Torchvision::nms couldn't run

1

I am trying to resolve this issue for almost 12 hrs. Sadly didn’t found that. Please, help me out :pray:.

Here’s what i did:

  1. Installed CUDA 12.4 toolkit and its corresponding pytorch. (Didn’t worked)
  2. Uninstalled pytorch and reinstalled that (didn’t worked)
  3. Uninstalled CUDA 12.4 and installed 12.1 and its corresponding pytorch (Didn’t worked)
  4. Then again reinstalled for that. Didn’t run that code.

My desktop specs: Nvidia RTX 3090 with AMD processor

I tested my installation with this result

1000077184
1000077184592×677 180 KB

(FYI, i tried the similar process for my laptop with 1050ti, it is still working just fine. But at the end of the video object detection my laptop gets blue screen error with Video TDR error)

Error:

“Could not run ‘torchvision::nms’ with arguments from the ‘CUDA’ backend. This could be because the operator doesn’t exist for this backend, or was omitted during the selective/custom build process (if using custom build). If you are a Facebook employee using PyTorch on mobile, please visit Internal Login for possible resolutions. ‘torchvision::nms’ is only available for these backends: [CPU, Meta, QuantizedCPU, BackendSelect, Python, FuncTorchDynamicLayerBackMode, Functionalize, Named, Conjugate, Negative, ZeroTensor, ADInplaceOrView, AutogradOther, AutogradCPU, AutogradCUDA, AutogradXLA, AutogradMPS, AutogradXPU, AutogradHPU, AutogradLazy, AutogradMeta, Tracer, AutocastCPU, AutocastXPU, AutocastCUDA, FuncTorchBatched, BatchedNestedTensor, FuncTorchVmapMode, Batched, VmapMode, FuncTorchGradWrapper, PythonTLSSnapshot, FuncTorchDynamicLayerFrontMode, PreDispatch, PythonDispatcher].”

2

You’ve installed the CPU-only torchvision binary, so uninstall it and install the right one with CUDA support.

1 Like
3

image
image1076×437 25.5 KB

i downloaded using this code.

how do I find the GPU accelerated one? Please help me. I will be eagerly looking forward to your result.

4

image
i have just freshly installed CUDA 12.4 removing the earlier one(even from cache). Above image is the version list showing from “pip list”

“import torch
import torchvision
print(torch.version)
print(torchvision.version)”

→ 2.4.0+cu124
0.19.0+cpu

This version is showing from Jupyter Notebook.

Result: Still the same.
i haven’t tried for CUDA 11.8 because it was from 2022. Should I try that one too???

Please, respond. Thanks in advance