has an _output_shapes attribute inconsistent with the GraphDef for output #3: Dimension 0 in both shapes must be equal, but are 0 and 1. Shapes are [0] and [1].

[FANzhaoxin666]

Hi! I can't load the pre-trained model. Can you help me?

[TimoBolkart]

Can you please give more details, why you cannot download the pre-trained model?

[FANzhaoxin666]

Can you please give more details, why you cannot download the pre-trained model? Thanks. I didn't make any changes of the source code except that I use tensorflow 2.4.0. My GPU is 3090, there are I must use cuda11 we tensorflow 2.4.0 I import tensorflow like:

import tensorflow.compat.v1 as tf tf.enable_resource_variables() tf.disable_v2_behavior()

However, the pretrained model can't be load. I suspect it is the tensorflow version caused the problem.

[zhanchao019]

Can you please give more details, why you cannot download the pre-trained model? Thanks. I didn't make any changes of the source code except that I use tensorflow 2.4.0. My GPU is 3090, there are I must use cuda11 we tensorflow 2.4.0 I import tensorflow like:

import tensorflow.compat.v1 as tf tf.enable_resource_variables() tf.disable_v2_behavior()

However, the pretrained model can't be load. I suspect it is the tensorflow version caused the problem.

You can use nvidia-tensorflow to run tf 1.x on 30series

[avacaondata]

Tensorflow always causes problems, it would be great to migrate this to PyTorch.

[sunkymepro]

I also meet the same question. I use tensorflow 2.7.2, my GPU is 3090, cuda is 11.2. Have you ever solve this problem?

[Langwenchong]

I also meet this question,have you ever solve this problem now?