks to do work during model training..
+
+    checkpointer = ModelCheckpoint(filepath='smile_regress.autosave.model.h5', verbose=1, save_weights_only=True, save_best_only=True)
+    csv_logger = CSVLogger('smile_regress.training.log')
+    reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.75, patience=20, verbose=1, mode='auto', epsilon=0.0001, cooldown=3, min_lr=0.000000001)
+    early_stop = EarlyStopping(monitor='val_loss', patience=100, verbose=1, mode='auto')
+
+    history = model.fit(x_train, y_train,
+                        batch_size=params['batch_size'],
+                        epochs=params['epochs'],
+                        verbose=1,
+                        validation_data=(x_val, y_val),
+                        callbacks=[checkpointer, csv_logger, reduce_lr, early_stop])
+
+    model.load_weights('smile_regress.autosave.model.h5')
+
+    return history
+
+def main():
+    params = initialize_parameters()
+    run(params)
+
+
+if __name__ == '__main__':
+    main()
+    if K.backend() == 'tensorflow':
+        K.clear_session()