Tensorflow saved model to tensorRT saved model ... variables directory is empty ?

AI & Data Science Deep Learning (Training & Inference) TensorRT
1

Hello

I have successfully optimized a tensorflow object detection saved model with tensorrt and loaded it into tensorflow serving 1.13.1.

I noticed the variables directory is empty. Is that to be expected?

Thanks in advance

BTW – I noticed this output during the optimization …

Use standard file APIs to check for files with this prefix.
WARNING:tensorflow:From /usr/local/lib/python2.7/dist-packages/tensorflow/contrib/tensorrt/python/trt_convert.py:301: convert_variables_to_constants (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.

2

Hi,

Please provide a small repro package with the original model and the conversion code so we can debug this further.

Thanks,
NVIDIA Enterprise Support

3

Hi Nvidia

I happened to attend the Ask the expert session on TensorRT at GTC last week and had my questions answered. Thanks!

(Yes, it is expected that the variables directory is empty.)

4

Good to hear! Thanks for the update.

NVIDIA Enterprise Support