Problem using different weights

[Nikolaiarn]

Hello, I am wondering if you could help me?
I changed the weights to another .pt file that I trained myself in yolov5. The code works fine when I use the default weights from coco128 but not the one I made myself. I have tested the weights in a detect.y script and it works, but not in this wrapper. I get this error when running the launch file.
[INFO] [launch]: All log files can be found below /home/nikolai/.ros/log/2023-11-13-20-37-17-700838-NikUbuntu22-41350
[INFO] [launch]: Default logging verbosity is set to INFO
[INFO] [yolov5_ros-1]: process started with pid [41351]
[yolov5_ros-1] /usr/lib/python3/dist-packages/requests/init.py:87: RequestsDependencyWarning: urllib3 (2.1.0) or chardet (5.2.0) doesn't match a supported version!
[yolov5_ros-1] warnings.warn("urllib3 ({}) or chardet ({}) doesn't match a supported "
[yolov5_ros-1] /usr/lib/python3/dist-packages/scipy/init.py:146: UserWarning: A NumPy version >=1.17.3 and <1.25.0 is required for this version of SciPy (detected version 1.26.0
[yolov5_ros-1] warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
[yolov5_ros-1] /opt/ros/humble/local/lib/python3.10/dist-packages/rclpy/node.py:441: UserWarning: when declaring parameter named 'classes', declaring a parameter only providing its name is deprecated. You have to either:
[yolov5_ros-1] - Pass a name and a default value different to "PARAMETER NOT SET" (and optionally a descriptor).
[yolov5_ros-1] - Pass a name and a parameter type.
[yolov5_ros-1] - Pass a name and a descriptor with `dynamic_typing=True
[yolov5_ros-1] warnings.warn(
[yolov5_ros-1] YOLOv5 🚀 2023-11-9 torch 2.1.0+cu121 CPU
[yolov5_ros-1]
[yolov5_ros-1] Traceback (most recent call last):
[yolov5_ros-1] File "/home/nikolai/yolov5/install/yolov5_ros/lib/yolov5_ros/yolov5_ros", line 33, in
[yolov5_ros-1] sys.exit(load_entry_point('yolov5-ros==0.2.0', 'console_scripts', 'yolov5_ros')())
[yolov5_ros-1] File "/home/nikolai/yolov5/install/yolov5_ros/lib/python3.10/site-packages/yolov5_ros/main.py", line 294, in ros_main
[yolov5_ros-1] yolov5_node = yolov5_ros()
[yolov5_ros-1] File "/home/nikolai/yolov5/install/yolov5_ros/lib/python3.10/site-packages/yolov5_ros/main.py", line 235, in init
[yolov5_ros-1] self.yolov5 = yolov5_demo(self.weights,
[yolov5_ros-1] File "/home/nikolai/yolov5/install/yolov5_ros/lib/python3.10/site-packages/yolov5_ros/main.py", line 64, in init
[yolov5_ros-1] self.load_model()
[yolov5_ros-1] File "/home/nikolai/yolov5/install/yolov5_ros/lib/python3.10/site-packages/yolov5_ros/main.py", line 71, in load_model
[yolov5_ros-1] self.model = DetectMultiBackend(self.weights, device=self.device, dnn=self.dnn, data=self.data)
[yolov5_ros-1] File "/home/nikolai/yolov5/install/yolov5_ros/lib/python3.10/site-packages/yolov5_ros/models/common.py", line 307, in init
[yolov5_ros-1] model = attempt_load(weights if isinstance(weights, list) else w, map_location=device)
[yolov5_ros-1] File "/home/nikolai/yolov5/install/yolov5_ros/lib/python3.10/site-packages/yolov5_ros/models/experimental.py", line 96, in attempt_load
[yolov5_ros-1] ckpt = torch.load(attempt_download(w), map_location=map_location) # load
[yolov5_ros-1] File "/home/nikolai/.local/lib/python3.10/site-packages/torch/serialization.py", line 1014, in load
[yolov5_ros-1] return _load(opened_zipfile,
[yolov5_ros-1] File "/home/nikolai/.local/lib/python3.10/site-packages/torch/serialization.py", line 1422, in _load
[yolov5_ros-1] result = unpickler.load()
[yolov5_ros-1] File "/home/nikolai/.local/lib/python3.10/site-packages/torch/serialization.py", line 1415, in find_class
[yolov5_ros-1] return super().find_class(mod_name, name)
[yolov5_ros-1] AttributeError: Can't get attribute 'DetectionModel' on <module 'models.yolo' from '/home/nikolai/yolov5/install/yolov5_ros/lib/python3.10/site-packages/yolov5_ros/models/yolo.py'>
[ERROR] [yolov5_ros-1]: process has died [pid 41351, exit code 1, cmd '/home/nikolai/yolov5/install/yolov5_ros/lib/yolov5_ros/yolov5_ros --ros-args --params-file /tmp/launch_params_c9wgt9re --params-file /tmp/launch_params_kqr_4gek'].