lessw2020/Ranger21

error when training with batch_size = 1

neuronflow opened this issue · 0 comments

error when training with batch_size = 1

  File "/home/florian/miniconda3/envs/msblob/lib/python3.8/site-packages/torch/autograd/grad_mode.py", line 26, in decorate_context
    return func(*args, **kwargs)
  File "/home/florian/miniconda3/envs/msblob/lib/python3.8/site-packages/ranger21/ranger21.py", line 680, in step
    raise RuntimeError("hit nan for variance_normalized")
RuntimeError: hit nan for variance_normalized