rafzi opened a new pull request, #11879:
URL: https://github.com/apache/tvm/pull/11879

   Reading the following model did not work for me: 
https://github.com/google/mediapipe/blob/master/mediapipe/modules/face_landmark/face_landmark.tflite
   
   with the error:
   
   ```
     File "tvm/python/tvm/relay/frontend/tflite.py", line 3092, in convert_prelu
       self.get_tensor_value(alpha_tensor), dtype=alpha_tensor_type_str
     File "tvm/python/tvm/relay/frontend/tflite.py", line 520, in 
get_tensor_value
       return np.frombuffer(data, dtype=dtype).reshape(shape)
   TypeError: memoryview: a bytes-like object is required, not 'int'
   ```
   
   The input of the PreLU op is the output of a Dequantize op, so not a 
constant, therefore its tensor buffer is not set.
   
   I was unable to test this successfully because the existing test failed with 
the following error, even before any changes.
   
   ```
   The Relay type checker is unable to show the following types match.
   In particular dimension 0 conflicts: 3 does not match 96.
   The Relay type checker is unable to show the following types match.
   In particular `Tensor[(96), float32]` does not match `Tensor[(3), float32]`
   ```
   
   on
   
   ```
   _test_prelu(
           np.random.uniform(-5, 5, size=(1, 32, 32, 3)).astype("float32"),
           np.full((32, 3), 0.2, dtype="float32"),
       )
   ```
   
   @FrozenGene @tqchen 


-- 
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