LiquidGRU-SSM

LiquidGRU-SSM is a multi-layer hybrid model that blends liquid state-space dynamics with gated recurrent units (GRU) to effectively capture both long-range dependencies and fine-grained temporal nuances in sequential data. Enhanced with rotary embeddings for robust temporal processing and featuring a liquid correction network for dynamic state updates, this versatile model is designed for tasks such as language modeling, time-series forecasting, and other advanced sequence modeling applications.