diff --git a/megatron/initialize.py b/megatron/initialize.py index bc4032649..dd752659b 100644 --- a/megatron/initialize.py +++ b/megatron/initialize.py @@ -57,6 +57,7 @@ def finish_mpu_init(): if ( neox_args.scaled_upper_triang_masked_softmax_fusion or neox_args.scaled_masked_softmax_fusion + or neox_args.rope_fusion ): fused_kernels.load_fused_kernels()