您好，我在训练你第二阶段代码的时候出现下面问题，仔细观察是因为在Progressive Graph Matching过程中，并不是RGB聚类所有的类别都能从infrared找到对应类别，然后计算对比loss出现bug，找不到id，怎么解决呢？

[xuclear]

Traceback (most recent call last): File "train_sysu.py", line 1128, in main() File "train_sysu.py", line 378, in main main_worker_stage2(args,log_s1_name,log_s2_name) # Stage 2 File "train_sysu.py", line 916, in main_worker_stage2 trainer.train(epoch, train_loader_ir, train_loader_rgb, optimizer, File "/home/ubuntu/Documents/USL-VI-ReID/clustercontrast/trainers.py", line 90, in train rgb2ir_labels = torch.tensor([r2i[key.item()] for key in labels_rgb]).cuda() File "/home/ubuntu/Documents/USL-VI-ReID/clustercontrast/trainers.py", line 90, in rgb2ir_labels = torch.tensor([r2i[key.item()] for key in labels_rgb]).cuda() KeyError: 547