Skip to content

There is a bug in the memory allocation algorithm under multi branch PAD operation #109

Description

@HomiKetalys

When there is a branch structure in the model as shown in the following figure, the input tensor of branch B should only be read before it is used by branch B.
image
As shown in the following figure, the tensor address of this branch is 49152. Branch A can use it correctly.
image
As shown in the figure below, the input tensor address of branch B is 0, which should be 49152. On the right side of the graph, it can be seen that the tensor at 49152 has been modified, which should not appear before branch B. In addition, the input tensor shape of branch B is also incorrect.
image

Here is my tflite file.
model.zip

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions