Peng Wang

icon indicating a website link https://peter-peng-w.github.io/ icon indicating an email [email protected]

icon company University of Virginia icon location Charlottesville, VA icon location Ph.D. Student @ UVa

Results 5 comments of Peng Wang

icdar2015上训练效果

Maybe you should refer to the issue #27 first and try to follow the way that the author was doing. I am also facing this problem and currently working on...

icdar2015上训练效果

@lijian10086 It seems that you did not change the parameters in the config file (well, except for the epoch number) but could not reach an eval loss under 0.7. Actually...

OSError: [Errno 12] Cannot allocate memory

I think this error occurs due to the replay memory which is allocated on the memory. The whole replay memory would be too large for memory though the os had...

Any plan to release the code?

Hi, will the code be released? I am really interested in this work and it would be very helpful to have the implementation.

dqn uses huberloss instead of mseloss

Correct! The same implementation of using Huber Loss and gradient clipping (using Module.parameters.grad.clamp_(-1,1) in PyTorch) could be seen in the PyTorch Tutorial's RL part.