Intel® DevCloud
Help for those needing help starting or connecting to the Intel® DevCloud
1628 Discussions

OpenVINO Speedup Training?

CLi37
Beginner
619 Views

I apply the codes below to enable openvino and generate results during training

 

import openvino_tensorflow as otf
print(otf.list_backends())
otf.set_backend('CPU')

2022-05-30 07:50:33.449595: OVTF Summary -> 113 out of 1041 nodes in the graph (10%) are now running with OpenVINO™ backend
2022-05-30 07:50:33.533068: OVTF Summary -> 0 out of 5 nodes in the graph (0%) are now running with OpenVINO™ backend
2022-05-30 07:50:33.535766: OVTF Summary -> 0 out of 6 nodes in the graph (0%) are now running with OpenVINO™ backend
2022-05-30 07:50:33.537710: OVTF Summary -> 0 out of 5 nodes in the graph (0%) are now running with OpenVINO™ backend
1562/1562 [==============================] - ETA: 0s - loss: 4.0426 - accuracy: 0.0826 - top-5-accuracy: 0.2652

Later the display is 

1562/1562 [==============================] - ETA: 0s - loss: 3.5136 - accuracy: 0.1585 - top-5-accuracy: 0.42112022-05-30 08:07:17.139680: OVTF Summary -> 0 out of 8 nodes in the graph (0%) are now running with OpenVINO™ backend
2022-05-30 08:07:17.144779: OVTF Summary -> 0 out of 14 nodes in the graph (0%) are now running with OpenVINO™ backend
2022-05-30 08:07:17.148742: OVTF Summary -> 0 out of 10 nodes in the graph (0%) are now running with OpenVINO™ backend
2022-05-30 08:07:17.151411: OVTF Summary -> 0 out of 6 nodes in the graph (0%) are now running with OpenVINO™ backend
2022-05-30 08:07:17.174284: OVTF Summary -> 0 out of 6 nodes in the graph (0%) are now running with OpenVINO™ backend
2022-05-30 08:07:17.178607: OVTF Summary -> 0 out of 6 nodes in the graph (0%) are now running with OpenVINO™ backend
2022-05-30 08:07:17.170349: W tensorflow/core/framework/dataset.cc:768] Input of GeneratorDatasetOp::Dataset will not be optimized because the dataset does not implement the AsGraphDefInternal() method needed to apply optimizations.

 

Does this mean that OpenVINO does not make any OpenVINO optimization?

0% run OpenVINO backend?

 

0 Kudos
1 Solution
Hari_B_Intel
Moderator
547 Views

Hi CLi37,


Thank you for your patience and understanding, For your information, OpenVINO integration for Tensorflow is to accelerate inference of the Tensorflow model and this is not designed for Tensorflow training.

Furthermore, Intel DevCloud is not recommended for training the mode.


Here is the documentation for OpenVINO™ Integration with TensorFlow* (intel.com) you can review.


Hope this information help


Thank you


View solution in original post

0 Kudos
3 Replies
VaradJ_Intel
Moderator
593 Views

Hi,

 

Thanks for posting in Intel Communities.

 

Can you share the Intel DevCloud that you are using? Is it Intel DevCloud for oneAPI or FPGA or Edge?

 

Thank You


0 Kudos
Hari_B_Intel
Moderator
570 Views

Hi CLi37

we might need some time to further investigate this issue. please allow us some time to investigate it.


Thank you


0 Kudos
Hari_B_Intel
Moderator
548 Views

Hi CLi37,


Thank you for your patience and understanding, For your information, OpenVINO integration for Tensorflow is to accelerate inference of the Tensorflow model and this is not designed for Tensorflow training.

Furthermore, Intel DevCloud is not recommended for training the mode.


Here is the documentation for OpenVINO™ Integration with TensorFlow* (intel.com) you can review.


Hope this information help


Thank you


0 Kudos
Reply