Jason Lee

icon indicating an email [email protected]

icon company BTBU icon location BEIJING icon location Object detection Image captioning vision language NLP

Results 36 comments of Jason Lee

RuntimeError: Expected object of type Variable but found type CUDAType for argument #0 'result'

Torch version bug but i need torch1.1.0 environment, any help

蓝灯速度慢不好用，建议试试这个SSR

ok？

Is there any tool to monitor the training process?

tensorboard is enough. Ref. https://github.com/yunjey/pytorch-tutorial/tree/master/tutorials/04-utils/tensorboard

Device index must be -1 or non-negative, got -1160

Maybe there is some problem with your GPU setting.

what is the meaning of chunk_sizes?

len(chunk_sizes ) equal to number of your GPUs

Should training suppercategory id and validation supercategory id be the same while training?

same is ok

当我使用tensorflow　包括warp-ctc的时候，如何拿到预测的结果？

@ghost 请问现在知道了吗

Why the recognition accuracy different from paper?

@Jyouhou @zobeirraisi Hi, can you tell us more about your pretrained model

Why the recognition accuracy different from paper?

@delveintodetail Is there wrong in the predict.py file?

testloss would lead to model update on eval mode

@Sanster Thank for answer to @gussmith's question😄