Changxu Cheng

I want the ctc loss_fn to return a tensor (with shape as [batch]) containing every loss in a batch, like the parameter `reduction` of CTCloss in Pytorch: `ctcloss = torch.nn.CTCLoss(reduction='none')`...

what is the attributes in the large-kernel CNN

2

Great analysis! I wonder the attributes of large-kernel CNN. In your paper, the basic 3x3 resnet is fully explored. 3x3 conv extracts detailed local patterns, thus may contribute to the...

Hello, I am interested in the great work. However, I am a little bit confused about the linking task in CORD. Is the entity with category as "menu.nm" linking to...

add forward patch for stable training with deepspeed zero3

1

A patch to fix the forward propagation module for **stable multimodal multitask training with deepspeed zero3**. The official implementation is going to get stuck while GPU utilization is always 100%...