SMELT: Scaling Laws for Compute-Matched MoE Looped Transformers

• URL: http://arxiv.org/abs/2609.01343v1
• Date: Tue, 01 Sep 2026 14:52:57 GMT
• Title: SMELT: Scaling Laws for Compute-Matched MoE Looped Transformers
• Abstract summary: Looped Transformers increase effective depth by iterating a shared block of layers, but most evaluations compare at fixed model size.<n>We study looping on Mixture-of-Experts Transformers while closely matching per-token FLOPs, total non-embedding parameters, and KV cache.<n>Through a series of ablations, we arrive at a recipe we call SMELT, which loops the middle half of layers twice while matching the unlooped Baseline on all three budgets.
• Score: 50.00050394179417
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Looped Transformers increase effective depth by iterating a shared block of layers, but most evaluations compare at fixed model size, conflating architectural advantage with extra FLOPs. We study looping on Mixture-of-Experts Transformers while closely matching per-token FLOPs, total non-embedding parameters, and KV cache. Through a series of ablations, we arrive at a recipe we call SMELT (Sparse MoE Transformer, middle layers Loop Twice), which loops the middle half of layers twice while matching the unlooped Baseline on all three budgets. We scale SMELT across four sizes up to 54B non-embedding parameters and fit a separate Chinchilla-style scaling law for each architecture. SMELT's loss drops faster with compute, saving 6.8--18.0\% of training FLOPs on the compute-optimal frontier. The advantage transfers to downstream benchmarks beyond what validation loss predicts, is largest on Code, and grows with sample length and the number of in-context examples. Mechanistic analysis shows that the second visit reduces the attention sink and redirects mass toward content-relevant tokens, an inductive bias that may underlie the observed performance gains. These results show that looping can improve Transformers even under budget matching, offering a practical recipe that turns depth reuse into measurable gains.

Related papers

• DeepLoop: Depth Scaling for Looped Transformers [91.7922038545625]
  Looped Transformers scale sequential computation by applying a compact stack of physical blocks for multiple rounds.<n>We formalize this tied-depth effect through a first-order perturbation bound controlled by a visit-alignment coefficient.<n>The resulting method, textbfDeepLoop, keeps the Post-LN DeepNorm architecture and sets $= (2N)1/2$ and $=(8N)-1/2$ for unrolled depth.
  arXiv  Detail & Related papers  (2026-07-15T06:37:05Z)
• Hidden Decoding at Scale: Latent Computation Scaling for Large Language Models [75.80275843320511]
  We study whether an existing backbone can keep improving by allocating more computation to each token while leaving the Transformer backbone fixed.<n>We propose Hidden Decoding, a sequence-length scaling method applied during continued pretraining.
  arXiv  Detail & Related papers  (2026-07-09T07:37:59Z)
• LoopCoder-v2: Only Loop Once for Efficient Test-Time Computation Scaling [72.71005779366162]
  This study studies loop-count selection through a gain--cost view.<n>We instantiate this study by training LoopCoder-v2 with different loop counts, from scratch on 18T tokens, followed by matched instruction tuning and evaluation.<n> Empirically, the two-loop variant delivers broad gains over the non-looped baseline across code generation, code reasoning, agentic software engineering, and tooluse benchmarks.
  arXiv  Detail & Related papers  (2026-06-16T15:03:05Z)
• N-vium: Mixture-of-Exits Transformer for Accelerated Exact Generation [68.47358899451255]
  N-vium is a mixture-of-exits transformer that partially parallelizes computation across depth on standard hardware.<n>N-vium attaches prediction heads at multiple depths and defines the next-token distribution as a learned mixture over these exits.
  arXiv  Detail & Related papers  (2026-05-13T08:46:17Z)
• Relaxed Recursive Transformers: Effective Parameter Sharing with Layer-wise LoRA [38.30350849992281]
  "Recursive" language models share parameters across layers with minimal loss of performance.<n>Recursive Transformers are efficiently from standard pretrained Transformers, but only use a single block of unique layers that is then repeated multiple times in a loop.<n>We show that our models outperform both similar-sized vanilla pretrained models and knowledge distillation baselines.
  arXiv  Detail & Related papers  (2024-10-28T02:15:45Z)
• Mixture-of-Modules: Reinventing Transformers as Dynamic Assemblies of Modules [96.21649779507831]
  We propose a novel architecture dubbed mixture-of-modules (MoM) MoM is motivated by an intuition that any layer, regardless of its position, can be used to compute a token. We show that MoM provides not only a unified framework for Transformers but also a flexible and learnable approach for reducing redundancy.
  arXiv  Detail & Related papers  (2024-07-09T08:50:18Z)
• SMILE: Scaling Mixture-of-Experts with Efficient Bi-level Routing [47.11171833082974]
  We introduce SMILE, which exploits heterogeneous network bandwidth and splits a single-step routing into bi-level routing. Our experimental results show that the proposed method obtains a 2.5x speedup over Switch Transformer in terms of pretraining throughput on the Colossal Clean Crawled Corpus without losing any convergence speed.
  arXiv  Detail & Related papers  (2022-12-10T03:44:16Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.