$$\nabla L = \begin{pmatrix}
\frac{\partial L}{\partial w_1} \\
\frac{\partial L}{\partial w_2} \\
\vdots \\
\frac{\partial L}{\partial w_n} \end{pmatrix}$$
This requires computing the derivatives of the terms like
$$\log {1 \over {1+e^{-\vec x \cdot \vec w}}} = \log {1 \over {1+e^{-(x_1 \cdot w_1 + x_2 \cdot w_2 + \, \dots \, + x_n \cdot w_n)}}}$$
where you can use
$$\frac{\partial}{\partial x} \left( \log\frac{1}{1+e^{-(a+bx)}} \right) = \frac{b}{1+e^{(a+bx)}}$$
and
$$\frac{\partial}{\partial x} \left( \log(1-\frac{1}{1+e^{-(a+bx)}} ) \right) = \frac{b}{1-e^{-(a+bx)}}$$
Filling that in you get
$$\frac{\partial}{\partial w_j} L = \frac{\bar{y_i} x_j}{{1+e^{\vec x \cdot \vec w}}} - \frac{(1-\bar{y_i}) x_j }{{1-e^{-\vec x \cdot \vec w}}}$$
and
$$\nabla L = \left( \frac{\bar{y_i}}{{1+e^{\vec x \cdot \vec w}}} - \frac{(1-\bar{y_i}) }{{1-e^{-\vec x \cdot \vec w}}} \right) \vec{x} $$