TensorFlow & PyTorch

API Tutorial

About

This repository provides hands-on tutorials about the Sequential API, Functional API, and Model Subclassing in TensorFlow and PyTorch.

Table of Contents

• TensorFlow API Tutorial
• PyTorch API Tutorial

Setting Up the Development Environment

Important: Ensure that the versions of TensorFlow or PyTorch you install are compatible with your Python version and GPU drivers/CUDA versions.

This setup has been tested with an NVIDIA L40S 48GB GPU running driver version 550.127.08, which is backward compatible with earlier CUDA versions (e.g., CUDA 11.8).

• Verify GPU availability:

nvidia-smi

PyTorch Setup

• Create and activate the environment:

conda env create -f torch_env.yml && conda activate torch-env

• Verify the GPU installation:

python -c "import torch; [print(f'GPU {i}: {torch.cuda.get_device_name(i)}') for i in range(torch.cuda.device_count())]"

Expected output:

GPU 0: NVIDIA L40S
GPU 1: NVIDIA L40S

TensorFlow Setup

• TensorFlow-CPU:

conda env create -f tf_cpu_env.yml && conda activate tf-cpu-env

• TensorFlow-GPU:

conda env create -f tf_gpu_env.yml && conda activate tf-gpu-env

• Verify the GPU installation:

python -c "import tensorflow as tf; print(tf.config.list_physical_devices('GPU'))"

Expected output:

[PhysicalDevice(name='/physical_device:GPU:0', device_type='GPU'), PhysicalDevice(name='/physical_device:GPU:1', device_type='GPU')]