Skip to content

Conversation

@jacquesqiao
Copy link
Member

design #3835

@jacquesqiao jacquesqiao requested a review from reyoung September 6, 2017 19:10
message LoDTensorDesc {
required DataType data_type = 1;
repeated int32 dims = 2; // [UNK, 640, 480] is saved as [-1, 640, 480]
optional int32 lod_level = 3 [ default = 0 ];
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

lod_level is not enough for InferShape. The whole lod information should be decided while inferring shape.

Copy link
Collaborator

@reyoung reyoung left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I am not sure whether lod_level itself is enough for inferring shape, but rest of this PR looks good to me.

So I just approve it. If we find lod_level is not enough later, we would change it later.

@jacquesqiao jacquesqiao merged commit 8162e2c into PaddlePaddle:develop Sep 6, 2017
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

2 participants