I know this not related about pytorch but possible about pytorch symblic to onnx, I have this model and I converted to onnx, but somehow the model can not converted to tensorrt using onnx2trt:
class MG(nn.Module): def __init__(self): super().__init__() # for test if torch.cat([bool, bool]) can convert def forward(self, x, b): preds = F.conv2d(x, b, stride=1) return preds torch_model = MG() x = torch.randn([1, 4, 24, 24]) b = torch.randn([8, 4, 3, 3]) torch_out = torch_model(x, b) # Export the model torch.onnx.export(torch_model, # model being run (x, b), "a.onnx", export_params=True, # store the trained parameter weights inside the model file opset_version=11, # the ONNX version to export the model to do_constant_folding=True, verbose=True) print('Done!')
It’s very simple model. I want accerlerate it with tensorrt, but using onnx2trt convert onnx, I got error:
 Assertion failed: ctx->network()->hasExplicitPrecision() && "TensorRT only supports multi-input conv for explicit precision QAT networks!"
A reference is  Assertion failed: ctx->network()->hasExplicitPrecision() && "TensorRT only supports multi-input conv for explicit precision QAT networks!" · Issue #645 · onnx/onnx-tensorrt · GitHub
I found the onnx-tensorrt is no one maintain it but I really need it, so ask to pytorch community doesn anyone able to solve it?
BTW: this model seems can converted in pytorch 1.6 but can not converted to pytorch 1.7.