Laughing

@s4ki3f > when I try this, !yolo task=classify mode=train model=yolov8n.pt data='/content/Cow-Identification-1' epochs=3 it says no labels found hi the `yolov8n.pt` is detection model so it'll raise errors when you're trying...

@s4ki3f oh that's a log issue I think. The training actually should be trained with `imgsz=224` in both cases you showed. Thanks for the reporting, actually we're planning to perfect...

@AyushExel @glenn-jocher added this to TODO

@classico09 can you share your training command?

@pepijnob @duynguyen1907 @jiyuwangbupt hey guys, can you try to replace the following line to `self.optimizer.step()`? and restart training to check if the losses are good. Thanks! [https://github.com/ultralytics/ultralytics/blob/2bc36d97ce7f0bdc0018a783ba56d3de7f0c0518/ultralytics/yolo/engine/trainer.py#L410](https://github.com/ultralytics/ultralytics/blob/2bc36d97ce7f0bdc0018a783ba56d3de7f0c0518/ultralytics/yolo/engine/trainer.py#L410)

@nikbobo @hdnh2006 hi guys, may I ask if there're corrupt labels in your datasets?  actually I just found that we got a mismatch issue if there's corrupt label in...

the nan loss issue has been solved in this PR #490, which we'll merge it later today. :)

@toulzx yes you're right! It is because the following code: ```python if self.args.imgsz == 640: self.args.imgsz = 224 ``` Let me explain it for you, the default.yaml is shared among...

@majnas @chdre @toulzx @knoppmyth ok guys I updated the logic here #2637 . And now we support classification training with `imgsz=640`. training with `imgsz=224` by default: ```bash yolo classify train...

@fcakyon @bit-scientist hi guys, may I ask if there're corrupt labels in your datasets?  actually I just found that we got a mismatch issue if there's corrupt label in...