Adam First Step

The Adam optimiser updates a parameter with $m_t = \beta_1 m_{t-1} + (1-\beta_1) g_t$, $v_t = \beta_2 v_{t-1} + (1-\beta_2) g_t^2$, bias-corrected $\hat m_t = m_t/(1-\beta_1^t)$, $\hat v_t = v_t/(1-\beta_2^t)$ and step $\theta_t = \theta_{t-1} - \eta\,\hat m_t/(\sqrt{\hat v_t} + \epsilon)$, with $m_0 = v_0 = 0$. Take $\beta_1 = 0.9$, $\beta_2 = 0.999$, learning rate $\eta = 0.01$, $\epsilon = 0$ and a first gradient $g_1 = 3.7$. What is the magnitude of the very first parameter update $|\theta_1 - \theta_0|$?