CHASE: Cache-Hole-Adapted Skip Exit for Looped State-Space Language Models
cs.AI
Submitted: 2026-07-11
Updated: 2026-09-02
Terminology
Sources
- Longformer: The Long-Document Transformer
- NVIDIA Nemotron 3: Efficient and Open Intelligence
- Mistral 7B
- Less is More: Recursive Reasoning with Tiny Networks
- BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questions
- Jamba: A Hybrid Transformer-Mamba Language Model
- Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge
- Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality
- DeepSeek-V3 Technical Report
- Universal Transformers
- Implicit Language Models are RNNs: Balancing Parallelization and Expressivity
- LT2: Linear-Time Looped Transformers
- Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters
- RoFormer: Enhanced Transformer with Rotary Position Embedding
- Mamba: Linear-Time Sequence Modeling with Selective State Spaces
- Llama 2: Open Foundation and Fine-Tuned Chat Models
- Training Compute-Optimal Large Language Models
- An Empirical Study of Mamba-based Language Models
- Tiny Recursive Reasoning with Mamba-2 Attention Hybrid
- Qwen3 Technical Report
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems
- VSAL: A Vision Solver with Adaptive Layouts for Graph Property Detection