Mentored Decoding: Faster Inference meets Boosting
cs.LG
Submitted: 2026-09-24
Updated: 2026-09-24
License: http://creativecommons.org/licenses/by/4.0/
The gist: Speculative decoding is a successful technique speeding up inference of a target autoregressive language model via a fast drafter model.
Terminology
Abstract
Speculative decoding is a successful technique speeding up inference of a target autoregressive language model via a fast drafter model. Lossy speculative decoding allows a drift with respect to the target to further improve speed. Interestingly, it has been observed experimentally that the resulting model can also beat the target quality-wise. Our paper formally proves how such a feat is possible with a formal approach to lossy speculative decoding called mentored decoding. To get there, we connect inference to a celebrated ML training theory, boosting, and proceed via the generalization of mentored decoding to the whole set of f-divergences. We uncover key properties of mentored decoding, among which (i) the particularly appealing geometric nature of the total variation case, (ii) simple approximations for any f-divergence in direct relation with boosting compliance, and (iii) a divergence independent O(n) space and O(sort(n)) time data structure built on drafter and target outputs, which allows to query the optimal parameters of the dual problem in O(n) time and constructing optimal mentored distributions in O(n) time for any f-divergence.
Sources
- Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads
- DFlash: Block Diffusion for Flash Speculative Decoding
- Break the Sequential Dependency of LLM Inference Using Lookahead Decoding
- Better & Faster Large Language Models via Multi-token Prediction
- Beyond the Target: From Imitation to Collaboration in Speculative Decoding
- A Practical Investigation of Training-free Relaxed Speculative Decoding
- Inference with Reference: Lossless Acceleration of Large Language Models
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks