KV Shifting Attention Enhances Language Modeling
This work addresses a bottleneck in making language models more efficient for AI researchers and practitioners, though it is incremental as it builds on existing induction head mechanisms.
The paper tackles the inefficiency of implementing induction heads in large language models by proposing KV shifting attention, which reduces the required model depth and width and leads to better performance or faster convergence in models with over 10 billion parameters.
The current large language models are mainly based on decode-only structure transformers, which have great in-context learning (ICL) capabilities. It is generally believed that the important foundation of its ICL capability is the induction heads mechanism, which requires at least two layers attention. In order to more efficiently implement the ability of the model's induction, we revisit the induction heads mechanism and proposed a KV shifting attention. We theoretically prove that the KV shifting attention reducing the model's requirements for the depth and width of the induction heads mechanism. Our experimental results demonstrate that KV shifting attention is beneficial to learning induction heads and language modeling, which lead to better performance or faster convergence from toy models to the pre-training models with more than 10 B parameters.