D-REAL: Acceptance-Length Optimization with Unbiased Gradients for Speculative Decoding
cs.AI
Submitted: 2026-11-07
Updated: 2026-11-07
Code: https://github.com/tatsu-lab/stanf
Terminology
Sources
- Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Models
- Program Synthesis with Large Language Models
- Accelerating Large Language Model Decoding with Speculative Sampling
- Evaluating Large Language Models Trained on Code
- DSpark: Confidence-Scheduled Speculative Decoding with Semi-Autoregressive Generation
- Training Verifiers to Solve Math Word Problems
- SpecForge: A Flexible and Efficient Open-Source Training Framework for Speculative Decoding
- Qwen3 Technical Report
- Beyond the Point Mass: Acceptance-Derived Weights and Distributional Losses for Speculative Decoding
- D-PACE: Dynamic Position-Aware Cross-Entropy for Parallel Speculative Drafting
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- 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
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents