bug in batchsize when unspecified in update()

        # Optimize policy for multiple epochs:

        mb_losses, mb_vlosses, mb_plosses = [], [], []

        batchsize = len(ret) if batchsize is not None else batchsize

        batchsize = len(ret) if batchsize is None else batchsize

        for e in range(epochs):

            permutations = torch.randperm(ret.size()[0])

            # iterate over minibatches (mb) for stochastic updates