mapillary dataset validation error

[kimin-yun]

Thank you for previously assisting me with the learning procedure. I'm currently experimenting with the training on RbA + Mapillary + COCO with outlier supervision. The training process seems fine and I can evaluate the trained model for Road Anomaly and Fishscape without issues. However, when I try to run validation on the Mapillary dataset in the training code, the following error occurs:

File "/home/kimin/PT-2.0/RbA/train_net_debug.py", line 378, in main
res = Trainer.test(cfg, model)
File "/home/kimin/PT-2.0/detectron2/detectron2/engine/defaults.py", line 617, in test
results_i = inference_on_dataset(model, data_loader, evaluator)
File "/home/kimin/PT-2.0/detectron2/detectron2/evaluation/evaluator.py", line 164, in inference_on_dataset
evaluator.process(inputs, outputs)
File "/home/kimin/PT-2.0/detectron2/detectron2/evaluation/sem_seg_evaluation.py", line 141, in process
).reshape(self._conf_matrix.shape)
ValueError: cannot reshape array of size 415 into shape (20,20)

Occasionally, instead of 415, a different number appears. I suspect there's an issue with computing the 20x20 confusion matrix in semantic segmentation. Is there any specific configuration I need to set for the Mapillary dataset? Given that the results from my training are quite similar to the provided model, I believe the dataloader isn't the issue. If you have any insights or suspect areas, I'd appreciate your assistance.