| from huggingface_hub import hf_hub_download | |
| import tensorflow as tf | |
| from tensorflow import keras | |
| # Replace 'your-username/your-model-name' with your actual Hugging Face model repository ID. | |
| repo_id = "your-username/your-model-name" | |
| # Replace 'my_keras_model.keras' with the name of the file you uploaded. | |
| filename = "my_keras_model.keras" | |
| # Download the model file from the Hugging Face Hub. | |
| model_path = hf_hub_download(repo_id=repo_id, filename=filename) | |
| # Load the model using Keras's built-in function. | |
| # The 'safe_mode=False' argument is often necessary when loading models saved from older TensorFlow versions | |
| # or if the model contains custom layers. | |
| model = keras.models.load_model(model_path, safe_mode=False) | |
| # Now you can use the loaded model for inference. | |
| # Example: Load a single MNIST test image and make a prediction. | |
| (x_train, y_train), (x_test, y_test) = keras.datasets.mnist.load_data() | |
| x_test = x_test.astype("float32") / 255.0 | |
| x_test = tf.expand_dims(x_test, -1) | |
| image_to_predict = x_test[0:1] | |
| # Get the model's prediction. | |
| predictions = model.predict(image_to_predict) | |
| # Print the predicted class (the one with the highest probability). | |
| predicted_class = tf.argmax(predictions[0]).numpy() | |
| print(f"Predicted class: {predicted_class}") | |
| # Display the model summary to confirm it's loaded correctly. | |
| model.summary() |