diff --git a/conftest.py b/conftest.py
index d8f0897ff8096e30f4b7f7e06f709eda11a629e2..207606e6ec111459302360f5f2c4f917771bf80d 100644
--- a/conftest.py
+++ b/conftest.py
@@ -27,7 +27,7 @@ def execute_removing(path):
if "data" in list_dir and path != os.path.dirname(__file__): # do not delete data folder in src
shutil.rmtree(os.path.join(path, "data"), ignore_errors=True)
# remove TestExperiment folders
- remove_files_from_regex(list_dir, path, re.compile(r"TestExperiment_.*"))
+ remove_files_from_regex(list_dir, path, re.compile(r"TestExperiment.*"))
# remove all tracking json
remove_files_from_regex(list_dir, path, re.compile(r"tracking_\d*\.json"))
# remove all tracking pdf
diff --git a/src/model_modules/model_class.py b/src/model_modules/model_class.py
index 5dd69608e0ad6e66b250505c3fffa037bba212a4..dab2e168c5a9f87d4aee42fc94489fd0fa67772a 100644
--- a/src/model_modules/model_class.py
+++ b/src/model_modules/model_class.py
@@ -119,7 +119,6 @@ from typing import Any, Callable, Dict
import keras
import tensorflow as tf
-import logging
from src.model_modules.inception_model import InceptionModelBase
from src.model_modules.flatten import flatten_tail
from src.model_modules.advanced_paddings import PadUtils, Padding2D
@@ -355,7 +354,6 @@ class MyLittleModel(AbstractModelClass):
self.channels = channels
self.dropout_rate = 0.1
self.regularizer = keras.regularizers.l2(0.1)
- self.batch_size = int(256)
self.activation = keras.layers.PReLU
# apply to model
@@ -428,7 +426,6 @@ class MyBranchedModel(AbstractModelClass):
self.channels = channels
self.dropout_rate = 0.1
self.regularizer = keras.regularizers.l2(0.1)
- self.batch_size = int(256)
self.activation = keras.layers.PReLU
# apply to model
@@ -502,7 +499,6 @@ class MyTowerModel(AbstractModelClass):
self.initial_lr = 1e-2
self.lr_decay = src.model_modules.keras_extensions.LearningRateDecay(base_lr=self.initial_lr, drop=.94,
epochs_drop=10)
- self.batch_size = int(256 * 4)
self.activation = keras.layers.PReLU
# apply to model
@@ -615,7 +611,6 @@ class MyPaperModel(AbstractModelClass):
self.initial_lr = 1e-3
self.lr_decay = src.model_modules.keras_extensions.LearningRateDecay(base_lr=self.initial_lr, drop=.94,
epochs_drop=10)
- self.batch_size = int(256 * 2)
self.activation = keras.layers.ELU
self.padding = "SymPad2D"
diff --git a/src/run.py b/src/run.py
index ac1d27b8a05a544bd623bcfd7fca5884986abf2a..eda0373c1e609e0818e98358d00a00beddb63cdf 100644
--- a/src/run.py
+++ b/src/run.py
@@ -25,7 +25,8 @@ def run(stations=['DEBW107', 'DEBY081', 'DEBW013', 'DEBW076', 'DEBW087', 'DEBW00
train_min_length=None, val_min_length=None, test_min_length=None,
evaluate_bootstraps=True, number_of_bootstraps=None, create_new_bootstraps=False,
plot_list=None,
- model=None):
+ model=None,
+ batch_size=None):
params = inspect.getfullargspec(ExperimentSetup).args
kwargs = {k: v for k, v in locals().items() if k in params}
diff --git a/src/run_modules/experiment_setup.py b/src/run_modules/experiment_setup.py
index 1f4c063415efc8adae088c965c33d41c36a08bc4..ff6fec842714d599696b8726e9d25aa22e55583f 100644
--- a/src/run_modules/experiment_setup.py
+++ b/src/run_modules/experiment_setup.py
@@ -233,12 +233,13 @@ class ExperimentSetup(RunEnvironment):
create_new_model: bool = None, bootstrap_path=None, permute_data_on_training: bool = None, transformation=None,
train_min_length=None, val_min_length=None, test_min_length=None, extreme_values: list = None,
extremes_on_right_tail_only: bool = None, evaluate_bootstraps=True, plot_list=None, number_of_bootstraps=None,
- create_new_bootstraps=None, data_path: str = None, login_nodes=None, hpc_hosts=None, model=None, epochs=None):
+ create_new_bootstraps=None, data_path: str = None, login_nodes=None, hpc_hosts=None, model=None,
+ batch_size=None, epochs=None):
# create run framework
super().__init__()
- # experiment setup
+ # experiment setup, hyperparameters
self._set_param("data_path", path_config.prepare_host(data_path=data_path, sampling=sampling))
self._set_param("hostname", path_config.get_host())
self._set_param("hpc_hosts", hpc_hosts, default=DEFAULT_HPC_HOST_LIST + DEFAULT_HPC_LOGIN_LIST)
@@ -257,6 +258,7 @@ class ExperimentSetup(RunEnvironment):
upsampling = self.data_store.get("upsampling", "train")
permute_data = False if permute_data_on_training is None else permute_data_on_training
self._set_param("permute_data", permute_data or upsampling, scope="train")
+ self._set_param("batch_size", batch_size, default=int(256 * 2))
self._set_param("epochs", epochs, default=20)
# set experiment name
diff --git a/src/run_modules/model_setup.py b/src/run_modules/model_setup.py
index 13a13bb72fd634b6ebbec20f46a3a08a9f0afa8e..f9683b953d85bacf6e452e0a1922e85dfe946cd1 100644
--- a/src/run_modules/model_setup.py
+++ b/src/run_modules/model_setup.py
@@ -33,6 +33,7 @@ class ModelSetup(RunEnvironment):
* `generator` [train]
* `window_lead_time` [.]
* `window_history_size` [.]
+ * `model_class` [.]
Optional objects
* `lr_decay` [model]
@@ -43,7 +44,7 @@ class ModelSetup(RunEnvironment):
* `hist` [model]
* `callbacks` [model]
* `model_name` [model]
- * all settings from model class like `dropout_rate`, `initial_lr`, `batch_size`, and `optimizer` [model]
+ * all settings from model class like `dropout_rate`, `initial_lr`, and `optimizer` [model]
Creates
* plot of model architecture `<model_name>.pdf`
diff --git a/src/run_modules/training.py b/src/run_modules/training.py
index 5df8a15c6e0beae8ea9b1504c654e18df46e7b00..1a0d7beb1ec37bb5e59a4129da58572d79a73636 100644
--- a/src/run_modules/training.py
+++ b/src/run_modules/training.py
@@ -33,8 +33,8 @@ class Training(RunEnvironment):
Required objects [scope] from data store:
* `model` [model]
- * `batch_size` [model]
- * `epochs` [model]
+ * `batch_size` [.]
+ * `epochs` [.]
* `callbacks` [model]
* `model_name` [model]
* `experiment_name` [.]
@@ -67,7 +67,7 @@ class Training(RunEnvironment):
self.train_set: Union[Distributor, None] = None
self.val_set: Union[Distributor, None] = None
self.test_set: Union[Distributor, None] = None
- self.batch_size = self.data_store.get("batch_size", "model")
+ self.batch_size = self.data_store.get("batch_size")
self.epochs = self.data_store.get("epochs")
self.callbacks: CallbackHandler = self.data_store.get("callbacks", "model")
self.experiment_name = self.data_store.get("experiment_name")
diff --git a/test/test_modules/test_model_setup.py b/test/test_modules/test_model_setup.py
index 541f8d17c8084a34a75b1907c5ff89b08bea9e03..60d140f8845b25432184de1f3890b3ee4d0b034e 100644
--- a/test/test_modules/test_model_setup.py
+++ b/test/test_modules/test_model_setup.py
@@ -88,7 +88,7 @@ class TestModelSetup:
setup_with_gen.build_model()
assert isinstance(setup_with_gen.model, AbstractModelClass)
expected = {"window_history_size", "window_lead_time", "channels", "dropout_rate", "regularizer", "initial_lr",
- "optimizer", "batch_size", "activation"}
+ "optimizer", "activation"}
assert expected <= self.current_scope_as_set(setup_with_gen)
def test_set_channels(self, setup_with_gen_tiny):
diff --git a/test/test_modules/test_training.py b/test/test_modules/test_training.py
index a26bd18c75ae34d200669141578b5fa3ea2bb7c8..66ba0709c21b105bd798cd35f20715e6c0a83177 100644
--- a/test/test_modules/test_training.py
+++ b/test/test_modules/test_training.py
@@ -155,7 +155,7 @@ class TestTraining:
obj.data_store.set("model", model, "general.model")
obj.data_store.set("model_path", model_path, "general")
obj.data_store.set("model_name", os.path.join(model_path, "test_model.h5"), "general.model")
- obj.data_store.set("batch_size", 256, "general.model")
+ obj.data_store.set("batch_size", 256, "general")
obj.data_store.set("epochs", 2, "general.model")
clbk, hist, lr = callbacks
obj.data_store.set("callbacks", clbk, "general.model")