REVIEW 28 cited by
Just Enough Thinking: Efficient Reasoning with Adaptive Length Penalties Reinforcement Learning
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
Just Enough Thinking: Efficient Reasoning with Adaptive Length Penalties Reinforcement Learning
read the original abstract
Large reasoning models (LRMs) achieve higher performance on challenging reasoning tasks by generating more tokens at inference time, but this verbosity often wastes computation on easy problems. Existing solutions, including supervised finetuning on shorter traces, user-controlled budgets, or RL with uniform penalties, either require data curation, manual configuration, or treat all problems alike regardless of difficulty. We introduce Adaptive Length Penalty (ALP), a reinforcement learning objective tailoring generation length to per-prompt solve rate. During training, ALP monitors each prompt's online solve rate through multiple rollouts and adds a differentiable penalty whose magnitude scales inversely with that rate, so confident (easy) prompts incur a high cost for extra tokens while hard prompts remain unhindered. Posttraining DeepScaleR-1.5B with ALP cuts average token usage by 50\% without significantly dropping performance. Relative to fixed-budget and uniform penalty baselines, ALP redistributes its reduced budget more intelligently by cutting compute on easy prompts and reallocating saved tokens to difficult ones, delivering higher accuracy on the hardest problems with higher cost.
Forward citations
Cited by 28 Pith papers
-
LEAD: Length-Efficient Adaptive and Dynamic Reasoning for Large Language Models
LEAD uses online adaptive mechanisms including Potential-Scaled Instability and symmetric efficiency rewards based on correct rollouts to achieve higher accuracy-efficiency scores with substantially shorter reasoning ...
-
DUET: Optimize Token-Budget Allocation for Reinforcement Learning with Verifiable Rewards
DUET improves RLVR by allocating tokens across both prompt selection and rollout length, outperforming full-budget baselines even when using only half the tokens.
-
Schoenfeld's Anatomy of Mathematical Reasoning by Language Models
ThinkARM abstracts LLM reasoning traces into Schoenfeld episodes and shows that exploration steps correlate with correctness while efficiency methods selectively suppress evaluative feedback.
-
Length Penalties Make Chain-of-Thought Less Monitorable
Length-penalized RL shortens chain-of-thought while preserving accuracy and hint influence, but preferentially removes the cues that let a monitor detect that influence.
-
From Reasoning Traces to Reusable Modules: Understanding Compositional Generalization in Language Model Reasoning
SFT supplies entangled compositional traces of atomic skills and routing modules; RL identifies those modules and enables recombination on novel compositions outside the SFT support.
-
Length Penalties Make Chain-of-Thought Less Monitorable
Length-penalized RL reduces the chance that a reasoning trace discloses a biasing hint, an effect beyond the shorter token budget alone.
-
Length Penalties Make Chain-of-Thought Less Monitorable
Length-penalized RL keeps hint influence intact but makes chains mention the hint less, beyond what shorter length alone predicts.
-
From Reasoning Traces to Reusable Modules: Understanding Compositional Generalization in Language Model Reasoning
Introduces a hierarchical latent selection model showing SFT supplies raw module materials in compound traces while RL decomposes them to identify atomic modules and enable recombination for new reasoning configurations.
-
CLORE: Content-Level Optimization for Reasoning Efficiency
CLORE augments correct on-policy rollouts by deleting repetitive and irrelevant segments then optimizes with auxiliary DPO to improve accuracy-efficiency trade-off on math benchmarks.
-
Taming the Thinker: Conditional Entropy Shaping for Adaptive LLM Reasoning
CES applies conditional bidirectional entropy control on top of DAPO to improve accuracy and shorten responses on mathematical benchmarks for 7B and 1.5B LLMs.
-
ZAYA1-8B Technical Report
ZAYA1-8B is a reasoning MoE model with 700M active parameters that matches larger models on math and coding benchmarks and reaches 91.9% on AIME'25 via Markovian RSA test-time compute.
-
Length Value Model: Scalable Value Pretraining for Token-Level Length Modeling
LenVM models token-level remaining generation length as a bounded discounted value function derived from constant negative per-token rewards, providing a scalable proxy for generation horizon.
-
Length Value Model: Scalable Value Pretraining for Token-Level Length Modeling
LenVM trains a token-level value head to predict discounted remaining length, enabling length control and efficiency steering on LLMs and VLMs.
-
CODA: Difficulty-Aware Compute Allocation for Adaptive Reasoning
CODA uses rollout-based difficulty signals to drive two gates that penalize verbosity on easy instances and promote deliberation on hard ones, cutting token use over 60% on simple tasks while maintaining accuracy.
-
ATTNPO: Attention-Guided Process Supervision for Efficient Reasoning
ATTNPO guides process-supervised RL with intrinsic attention signals to shorten reasoning traces while raising accuracy on nine benchmarks.
-
On the Optimal Reasoning Length for RL-Trained Language Models
For RL-trained language models, reasoning accuracy peaks at an intermediate output length; longer outputs mainly add dispersion around an improving mode, while too-short outputs under-think.
-
Neural Chain-of-Thought Search: Searching the Optimal Reasoning Path to Enhance Large Language Models
NCoTS treats chain-of-thought reasoning as a search problem and uses a dual-factor heuristic to find paths that are over 3.5% more accurate and 22% shorter on benchmarks.
-
CASHEW: Stabilizing Multimodal Reasoning via Iterative Trajectory Aggregation
Iteratively aggregating multiple reasoning trajectories with Grounding-DINO-verified visual evidence improved VLM accuracy on 13 multimodal benchmarks, and an RL-trained version internalizes the aggregation.
-
Learning to Reason Efficiently with Discounted Reinforcement Learning
Discounted correctness rewards in GRPO post-training shorten LLM reasoning traces by roughly 13-40% while matching pass@1 accuracy, supported by Blackwell-optimality analysis for restricted policy classes.
-
ZUNA1.1: A more flexible EEG foundation model for Denoising and Super-resolution
ZUNA1.1, an open-source 380M EEG diffusion autoencoder, reconstructs variable-length, flexibly masked EEG at least as well as its predecessor and far better than spherical spline interpolation.
-
Beyond Penalizing Mistakes: Stabilizing Efficiency Training in Large Reasoning Models via Adaptive Correct-Only Rewards
ACOER applies adaptive correct-only efficiency rewards in GRPO to avoid reward collapse, yielding higher accuracy and over 60% fewer tokens on math reasoning benchmarks.
-
Modularized Reinforcement Learning on LLMs: From MDP Creation to Exploration and Learning
Survey mapping RL techniques onto LLM training and highlighting gaps in value-based, off-policy, and bootstrapping methods.
-
Trust Region On-Policy Distillation
TrOPD stabilizes on-policy distillation for LLMs with trust-region learning, outlier estimation, and off-policy guidance, outperforming prior OPD methods on reasoning and code benchmarks.
-
ZONOS2 Technical Report
ZONOS2 8B is a scaled MoE TTS model with 900M active parameters trained on 6M hours of data that reports competitive SOTA results on naturalness, speaker similarity, WER, and a new ZTTS1-Eval benchmark while releasing...
-
Libra: Efficient Resource Management for Agentic RL Post-Training
Libra optimizes GPU allocation across rollout and training in agentic RL via an elastic hybrid pool and C-MLFQ scheduler based on tool-return causal signals, claiming up to 3.0x throughput and 2.5x faster reward conve...
-
ZONOS2 Technical Report
ZONOS2 8B scales a prior TTS system to 8B parameters with MoE architecture and 6M hours of data, reporting competitive benchmark performance on naturalness and speaker similarity while releasing weights.
-
Token-Operations-Oriented Inference Optimization Techniques for Large Models
A survey of large-model inference optimization, organized as a four-layer 'token-operations' taxonomy: multi-model fusion, model optimization, compute-model fusion, and compute-network-model fusion.
-
Token-Operations-Oriented Inference Optimization Techniques for Large Models
The paper introduces a four-layer technical architecture for token-operations-oriented inference optimization in large models and reviews key technologies and industry status at each layer.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.