Hello, we have previously had success uploading locally trained yolov12 models to roboflow. Using the same training script, we are now unable to get any uploads to work properly, all show this error in roboflow. We have tried re-training, different configs, different versions of cuda and torch, nothing works properly. We are using the GitHub - sunsmarterjie/yolov12: YOLOv12: Attention-Centric Real-Time Object Detectors repo as the docs suggest.
Hi @rstajduhar-dwt!
I’m sorry you’ve run into this issue! I am investigating this issue on my end. To help me triage faster, do you give Roboflow Support permission to access your workspace?
Additionally, do you mind clarifying when you are encountering this error message? Is it when you are uploading model weights or uploading a video to test a model?