Use a local chaining complexity computed from an empirical input metric to predict stochastic output error for each individual input. Easy, locally concentrated inputs can use fewer dropout, ensemble, or diffusion samples, while high-complexity inputs receive additional computation; unlike a global confidence threshold, the allocation varies with the input.
Precondition activation or cache blocks with a fixed product U = A Sigma B of orthogonal transforms and a random signed permutation before quantization or coordinate sampling. The random permutation makes the product incoherent, so energy is less concentrated in a few coordinates and lossy compression should introduce less worst-case distortion.
Freeze a randomly initialized single-layer transformer and use a constructed soft prompt to make its attention weights equal Gaussian-kernel weights over support examples. The resulting model performs Nadaraya-Watson regression in one forward pass, so task adaptation stores prompt tokens rather than modifying network weights. Prompt length becomes the number of kernel centers, while hidden dimension and prompt norm determine whether the required logits can be represented accurately.
Replace selected dense neural-network operators by low-rank factors whose rank is selected by a randomized residual test at a user-specified tolerance. Construct candidate bases in large blocks for efficient matrix operations, then prune the block to the smallest rank that passes the residual criterion instead of treating the block size as the final rank.
Use attention-graph hitting times to identify tokens whose information has not mixed through the network, then route only those tokens through additional Transformer blocks. Tokens with fast reachability exit early, while slow or isolated tokens receive more computation.
Augment a causal sequence model with a small hierarchy of prefix summaries weighted by powers of the logarithmic rank of each preceding token. The summaries retain order-sensitive deviations from a baseline representation while costing O(KNd) for sequence length N, hierarchy width K, and hidden dimension d, instead of O(N^2d) dense attention.
Replace dense token-to-token attention with attention over connected token groups, called polymers, while forbidding nearby polymers from being simultaneously selected. Each candidate group receives an exponentially decaying size and boundary penalty, and the layer sums or samples only compatible collections of groups. The construction should create structured sparsity and prevent redundant overlapping attention regions.
Replace an unrestricted collection of nested dyadic attention windows on a 2D token grid by a sparse antichain: no selected window may contain another selected window. Use the paper's exponential occupancy guarantee to control how many attention blocks reuse the same token, and add a differentiable log-moment penalty during training when exact antichain selection is relaxed. The expected benefit is bounded peak KV reuse and more predictable sparse-attention cost without discarding multiscale…
Replace a fixed top-k expert count with a stochastic, token-specific fanout generated by a supercritical binary branching process stopped at a geometric time. The resulting number of active experts has finite mean but a power-law tail with log-periodic modulation, allowing most tokens to use little compute while reserving larger computation for difficult or ambiguous tokens.