sine_model

Sine approximation model

• Precision: FP32
• Input: [0, 2π]
• Output: [-1, 1]

Run:

$ make
$ ./sine_model

How was this model trained and inferenced ?

1. Train model on tensorflow, and export the tflite model.
2. Using netron to extract the weights and biases as .npy files
3. Use the npy_convert script to convert the npy files into raw bin files.
4. Load the weights into memory allocated during runtime.
5. run the necessary microkernel operations and inference.