AI & ML
Exact Sequence Interpolation with Transformers
arXiv stat.ML0 views
arXiv:2502.02270v4 Announce Type: replace-cross Abstract: We prove that transformers can exactly interpolate datasets of finite input sequences in $\mathbb{R}^d$, $d\geq 2$, with corresponding output sequences of smaller or equal length. Specifically, given $N$ sequences of arbitrary but finite lengths in $\mathbb{R}^d$ and output sequences of lengths $m^1, \dots, m^N \in \mathbb{N}$, we construct a transformer with $\mathcal{O}(\sum_{j=1}^N m^j)$ blocks and $\smash{\mathcal{O}(d \sum_{j=1}^N m^j)}$ parameters that exactly interpolates the dataset. Our construction provides complexity estimates that are independent of the input sequence length, by alternating feed-forward and self-attention layers and by capitalizing on the clustering effect inherent to the latter. Our novel constructive method also uses low-rank parameter matrices in the self-attention mechanism, a common feature of practical transformer implementations. These results are first established in the hardmax self-attention setting, where the geometric structure permits an explicit and quantitative analysis, and are then extended to the softmax setting. Finally, we demonstrate the applicability of our exact interpolation construction to learning problems, in particular by providing convergence guarantees to a global minimizer under regularized training strategies. Our analysis contributes to the theoretical understanding of transformer models, offering an explanation for their excellent performance in exact sequence-to-sequence interpolation tasks.
Source:https://arxiv.org/abs/2502.02270PreviousClustering in pure-attention hardmax transformers and its role in sentiment analysisNextFunctional Compression as Occam's Razor: Algebraic Complexity from Sensitivity Minimization
Related posts
Sam Altman says his life is so weird that he’s unfazed by Andrew Garfield portraying him in upcoming movie
0TechCrunch
Account pools for multi-platform posting #lsdxb
0dev.to

Mathematical proof, powered by AI. And no hallucinations!
0Amazon Science

Qwen4-Max vs. Claude Opus 5.5: The cheaper model costs you somewhere else.
0The New Stack

“The best escape room players that you’ve ever met”: What Red Hat’s AI architect really thinks about agents.
0The New Stack

Microsoft says Bing serves 1B people. Its next search engine isn’t for them.
0The New Stack