Basic torch nn training on simple signals

[tmolteno]

Using the tmolteno/toynn repository. Create a small torch-based python project in a subfolder, that tries to measure step changes in amplitude of a sine wave signal with additive noise.

• Copy the torch example and get it going.
• Use a pyproject.toml based design with a command line interface
• Modify it to predict the amplitude and frequency of an input signal (regular samples in time).
• Try and train it to detect when the amplitude has changed (as fast as possible).