mlsanityData to 14 Oct 2026

RADAR: Accelerate Large Language Model Inference With RL-Based Dynamic Draft Trees

arXiv 2512.14069 ↗First posted 16 Dec 2025cs.AIcited by 0

On the map: Efficient LLM inference and serving → Speculative decoding

Abstract (arXiv)

Inference with modern Large Language Models (LLMs) is expensive and slow, and speculative sampling has emerged as an effective solution to this problem. However, the number of calls to the draft model for generating candidate tokens in speculative sampling is a preset hyperparameter, lacking flexibility. To generate and utilize the candidate tokens more effectively, we propose RADAR, a novel speculative sampling method with RL-based dynamic draft trees. RADAR formulates the draft tree generation process as a Markov Decision Process (MDP) and employs offline reinforcement learning to train a prediction model, which enables real-time decisions on calls to the draft model, reducing redundant computations and further accelerating inference. Evaluations across three LLMs and four tasks show that RADAR achieves a speedup of 3.17x-4.82x over the auto-regressive decoding baseline. The code is available at https://github.com/minaduki-sora/RADAR.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Speculative decoding, one of 12 directions in Efficient LLM inference and serving. Builds on 18 of its 24 references that are in the map. First posted on arXiv 16 Dec 2025.

Builds on

18 of 24 references.

Closest ancestors · 12

Same direction, newest first

  1. 2025EAGLE-3: Scaling up Inference Acceleration of Large Language Models via Training-Time Test3 Mar
  2. 2024Falcon: Faster and Parallel Inference of Large Language Models through Enhanced Semi-Autoregressive Drafting and Custom-Designed Decoding Tree17 Dec
  3. 2024Learning Harmonized Representations for Speculative Sampling28 Aug
  4. 2024PEARL: Parallel Speculative Decoding with Adaptive Draft Length13 Aug
  5. 2024EAGLE-2: Faster Inference of Language Models with Dynamic Draft Trees24 Jun
  6. 2024SpecDec++: Boosting Speculative Decoding via Adaptive Candidate Lengths30 May
  7. 2024Kangaroo: Lossless Self-Speculative Decoding via Double Early Exiting29 Apr
  8. 2024EAGLE: Speculative Sampling Requires Rethinking Feature Uncertainty26 Jan
  9. 2024Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads19 Jan
  10. 2023SpecInfer: Accelerating Generative Large Language Model Serving with Tree-based Speculative Inference and Verification16 May
  11. 2023Accelerating Large Language Model Decoding with Speculative Sampling2 Feb
  12. 2022Fast Inference from Transformers via Speculative Decoding30 Nov

Other references · 6

  1. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
  2. 2021Program Synthesis with Large Language Modelscited by 3,637
  3. 2022GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformerscited by 1,919
  4. 2018Off-Policy Deep Reinforcement Learning without Explorationcited by 1,241
  5. 2023On-Policy Distillation of Language Models: Learning from Self-Generated Mistakescited by 939
  6. 2023Efficient Prompting via Dynamic In-Context Learningcited by 18

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Speculative Sampling with Reinforcement Learning
  2. 2026Learning to Draft: Adaptive Speculative Decoding with Reinforcement Learning
  3. 2025READER: Retrieval-Assisted Drafter for Efficient LLM Inference
  4. 2024Recursive Speculative Decoding: Accelerating LLM Inference via Sampling Without Replacement
  5. 2024Speculative Decoding with CTC-based Draft Model for LLM Inference Acceleration
  6. 2026Training Parallel Speculative Draft Models by Directly Minimizing Expected Decoding Rounds
  7. 2025FR-Spec: Accelerating Large-Vocabulary Language Models via Frequency-Ranked Speculative Sampling
  8. 2026PRISM: Parametrically Refactoring Inference for Speculative Sampling Draft Models