Dynamic Semantic Compression for Efficient Latent-Space Inference in Large Language Models
cs.CL, cs.AI
Submitted: 2026-09-14
Updated: 2026-09-14
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Think Silently, Think Fast: Dynamic Latent Compression of LLM Reasoning Chains
- SentenceVAE: Enable Next-sentence Prediction for Large Language Models with Faster Speed, Higher Accuracy and Longer Context
- WanJuan: A Comprehensive Multimodal Dataset for Advancing English and Chinese Large Models
- Context Cascade Compression: Exploring the Upper Limits of Text Compression
- SeqDiffuSeq: Text Diffusion with Encoder-Decoder Transformers
- From Explicit CoT to Implicit CoT: Learning to Internalize CoT Step by Step
- TwT: Thinking without Tokens by Habitual Reasoning Distillation with Multi-Teachers' Guidance
- Think before you speak: Training Language Models With Pause Tokens
- Compressed Chain of Thought: Efficient Reasoning Through Dense Representations
- PonderLM: Pretraining Language Models to Ponder in Continuous Space
- Latent Thinking Optimization: Your Latent Reasoning Language Model Secretly Encodes Reward Signals in Its Latent Thoughts
- Sequoia: Scalable, Robust, and Hardware-aware Speculative Decoding
- SIM-CoT: Supervised Implicit Chain-of-Thought
- OPT: Open Pre-trained Transformer Language Models
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