Chuanwise

Results 11 comments of Chuanwise
trafficstars

If run codes with autonomy detection enabled: ```py with torch.autograd.detect_anomaly(): vqa_loss, vaq_loss, qav_loss = model(data) ``` It can't detect where the `nan` come from. ``` (video-question-answering) [23031212503@login01 Flipped-VQA]$ tail err.90258*...