arXiv Computation and Language November 4, 2026 By Adam Y. Ahmed, Abdullah G. Elafifi

Test-Time Scaling for Open-Ended Reasoning at Small Model Scale

Read the original on arXiv Computation and Language →

The paper investigates test-time scaling for small-scale open-ended reasoning models by introducing semantic cluster voting, a judge-free method that groups model outputs by meaning and selects the largest cluster’s representative. Experiments on a 250-prompt benchmark across three 3‑B models show that voting improves accuracy on math tasks (10–29 point gains) but only marginally on open-ended text (+0.30 points). A larger 200‑problem MATH‑500 run confirms a +8.5 point math improvement, while most open-ended samples fall into a single cluster, indicating limited diversity and that higher decoding temperatures reduce quality.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computation and Language.

arXiv AI

Aug 25

More Accurate or More Efficient? Evaluating Locally Deployed Compact Open-Weight Language Models for Mathematical Reasoning

arXiv:2608.22048v1 Announce Type: new Abstract: Large language models are increasingly deployed on local hardware for privacy, cost, and accessibility reasons. Yet many evaluations emphasize accuracy...

By Orion Powers, Daniella Seum, Khaled Slhoub

llms safety

arXiv AI

Jun 30

When More Sampling Hurts: The Modal Ceiling and Correlation Ceiling of Test-Time Scaling

arXiv:2606. 28661v1 Announce Type: cross Abstract: People overthink; language models over-sample, and the extra effort can talk both into a worse answer.

By Yong Yi Bay, Kathleen A. Yearick

llms benchmarks

arXiv AI

Aug 7

Refining Over Resampling: Test-Time Self-Correction for LLM Reasoning

arXiv:2608. 05643v1 Announce Type: new Abstract: Test-time scaling improves LLM reasoning by using additional inference compute, but wider sampling alone can suffer from diminishing returns: new rollouts often repeat existing answer patterns instead of adding useful reasoning diversity.

By Ahsan Bilal, Muhammad Ahmed Mohsin, Muhammad Umer, Lena Trigg, Ali Subhan, Muhammad Ali, Dean F. Hougen

llms reinforcement-learning

arXiv AI

Jul 14

LLMs as a Jury: Cross-Model Consensus Can Outperform Process Reward Models for LLM Reasoning

arXiv:2607. 10139v1 Announce Type: cross Abstract: Selecting the correct answer from a pool of candidate reasoning chains is the engine of test-time scaling, yet the standard selectors each carry a cost: self-consistency inherits the errors of the single model it resamples, and trained reward models need labeled data and transfer poorly off-distribution.

By Ning Liu

llms benchmarks

arXiv AI

Aug 18

Measuring Reward Hacking and Reasoning-Answer Decoupling Under Position-Confounded Optimization

arXiv:2608. 15445v1 Announce Type: new Abstract: When a reward is correct on every training example yet consistent with more than one goal, a model can acquire an unintended one, a failure known as goal misgeneralization.

By Suyash Maniyar, Armaan Sandhu, Abhishek Mishra

llms benchmarks

arXiv Machine Learning

Jul 31

Sample More, Reflect Less: Self-Refine and Reflexion Lose to Repeated Sampling at Equal Token Cost, from 1.5B to 7B

arXiv:2607. 28576v1 Announce Type: cross Abstract: Methods that make a language model plan, criticise and rewrite its own answer, reflect on mistakes, pick the best of several attempts, or debate with copies of itself nearly all make it generate far more text than a single chain of thought.

By Iliya Mirzaei

llms benchmarks

One email a morning, machine-written

Related stories

More Accurate or More Efficient? Evaluating Locally Deployed Compact Open-Weight Language Models for Mathematical Reasoning

When More Sampling Hurts: The Modal Ceiling and Correlation Ceiling of Test-Time Scaling

Refining Over Resampling: Test-Time Self-Correction for LLM Reasoning

LLMs as a Jury: Cross-Model Consensus Can Outperform Process Reward Models for LLM Reasoning

Measuring Reward Hacking and Reasoning-Answer Decoupling Under Position-Confounded Optimization

Sample More, Reflect Less: Self-Refine and Reflexion Lose to Repeated Sampling at Equal Token Cost, from 1.5B to 7B