manuel-loew-infineon opened a new issue, #16142:
URL: https://github.com/apache/tvm/issues/16142

   ### Expected behavior
   
   AoT Executor and Graph Executor should treat the names in the same way.
   
   ### Actual behavior
   
   AoT Executor replaces non-alphanumeric characters in the input name while 
Graph Executor does not.
   Setting the input data using the original name e.g. 
   
   `aot_executor.set_input(‘predict_images:0’, input_data)`
   
   results in error ‘cannot find ‘predict_images:0’ among input’ while
   
   `graph_executor.set_input(‘predict_images:0’, input_data)`
   
   is fine. The different behavior is not intuitive.
   
   See 
https://discuss.tvm.apache.org/t/aot-problem-in-microtvm-utvm-host-driven-application/13100
   
   ### Environment
   
   TVM version: 
   
   ### Steps to reproduce
   
   Use a network with an input tensor name containing a non-alphanumeric 
character, e.g. 'serving_default_input_2:0' in
   
[conv1d_batchnorm_pruned.zip](https://github.com/apache/tvm/files/13389703/conv1d_batchnorm_pruned.zip)
   
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to