You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@LIZHICHAOUNICORN Thanks for using DeepSpeed. Please see responses to your questions below
'lr_this_step' referenced before assignment. : This is a bug in fp32 mode. We will fix asap.
the log warning: OVERFLOW: Overflow is normal at the beginning of training in fp16 mode, aka mixed-precision training. The training will automatically adjust with loss-scaling to stop the overflows.
ENV:
two GPU: Tesla P40
GPU Memory: 22919MiB
os: ubuntu 16.04
Driver Version: 440.33.01
CUDA Version: 10.2
Python 3.7.0
I run DeepSpeedExamples/bing_bert: ds_train_bert_bsz32k_seq512.sh, I modified some configure:
Notice bert_base.json change fp16 from true to false.
then, i got error: 'lr_this_step' referenced before assignment.
second:
when i enable , i found the log warning: OVERFLOW.
so does Deepspeed support disable fp16 or my operations was wrong?
The text was updated successfully, but these errors were encountered: