2bf236792fee65bd2ceab922451a230a3a986cf6,memcnn/models/tests/test_revop.py,,test_reversible_block_fwd_bwd,#,25

Before Change

                        keep_input = keep_input_sub or implementation_bwd == -1 or implementation_fwd == -1
                        // print(bwd, coupling, keep_input, implementation_fwd, implementation_bwd)
                        // test with zero padded convolution
                        X = Variable(torch.from_numpy(data.copy()))
                        Ytarget = Variable(torch.from_numpy(target_data.copy()))
                        Xshape = X.shape
                        Gm2 = copy.deepcopy(Gm)
                        rb = revop.ReversibleBlock(Gm2, coupling=coupling, implementation_fwd=implementation_fwd,
                                                   implementation_bwd=implementation_bwd, keep_input=keep_input)
                        rb.train()
                        rb.zero_grad()

                        optim = torch.optim.RMSprop(rb.parameters())
                        optim.zero_grad()
                        if not bwd:
                            Xin = X.clone()
                            Y = rb(Xin)
                            Yrev = Y.clone()
                            Xinv = rb.inverse(Yrev)
                        else:
                            Xin = X.clone()
                            Y = rb.inverse(Xin)
                            Yrev = Y.clone()
                            Xinv = rb(Yrev)
                        loss = torch.nn.MSELoss()(Y, Ytarget)

                        // has input been retained/discarded after forward (and backward) passes?
                        if keep_input:

After Change

                        keep_input = keep_input_sub or implementation_bwd == -1 or implementation_fwd == -1
                        // print(bwd, coupling, keep_input, implementation_fwd, implementation_bwd)
                        // test with zero padded convolution
                        X = torch.from_numpy(data.copy())
                        Ytarget = torch.from_numpy(target_data.copy())
                        Xshape = X.shape
                        Gm2 = copy.deepcopy(Gm)
                        rb = revop.ReversibleBlock(Gm2, coupling=coupling, implementation_fwd=implementation_fwd,
                                                   implementation_bwd=implementation_bwd, keep_input=keep_input)
                        rb.train()
                        rb.zero_grad()

                        optim = torch.optim.RMSprop(rb.parameters())
                        optim.zero_grad()
                        if not bwd:
                            Xin = X.clone()
                            Y = rb(Xin)
                            Yrev = Y.clone()
                            Xinv = rb.inverse(Yrev)
                        else:
                            Xin = X.clone()
                            Y = rb.inverse(Xin)
                            Yrev = Y.clone()
                            Xinv = rb(Yrev)
                        loss = torch.nn.MSELoss()(Y, Ytarget)

                        // has input been retained/discarded after forward (and backward) passes?
                        if keep_input:

Instances