tensorflow-patterns

This plugin provides patterns and best practices for production TensorFlow workflows, including Keras model building with multi-input support, tf.data pipeline construction with correct operation ordering, custom layers and loss functions, SavedModel export with serving signatures, and TFLite quantization (both PTQ and QAT) for edge deployment. It covers TensorFlow Serving integration, TFRecord parsing, mixed precision training, and TFX Transform components for train-serve consistency.