sppu-be-comp-content/DeepLearning
10
0
Fork 0
Code Issues Activity

fix: remove function for loading dataset locally and add basic lines of code to load csv instead for practical 3b notebook.

Browse Source
This commit is contained in:
K
2026-05-04 16:27:37 +05:30
parent 73f8c867b7
commit deb41dfdc8
1 changed files with 4 additions and 51 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Notebooks/Notebook-3b.ipynb
+3 -50
View File
@@ -51,58 +51,11 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 5, "execution_count": null,
"id": "859cbc0f", "id": "859cbc0f",
"metadata": {}, "metadata": {},
"outputs": [ "outputs": [],
{ "source": "# 2. Load Dataset\n# Fashion MNIST is built into Keras — downloads automatically on first run\n(X_train, y_train), (X_test, y_test) = tf.keras.datasets.fashion_mnist.load_data()\n\n# --- Offline alternative (comment out tf.keras line above and use this instead) ---\n# import pandas as pd\n# train_df = pd.read_csv('fashion-mnist_train.csv')\n# test_df = pd.read_csv('fashion-mnist_test.csv')\n# y_train = train_df['label'].values\n# y_test = test_df['label'].values\n# X_train = train_df.drop('label', axis=1).values.reshape(-1, 28, 28) # unflatten pixels to 28x28\n# X_test = test_df.drop('label', axis=1).values.reshape(-1, 28, 28)\n\nprint(\"Training set shape:\", X_train.shape) # (60000, 28, 28)\nprint(\"Test set shape: \", X_test.shape) # (10000, 28, 28)\nprint(\"Classes:\", np.unique(y_train))"
"name": "stdout",
"output_type": "stream",
"text": [
"Training set shape: (60000, 28, 28)\n",
"Test set shape: (10000, 28, 28)\n",
"Classes: [0 1 2 3 4 5 6 7 8 9]\n"
]
}
],
"source": [
"# 2. Load Dataset\n",
"# Fashion MNIST is built into Keras, downloads automatically on first run\n",
"(X_train, y_train), (X_test, y_test) = tf.keras.datasets.fashion_mnist.load_data()\n",
"\n",
"'''\n",
"import numpy as np\n",
"import gzip\n",
"import os\n",
"\n",
"def load_fashion_mnist(path):\n",
" \"\"\"Load Fashion MNIST from local .gz files (Kaggle Zalando format).\"\"\"\n",
" files = {\n",
" 'X_train': 'train-images-idx3-ubyte.gz',\n",
" 'y_train': 'train-labels-idx1-ubyte.gz',\n",
" 'X_test': 't10k-images-idx3-ubyte.gz',\n",
" 'y_test': 't10k-labels-idx1-ubyte.gz',\n",
" }\n",
"\n",
" with gzip.open(os.path.join(path, files['X_train'])) as f:\n",
" X_train = np.frombuffer(f.read(), np.uint8, offset=16).reshape(-1, 28, 28)\n",
" with gzip.open(os.path.join(path, files['y_train'])) as f:\n",
" y_train = np.frombuffer(f.read(), np.uint8, offset=8)\n",
" with gzip.open(os.path.join(path, files['X_test'])) as f:\n",
" X_test = np.frombuffer(f.read(), np.uint8, offset=16).reshape(-1, 28, 28)\n",
" with gzip.open(os.path.join(path, files['y_test'])) as f:\n",
" y_test = np.frombuffer(f.read(), np.uint8, offset=8)\n",
"\n",
" return (X_train, y_train), (X_test, y_test)\n",
"\n",
"# Replace the Keras load line with:\n",
"(X_train, y_train), (X_test, y_test) = load_fashion_mnist('./fashion-mnist/')\n",
"'''\n",
"\n",
"print(\"Training set shape:\", X_train.shape) # (60000, 28, 28)\n",
"print(\"Test set shape: \", X_test.shape) # (10000, 28, 28)\n",
"print(\"Classes:\", np.unique(y_train))"
]
}, },
{ {
"cell_type": "code", "cell_type": "code",