When quantizing the YOLOv8 model and calling the code optimizer = oto.dhspg(), an error is thrown.

Question

When quantizing the YOLOv8 model and calling the code optimizer = oto.dhspg(), an error is thrown.

Gooddz1 opened this issue a year ago · comments

When quantizing the YOLOv8 model and calling the code optimizer = oto.dhspg(), it throws an error:
"Traceback (most recent call last): File "yolov8_test.py", line 32, in optimizer = oto.dhspg( File "/home/sta/ldz/project/quantization/yolov8/only_train_once/only_train_once/init.py", line 42, in dhspg self._optimizer = DHSPG( File "/home/sta/ldz/project/quantization/yolov8/only_train_once/only_train_once/optimizer/dhspg.py", line 62, in init super(DHSPG, self).init(params, defaults) File "/home/sta/anaconda3/envs/yolov8/lib/python3.8/site-packages/torch/optim/optimizer.py", line 187, in init raise ValueError("optimizer got an empty parameter list") ValueError: optimizer got an empty parameter list".

Gooddz1 · Answer 1 · Tue Oct 24 2023 16:28:06 GMT+0800 (China Standard Time)

When using model = resnet18_cifar10() and calling the code optimizer = oto.dhspg(), it does not throw an error.

Tianyi Chen · Answer 2 · Fri Dec 08 2023 02:52:01 GMT+0800 (China Standard Time)

Thanks for reaching out. We have an internal OTO version that will be released by the end of this month. The internal version has significantly major updates and improvements. Please retry on that if possible. @Gooddz1