Hello, I am trying to train my YOLO11s model for object detection using my imported dataset on the Ultralytics platform, but I keep receiving the same error message.
Could someone please tell me how I can resolve this issue?
Can you provide your username?
@Abir we fixed an issue around rate limiting today, I think your training should work now. Can you please try again?
Hello, thank you for your reply.
I just tried again, but I still receive the same message. I think the issue may be related to my imported pretrained model. In fact, when I choose one of the models directly proposed by the platform, such as YOLOv26s, the training works correctly.
I’m not sure whether it’s possible to select the YOLOv11 version from the list of available models, because I only see YOLOv26 in the options.
Yes — Ultralytics Platform supports Ultralytics YOLO11 training, but the base-model list you see depends on the project/architecture you’re currently in. If you’re inside a YOLO26 project you’ll only see YOLO26 models; switch/create a YOLO11 project and you should see the official YOLO11 n/s/m/l/x options as described in the Platform Models docs.
For the imported pretrained model: if training works with an official YOLO26s but fails only when you select your uploaded .pt, that strongly suggests the uploaded weights aren’t in a compatible Ultralytics checkpoint format (or they’re a different task like segment/pose/obb). Platform expects a native Ultralytics YOLO PyTorch .pt checkpoint (typically a best.pt from Ultralytics training), and task must match the dataset (task mismatches will fail, as noted in the Platform Cloud Training docs).
If you can share your Platform username (like we asked earlier) + the exact error text (copy/paste) and what created that pretrained file (Ultralytics version / which repo / which task), we can check the backend logs and tell you precisely why that .pt is being rejected.
Is it your own custom model? I tested uploading pretrained YOLO11s and training works fine.