Replace correlation-based memory pruning in an RNN or state-space model by measuring how hidden-state history changes the response to individual past input events. Train a compressed memory coordinate only if it preserves the event-consequence kernel for the target observable, such as future loss, prediction, or control return. A memory representation is accepted when the conditional variance of this kernel within compressed-state groups is small, even if dwell-time or autocorrelation…
Paper: The Memory Hidden in Response Fluctuations: Trajectory-Level Fluctuation-Response Theory and Inequalities for Non-Markovian Jump DynamicsarXiv:2608.20328
Replace consecutive or randomly assigned transformed KV coefficients with groups whose variance-volume is approximately equal. Train one equal-size vector-quantizer codebook per group, so a fixed-width cache does not waste its low-rate budget by forcing high-variance and low-variance coordinates into badly mismatched groups. This is a drop-in quantization-layout change that can be applied to keys, values, or both.
Replace the usual JL dimension choice with a task-specific dimension budget based on the paper's asymptotic ranking law. For retrieval, nearest-neighbor search, or compressed attention keys, choose m from a target Kendall correlation and automatically expand the projection when measured ranking quality falls below the target.
Paper: Exact Limits of Random Projections for Preserving Geometry: Distance Recovery, Nearest-Neighbor Rankings, and Covariance Shape in Gaussian ModelsarXiv:2609.02155
Use the paper's effective-rank and eigengap-dependent covariance estimation rates to construct a confidence-aware low-rank bottleneck for transformer activations or key/value tensors. The bottleneck is enabled only when the top-p empirical eigenspace can be estimated more accurately than the desired compression error; otherwise the layer remains full-rank.
Choose the retained context length by balancing the statistical complexity of adding lag j against the squared prediction bias from discarding it. Unlike a fixed context window, the rule uses both the remaining-horizon spectrum and the estimated tail energy, allowing a model to expand or shrink its memory online.