2 Issues (Export, Predict)

[Fotic]

1st issue is that i cant export on tflite only on onnx so i must use second version of ultralytics to make tflite 2nd issue is after training, i use on console the command yolo predict task=detect model=runs\detect\train\weights\best_saved_model\best_int8.tflite imgsz=640 but it doesnt detect anything (my code bellow)

Code:

from ultralytics8_custom import YOLO as yolo_custom
from ultralytics import YOLO as yolo_original

def main():
    model_t = yolo_custom('yolov8l.pt')
    model_t.train(data="coco8.yaml", epochs=15, imgsz=640, device='0', Distillation = None)
    model_t.model.model[-1].set_Distillation = True

    model_s = yolo_custom('yolov8n.pt')
    model_s.train(data="coco8.yaml", epochs=15, imgsz=640, device='0', Distillation=model_t.model)

    model_t = yolo_original('runs/detect/train/weights/best.pt')
    model_t.export(format="tflite", int8=True, device='0')

    model_s = yolo_original('runs/detect/train2/weights/best.pt')
    model_s.export(format="tflite", int8=True, device='0')

if __name__ == "__main__":
    main()

NOTE: ultralytics8_custom is your custom version of ultralytics YOLOv8.0.83 ultralytics is latest version of yolo

Am i doing something wrong? Any proposal