mlsanityData to 7 Oct 2026

LLM Latent Reasoning as Chain of Superposition

arXiv 2510.15522 ↗First posted 17 Oct 2025cs.CLcited by 37

On the map: LLM reasoning → Test-time compute and efficient thinking

Abstract (arXiv)

Latent reasoning offers a computation-efficient alternative to Chain-of-Thought but often suffers from performance degradation due to distributional misalignment and ambiguous chain definitions. Ideally, latent reasoning should function as a superposition of multiple reasoning paths. To realize this, we introduce Latent-SFT, a unified framework addressing challenges at three levels: token, chain, and learning. First, we define the Latent-Vocab to constrain hidden states within the pre-trained vocab-space. Second, we construct the Latent-Chain via Induction-Supervision Masking to ensure semantic compactness and sufficiency. Third, we employ Latent-Optim with stochastic Gumbel-Softmax to guide the model toward generalizable solutions. Empirical results demonstrate that Latent-SFT consistently outperforms explicit SFT across six mathematical benchmarks (e.g., GSM8k, AIME24) while achieving a 2.7x to 5.5x reduction in reasoning length. Analysis confirms that our method effectively captures a superposition of diverse reasoning trajectories rather than merely compressing a single path.

Key facts. Cited by 37 papers in the map (data through 7 Oct 2026). Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning. Builds on 17 of its 29 references that are in the map. First posted on arXiv 17 Oct 2025.

Builds on

17 of 29 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025A Survey on Latent Reasoning8 Jul
  2. 2025Think Silently, Think Fast: Dynamic Latent Compression of LLM Reasoning Chains22 May
  3. 2024Compressed Chain of Thought: Efficient Reasoning Through Dense Representations17 Dec

Other references · 14

  1. 2024The Llama 3 Herd of Modelscited by 14,192
  2. 2025Qwen3 Technical Reportcited by 8,195
  3. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,146
  4. 2024Training Large Language Models to Reason in a Continuous Latent Spacecited by 763
  5. 2024From Explicit CoT to Implicit CoT: Learning to Internalize CoT Step by Stepcited by 225
  6. 2025CODI: Compressing Chain-of-Thought into Continuous Space via Self-Distillationcited by 216
  7. 2025Soft Thinking: Unlocking the Reasoning Potential of LLMs in Continuous Concept Spacecited by 109
  8. 2025LLMs are Single-threaded Reasoners: Demystifying the Working Mechanism of Soft Thinkingcited by 34
  9. 2025Text Generation Beyond Discrete Token Samplingcited by 21
  10. 2025Parallel Continuous Chain-of-Thought with Jacobi Iterationcited by 21
  11. 2025Following the Autoregressive Nature of LLM Embeddings via Compression and Alignmentcited by 14
  12. 2025RLKD: Distilling LLMs' Reasoning via Reinforcement Learningcited by 10
2 more
  1. 2025Pretraining on the Test Set Is No Longer All You Need: A Debate-Driven Approach to QA Benchmarkscited by 7
  2. 2025Forewarned is Forearmed: Pre-Synthesizing Jailbreak-like Instructions to Enhance LLM Safety Guardrail to Potential Attackscited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 2; 2026: 35.

37 papers.

Sort:

By area:

  1. 2025Reasoning Within the Mind: Dynamic Multimodal Interleaving in Latent Spacecited by 25
  2. 2026Render-of-Thought: Rendering Textual Chain-of-Thought as Images for Visual Latent Reasoningcited by 19
  3. 2025SofT-GRPO: Surpassing Discrete-Token LLM Reinforcement Learning via Gumbel-Reparameterized Soft-Thinking Policy Optimizationcited by 18
  4. 2026Solve the Loop: Attractor Models for Language and Reasoningcited by 11
  5. 2026Bridging the Gap Between Latent and Explicit Reasoning with Looped Transformerscited by 11
  6. 2026Latent-GRPO: Group Relative Policy Optimization for Latent Reasoningcited by 10
  7. 2026Are Latent Reasoning Models Easily Interpretable?cited by 6
  8. 2026Latent On-Policy Self-Distillationcited by 6
  9. 2026Circular Reasoning: Understanding Self-Reinforcing Loops in Large Reasoning Modelscited by 5
  10. 2026Retrieve, Integrate, and Synthesize: Spatial-Semantic Grounded Latent Visual Reasoningcited by 5
  11. 2026Latent Reasoning with Normalizing Flowscited by 4
  12. 2026Representational Collapse in Multi-Agent LLM Committees: Measurement and Diversity-Aware Consensuscited by 3
25 more
  1. 2026Semantic-Enriched Latent Visual Reasoningcited by 3
  2. 2026When Shallow Wins: Silent Failures and the Depth-Accuracy Paradox in Latent Reasoningcited by 2
  3. 2026SPOT: Span-level Pause-of-Thought for Efficient and Interpretable Latent Reasoning in Large Language Modelscited by 2
  4. 2026Unlocking the Black Box of Latent Reasoning: An Interpretability-Guided Approach to Interventioncited by 2
  5. 2026What Makes Effective Supervision in Latent Chain-of-Thought: An Information-Theoretic Analysiscited by 2
  6. 2026The Gradient Does Not See Rank: Rank-Indifference in Matrix-CODI on ProsQAcited by 2
  7. 2026Demystifying Hidden-State Recurrence: Switchable Latent Reasoning with On-Policy Reinforcement Learningcited by 1
  8. 2026Formalizing Latent Thoughts: Four Axioms of Thought Representation in LLMscited by 1
  9. 2026MUX: Continuous Reasoning via Multiplexed Tokenscited by 1
  10. 2026LatentRAG: Latent Reasoning and Retrieval for Efficient Agentic RAGcited by 0
  11. 2026LaTER: Efficient Test-Time Reasoning via Latent Exploration and Explicit Verificationcited by 0
  12. 2026Dropout-GRPO: Variational Stochasticity for Continuous Latent Reasoningcited by 0
  13. 2026SuperThoughts: Reasoning Tokens in Superpositioncited by 0
  14. 2026Why Struggle with Continuous Latents? Interpretable Discrete Latent Reasoning via Rendered Compressioncited by 0
  15. 2026LatentMT: Machine Translation with Latent Reasoningcited by 0
  16. 2026SLPO: Scaling Latent Reasoning via a Surrogate Policycited by 0
  17. 2026The Dynamics of Continuous Mixture Collapse in Language Modelscited by 0
  18. 2026Structural Process Supervision for Latent Chain-of-Thought Reasoningcited by 0
  19. 2026Uni-LaDiR: Latent Diffusion Unifies Multimodal Reasoningcited by 0
  20. 2026Schema-Anchored Latent Reasoning for Semantic Parsing-Based Knowledge Base Question Answeringcited by 0
  21. 2026Principled Thoughts for Latent Recursive LLM Systemscited by 0
  22. 2026SkillFM: Generating Skills for LLM Agents via Latent Flow Matchingcited by 0
  23. 2026AURAL: Adaptive Latent Reasoning with Joint Chunk for Speech Language Modelscited by 0
  24. 2026Judging in Latent Space: Efficient Generative Reward Modeling via Semantics-Preserving Compressioncited by 0
  25. 2026Lapras: Latent Reasoning for Time Series Language Modelscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Reasoning Beyond Language: A Comprehensive Survey on Latent Chain-of-Thought Reasoning
  2. 2026Geometric Latent Reasoning Induces Shorter Generations in LLMs
  3. 2026Efficient Reasoning via Constrained Optimization in Latent Space
  4. 2026ReGuLaR: Variational Latent Reasoning Guided by Rendered Chain-of-Thought
  5. 2026RuPLaR : Efficient Latent Compression of LLM Reasoning Chains with Rule-Based Priors From Multi-Step to One-Step
  6. 2025Learning When to Stop: Adaptive Latent Reasoning via Reinforcement Learning
  7. 2026The Illusion of Superposition? A Principled Analysis of Latent Thinking in Language Models
  8. 2026CoLT: Reasoning with Chain of Latent Tool Calls