zhangxiao-stack opened a new issue, #15336:
URL: https://github.com/apache/tvm/issues/15336
I'm translating dynamic shape resnet.onnx file to relax IR format using code
like below. the input shape is [-1*3*-1*-1]
```
onnx_file = "./resnet50.onnx"
model = onnx.load(onnx_file)
ir_mod = from_onnx(model, keep_params_in_input=True)
relax_mod, params = relax.frontend.detach_params(ir_mod)
```
### Expected behavior
imported successfully.
### Actual behavior
It shows error:
```
Traceback (most recent call last):
File "resnet.py", line 34, in <module>
ir_mod = from_onnx(model, keep_params_in_input=True)
File
"/root/.local/lib/python3.8/site-packages/tvm-0.14.dev441+g5dcc25a-py3.8-linux-x86_64.egg/tvm/relax/frontend/onnx/onnx_frontend.py",
line 2175, in from_onnx
return g.from_onnx(graph, opset)
File
"/root/.local/lib/python3.8/site-packages/tvm-0.14.dev441+g5dcc25a-py3.8-linux-x86_64.egg/tvm/relax/frontend/onnx/onnx_frontend.py",
line 1832, in from_onnx
self._construct_nodes(graph)
File
"/root/.local/lib/python3.8/site-packages/tvm-0.14.dev441+g5dcc25a-py3.8-linux-x86_64.egg/tvm/relax/frontend/onnx/onnx_frontend.py",
line 1983, in _construct_nodes
op = self._convert_operator(op_name, inputs, attr, self.opset)
File
"/root/.local/lib/python3.8/site-packages/tvm-0.14.dev441+g5dcc25a-py3.8-linux-x86_64.egg/tvm/relax/frontend/onnx/onnx_frontend.py",
line 2079, in _convert_operator
sym = op_function(self.bb, inputs, attrs, [self._nodes, self._params])
File
"/root/.local/lib/python3.8/site-packages/tvm-0.14.dev441+g5dcc25a-py3.8-linux-x86_64.egg/tvm/relax/frontend/onnx/onnx_frontend.py",
line 1360, in _impl_v13
data_shape = [i.value for i in inputs[0].struct_info.shape]
File
"/root/.local/lib/python3.8/site-packages/tvm-0.14.dev441+g5dcc25a-py3.8-linux-x86_64.egg/tvm/relax/frontend/onnx/onnx_frontend.py",
line 1360, in <listcomp>
data_shape = [i.value for i in inputs[0].struct_info.shape]
File
"/root/.local/lib/python3.8/site-packages/tvm-0.14.dev441+g5dcc25a-py3.8-linux-x86_64.egg/tvm/runtime/object.py",
line 78, in __getattr__
raise AttributeError(f"{type(self)} has no attribute {name}") from None
AttributeError: <class 'tvm.tir.expr.Var'> has no attribute value
[03:44:24] /root/tvm/src/relax/ir/block_builder.cc:64: Warning: BlockBuilder
destroyed with remaining blocks!
```
### Environment
Operating System:linux
TVM version,:unity 5dcc25a [Unity][Relax][Transform] Do not remove MatchCast
for RemoveAllUnused (#15290)
### Steps to reproduce
```
import tvm
import tvm.testing
from tvm.relay import testing
from tvm import relax, relay
from tvm.relax.testing import relay_translator, nn
from tvm.runtime import vm as vm_rt
from tvm.script import relax as R
import numpy as np
import onnx
from tvm.relax.frontend.onnx import from_onnx
from tvm import relax
if __name__ == "__main__":
onnx_file = "./resnet50.onnx"
model = onnx.load(onnx_file)
ir_mod = from_onnx(model, keep_params_in_input=True)
relax_mod, params = relax.frontend.detach_params(ir_mod)
```
### Triage
Please refer to the list of label tags
[here](https://github.com/apache/tvm/wiki/Issue-Triage-Labels) to find the
relevant tags and add them below in a bullet format (example below).
* needs-triage
--
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]
