A Token Is Not a Unit of Difficulty: Segmentation Controls Where Adaptive Computation Goes Inside a Word
cs.CL, cs.AI
Submitted: 2026-10-31
Updated: 2026-10-31
Terminology
Sources
- Eliciting Latent Predictions from Transformers with the Tuned Lens
- BLOOM: A 176B-Parameter Open-Access Multilingual Language Model
- Optimal ANN-SNN Conversion for High-accuracy and Ultra-low-latency Spiking Neural Networks
- SkipDecode: Autoregressive Skip Decoding with Batching and Caching for Efficient LLM Inference
- Depth-Adaptive Transformer
- TIDE: Token-Informed Depth Execution for Per-Token Early Exit in LLM Inference
- Rethinking Calibration for Early-Exit Neural Networks
- Input-Aware Dynamic Timestep Spiking Neural Networks for Efficient In-Memory Computing
- Pointer Sentinel Mixture Models
- Language Model Tokenizers Introduce Unfairness Between Languages
- Confident Adaptive Language Modeling
- mGPT: Few-Shot Learners Go Multilingual
- Qwen3 Technical Report
- ADEPT: Adaptive Dynamic Early-Exit Process for Transformers
Related papers
- ALARM: Audio-Language Alignment for Reasoning Models
- ThinkSum: Probabilistic reasoning over sets using large language models
- Coherence boosting: When your pretrained language model is not paying enough attention
- From Facts to Insights: A Persona-Driven Dual Memory Framework and Dataset for Role-Playing Agents
- Improving General Role-Playing Agents via Psychology-Grounded Reasoning and Role-Aware Policy Optimization
- The Art of Asking: Multilingual Prompt Optimization for Synthetic Data