Hi,all,I have some problem about ONNX Prelu.I convert my pytorch model to onnx
,but got error:
In `main`:
#[version = "0.0.5"]
fn (%inputL: Tensor[(1, 3, 512, 960), float32],
%feature_extraction.firstconv.conv1.0.0.weight: Tensor[(48, 3, 1, 1), float32],
%feature_extraction.firstconv.conv1.0.1.weight: Tensor[(48), float32],
%feature_extraction.firstconv.conv1.0.1.bias: Tensor[(48), float32],
%feature_extraction.firstconv.conv1.0.1.running_mean: Tensor[(48), float32],
%feature_extraction.firstconv.conv1.0.1.running_var: Tensor[(48), float32],
%v1552: Tensor[(1, 1, 1), float32]) {
%0 = nn.conv2d(%inputL, %feature_extraction.firstconv.conv1.0.0.weight,
padding=[0, 0, 0, 0], kernel_size=[1, 1]);
%1 = nn.batch_norm(%0, %feature_extraction.firstconv.conv1.0.1.weight,
%feature_extraction.firstconv.conv1.0.1.bias,
%feature_extraction.firstconv.conv1.0.1.running_mean,
%feature_extraction.firstconv.conv1.0.1.running_var);
%2 = %1.0;
%3 = reshape(%v1552, newshape=[-1]);
nn.prelu(%2, %3) in particular dimension 0 conflicts 48 does not match 1;
unable to unify: Tensor[(48), float32]` and `Tensor[(1), float32]`;
I have read the doc about prelu:
> ### **PRelu**
>
> PRelu takes input data (Tensor) and slope tensor as input, and produces one
> output data (Tensor) where the function `f(x) = slope * x for x < 0` ,
> `f(x) = x for x >= 0` ., is applied to the data tensor elementwise. This
> operator supports **unidirectional broadcasting** (tensor slope should be
> unidirectional broadcastable to input tensor X); for more details please
> check [the
> doc](https://github.com/onnx/onnx/blob/master/docs/Broadcasting.md).
>
> #### Version
>
> This version of the operator has been available since version 9 of the
> default ONNX operator set.
>
> Other versions of this operator:
> [1](https://github.com/onnx/onnx/blob/master/docs/Changelog.md#PRelu-1),
> [6](https://github.com/onnx/onnx/blob/master/docs/Changelog.md#PRelu-6),
> [7](https://github.com/onnx/onnx/blob/master/docs/Changelog.md#PRelu-7)
>
> #### Inputs
>
> X : T
>
> Input tensor
>
> slope : T
>
> Slope tensor. The shape of slope can be smaller then first input X; if so,
> its shape must be unidirectional broadcastable to X
>
> #### Outputs
>
> Y : T
>
> Output tensor (same size as X)
my model have the prelu op,the slope shape is (1,1,1)
the Input tensor shape is (1,48,512,960)
Maybe tvm prelu can not support broadcast?
Thanks!
---
[Visit
Topic](https://discuss.tvm.apache.org/t/prelu-op-can-not-support-broadcast/7880/1)
to respond.
You are receiving this because you enabled mailing list mode.
To unsubscribe from these emails, [click
here](https://discuss.tvm.apache.org/email/unsubscribe/422b0b2317f6c5c8f229909f87b5a6b19095559f8605fd92ba508096468be232).
