How can use this repo to train custom dataset i.e YOLO Objection Detection on custom dataset but by using this repo

If it's a real time detection code, it would be better to upload a .py file along with the notebook file.

I want to extract the coordinates of boundary boxes. For this I use a function of YoloLayer -> forward.
When i want to use this function I have this error

TypeError Traceback (most recent call last)

in ()
----> 1 x = y.forward(resized_image, nms_thresh)

/content/darknet.py in forward(self, output, nms_thresh)
25
26 for m in self.anchor_mask:
---> 27 masked_anchors += self.anchors[m*self.anchor_step:(m+1)*self.anchor_step]
28
29 masked_anchors = [anchor/self.stride for anchor in masked_anchors]

TypeError: can't multiply sequence by non-int of type 'float'

If someone have an idea to solve this. I attached my code below. For use it you just need to download yolov3.weights on https://github.com/pjreddie/darknet.git
YOLO-Object-Detection.tar.gz

In the detect_obects function we have :

list_boxes = model(img, nms_thresh)

# Make a new list with all the bounding boxes returned by the neural network
boxes = list_boxes[0][0] + list_boxes[1][0] + list_boxes[2][0]

In the last line, why do we need to concatenate the 3 terms ? What do they represent ?