Search papers, labs, and topics across Lattice.
The authors investigate "recurrent-state write-back"鈥攖he storage interface governing how quantized recurrent states propagate across time steps鈥攊n low-precision GRU and LSTM architectures. Deterministic 4-bit state storage triggers catastrophic failure, increasing parameter estimation error by up to 300x because small recurrent updates persistently fall below the quantization threshold, freezing state trajectories. By incorporating temporal error feedback, residual memory, or direction memory, the authors demonstrate that full baseline accuracy can be recovered post-training without modifying model weights.
Sub-threshold recurrent updates silently freeze low-precision hidden states鈥攃ausing up to a 300x error spike鈥攗nless historical quantization residuals are actively fed forward across time.
Quantization is widely used to reduce the computational and memory demands of neural-network inference. In recurrent networks, however, the quantized state is stored and returned at the next time step, so the rule used to store that state can alter subsequent computations. Here, we introduce recurrent-state write-back to denote this rule and isolate its effect in a compact GRU encoder--decoder for fluorescence lifetime imaging, a molecular imaging modality used in quantitative biological imaging. A central task is estimating two lifetime parameters, the short-lived component {\tau}1 and the long-lived component {\tau}2, from high-noise time-resolved fluorescence signals. Holding the trained model fixed, replacing continuous state propagation with deterministic 4-bit state storage increases estimation errors for {\tau}1 and {\tau}2 by approximately 70x and 300x, respectively. Failure occurs when repeated small updates remain below the write threshold, leaving the stored state nearly fixed while the network continues to propose change. Error feedback, residual memory, and direction memory carry information from these suppressed updates across time and recover accuracy without retraining. Precision sweeps show that increasing state precision can worsen a fixed recurrent solution, while matched training shows that compatibility with the state interface can be learned. To test whether this behavior extends beyond the GRU, we repeat the post-training intervention in an independently trained LSTM, where coarse write-back reproduces the failure, error feedback restores accuracy, and state-specific interventions reveal greater sensitivity of the cell state than the hidden state. Our results establish recurrent-state write-back as a key determinant of low-precision recurrent dynamics and identify the state-storage interface as a central design consideration for quantized recurrent inference.
