TensorFlow Environment

Deploy GPU instances with TensorFlow pre-configured for immediate AI/ML development.

What's Included

Core Frameworks:

TensorFlow 2.x with GPU support
Keras (high-level neural networks API)
CUDA and cuDNN pre-configured

Development Tools:

Jupyter Notebook for interactive development
TensorBoard for visualization
NumPy, Pandas, Matplotlib
scikit-learn for traditional ML

System:

Ubuntu 22.04 or 24.04 LTS
NVIDIA drivers and CUDA toolkit
Python 3.9-3.11
pip and conda package managers

Deploying TensorFlow Environment

Select Environment

Go to app.spheron.ai → Deploy
Choose your GPU
Select OS: Ubuntu 24.04 LTS ML TensorFlow or Ubuntu 22.04 LTS + TensorFlow
Deploy

Instance ready in 30-60 seconds.

Connect via SSH

ssh root@your-instance-ip

Verify Installation

# Check TensorFlow
python3 -c "import tensorflow as tf; print(f'TensorFlow version: {tf.__version__}')"
python3 -c "import tensorflow as tf; print(f'GPU available: {tf.config.list_physical_devices('GPU')}')"
 
# Check CUDA
nvcc --version
 
# Check GPU
nvidia-smi

Quick Start Examples

Run TensorFlow Test

import tensorflow as tf
 
# Verify GPU
print("Num GPUs Available:", len(tf.config.list_physical_devices('GPU')))
 
# Simple computation test
with tf.device('/GPU:0'):
    a = tf.constant([[1.0, 2.0], [3.0, 4.0]])
    b = tf.constant([[1.0, 1.0], [0.0, 1.0]])
    c = tf.matmul(a, b)
    print(c)

Train Simple Model

import tensorflow as tf
from tensorflow import keras
 
# Load dataset
mnist = keras.datasets.mnist
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train, x_test = x_train / 255.0, x_test / 255.0
 
# Build model
model = keras.Sequential([
    keras.layers.Flatten(input_shape=(28, 28)),
    keras.layers.Dense(128, activation='relu'),
    keras.layers.Dropout(0.2),
    keras.layers.Dense(10)
])
 
# Compile and train
model.compile(optimizer='adam',
              loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
              metrics=['accuracy'])
 
model.fit(x_train, y_train, epochs=5)

Start Jupyter Notebook

# Install Jupyter if not included
pip install jupyter
 
# Start Jupyter (accessible from browser)
jupyter notebook --ip=0.0.0.0 --port=8888 --no-browser --allow-root

Access at: http://your-instance-ip:8888

Common Packages

Install additional packages as needed:

# Computer vision
pip install opencv-python pillow
 
# Data processing
pip install pandas numpy scipy
 
# Model optimization
pip install tensorflow-model-optimization
 
# TensorFlow Datasets
pip install tensorflow-datasets
 
# Visualization
pip install tensorboard seaborn

Use Cases

Production Deployment - Enterprise-grade model serving
Computer Vision - Image classification, object detection
NLP - Text analysis, sentiment analysis
Time Series - Forecasting and prediction
Mobile AI - TensorFlow Lite model development

Troubleshooting

GPU not detected by TensorFlow:

# Check GPU visibility
python3 -c "import tensorflow as tf; print(tf.config.list_physical_devices('GPU'))"
 
# Reinstall TensorFlow with GPU support
pip uninstall tensorflow
pip install tensorflow[and-cuda]

CUDA version mismatch:

Check TensorFlow compatibility: TensorFlow GPU support
Use compatible CUDA version for your TensorFlow release

Out of memory errors:

Reduce batch size
Enable memory growth: tf.config.experimental.set_memory_growth(gpu, True)
Monitor with: nvidia-smi -l 1

Additional Resources

TensorFlow Official Docs
Keras Documentation
Getting Started - Deploy your first instance
Quick Start - Fast deployment guide
API Reference - Automate deployments