Civitasv opened a new issue, #15180: URL: https://github.com/apache/tvm/issues/15180
### Expected behavior I'm translating onnx file to relax IR format using code snippets like below. ```py from tvm.relax.frontend.onnx import from_onnx onnx_file = "/path/to/sigmoid.onnx" model = onnx.load(onnx_file) ir_mod, params = from_onnx(model, keep_params_in_input=True) ``` The `sigmoid.onnx` is generated using `onnx.save_model(model, "sigmoid.onnx")` using this https://github.com/apache/tvm/blob/0dba0e3ed297a5f80dcd02031bb8c2664a3b5c4b/tests/python/relax/test_frontend_onnx.py#L399 and this https://github.com/apache/tvm/blob/0dba0e3ed297a5f80dcd02031bb8c2664a3b5c4b/tests/python/relax/test_frontend_onnx.py#L188. And this test works as expected. But after I save it to a file and load from this file using `onnx.load`, it does not work. ### Actual behavior It will show error: ```txt --------------------------------------------------------------------------- TVMError Traceback (most recent call last) Cell In[83], line 123 119 model = onnx.load(onnx_file) --> 123 ir_mod, params = from_onnx(model, keep_params_in_input=True) File [~/miniconda3/envs/tvm-venv/lib/python3.10/site-packages/tvm-0.13.dev667](https://vscode-remote+ssh-002dremote-002b10-002e224-002e105-002e236.vscode-resource.vscode-cdn.net/home/wyc/husen/sandbox/~/miniconda3/envs/tvm-venv/lib/python3.10/site-packages/tvm-0.13.dev667)+g730d380ff-py3.10-linux-x86_64.egg/tvm/ir/module.py:127, in IRModule.__getitem__(self, var) 125 if isinstance(var, _expr.GlobalVar): 126 return _ffi_api.Module_Lookup(self, var) --> 127 return _ffi_api.Module_LookupDef(self, var) File [~/miniconda3/envs/tvm-venv/lib/python3.10/site-packages/tvm-0.13.dev667](https://vscode-remote+ssh-002dremote-002b10-002e224-002e105-002e236.vscode-resource.vscode-cdn.net/home/wyc/husen/sandbox/~/miniconda3/envs/tvm-venv/lib/python3.10/site-packages/tvm-0.13.dev667)+g730d380ff-py3.10-linux-x86_64.egg/tvm/_ffi/_ctypes/packed_func.py:238, in PackedFuncBase.__call__(self, *args) 226 ret_tcode = ctypes.c_int() 227 if ( 228 _LIB.TVMFuncCall( 229 self.handle, (...) 236 != 0 237 ): --> 238 raise get_last_ffi_error() 239 _ = temp_args 240 _ = args TVMError: Traceback (most recent call last): 2: TVMFuncCall 1: tvm::runtime::PackedFuncObj::Extractor::AssignTypedLambda(tvm::{lambda(tvm::IRModule, tvm::GlobalTypeVar)#6}, std::__cxx11::basic_string, std::allocator >)::{lambda(tvm::runtime::TVMArgs const&, tvm::runtime::TVMRetValue*)#1}> >::Call(tvm::runtime::PackedFuncObj const*, tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) 0: tvm::runtime::TVMMovableArgValueWithContext_::operator tvm::GlobalTypeVar() const 3: TVMFuncCall 2: tvm::runtime::PackedFuncObj::Extractor::AssignTypedLambda(tvm::{lambda(tvm::IRModule, tvm::GlobalTypeVar)#6}, std::__cxx11::basic_string, std::allocator >)::{lambda(tvm::runtime::TVMArgs const&, tvm::runtime::TVMRetValue*)#1}> >::Call(tvm::runtime::PackedFuncObj const*, tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) 1: tvm::runtime::TVMMovableArgValueWithContext_::operator tvm::GlobalTypeVar() const 0: tvm::GlobalTypeVar tvm::runtime::TVMPODValue_::AsObjectRef() const File "/home/wyc/husen/sandbox/tvm/include/tvm/runtime/packed_func.h", line 777 TVMError: In function ir.Module_LookupDef(0: IRModule, 1: GlobalTypeVar) -> relay.TypeData: error while converting argument 1: [16:02:13] /home/wyc/husen/sandbox/tvm/include/tvm/runtime/packed_func.h:1890: --------------------------------------------------------------- An error occurred during the execution of TVM. For more information, please see: https://tvm.apache.org/docs/errors.html --------------------------------------------------------------- Check failed: type_code_ == kTVMObjectHandle (0 vs. 8) : expected Object but got int ``` ### Environment OS: Ubuntu 20.04 TVM version: tvm-0.13.dev667+g730d380ff-py3.10-linux-x86_64 ### Steps to reproduce ```py from tvm.relax.frontend.onnx import from_onnx onnx_file = "/path/to/sigmoid.onnx" model = onnx.load(onnx_file) ir_mod, params = from_onnx(model, keep_params_in_input=True) ``` ### Triage * 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]
