diff --git a/examples/41_multi_head_attention/fused_multihead_attention.cu b/examples/41_multi_head_attention/fused_multihead_attention.cu index 455e6284..ae3b66d3 100644 --- a/examples/41_multi_head_attention/fused_multihead_attention.cu +++ b/examples/41_multi_head_attention/fused_multihead_attention.cu @@ -43,6 +43,8 @@ # Run an attention example with batch size = 64 and head number = 16 without checking the correctness $ ./examples/41_multi_head_attention/41_multi_head_attention --head_number=16 --batch_size=64 --reference-check=false + Acknowledgement: this example is inspired by the idea originally prototyped by ByteDance Inc. + */ /////////////////////////////////////////////////////////////////////////////////////////////////