tensorflow 2.0

import numpy as np

# Use 'import tensorflow as tf' if full TF is installed, 

# otherwise use the standalone runtime import:

import tflite_runtime.interpreter as tflite

# 1. Load the TFLite model and allocate tensors

interpreter = tflite.Interpreter(model_path="model.tflite")

interpreter.allocate_tensors()

# 2. Get details about the model's expected inputs and outputs

input_details = interpreter.get_input_details()

output_details = interpreter.get_output_details()

# 3. Prepare your dummy input data (must match the model's expected shape and type)

# For example: 1 image, 28x28 pixels, float32 format

input_shape = input_details[0]['shape']

input_data = np.array(np.random.random_sample(input_shape), dtype=np.float32)

# 4. Set the input tensor data

interpreter.set_tensor(input_details[0]['index'], input_data)

# 5. Invoke the interpreter to run the math/prediction

interpreter.invoke()

# 6. Retrieve the prediction results

output_data = interpreter.get_tensor(output_details[0]['index'])

print("Inference completed successfully!")

print("Output prediction array shape:", output_data.shape)

print("Raw prediction scores:", output_data)

import tensorflow as tf

# 1. Load your existing trained Keras model

# (Or replace with tf.keras.models.load_model('my_model.h5'))

model = tf.keras.Sequential([

    tf.keras.layers.Flatten(input_shape=(28, 28)),

    tf.keras.layers.Dense(128, activation='relu'),

    tf.keras.layers.Dense(10, activation='softmax')

])

# 2. Initialize the TFLite Converter

converter = tf.lite.TFLiteConverter.from_keras_model(model)

# 3. Optional: Apply post-training quantization to compress the model size

converter.optimizations = [tf.lite.Optimize.DEFAULT]

# 4. Convert the model

tflite_model = converter.convert()

# 5. Save the optimized .tflite file

with open('model.tflite', 'wb') as f:

    f.write(tflite_model)

print("Model successfully converted and saved as 'model.tflite'.")