Selective Latent Thinking: Adaptive Compression of LLM Reasoning Chains

arXiv:2605.25745 · cs.CL · Submitted 2026-05-25 · Read on arXiv

cs.CL

Submitted: 2026-05-25

Updated: 2026-10-13

Code: https://github.com/hunshi34/SLT

Terminology

• Accuracy — 4× in this paper · explained in 1 episode
• Latent Reasoning — 2× in this paper · explained in 2 episodes
• Latent Representation — 2× in this paper · explained in 2 episodes
• Selective Latent Thinking — 2× in this paper
• Benchmark — 1× in this paper · explained in 2 episodes
• Chain-of-Thought (CoT) — 1× in this paper · explained in 4 episodes
• Chain-of-Thought (CoT) Reasoning — 1× in this paper · explained in 4 episodes
• Large Language Models (LLMs) — 1× in this paper · explained in 6 episodes
• Latent Representations — 1× in this paper · explained in 3 episodes
• Reinforcement Learning — 1× in this paper · explained in 4 episodes
• Trace — 1× in this paper · explained in 2 episodes

Sources

• L1: Controlling How Long A Reasoning Model Thinks With Reinforcement Learning
• Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads
• Efficient Reasoning Models: A Survey
• Better & Faster Large Language Models via Multi-token Prediction
• DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
• Training Large Language Models to Reason in a Continuous Latent Space
• OpenAI o1 System Card
• EAGLE-3: Scaling up Inference Acceleration of Large Language Models via Training-Time Test
• O1-Pruner: Length-Harmonizing Fine-Tuning for O1-Like Reasoning Pruning
• OneLatent: Single-Token Compression for Visual Latent Reasoning
• Your LLM Knows the Future: Uncovering Its Multi-Token Prediction Potential
• Efficient Reasoning with Hidden Thinking
• CODI: Compressing Chain-of-Thought into Continuous Space via Self-Distillation
• Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models
• Think Silently, Think Fast: Dynamic Latent Compression of LLM Reasoning Chains
• Kimi k1.5: Scaling Reinforcement Learning with LLMs
• Render-of-Thought: Rendering Textual Chain-of-Thought as Images for Visual Latent Reasoning
• TokenSkip: Controllable Chain-of-Thought Compression in LLMs
• Towards System 2 Reasoning in LLMs: Learning How to Think With Meta Chain-of-Thought
• Chain of Draft: Thinking Faster by Writing Less

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering