Training Trains Internship
Student login
← TensorFlow & Keras learning path
TensorFlow & Keras · Module 9 of 10

Saving and deploying models

Save models and serve them in apps.

📖 Key ideas

  1. model.save("model.keras") saves architecture and weights.
  2. keras.models.load_model loads it back.
  3. Export SavedModel for TF Serving.
  4. Convert to TFLite for Android/iOS/edge.
  5. TensorFlow.js runs models in the browser.
💡 Real-life example: Saving a model is bottling a recipe so any kitchen (server, phone, browser) can reuse it.
.kerasKeras model file format
SavedModelTensorFlow serving format
TFLiteLightweight mobile format
TensorFlow.jsML in JavaScript/browser