builder/save-keras-archive-default (#50)

* save model using keras archive format by default

* remove extra lines

* turn off keras archive for older mods

CHANGES.rst

@@ -1,3 +1,12 @@
+1.1.0 (unreleased)
+==================
+
+new features
+------------
+
+- `architect.builder.Builder.save_model` uses preferred keras archive format by default [#50]
+
+
 1.0.1 (2024-04-03)
 ==================
 

spacekit/builder/architect.py

@@ -145,7 +145,7 @@ def load_pretrained_network(self, arch=None):
             self.log.error(err)
             sys.exit(1)
         model_src = "spacekit.builder.trained_networks"
-        archive_file = f"{arch}.zip"  # hst_cal.zip | jwt_cal.zip | svm_align.zip
+        archive_file = f"{arch}.zip"  # hst_cal.zip | jwst_cal.zip | svm_align.zip
         with importlib.resources.path(model_src, archive_file) as mod:
             self.model_path = mod
         if self.blueprint is None:
@@ -338,7 +338,16 @@ def set_callbacks(self, patience=15):
         self.callbacks = [checkpoint_cb, early_stopping_cb]
         return self.callbacks
 
-    def save_model(self, weights=True, output_path="."):
+    def save_keras_model(self, model_path):
+        dpath = os.path.dirname(model_path)
+        name = os.path.basename(model_path)
+        if not name.endswith("keras"):
+            name += ".keras"
+        keras_model_path = os.path.join(dpath, name)
+        self.model.save(keras_model_path)
+        self.model_path = keras_model_path
+
+    def save_model(self, weights=True, output_path=".", keras_archive=True):
         """The model architecture, and training configuration (including the optimizer, losses, and metrics)
         are stored in saved_model.pb. The weights are saved in the variables/ directory.
 
@@ -348,6 +357,8 @@ def save_model(self, weights=True, output_path="."):
             save weights learned by the model separately also, by default True
         output_path : str, optional
             where to save the model files, by default "."
+        keras_archive : bool, optional
+            save model using new (preferred) keras archive format, by default True
         """
         if self.name is None:
             self.name = str(self.model.name_scope().rstrip("/"))
@@ -357,16 +368,20 @@ def save_model(self, weights=True, output_path="."):
             model_name = self.name
 
         model_path = os.path.join(output_path, "models", model_name)
-        weights_path = f"{model_path}/weights/ckpt"
-        self.model.save(model_path)
-        if weights is True:
-            self.model.save_weights(weights_path)
-        for root, _, files in os.walk(model_path):
-            indent = "    " * root.count(os.sep)
-            print("{}{}/".format(indent, os.path.basename(root)))
-            for filename in files:
-                print("{}{}".format(indent + "    ", filename))
-        self.model_path = model_path
+
+        if keras_archive is True:
+            self.save_keras_model(model_path)
+        else:
+            self.model.save(model_path)
+            if weights is True:
+                weights_path = f"{model_path}/weights/ckpt"
+                self.model.save_weights(weights_path)
+            for root, _, files in os.walk(model_path):
+                indent = "    " * root.count(os.sep)
+                print("{}{}/".format(indent, os.path.basename(root)))
+                for filename in files:
+                    print("{}{}".format(indent + "    ", filename))
+            self.model_path = model_path
 
     def model_diagram(
         self,

spacekit/skopes/hst/svm/train.py

@@ -164,7 +164,7 @@ def load_ensemble_data(
 
 
 def train_ensemble(
-    XTR, YTR, XTS, YTS, model_name="ensembleSVM", params=None, output_path=None
+    XTR, YTR, XTS, YTS, model_name="ensembleSVM", params=None, output_path=None, keras=False,
 ):
     """Build, compile and fit an ensemble model with regression test data and image input arrays.
 
@@ -215,7 +215,7 @@ def train_ensemble(
     if output_path is None:
         output_path = os.getcwd()
     model_outpath = os.path.join(output_path, os.path.dirname(model_name))
-    ens.save_model(weights=True, output_path=model_outpath)
+    ens.save_model(weights=True, output_path=model_outpath, keras_archive=keras)
     return ens
 
 
@@ -267,6 +267,7 @@ def run_training(
     model_name="ensembleSVM",
     params=None,
     output_path=None,
+    keras=False,
 ):
     """Main calling function to load and prep the data, train the model, compute results and save to disk.
 
@@ -305,6 +306,7 @@ def run_training(
         model_name=model_name,
         params=params,
         output_path=output_path,
+        keras=keras,
     )
     com, val = compute_results(ens, tv_idx, val_set=(XVL, YVL), output_path=output_path)
     return ens, com, val