fidelity/stoke

Issues

• How to use a loss function which itself is a CNN!?? DDP training issue

  #32 opened 3 years ago by rushi-the-neural-arch
  3

• DDP training gets terminated in the middle of the training because of some SIGKILL received by a PID (forked child process)

  #31 opened 3 years ago by rushi-the-neural-arch
  2

• Add StokeScheduler

  #20 opened 3 years ago by zaksemenov
  1

• Update Docs from Portray to Docusaurus

  #29 opened 3 years ago by ncilfone
  0

• TypeError: intercept_args() got an unexpected keyword argument 'multiprocessing_context'

  #23 opened 3 years ago by rushi-the-neural-arch
  37

• Add support/examples for GANs

  #19 opened 3 years ago by zaksemenov
  1

• ValueError: Stoke -- Fairscale extensions (currently: oss: True, sddp: True) requires CUDA (currently: True), GPU (currently: True), DDP (currently: False) and NCCL (currently: True)

  #26 opened 3 years ago by rushi-the-neural-arch
  2

• attr.exceptions.NotAnAttrsClassError: <class 'float'> is not an attrs-decorated class.

  #25 opened 3 years ago by rushi-the-neural-arch
  3

• Support GPU device selection

  #21 opened 3 years ago by ncilfone
  0

• [Known Issue] Deepspeed multiple loss support

  #4 opened 4 years ago by ncilfone
  2