8e12624641eab7a7e73497ea96124bb39aca1c01,memcnn/models/revop.py,InvertibleCheckpointFunction,forward,#,13

Before Change

        ctx.num_bwd_passes = num_bwd_passes

        with torch.no_grad():
            x = input_t.detach()  // Makes a detached copy which shares the storage
            output = ctx.fn(x)

        detached_output = output.detach_()  // Detaches y in-place (inbetween computations can now be discarded)

After Change

        ctx.num_bwd_passes = num_bwd_passes
        ctx.num_inputs = num_inputs

        input_t = inputs_and_weights[:num_inputs]
        ctx.input_requires_grad = [element.requires_grad for element in input_t]

        with torch.no_grad():
            // Makes a detached copy which shares the storage

Instances