RuntimeError: The size of tensor a (6) must match the size of tensor b (7) at non-singleton dimension 2

[YounghoJo01]

Search before asking

• I have searched the YOLOv5 issues and discussions and found no similar questions.

Question

I installed CUDA 12.6, cuDNN, and YOLOv5 to train a model for traffic light detection. However, I'm encountering an error while running tests on a subset of my dataset. Due to this error, I'm unable to resolve the issue. Please, I need your help to find a solution.

Additional

No response

[UltralyticsAssistant]

👋 Hello @YounghoJo01, thank you for reaching out and using YOLOv5 🚀!

We're here to help! It seems you're encountering a RuntimeError related to mismatched tensor sizes. To assist you better, please provide a minimum reproducible example of the issue. This will help us debug the problem more effectively. 🐛

In the meantime, I recommend checking out our ⭐️ Tutorials for guidance. Make sure your dataset is formatted correctly and your environment meets our Requirements with Python>=3.8.0 and PyTorch>=1.8.

Environments

YOLOv5 can be run in these verified environments:

• Notebooks with free GPU:
• Google Cloud, Amazon, and Docker. Refer to our Environments Guide for setup.

Status

Ensure this badge is green to confirm all CI tests are passing.

We're excited to also introduce YOLOv8 for your future projects! 🚀

Remember, this is an automated response, and an Ultralytics engineer will review your issue shortly. 😊