vicalloy opened a new issue #7908:
URL: https://github.com/apache/tvm/issues/7908


   1. The follow code is used to run the model by vm.
   ```python
   import onnx
   import time
   import tvm
   import numpy as np
   import tvm.relay as relay
   
   target = 'llvm'
   
   # model from openvino. 
https://github.com/openvinotoolkit/open_model_zoo/blob/master/models/intel/face-detection-0200/description/face-detection-0200.md
   model_path = 'face-detection-0200.org.onnx'
   onnx_model = onnx.load(model_path)
   
   shape = [1, 3, 256, 256]
   
   input_name = "image"
   shape_dict = {
           "image": shape,
           }
   
   mod, params = relay.frontend.from_onnx(onnx_model, shape_dict)
   with tvm.transform.PassContext(opt_level=3):
       exe = relay.vm.compile(mod, target="llvm", target_host=None, 
params=params)
   
   data = np.random.uniform(size=shape).astype("float32")
   ctx = tvm.cpu()
   vm = tvm.runtime.vm.VirtualMachine(exe, ctx)
   out = vm.run(data)
   print(out.asnumpy())
   ```
   
   2. The error
   ```
     File "run-tvm.py", line 30, in <module>
       print(out.asnumpy())
     File "/home/hu/tvm/python/tvm/runtime/object.py", line 63, in __getattr__
       return _ffi_node_api.NodeGetAttr(self, name)
     File "/home/hu/tvm/python/tvm/_ffi/_ctypes/packed_func.py", line 237, in 
__call__
       raise get_last_ffi_error()
   TypeError: Traceback (most recent call last):
     3: TVMFuncCall
     2: _ZNSt17_Function_handlerIFvN3t
     1: tvm::NodeGetAttr(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*)
     0: tvm::ReflectionVTable::GetAttr(tvm::runtime::Object*, 
tvm::runtime::String const&) const
     File "/home/hu/tvm/include/tvm/node/reflection.h", line 390
   TypeError: runtime.ADT is not registered via TVM_REGISTER_NODE_TYPE
   ```
   
   3. The model: 
[face-detection-0200.org.zip](https://github.com/apache/tvm/files/6355337/face-detection-0200.org.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.

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


Reply via email to