PAY AT THE EVENT: EVENT-CONDITIONED RECURRENT STATE TRACKING AND ASSOCIATIVE RECALL

We introduce an event-conditioned recurrent block that explicitly separates transition computation from associative storage. FLINT maintains a low-cost diagonal carry at every token while applying expressive Householder transformations only at selected events. When event gates are determined before scanning, the recurrent dependency chain can be reorganized around event boundaries, making expensive transition work proportional to the number of events. EGAM provides fixed-size associative memory through binary addressing, gated additive writes, and normalized slot reads. In a shared-backbone block benchmark, the combined block achieves $1.2\times$ faster forward execution than Gated DeltaNet-2 at known-mask density $0.1$ and length $2048$. Replacing memory-projection multiplications with additive and bitwise operations reduces mixer multiply-accumulates (MACs) from $0.865$M to $0.469$M per token and layer, with $1.262$M accumulate/bit operations, excluding the MLP. Across three seeds, mean WikiText-103 perplexity changes from $26.705$ to $26.709$. On FineWeb-Edu, a 122.54M-parameter model trained on 1.007B tokens retains validation perplexity ($29.328$ to $29.355$, one seed) while reducing mixer MACs from $1.838$M to $1.047$M. The combined task state is independent of context length and, at length $1024$, is approximately one-eighth the size of attention’s KV cache. Learned-gate and joint tasks evaluate tracking and recall within these budgets. We will release our code upon acceptance.