Abstract:Attention heads retrieve: given a query, they return a weighted average of stored values. We showed that this computation is one step of gradient descent on the modern Hopfield energy, and that Langevin sampling from the corresponding Boltzmann distribution yielded stochastic attention, a training-free sampler controlled by a single temperature parameter. Lowering the temperature gave exact retrieval; raising it gave open-ended generation. Because the energy gradient equals the attention map, no score network, training loop, or learned model was required, making the approach particularly suited to the low-data regime where learned generative models are starved of training signal. We derived an entropy inflection condition that identified the retrieval-to-generation transition temperature for any memory geometry and validated the sampler on five domains spanning two orders of magnitude in dimension. A single Boolean mask on the attention softmax, identical to the causal mask used in transformers but applied along the memory axis rather than the sequence axis, turned the sampler into a zero-shot class-conditional generator on Olivetti faces with no retraining and no learned classifier. On MNIST digit images, stochastic attention produced samples that were markedly more novel and more diverse than the best learned baseline while matching a Metropolis-corrected gold standard. On protein sequences from a small Pfam family, the generation regime preserved amino acid composition far more faithfully than a variational autoencoder at matched novelty, indicating that the training-free score function retained family-level fidelity that learned models lost. A denoising diffusion baseline failed across all memory sizes tested, producing samples indistinguishable from isotropic noise. The approach required no architectural changes to the underlying attention mechanism.
| Comments: | Main body (including references excluding the appendix): 11 pages, 2 figures and 1 table. Total paper: 26 pages, 13 figures and 7 pages |
| Subjects: | Machine Learning (cs.LG); Computational Finance (q-fin.CP) |
| Cite as: | arXiv:2603.06875 [cs.LG] |
| (or arXiv:2603.06875v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2603.06875 arXiv-issued DOI via DataCite |
From: Jeffrey Varner [view email]
[v1]
Fri, 6 Mar 2026 20:50:30 UTC (5,451 KB)
[v2]
Tue, 10 Mar 2026 23:55:26 UTC (5,679 KB)
[v3]
Thu, 14 May 2026 14:55:42 UTC (5,433 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。