I think it is (or should be) the latter (reasoning below). But many writers (e.g. Stewart, Calculus) use the former instead.
So I'm confused and hence this question. (Maybe this is just one of those phrases/terms that have different definitions according to the writer?)
Say $x_t$ is population, $x_{2021}=100000$, and $x_{2022}=101000$. Then most people would say that the annual growth rate of population was $1\%$: $$g_x=\frac{\frac{x_{2022}-x_{2021}}{1\ \text{year}}}{x_{2021}}=\frac{\frac{\Delta x}{\Delta t}}{x}.$$
(I don't think many would say that the growth rate of population was $1000$.)
Now going from the above discrete case to the continuous case, shouldn't we similarly have that
$$g_x=\frac{\frac{\mathrm{d}x}{\mathrm{d}t}}{x}?$$
Many calculus textbooks define $\frac{\mathrm{d}x}{\mathrm{d}t}$ as the (absolute) growth rate and $\frac{\frac{\mathrm{d}x}{\mathrm{d}t}}{x}$ as the relative growth rate.
But many social scientists (economists, demographers) often simply call the latter the "growth rate".
So, growth rate is $\frac{\mathrm{d}x}{\mathrm{d}t}$ in math but $\frac{\frac{\mathrm{d}x}{\mathrm{d}t}}{x}$ in the social sciences.
Briggs, Cochran, Gillett, and Schulz (Calculus: Early Transcendentals, 2019), p. 493:
Hass, Heil, and Weir (Thomas' Calculus, 2018), p. 541:
Our World in Data:
Mankiw (Principles of Economics, 2020):