Add attention_bias
argument in transformer block and transformer layer modules, addressing change in MCore
#6361
This job succeeded
Loading