# DEL: Context-Aware Dynamic Exit Layer for Efficient Self-Speculative Decoding

arXiv: https://arxiv.org/abs/2504.05598 · HTML: https://mlsanity.com/paper/2504.05598/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- Placed in Speculative decoding, one of 12 directions in Efficient LLM inference and serving.
- Builds on 22 of its 29 references that are in the map.
- First posted on arXiv 8 Apr 2025.

## Placed in

- Efficient LLM inference and serving / [Speculative decoding](https://mlsanity.com/inference/speculative-decoding/)

## Builds on (most cited first, 22 of 22)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022)
- [Fast Inference from Transformers via Speculative Decoding](https://mlsanity.com/paper/2211.17192/) (2022)
- [Don't Give Me the Details, Just the Summary! Topic-Aware Convolutional Neural Networks for Extreme Summarization](https://mlsanity.com/paper/1808.08745/) (2018)
- [Accelerating Large Language Model Decoding with Speculative Sampling](https://mlsanity.com/paper/2302.01318/) (2023)
- [Abstractive Text Summarization Using Sequence-to-Sequence RNNs and Beyond](https://mlsanity.com/paper/1602.06023/) (2016)
- [Program Induction by Rationale Generation : Learning to Solve and Explain Algebraic Word Problems](https://mlsanity.com/paper/1705.04146/) (2017)
- [LayerSkip: Enabling Early Exit Inference and Self-Speculative Decoding](https://mlsanity.com/paper/2404.16710/) (2024)
- [Unlocking Efficiency in Large Language Model Inference: A Comprehensive Survey of Speculative Decoding](https://mlsanity.com/paper/2401.07851/) (2024)
- [Online Speculative Decoding](https://mlsanity.com/paper/2310.07177/) (2023)
- [GliDe with a CaPE: A Low-Hassle Method to Accelerate Speculative Decoding](https://mlsanity.com/paper/2402.02082/) (2024)
- [SpecDec++: Boosting Speculative Decoding via Adaptive Candidate Lengths](https://mlsanity.com/paper/2405.19715/) (2024)
- [SWIFT: On-the-Fly Self-Speculative Decoding for LLM Inference Acceleration](https://mlsanity.com/paper/2410.06916/) (2024)
- [ROUGE 2.0: Updated and Improved Measures for Evaluation of Summarization Tasks](https://mlsanity.com/paper/1803.01937/) (2018)
- [Kangaroo: Lossless Self-Speculative Decoding via Double Early Exiting](https://mlsanity.com/paper/2404.18911/) (2024)
- [AdaEDL: Early Draft Stopping for Speculative Decoding of Large Language Models via an Entropy-based Lower Bound on Token Acceptance Probability](https://mlsanity.com/paper/2410.18351/) (2024)
- [Direct Alignment of Draft Model for Speculative Decoding with Chat-Fine-Tuned LLMs](https://mlsanity.com/paper/2403.00858/) (2024)
- [Training Domain Draft Models for Speculative Decoding: Best Practices and Insights](https://mlsanity.com/paper/2503.07807/) (2025)
- [Draft on the Fly: Adaptive Self-Speculative Decoding using Cosine Similarity](https://mlsanity.com/paper/2410.01028/) (2024)
- [S3D: A Simple and Cost-Effective Self-Speculative Decoding Scheme for Low-Memory GPUs](https://mlsanity.com/paper/2405.20314/) (2024)

## Built on by (most cited first, 8 of 8)

- [KnapSpec: Self-Speculative Decoding via Adaptive Layer Selection as a Knapsack Problem](https://mlsanity.com/paper/2602.20217/) (2026)
- [ConfLayers: Adaptive Confidence-based Layer Skipping for Self-Speculative Decoding](https://mlsanity.com/paper/2604.14612/) (2026)
- [DSDE: Dynamic Speculative Decoding with KLD Stability for Real-World Serving](https://mlsanity.com/paper/2509.01083/) (2025)
- [Pipeline Parallelism is All You Need for Optimized Early-Exit Based Self-Speculative Decoding](https://mlsanity.com/paper/2509.19368/) (2025)
- [BitSkip: An Empirical Analysis of Quantization and Early Exit Composition in Transformers](https://mlsanity.com/paper/2510.23766/) (2025)
- [ASPIRE: Asynchronous Batched Self-Speculative Decoding for Long-Context LLM Inference](https://mlsanity.com/paper/2609.17943/) (2026)
- [Depth Exploration for LLM Decoding](https://mlsanity.com/paper/2606.29223/) (2026)
- [SEED: Self-Speculative Decoding via Implicit Encoder-Decoder](https://mlsanity.com/paper/2609.36590/) (2026)

## Abstract (arXiv)

Speculative Decoding (SD) is a widely used approach to accelerate the inference of large language models (LLMs) without reducing generation quality. It operates by first using a compact model to draft multiple tokens efficiently, followed by parallel verification using the target LLM. This approach leads to faster inference compared to auto-regressive decoding. While there are multiple approaches to create a draft model, one promising approach is to use early-exit methods. These methods draft candidate tokens by using a subset of layers of the primary model and applying the remaining layers for verification, allowing a single model to handle both drafting and verification. While this technique reduces memory usage and computational cost, its performance relies on the choice of the exit layer for drafting and the number of tokens drafted (speculation length) in each SD round. Prior works use hyperparameter exploration to statically select these values. However, our evaluations show that these hyperparameter values are task-specific, and even within a task they are dependent on the current sequence context. We introduce DEL (Dynamic Exit Layer), a plug-and-play method that adaptively selects the exit layer and speculation length during inference. DEL dynamically tracks the token acceptance rate if the tokens are drafted at each layer of an LLM and uses that knowledge to heuristically select the optimal exit layer and speculation length. Our experiments across a broad range of models and downstream tasks show that DEL achieves overall speedups of 2.16×~2.62× over vanilla auto-regressive decoding and improves upon state-of-the-art SD methods, which peak at 2.43×, by up to 0.19×. The code is available at https://github.com/hoenza/DEL.