TensorFlow Lite for Microcontrollers is designed to work with as little as ___ KB of flash and ___ KB of SRAM

 __________ involves pruning at the level of individual neurons, connections, or weights within a layer of the neural network.

In the program (sine model) the input and output are treated as tensor of what dimension?

_____________________integrating the pruning process directly into the training phase of the neural network.

When harvesting power to run a TinyML model which approach would yield the least effective means?

What was the ration used for training, validation, and testing for the sine application?

Which of these is not a benefit of pruning?

Which of these components in your embedded platform will use the most amount of power?

If you have a model that takes 50 million operations to execute, and you have a chip like an Arm Cortex-M4 running at 24 MHz, How many Joules are required per inference if your chip set used 1 mW?

A ___________ value indicates less randomness or information content, suggesting that a connection or feature is less critical to the network’s overall function and can be pruned