fuck you facebook

def plot_grad_flow(model):
    """
    Plots the gradients flowing through different layers in the net during training.
    Can be used for checking for possible gradient vanishing / exploding problems.
    """
    ave_grads = []
    max_grads = []
    layers = []
    for n, p in model.named_parameters():
        if (p.requires_grad) and ("bias" not in n and "norm" not in n):
            layers.append(n[:n.find('.weight')])
            ave_grads.append(p.grad.abs().mean())
            max_grads.append(p.grad.abs().max())
    plt.bar(np.arange(len(max_grads)), max_grads, alpha=0.1, lw=1, color="c")
    plt.bar(np.arange(len(max_grads)), ave_grads, alpha=0.1, lw=1, color="b")
    plt.hlines(0, 0, len(ave_grads) + 1, lw=2, color="k")
    plt.xticks(range(0, len(ave_grads), 1), layers, rotation="vertical")
    plt.xlim(left=0, right=len(ave_grads))
    plt.ylim(bottom=-0.001, top=0.02)  # zoom in on the lower gradient regions
    plt.xlabel("Layers")
    plt.ylabel("average gradient")
    plt.title("Gradient flow")
    plt.grid(True)
    plt.legend([Line2D([0], [0], color="c", lw=4),
                Line2D([0], [0], color="b", lw=4),
                Line2D([0], [0], color="k", lw=4)], ['max-gradient', 'mean-gradient', 'zero-gradient'])
    plt.tight_layout()