nn_equal_f16()

static uint8_t nn_equal_f16 ( float16_t  golden, float16_t  actual, float  rel_err  )

inlinestatic

nn_equal_f32

Checks if two half-precision floating-point numbers are equal within a relative error.

Parameters

golden	The expected value.
actual	The actual value.
rel_err	The relative error tolerance.

Returns

1 if the numbers are equal within the relative error, 0 otherwise.