NVMLError_NotSupported when creating Trainer() object

Beginners
1

Hello, I am trying to finetune the CLIP model on my local computer for the DLRSD dataset. I created this jupyter notebook.

I have two problems:

  1. As you can see, when creating the Trainer() object in cell 16, I get the error: NVMLError_NotSupported: Not Supported. And this happens irrespective of whether I use CUDA:1 or CPU as my device for my model (cells 2, 3, 4). What am I doing incorrectly?
  2. Also, is it possible to finetune only a few layers of the CLIP model?

Any advice would be truly appreciated.

For reference, I am running an NVIDIA Quadro K1200 GPU with CUDA v12.

PS C:\Users\admin> nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2024 NVIDIA Corporation
Built on Thu_Mar_28_02:30:10_Pacific_Daylight_Time_2024
Cuda compilation tools, release 12.4, V12.4.131
Build cuda_12.4.r12.4/compiler.34097967_0
2

The error stemmed from codecarbon. Doing pip uninstall codecarbon -y got me up and running again.

Reference

3

This topic was automatically closed 12 hours after the last reply. New replies are no longer allowed.