edgeyolo icon indicating copy to clipboard operation
edgeyolo copied to clipboard

Training Model gets stuck in loop

Open SolSearcher opened this issue 4 months ago • 1 comments

I used the readme_en guide and example dataset of visdrone2019 and get to the point where it starts running, however shortly after, I get error 'info' on line 481 and it keeps repeating that line over and over again. When I tried it on other datasets that are much smaller, I get that error almost immediately and the code stalls.

Can you help me get this up and running? Thank You.

Edit: I also thought it would be worth trying Docker in case some of my setup was wrong but the file is unavailable internationally.

Image

SolSearcher avatar Sep 01 '25 14:09 SolSearcher

please add one line after line 482 in edgeyolo/train/trainer.py and then run and show me the error

                if (self.now_epoch + 1) % self.eval_interval == 0 or self.no_aug:
                    while True:
                        try:
                            self.evaluate()
                            break
                        except Exception as e:
                            logger.error(f"error: {e}")
                            torch.cuda.empty_cache()
                            raise    # add this line

LSH9832 avatar Sep 03 '25 09:09 LSH9832