Add attention_bias
argument in transformer block and transformer layer modules, addressing change in MCore
#11289
+7
−0
Loading