0.23.6

get_tuned_architecture() simplified

Author: bartzbeielstein

notebooks/00_spotPython_tests.ipynb

@@ -8142,7 +8142,18 @@
      "text": [
       "module_name: light\n",
       "submodule_name: regression\n",
-      "model_name: NNLinearRegressor\n"
+      "model_name: NNLinearRegressor\n",
+      "Experiment saved to get_one_config_from_X_exp.pkl\n",
+      "Result file get_one_config_from_X_res.pkl exists. Loading the result.\n",
+      "Loaded experiment from get_one_config_from_X_res.pkl\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/Users/bartz/miniforge3/envs/spot312/lib/python3.12/site-packages/torch/storage.py:414: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.\n",
+      "  return torch.load(io.BytesIO(b))\n"
      ]
     },
     {
@@ -8153,8 +8164,8 @@
        " 'batch_size': 2,\n",
        " 'act_fn': LeakyReLU(),\n",
        " 'optimizer': 'SGD',\n",
-       " 'dropout_prob': 0.03480567,\n",
-       " 'lr_mult': 7.98104224,\n",
+       " 'dropout_prob': 0.034805674424520705,\n",
+       " 'lr_mult': 7.981042243396318,\n",
        " 'patience': 2,\n",
        " 'batch_norm': True,\n",
        " 'initialization': 'Default'}"
@@ -8172,7 +8183,7 @@
     "from spotpython.utils.init import fun_control_init, design_control_init\n",
     "from spotpython.spot import Spot\n",
     "import numpy as np\n",
-    "from spotpython.hyperparameters.values import set_hyperparameter, get_one_config_from_X\n",
+    "from spotpython.hyperparameters.values import set_hyperparameter, get_tuned_architecture\n",
     "\n",
     "\n",
     "fun_control = fun_control_init(\n",
@@ -8187,16 +8198,14 @@
     "    _L_in=10,\n",
     "    _L_out=1)\n",
     "\n",
-    "# set_hyperparameter(fun_control, \"epochs\", [2,3])\n",
-    "# set_hyperparameter(fun_control, \"patience\", [1,2])\n",
-    "# design_control = design_control_init(init_size=5)\n",
+    "set_hyperparameter(fun_control, \"epochs\", [2,2])\n",
+    "set_hyperparameter(fun_control, \"patience\", [1,2])\n",
+    "design_control = design_control_init(init_size=5)\n",
     "\n",
-    "# fun = HyperLight().fun\n",
-    "# S = Spot(fun=fun,fun_control=fun_control, design_control=design_control)\n",
-    "# S.run()\n",
-    "# X = S.to_all_dim(S.min_X.reshape(1, -1))\n",
-    "X = np.array([[ 4., 2., 1., 3.,  11., 0.03480567, 7.98104224,  1., 1., 0.]])\n",
-    "get_one_config_from_X(X, fun_control)"
+    "fun = HyperLight().fun\n",
+    "S = Spot(fun=fun,fun_control=fun_control, design_control=design_control)\n",
+    "S.run()\n",
+    "get_tuned_architecture(S)"
    ]
   },
   {

pyproject.toml

@@ -7,7 +7,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "spotpython"
-version = "0.23.5"
+version = "0.23.6"
 authors = [
   { name="T. Bartz-Beielstein", email="tbb@bartzundbartz.de" }
 ]

src/spotpython/hyperparameters/values.py

@@ -464,7 +464,9 @@ def get_tuned_architecture(spot_tuner, force_minX=False) -> dict:
             from spotpython.utils.init import fun_control_init, design_control_init
             from spotpython.spot import Spot
             import numpy as np
-            from spotpython.hyperparameters.values import set_hyperparameter, get_one_config_from_X
+            from spotpython.hyperparameters.values import set_hyperparameter, get_tuned_architecture
+
+
             fun_control = fun_control_init(
                 force_run=False,
                 PREFIX="get_one_config_from_X",
@@ -476,24 +478,22 @@ def get_tuned_architecture(spot_tuner, force_minX=False) -> dict:
                 hyperdict=LightHyperDict,
                 _L_in=10,
                 _L_out=1)
-            # Uncomment the following lines to perform the hyperparameter optimization:
-            # set_hyperparameter(fun_control, "epochs", [2,3])
-            # set_hyperparameter(fun_control, "patience", [1,2])
-            # design_control = design_control_init(init_size=5)
-            # fun = HyperLight().fun
-            # S = Spot(fun=fun,fun_control=fun_control, design_control=design_control)
-            # S.run()
-            # X = S.to_all_dim(S.min_X.reshape(1, -1))
-            # Otherwise, only take the result:
-            X = np.array([[ 4., 2., 1., 3.,  11., 0.03480567, 7.98104224,  1., 1., 0.]])
-            get_one_config_from_X(X, fun_control)
-               {'l1': 16,
+
+            set_hyperparameter(fun_control, "epochs", [2,3])
+            set_hyperparameter(fun_control, "patience", [1,2])
+            design_control = design_control_init(init_size=5)
+
+            fun = HyperLight().fun
+            S = Spot(fun=fun,fun_control=fun_control, design_control=design_control)
+            S.run()
+            get_tuned_architecture(S)
+                {'l1': 16,
                 'epochs': 4,
                 'batch_size': 2,
                 'act_fn': LeakyReLU(),
                 'optimizer': 'SGD',
-                'dropout_prob': 0.03480567,
-                'lr_mult': 7.98104224,
+                'dropout_prob': 0.034805674424520705,
+                'lr_mult': 7.981042243396318,
                 'patience': 2,
                 'batch_norm': True,
                 'initialization': 'Default'}
@@ -503,7 +503,7 @@ def get_tuned_architecture(spot_tuner, force_minX=False) -> dict:
     else:
         # noise or force_minX is False:
         X = spot_tuner.to_all_dim(spot_tuner.min_mean_X.reshape(1, -1))
-        fun_control = copy(spot_tuner.fun_control)
+    fun_control = copy.copy(spot_tuner.fun_control)
     config = get_one_config_from_X(X, fun_control)
     return config
 

test/test_get_tuned_architecture.py

@@ -1,39 +1,32 @@
 from spotpython.data.diabetes import Diabetes
 from spotpython.hyperdict.light_hyper_dict import LightHyperDict
 from spotpython.fun.hyperlight import HyperLight
-from spotpython.utils.init import fun_control_init
+from spotpython.utils.init import fun_control_init, design_control_init
 from spotpython.spot import Spot
 import numpy as np
-from spotpython.hyperparameters.values import get_one_config_from_X
+from spotpython.hyperparameters.values import set_hyperparameter, get_tuned_architecture
 
 
 def test_get_tuned_architecture():
     fun_control = fun_control_init(
-        force_run=False,
+        force_run=True,
         PREFIX="get_one_config_from_X",
-        save_experiment=True,
-        fun_evals=10,
+        fun_evals=5,
         max_time=1,
         data_set = Diabetes(),
         core_model_name="light.regression.NNLinearRegressor",
         hyperdict=LightHyperDict,
         _L_in=10,
         _L_out=1)
-    X = np.array([[ 4., 2., 1., 3.,  11., 0.03480567, 7.98104224,  1., 1., 0.]])
-    values = get_one_config_from_X(X, fun_control)
-    
-    # Replace values['act_fn'] with the class name, i.e. .__class__.__name__ to get the class name
-    values['act_fn'] = values['act_fn'].__class__.__name__
-    # This is necessary for this comparison, because the act_fn is a class object and not a string
+
+    set_hyperparameter(fun_control, "epochs", [2,2])
+    set_hyperparameter(fun_control, "patience", [1,1])
+    design_control = design_control_init(init_size=4)
+
+    fun = HyperLight().fun
+    S = Spot(fun=fun,fun_control=fun_control, design_control=design_control)
+    S.run()
+    conf = get_tuned_architecture(S)
 
-    expected = {'l1': 16,
-                'epochs': 4,
-                'batch_size': 2,
-                'act_fn': 'LeakyReLU',
-                'optimizer': 'SGD',
-                'dropout_prob': 0.03480567,
-                'lr_mult': 7.98104224,
-                'patience': 2,
-                'batch_norm': True,
-                'initialization': 'Default'}
-    assert values == expected
+    # Consider transformation of the values
+    assert conf['epochs'] == 4