Poor error message when trying to jit a function instead of a module (RuntimeError: Cannot insert a Tensor that requires grad as a constant.) #55282
Description
🐛 Bug
Confusing error when using a function which runs a model.
RuntimeError: Cannot insert a Tensor that requires grad as a constant. Consider making it a parameter or input, or detaching the gradient
Better would be to recommend using a nn.Module.
To Reproduce
import torch.nn as nn
import torch
model = nn.Conv2d(5, 10, (3, 3))
def run(batch):
return model(batch)
traced = torch.jit.trace(run, torch.zeros(1, 5, 64, 64))
Expected behavior
Better error message for how to fix the problem.
Environment
Please copy and paste the output from our
environment collection script
(or fill out the checklist below manually).
You can get the script and run it with:
Collecting environment information...
PyTorch version: 1.8.1
Is debug build: False
CUDA used to build PyTorch: 11.1
ROCM used to build PyTorch: N/A
OS: Ubuntu 20.04.1 LTS (x86_64)
GCC version: (Ubuntu 9.3.0-17ubuntu1~20.04) 9.3.0
Clang version: 9.0.0 (https://github.com/conda-forge/clangdev-feedstock 284a3d5d88509307bcfba64b055653ee347371db)
CMake version: version 3.16.0
Python version: 3.7 (64-bit runtime)
Is CUDA available: True
CUDA runtime version: 11.1.105
GPU models and configuration: GPU 0: GeForce RTX 2070
Nvidia driver version: 455.32.00
cuDNN version: Probably one of the following:
/usr/lib/x86_64-linux-gnu/libcudnn.so.8.1.1
/usr/lib/x86_64-linux-gnu/libcudnn_adv_infer.so.8.1.1
/usr/lib/x86_64-linux-gnu/libcudnn_adv_train.so.8.1.1
/usr/lib/x86_64-linux-gnu/libcudnn_cnn_infer.so.8.1.1
/usr/lib/x86_64-linux-gnu/libcudnn_cnn_train.so.8.1.1
/usr/lib/x86_64-linux-gnu/libcudnn_ops_infer.so.8.1.1
/usr/lib/x86_64-linux-gnu/libcudnn_ops_train.so.8.1.1
HIP runtime version: N/A
MIOpen runtime version: N/A
Versions of relevant libraries:
[pip3] numpy==1.19.2
[pip3] numpy-quaternion==2020.9.5.14.42.2
[pip3] pytorch-tools==0.1
[pip3] torch==1.8.1
[pip3] torch2trt==0.1.0
[pip3] torchaudio==0.8.0a0+e4e171a
[pip3] torchvision==0.2.2
[pip3] trtorch==0.3.0a0
[conda] Could not collect
cc @gmagogsfm