REVIEW 17 cited by
Reward-Guided Speculative Decoding for Efficient LLM Reasoning
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
We introduce Reward-Guided Speculative Decoding (RSD), a novel framework aimed at improving the efficiency of inference in large language models (LLMs). RSD synergistically combines a lightweight draft model with a more powerful target model, incorporating a controlled bias to prioritize high-reward outputs, in contrast to existing speculative decoding methods that enforce strict unbiasedness. RSD employs a process reward model to evaluate intermediate decoding steps and dynamically decide whether to invoke the target model, optimizing the trade-off between computational cost and output quality. We theoretically demonstrate that a threshold-based mixture strategy achieves an optimal balance between resource utilization and performance. Extensive evaluations on challenging reasoning benchmarks, including Olympiad-level tasks, show that RSD delivers significant efficiency gains against decoding with the target model only (up to 4.4x fewer FLOPs), while achieving significant better accuracy than parallel decoding method on average (up to +3.5). These results highlight RSD as a robust and cost-effective approach for deploying LLMs in resource-intensive scenarios. The code is available at https://github.com/BaohaoLiao/RSD.
Forward citations
Cited by 17 Pith papers
-
When to Plan: Learning to Select Between Reactive Control and Deliberative Planning
An RL-trained meta-policy that uses ensemble uncertainty to choose between a cheap reactive policy and costly planning reaches goals faster than fixed baselines and adapts as the reactive policy improves.
-
Multi-Head Latent Control: A Unified Interface for LLM Agent Decision Making
Hidden-state traces of frozen LLMs/VLMs can be read by lightweight trained heads to predict when to defer, clarify, call tools, or abstain, cutting routed inference cost 27–90%.
-
Simultaneous Multi-objective Alignment Across Verifiable and Non-verifiable Rewards
MAHALO aligns LLMs to multiple objectives in one model via per-objective action heads and PRM-guided decoding, improving math, value, and tutoring metrics jointly.
-
AALC: Large Language Model Efficient Reasoning via Adaptive Accuracy-Length Control
AALC, an accuracy-aware length reward, reduces reasoning-model response length by over 50% on math benchmarks while preserving or improving accuracy.
-
How Far Are We from Optimal Reasoning Efficiency?
The authors define a reasoning efficiency frontier and a gap metric (REG), then train models with REO-RL to shrink the gap by at least 50% with only small accuracy losses.
-
What makes Reasoning Models Different? Follow the Reasoning Leader for Efficient Decoding
FoReaL-Decoding lets a strong reasoning model generate the first few tokens of each sentence and a weaker model complete the sentence, cutting theoretical FLOPs by 30-55% while retaining 86-100% of accuracy on four ma...
-
Is Your Model Thinking or Just Stagnating? PUMA: Diagnosing Reasoning Pathology via Phase-Momentum Alignment
PUMA detects when a reasoning model's entropy drop aligns with hidden-state momentum, truncates at that point, and reports improved accuracy-efficiency on 1.5B-32B reasoning models.
-
MixReasoning: Switching Modes to Think
A reasoning model can cut token usage by roughly a third to a half without losing accuracy if it switches to brief output on low-uncertainty steps and expands only at high-uncertainty forks.
-
Rethinking Visual Autoregressive Sampling with Information-Grounding Guidance
IGG, an attention-based reweighting of classifier-free guidance, concentrates guidance on important tokens and modestly improves FID/IS in scale-wise autoregressive image generation.
-
CoRE: Enhancing Metacognition with Label-free Self-evaluation in LRMs
A training-free and label-free detector of cyclic hidden-state patterns triggers early exit during chain-of-thought reasoning, reducing token length while mostly preserving or improving accuracy.
-
Efficient Reasoning Through Suppression of Self-Affirmation Reflections in Large Reasoning Models
Suppressing low-probability 'wait' tokens that trigger self-affirmation reflections shortens reasoning-model output by 8-18% with roughly unchanged accuracy.
-
From Long to Short: LLMs Excel at Trimming Own Reasoning Chains
EDIT searches across step-count prompts to find the shortest reasoning path the model consistently answers correctly.
-
Collaborative Inference and Learning between Edge SLMs and Cloud LLMs: A Survey of Algorithms, Execution, and Open Challenges
A survey that builds a taxonomy of edge-cloud LLM-SLM collaboration for inference and training, claiming to be the first to unify both phases.
-
Logit Arithmetic Elicits Long Reasoning Capabilities Without Training
ThinkLogit blends logits from a small reasoning guider into a frozen 32B model, improving math pass@1 by up to 29% without training the large model.
-
PREMISE: Scalable and Strategic Prompt Optimization for Efficient Mathematical Reasoning in Large Models
The paper reports that a concise optimized prompt reduces reasoning tokens by up to 87.5% while roughly preserving accuracy on GSM8K, SVAMP, and MATH-500.
-
Wait, We Don't Need to "Wait"! Removing Thinking Tokens Improves Reasoning Efficiency
Suppressing "Wait"-like reflection tokens at decode time reduces reasoning token counts by 27-51% across five R1-style model families, with mixed accuracy effects.
-
Inverse Reinforcement Learning Meets Large Language Model Post-Training: Basics, Advances, and Opportunities
A tutorial reviewing LLM alignment through the lens of inverse reinforcement learning, arguing that neural reward models learned from human data are central to post-training.
Discussion (0). Sign in to comment.