Week 2:
i Setup of TensorFlow, TensorFlow with GPU support.
TensorFlow is an open-source deep learning framework developed by
Google. It allows building and training machine learning and deep learning
models in a flexible and scalable way.
Installation of TensorFlow (CPU Version)
Step-by-Step Guide:
➤ Step 1: Install Python
Recommended: Python 3.8 to 3.11
Download from: https://www.python.org/downloads/
➤ Step 2: Create a Virtual Environment
python -m venv tf_env
➤ Step 3: Activate the Environment
Windows:
.\tf_env\Scripts\activate
Linux/macOS:
source tf_env/bin/activate
➤ Step 4: Install TensorFlow (CPU version)
pip install tensorflow
Installation of TensorFlow with GPU Support
Component Version Requirement
Python 3.8–3.11
TensorFlow 2.12.0 or later
CUDA Toolkit 11.8 (for TF 2.12), or 12.2+ for TF 2.15+
cuDNN 8.6 or 8.9
NVIDIA GPU With compute capability ≥ 3.5 (e.g., GTX 10xx, RTX)
NVIDIA Driver Compatible with your CUDA version
Step-by-Step GPU Setup (Windows/Linux)
➤ Step 1: Install Python and Virtual Environment
(Same as above)
➤ Step 2: Install GPU-compatible TensorFlow
pip install tensorflow
Since TensorFlow 2.11+, GPU support is included in tensorflow package (no need for tensorflow-
gpu separately).
➤ Step 3: Install CUDA Toolkit and cuDNN
🟢 Option A: Use NVIDIA CUDA Installer
Install CUDA 11.8:
https://developer.nvidia.com/cuda-11-8-0-download-archive
Install cuDNN 8.6:
https://developer.nvidia.com/rdp/cudnn-archive
✅ Add these to Environment Variables:
CUDA_HOME
Add to PATH:
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\bin
...\CUDA\v11.8\libnvvp
🟢 Option B: Use conda for Simpler Setup
conda create -n tf_gpu python=3.10
conda activate tf_gpu
conda install -c nvidia cuda-toolkit=11.8 cudnn=8.6
pip install tensorflow
✅ 4. Verify GPU Setup
Run the following script:
import tensorflow as tf
print("TensorFlow version:", tf.__version__)
print("Num GPUs Available:", len(tf.config.list_physical_devices('GPU')))
You should see:
TensorFlow version: 2.12.0
Num GPUs Available: 1
ii Setup to Keras.
Keras is a high-level API for building and training deep learning models.
Originally developed as an independent library, Keras is now tightly
integrated into TensorFlow as tf.keras.
Keras Installation
🔹 Step-by-Step:
Keras is included with TensorFlow (since TF 2.0+), so you don't need to
install it separately.
✅ To install:
pip install tensorflow
This will install tensorflow along with tf.keras.
✅ 2. Importing and Using Keras
You should always use tf.keras, not the old standalone keras:
from tensorflow import keras
from tensorflow.keras import layers
✅ 3. Basic Keras Workflow
🔷 Step 1: Build the Model
model = keras.Sequential([
layers.Dense(128, activation='relu', input_shape=(784,)),
layers.Dense(10, activation='softmax')
])
🔷 Step 2: Compile the Model
model.compile(optimizer='adam',
loss='sparse_categorical_crossentropy',
metrics=['accuracy'])
🔷 Step 3: Train the Model
model.fit(x_train, y_train, epochs=5, batch_size=32)
🔷 Step 4: Evaluate
model.evaluate(x_test, y_test)
iii Learning the other deep learning libraries like Caffe,
Theano, etc.
These are not actively maintained for new projects but are useful for understanding legacy
systems and framework design.
🔸 Caffe (Convolutional Architecture for Fast Feature Embedding)
Written in C++
Developed by Berkeley AI Research
Pretrained models available
Installation (Linux):
sudo apt install caffe-cpu
Documentation: https://caffe.berkeleyvision.org/
Theano
Python library for defining, optimizing, and evaluating mathematical expressions
Was foundational to many other libraries (e.g., Keras originally used Theano)
pip install Theano
import theano
print(theano.config.device)
🔸 Other Libraries (for awareness):
Library Notes
PyTorch Widely used in academia and research
MXNet Backed by Apache; used by AWS
CNTK Microsoft Cognitive Toolkit
JAX High-performance computing; by Google
