{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Found 2199 images belonging to 2 classes.\n",
"Found 549 images belonging to 2 classes.\n",
"Training samples: 2199\n",
"Validation samples: 549\n"
]
}
],
"source": [
"import tensorflow as tf\n",
"from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
"\n",
"dataset_dir = 'C:/Users/vsavelev/GITHUB/DS_projet/jan24_cds_mushrooms/data'\n",
"\n",
"# Create ImageDataGenerator with validation split\n",
"datagen = ImageDataGenerator(rescale=1.0/255, validation_split=0.2)\n",
"\n",
"train_generator = datagen.flow_from_directory(\n",
" dataset_dir,\n",
" target_size=(224, 224),\n",
" batch_size=32,\n",
" class_mode='categorical',\n",
" subset='training' # Set as training data\n",
")\n",
"\n",
"validation_generator = datagen.flow_from_directory(\n",
" dataset_dir,\n",
" target_size=(224, 224),\n",
" batch_size=32,\n",
" class_mode='categorical',\n",
" subset='validation' # Set as validation data\n",
")\n",
"\n",
"print(f'Training samples: {train_generator.samples}')\n",
"print(f'Validation samples: {validation_generator.samples}')"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"Model: \"sequential\"\n",
"\n"
],
"text/plain": [
"\u001b[1mModel: \"sequential\"\u001b[0m\n"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
"┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓\n",
"┃ Layer (type) ┃ Output Shape ┃ Param # ┃\n",
"┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩\n",
"│ resnet50 (Functional) │ ? │ 23,587,712 │\n",
"├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
"│ global_average_pooling2d │ ? │ 0 (unbuilt) │\n",
"│ (GlobalAveragePooling2D) │ │ │\n",
"├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
"│ dense (Dense) │ ? │ 0 (unbuilt) │\n",
"├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
"│ dropout (Dropout) │ ? │ 0 (unbuilt) │\n",
"├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
"│ dense_1 (Dense) │ ? │ 0 (unbuilt) │\n",
"└─────────────────────────────────┴────────────────────────┴───────────────┘\n",
"\n"
],
"text/plain": [
"┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓\n",
"┃\u001b[1m \u001b[0m\u001b[1mLayer (type) \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1mOutput Shape \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1m Param #\u001b[0m\u001b[1m \u001b[0m┃\n",
"┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩\n",
"│ resnet50 (\u001b[38;5;33mFunctional\u001b[0m) │ ? │ \u001b[38;5;34m23,587,712\u001b[0m │\n",
"├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
"│ global_average_pooling2d │ ? │ \u001b[38;5;34m0\u001b[0m (unbuilt) │\n",
"│ (\u001b[38;5;33mGlobalAveragePooling2D\u001b[0m) │ │ │\n",
"├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
"│ dense (\u001b[38;5;33mDense\u001b[0m) │ ? │ \u001b[38;5;34m0\u001b[0m (unbuilt) │\n",
"├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
"│ dropout (\u001b[38;5;33mDropout\u001b[0m) │ ? │ \u001b[38;5;34m0\u001b[0m (unbuilt) │\n",
"├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
"│ dense_1 (\u001b[38;5;33mDense\u001b[0m) │ ? │ \u001b[38;5;34m0\u001b[0m (unbuilt) │\n",
"└─────────────────────────────────┴────────────────────────┴───────────────┘\n"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
" Total params: 23,587,712 (89.98 MB)\n",
"\n"
],
"text/plain": [
"\u001b[1m Total params: \u001b[0m\u001b[38;5;34m23,587,712\u001b[0m (89.98 MB)\n"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
" Trainable params: 0 (0.00 B)\n",
"\n"
],
"text/plain": [
"\u001b[1m Trainable params: \u001b[0m\u001b[38;5;34m0\u001b[0m (0.00 B)\n"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
" Non-trainable params: 23,587,712 (89.98 MB)\n",
"\n"
],
"text/plain": [
"\u001b[1m Non-trainable params: \u001b[0m\u001b[38;5;34m23,587,712\u001b[0m (89.98 MB)\n"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"from tensorflow.keras.applications import ResNet50\n",
"from tensorflow.keras import layers, models\n",
"\n",
"# Load and Configure the Pre-trained ResNet50 Model\n",
"base_model = ResNet50(weights='imagenet', include_top=False, input_shape=(224, 224, 3))\n",
"\"\"\"\n",
"weights='imagenet': Loads the pre-trained weights from the ImageNet dataset.\n",
"include_top=False: Excludes the top fully-connected layers of the ResNet50 model, enabling you to add your own custom layers.\n",
"input_shape=(224, 224, 3): Specifies the input shape of the images (224x224 pixels, with 3 color channels - RGB).\n",
"\"\"\"\n",
"\n",
"# Freeze the base model (to freeze the pre-trained layers)\n",
"base_model.trainable = False\n",
"\n",
"# Add custom layers on top of the base model\n",
"model = models.Sequential([ #allows to stack layers linearly\n",
" base_model,\n",
" layers.GlobalAveragePooling2D(),\n",
" layers.Dense(1024, activation='relu', kernel_regularizer=tf.keras.regularizers.l2(0.01)), # L2 regularization\n",
" layers.Dropout(0.5),\n",
" layers.Dense(train_generator.num_classes, activation='softmax')\n",
"])\n",
"\n",
"\"\"\"\n",
"GlobalAveragePooling2D(): Reduces each feature map to a single number by taking the average, \n",
"which helps to reduce the size of the model and prevent overfitting.\n",
"Dense(1024, activation='relu'): Adds a fully connected layer with 1024 units and ReLU activation function.\n",
"Dropout(0.5): Adds a dropout layer with a 50% dropout rate to prevent overfitting by randomly setting half of the input units \n",
"to 0 at each update during training.\n",
"Dense(train_generator.num_classes, activation='softmax'): \n",
"Adds the final output layer with units equal to the number of classes in your dataset, using the softmax activation function for multi-class classification.\n",
"\"\"\"\n",
"model.compile(optimizer=tf.keras.optimizers.Adam(),\n",
" loss='categorical_crossentropy',\n",
" metrics=['accuracy'])\n",
"\n",
"\"\"\"\n",
"optimizer=tf.keras.optimizers.Adam(): Uses the Adam optimizer, which is an adaptive learning rate optimization algorithm.\n",
"loss='categorical_crossentropy': Uses categorical cross-entropy as the loss function, suitable for multi-class classification.\n",
"metrics=['accuracy']: Tracks accuracy as the metric to evaluate the model's performance during training and testing.\n",
"\"\"\"\n",
"model.summary()\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/10\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"c:\\Users\\vsavelev\\AppData\\Local\\anaconda3\\Lib\\site-packages\\keras\\src\\trainers\\data_adapters\\py_dataset_adapter.py:121: UserWarning: Your `PyDataset` class should call `super().__init__(**kwargs)` in its constructor. `**kwargs` can include `workers`, `use_multiprocessing`, `max_queue_size`. Do not pass these arguments to `fit()`, as they will be ignored.\n",
" self._warn_if_super_not_called()\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m143s\u001b[0m 2s/step - accuracy: 0.8540 - loss: 6.8043 - val_accuracy: 0.8860 - val_loss: 0.6636\n",
"Epoch 2/10\n",
"\u001b[1m 1/68\u001b[0m \u001b[37m━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[1m2:09\u001b[0m 2s/step - accuracy: 0.8438 - loss: 0.8550"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"c:\\Users\\vsavelev\\AppData\\Local\\anaconda3\\Lib\\contextlib.py:158: UserWarning: Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches. You may need to use the `.repeat()` function when building your dataset.\n",
" self.gen.throw(typ, value, traceback)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m3s\u001b[0m 11ms/step - accuracy: 0.8438 - loss: 0.8550 - val_accuracy: 1.0000 - val_loss: 0.4526\n",
"Epoch 3/10\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m146s\u001b[0m 2s/step - accuracy: 0.8873 - loss: 0.6507 - val_accuracy: 0.8879 - val_loss: 0.4938\n",
"Epoch 4/10\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 6ms/step - accuracy: 0.8750 - loss: 0.5112 - val_accuracy: 0.8000 - val_loss: 0.6924\n",
"Epoch 5/10\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m149s\u001b[0m 2s/step - accuracy: 0.8857 - loss: 0.5040 - val_accuracy: 0.8879 - val_loss: 0.4471\n",
"Epoch 6/10\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 6ms/step - accuracy: 0.8750 - loss: 0.4322 - val_accuracy: 0.8000 - val_loss: 0.6413\n",
"Epoch 7/10\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m166s\u001b[0m 2s/step - accuracy: 0.8926 - loss: 0.4378 - val_accuracy: 0.8860 - val_loss: 0.4396\n",
"Epoch 8/10\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 7ms/step - accuracy: 0.9375 - loss: 0.3477 - val_accuracy: 1.0000 - val_loss: 0.2697\n",
"Epoch 9/10\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m173s\u001b[0m 3s/step - accuracy: 0.8890 - loss: 0.4132 - val_accuracy: 0.8879 - val_loss: 0.4444\n",
"Epoch 10/10\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 7ms/step - accuracy: 0.9688 - loss: 0.3492 - val_accuracy: 0.8000 - val_loss: 0.5367\n"
]
}
],
"source": [
"history = model.fit(\n",
" train_generator,\n",
" steps_per_epoch=train_generator.samples // train_generator.batch_size,\n",
" validation_data=validation_generator,\n",
" validation_steps=validation_generator.samples // validation_generator.batch_size,\n",
" epochs=10\n",
")\n",
"\n",
"#This specifies the number of complete passes through the training dataset. Here, the model will train for 10 epochs."
]
},
{
"cell_type": "code",
"execution_count": 12,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Model compiled successfully.\n",
"Callbacks created successfully.\n",
"Epoch 1/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m267s\u001b[0m 4s/step - accuracy: 0.8686 - loss: 0.6860 - val_accuracy: 0.8897 - val_loss: 0.5719\n",
"Epoch 2/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 9ms/step - accuracy: 0.8750 - loss: 0.6368 - val_accuracy: 0.6000 - val_loss: 0.7365\n",
"Epoch 3/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m286s\u001b[0m 4s/step - accuracy: 0.8856 - loss: 0.6046 - val_accuracy: 0.8897 - val_loss: 0.6145\n",
"Epoch 4/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m5s\u001b[0m 9ms/step - accuracy: 0.8438 - loss: 0.5484 - val_accuracy: 0.6000 - val_loss: 0.7114\n",
"Epoch 5/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m323s\u001b[0m 5s/step - accuracy: 0.8840 - loss: 0.5082 - val_accuracy: 0.8860 - val_loss: 0.5360\n",
"Epoch 6/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 7ms/step - accuracy: 0.8750 - loss: 0.4655 - val_accuracy: 1.0000 - val_loss: 0.4314\n",
"Epoch 7/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m299s\u001b[0m 4s/step - accuracy: 0.8770 - loss: 0.4487 - val_accuracy: 0.8879 - val_loss: 0.4546\n",
"Epoch 8/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 6ms/step - accuracy: 0.9375 - loss: 0.3386 - val_accuracy: 0.8000 - val_loss: 0.5507\n",
"Epoch 9/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m276s\u001b[0m 4s/step - accuracy: 0.8946 - loss: 0.3906 - val_accuracy: 0.8879 - val_loss: 0.4199\n",
"Epoch 10/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 7ms/step - accuracy: 0.8438 - loss: 0.4413 - val_accuracy: 0.8000 - val_loss: 0.5716\n",
"Epoch 11/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m293s\u001b[0m 4s/step - accuracy: 0.8840 - loss: 0.3867 - val_accuracy: 0.8860 - val_loss: 0.4153\n",
"Epoch 12/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m3s\u001b[0m 7ms/step - accuracy: 0.9565 - loss: 0.2604 - val_accuracy: 1.0000 - val_loss: 0.2164\n",
"Epoch 13/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m290s\u001b[0m 4s/step - accuracy: 0.8951 - loss: 0.3485 - val_accuracy: 0.8915 - val_loss: 0.3989\n",
"Epoch 14/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 7ms/step - accuracy: 0.9375 - loss: 0.2921 - val_accuracy: 0.4000 - val_loss: 1.3309\n",
"Epoch 15/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m273s\u001b[0m 4s/step - accuracy: 0.8950 - loss: 0.3413 - val_accuracy: 0.8860 - val_loss: 0.4095\n",
"Epoch 16/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 7ms/step - accuracy: 0.9062 - loss: 0.3136 - val_accuracy: 1.0000 - val_loss: 0.1990\n",
"Epoch 17/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m270s\u001b[0m 4s/step - accuracy: 0.8919 - loss: 0.3295 - val_accuracy: 0.8860 - val_loss: 0.4054\n",
"Epoch 18/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 8ms/step - accuracy: 0.8750 - loss: 0.3683 - val_accuracy: 1.0000 - val_loss: 0.1749\n",
"Epoch 19/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m279s\u001b[0m 4s/step - accuracy: 0.8906 - loss: 0.3170 - val_accuracy: 0.8860 - val_loss: 0.4067\n",
"Epoch 20/20\n",
"\u001b[1m68/68\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 8ms/step - accuracy: 0.9375 - loss: 0.2421 - val_accuracy: 1.0000 - val_loss: 0.2140\n"
]
}
],
"source": [
"from tensorflow.keras.callbacks import EarlyStopping, ModelCheckpoint\n",
"\n",
"\n",
"# Unfreeze some layers\n",
"base_model.trainable = True\n",
"fine_tune_at = 100 # fine-tune from this layer onwards\n",
"\n",
"for layer in base_model.layers[:fine_tune_at]:\n",
" layer.trainable = False\n",
"\n",
"# # Unfreeze more layers gradually\n",
"# for layer in base_model.layers[:-10]: # Unfreeze all layers except the last 10 layers\n",
"# layer.trainable = False\n",
"\n",
"model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=1e-6), #learning_rate=1e-5\n",
" loss='categorical_crossentropy', #The categorical cross-entropy loss function is used because this is a multi-class classification problem\n",
" metrics=['accuracy'])\n",
"\n",
"print(\"Model compiled successfully.\")\n",
"\n",
"\"\"\"\n",
"The Adam optimizer is used with a very small learning rate (1e-5). Fine-tuning typically \n",
"uses a smaller learning rate to prevent large updates to the weights, which could potentially destroy the learned features in the pre-trained model.\n",
"\"\"\"\n",
"\n",
"early_stopping = EarlyStopping(monitor='val_loss', patience=5, restore_best_weights=True)\n",
"model_checkpoint = ModelCheckpoint('best_model.keras', save_best_only=True, monitor='val_loss')\n",
"\n",
"print(\"Callbacks created successfully.\")\n",
"\n",
"\n",
"history_fine = model.fit(\n",
" train_generator,\n",
" steps_per_epoch=train_generator.samples // train_generator.batch_size,\n",
" validation_data=validation_generator,\n",
" validation_steps=validation_generator.samples // validation_generator.batch_size,\n",
" epochs=20,\n",
" #callbacks=[early_stopping, model_checkpoint]\n",
")\n"
]
},
{
"cell_type": "code",
"execution_count": 13,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\u001b[1m18/18\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m22s\u001b[0m 1s/step - accuracy: 0.8704 - loss: 0.4312\n",
"Validation loss: 0.4050602614879608\n",
"Validation accuracy: 0.8870673775672913\n"
]
}
],
"source": [
"loss, accuracy = model.evaluate(validation_generator)\n",
"print(f'Validation loss: {loss}')\n",
"print(f'Validation accuracy: {accuracy}')\n"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "base",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.11.7"
}
},
"nbformat": 4,
"nbformat_minor": 2
}