Is this assertion for mask wrong?

[yinfangchen]

I got an AssertionError: Mask is silently ignored due to the use of a custom kernel when training GPT-2 with examples/pretrain_gpt.sh.

This line leads to the assertion error:

Megatron-DeepSpeed/megatron/model/fused_softmax.py

Line 191 in 8387ae1

assert mask is None, "Mask is silently ignored due to the use of a custom kernel"

Is this assertion necessary? And is it even correct?

[LordEdison]

same puzzlement