YuhengHuang42 opened a new issue #8184: URL: https://github.com/apache/tvm/issues/8184
## Description A similar bug was found in ONNX model, and fixed by this PR: https://github.com/apache/tvm/pull/7208 However, for Pytorch importation, the bug still exists. By definition: [PRelu definition in pytorch](https://pytorch.org/docs/stable/generated/torch.nn.PReLU.html?highlight=prelu#torch.nn.PReLU), num_parameters can be set at 1 or the number of channels at input. However, currently PReLU in TVM seems to support num_parameters = number of channels only. So there will be an error if you set num_parameters = 1 while give input channels > 1. And please notice that **by default** Pytorch set the num_parameters = 1. (Currently, the workaround is to turn Pytorch model to ONNX model first, and then import the ONNX model.) ## Code to reproduce ```python import torch import tvm from tvm import relay minimal_example = torch.nn.Sequential( torch.nn.PReLU(num_parameters=1) ) minimal_example.eval() input_shape = (1, 6, 10, 10) random_input = torch.randn(input_shape) trace = torch.jit.trace(minimal_example, random_input) input_info = [("input0", input_shape)] mod, params = tvm.relay.frontend.from_pytorch(trace, input_info) ``` ## Environment TVM version: 0.8.dev0 at cc3d60e6cd4ec674070fd4a5705723c17ee59389 Pytorch version: 1.8.1 OS version: macOS 10.15.7 -- 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]
