Saved in:
Bibliographic Details
Main Authors: Kim, Hongbeen, Lee, Juhyun, Lee, Sanghyeon, Choi, Kwanghoon, Huh, Jaehyuk
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.00510
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918422963552256
author Kim, Hongbeen
Lee, Juhyun
Lee, Sanghyeon
Choi, Kwanghoon
Huh, Jaehyuk
author_facet Kim, Hongbeen
Lee, Juhyun
Lee, Sanghyeon
Choi, Kwanghoon
Huh, Jaehyuk
contents Monte Carlo Tree Search (MCTS) is an effective test-time compute scaling (TTCS) method for improving the reasoning performance of large language models, but its highly variable execution time leads to severe long-tail latency in practice. Existing optimizations such as positive early exit, reduce latency in favorable cases but are less effective when search continues without meaningful progress. We introduce {\it negative early exit}, which prunes unproductive MCTS trajectories, and an {\it adaptive boosting mechanism} that reallocates reclaimed computation to reduce resource contention among concurrent searches. Integrated into vLLM, these techniques substantially reduce p99 end-to-end latency while improving throughput and maintaining reasoning accuracy.
format Preprint
id arxiv_https___arxiv_org_abs_2604_00510
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Adaptive Parallel Monte Carlo Tree Search for Efficient Test-time Compute Scaling
Kim, Hongbeen
Lee, Juhyun
Lee, Sanghyeon
Choi, Kwanghoon
Huh, Jaehyuk
Artificial Intelligence
Monte Carlo Tree Search (MCTS) is an effective test-time compute scaling (TTCS) method for improving the reasoning performance of large language models, but its highly variable execution time leads to severe long-tail latency in practice. Existing optimizations such as positive early exit, reduce latency in favorable cases but are less effective when search continues without meaningful progress. We introduce {\it negative early exit}, which prunes unproductive MCTS trajectories, and an {\it adaptive boosting mechanism} that reallocates reclaimed computation to reduce resource contention among concurrent searches. Integrated into vLLM, these techniques substantially reduce p99 end-to-end latency while improving throughput and maintaining reasoning accuracy.
title Adaptive Parallel Monte Carlo Tree Search for Efficient Test-time Compute Scaling
topic Artificial Intelligence
url https://arxiv.org/abs/2604.00510