arXiv Machine Learning

Temporal Recurrence Favors Fewer Layers

The paper investigates how temporal recurrence affects the required depth of neural networks in streaming tasks. By treating depth, expert width, and parallel experts as a compute‑allocation problem, the authors compare recurrent and non‑recurrent models across various compute budgets. Experiments on Sokoban and FineWeb language modeling show that recurrence shifts the optimal compute allocation toward fewer layers while maintaining or improving performance.

arXiv Machine Learning
Sep 24

Attention Routing Stabilizes Early: Working-Set Inference for Recurrent Language Models

The paper investigates how attention dynamics evolve across recurrent depth in language models, finding that attention support stabilizes early while hidden states and outputs take longer. It proposes WISE, a training‑free method that uses full attention in early steps and then reuses the discovered sparse working set for later steps, preserving performance on multi‑hop QA tasks. Experiments show that WISE maintains quality up to 2K context, offers measurable speedups, and highlights the importance of recurrent discovery of attention support.

By Ke Wan, Chen Chen
arXiv Machine Learning
Aug 4

Structured Recurrent Mixers for Massively Parallelized Sequence Generation

arXiv:2605. 08696v4 Announce Type: replace-cross Abstract: Over the last two decades, language modeling has experienced a shift from the use of predominantly recurrent architectures that process tokens sequentially during training and inference to non-recurrent models that process sequence elements in parallel during training, which results in greater training efficiency and stability at the expense of lower inference throughput.

By Benjamin L. Badger
arXiv Machine Learning
Aug 19

Recirculation

The paper introduces recirculation, an inference‑time architectural enhancement for foundation models that reduces perplexity and improves accuracy on generation and reasoning tasks without adding significant latency. Recirculation adds a specific form of recurrence, enabling the model to function as a dynamical system that tracks belief states, and is distinct from chain‑of‑thought or depth‑recurrence methods. An adaptive variant requires minimal hyperparameter tuning and achieves notable gains on the Gemma3 family, including a 23% perplexity drop and a 21% accuracy increase on GSM8k.

By Michael C. Mozer, Shoaib Ahmed Siddiqui, Danny Sawyer, Sunny Sanyal, Rosanne Liu
arXiv Machine Learning
Sep 21

Trading Depth for Time in Recurrent Transformers

The paper investigates whether extra computation in recurrent Transformers should be allocated to more temporal steps or greater physical depth. Using Latent Recurrent Transformers (LRTs), the authors insert a latent thought token between vocabulary tokens, allowing each token to pass through the same $L$ layers twice while sharing parameters. Experiments on 16‑ and 20‑layer mixture‑of‑experts NanoChat backbones show that a single thought token brings a shallower model within 0.006–0.004 bits per byte of a double‑depth counterpart, recovering 67–81% of the improvement with roughly 48% fewer parameters.

By Zeyi Huang, Xuehai He, Yong Jae Lee, Yelong Shen
arXiv AI
Oct 2

Reshape and Recur: Improving SSMs with Input Reshaping and Depth Recurrence

The paper proposes two extensions to State Space Models (SSMs) to reduce memory usage and improve performance. First, it introduces depth recurrence, allowing a looped SSM with fewer parameters to match the performance of a larger, non-recurrent model. Second, it advocates using a fixed time granularity across tasks by reshaping input sequences, which enhances how information is presented to the model. Both techniques consistently benefit four representative SSM architectures (LRU, S5, LinOSS, LrcSSM).

By M\'onika Farsang, Ramin Hasani, Daniela Rus, Radu Grosu
arXiv AI
Jun 4

MesaNet: Sequence Modeling by Locally Optimal Test-Time Training

arXiv:2506. 05233v2 Announce Type: replace-cross Abstract: Sequence modeling is currently dominated by causal transformer architectures that use softmax self-attention.

By Johannes von Oswald, Nino Scherrer, Seijin Kobayashi, Luca Versari, Songlin Yang, Sarthak Mittal, Maximilian Schlegel, Kaitlin Maile, Yanick Schimpf, Oliver Sieberling, Alexander Meulemans, Rif A. Saurous, Guillaume Lajoie, Charlotte Frenkel, Razvan Pascanu, Blaise Ag\"uera y Arcas, Jo\~ao Sacramento
arXiv Machine Learning
Aug 20

Allocating Recurrent Compute in Looped Language Models

The paper proposes MixerLoop, a looped language model that repeats only the Gated DeltaNet mixer while applying the dense feed‑forward network (FFN) once, instead of repeating the entire layer stack. Using Iterative Transport Rank (ITR) to measure cross‑position influence, the authors show that MixerLoop retains most of the performance gains of full recurrence while reducing FLOPs by 45.9%. Experiments on 15M and 110M parameter models demonstrate that MixerLoop outperforms full‑block recurrence on aggregate CORE metrics.

By Ruhai Lin, Yiyang Guo, Rui-Jie Zhu, Hao Ye, Jason K. Eshraghian