mlsanityData to 14 Oct 2026

LatentMT: Machine Translation with Latent Reasoning

arXiv 2607.18618 ↗First posted 21 Jul 2026cs.CL cs.AI cs.LGcited by 0

On the map: LLM reasoning → Reasoning benchmarks and evaluation

Abstract (arXiv)

Latent-reasoning looped language models (LoopLMs) offer a different scaling path for machine translation (MT): instead of increasing parameter count or emitting explicit chain-of-thought tokens, they spend additional recurrent computation inside hidden states. We introduce LatentMT, the first systematic study of latent-reasoning LoopLMs for machine translation. LatentMT adapts a small 2.6B-parameter backbone model with lightweight training. Across 32 translation directions spanning high-, mid-, and low-resource languages, LatentMT achieves performance comparable to models three to five times larger. It is competitive in a high-resource language and achieves state-of-the-art performance on both mid-resource and low-resource languages. Studying the behavior of scaling the number of recurrent reasoning steps, we find that recurrent computation consistently improves translation quality in early steps, then saturates quickly afterwards. Our mechanistic analysis shows that hidden-representation differences shrink along the recurrent reasoning-step axis, supporting the observed saturation in performance. Finally, our efficiency analysis shows that LatentMT requires lower training and inference compute than much larger non-latent-reasoning models with similar performance, making latent recurrent computation a promising path toward compact, efficient, and strong machine translation.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning. Builds on 19 of its 37 references that are in the map. First posted on arXiv 21 Jul 2026.

Builds on

19 of 37 references.

Closest ancestors · 6

Same direction, newest first

  1. 2026Unlocking Reasoning Capability on Machine Translation in Large Language Models16 Feb
  2. 2025Chain-of-Thought Reasoning Improves Context-Aware Translation with Large Language Models20 Oct
  3. 2025ExTrans: Multilingual Deep Reasoning Translation via Exemplar-Enhanced Reinforcement Learning19 May
  4. 2025Qwen3 Technical Report14 May
  5. 2025DeepTrans: Deep Reasoning Translation via Reinforcement Learning14 Apr
  6. 2025New Trends for Modern Machine Translation with Large Reasoning Models13 Mar

Other references · 13

  1. 2020Scaling Laws for Neural Language Modelscited by 6,586
  2. 2022No Language Left Behind: Scaling Human-Centered Machine Translationcited by 1,312
  3. 2024Training Large Language Models to Reason in a Continuous Latent Spacecited by 797
  4. 2025Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approachcited by 415
  5. 2025Scaling Latent Reasoning via Looped Language Modelscited by 165
  6. 2025SIM-CoT: Supervised Implicit Chain-of-Thoughtcited by 70
  7. 2025LLM Latent Reasoning as Chain of Superpositioncited by 39
  8. 2025Latent Chain-of-Thought? Decoding the Depth-Recurrent Transformercited by 27
  9. 2025Compositional Translation: A Novel LLM-based Approach for Low-resource Machine Translationcited by 14
  10. 2025Massively Multilingual Adaptation of Large Language Models Using Bilingual Translation Datacited by 4
  11. 2020Edge Intelligence: Architectures, Challenges, and Applicationscited by 2
  12. 2023On Achieving Privacy-Preserving State-of-the-Art Edge Intelligencecited by 1
1 more
  1. 2023MobileNMT: Enabling Translation in 15MB and 30mscited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Test-Time Scaling of Reasoning Models for Machine Translation
  2. 2026Monitoring latent reasoning in looped language models
  3. 2025R1-T1: Fully Incentivizing Translation Capability in LLMs via Reasoning Learning
  4. 2026Closing the Loop: Practical Training Recipes for Looped Language Models
  5. 2025How Well Do Large Reasoning Models Translate? A Comprehensive Evaluation for Multi-Domain Machine Translation
  6. 2026SPOT: Span-level Pause-of-Thought for Efficient and Interpretable Latent Reasoning in Large Language Models
  7. 2026Reasoning over Grammar: Can Synthetic Linguistic Reasoning Traces Enhance Low-Resource Machine Translation?
  8. 2026Tyler: Typed Latent Reasoning for Language Models -- When to Think, What to Compute, and How Much to Allocate