RRWKV: Capturing Long-range Dependencies in RWKV
This addresses a limitation in efficient non-transformer models for NLP tasks, but appears incremental as it builds directly on RWKV.
The paper tackles the problem of RWKV's inability to capture long-range dependencies by proposing the RRWKV architecture, which incorporates retrospecting ability while maintaining memory and computational efficiency.
Owing to the impressive dot-product attention, the Transformers have been the dominant architectures in various natural language processing (NLP) tasks. Recently, the Receptance Weighted Key Value (RWKV) architecture follows a non-transformer architecture to eliminate the drawbacks of dot-product attention, where memory and computational complexity exhibits quadratic scaling with sequence length. Although RWKV has exploited a linearly tensor-product attention mechanism and achieved parallelized computations by deploying the time-sequential mode, it fails to capture long-range dependencies because of its limitation on looking back at previous information, compared with full information obtained by direct interactions in the standard transformer. Therefore, the paper devises the Retrospected Receptance Weighted Key Value (RRWKV) architecture via incorporating the retrospecting ability into the RWKV to effectively absorb information, which maintains memory and computational efficiency as well.