yangwenjie comments

Repositories
Issues
Comments

Results 3 comments of


                                            yangwenjie

The highest result of the benchmark network I trained was 90.83, how to solve?

The result in the table of this repository is obtained with one time experiment. For a more stable result, you can refer to [pytorch reproducibility](https://pytorch.org/docs/stable/notes/randomness.html)

Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained

I have the same issue

get error when run demo _dcn_v2.so: undefined symbol: __cudaRegisterFatBinaryEnd

As @uniquezhengjie said, this error raised when PyTorch CUDA and system CUDA version do not match. The Deformable_Convolution (DCN) in this repository requires CUDA version =10.0. - conda install pytorch=1.0...