<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="tran-etal-2026-exploiting">
    <titleInfo>
        <title>Exploiting Tree Structure for Credit Assignment in Reinforcement Learning with Large Language Models</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Hieu</namePart>
        <namePart type="family">Tran</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Zonghai</namePart>
        <namePart type="family">Yao</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Hong</namePart>
        <namePart type="family">yu</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026-07</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Findings of the Association for Computational Linguistics: ACL 2026</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Maria</namePart>
            <namePart type="family">Liakata</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Viviane</namePart>
            <namePart type="given">P</namePart>
            <namePart type="family">Moreira</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Jiajun</namePart>
            <namePart type="family">Zhang</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">David</namePart>
            <namePart type="family">Jurgens</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">San Diego, California, United States</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-395-1</identifier>
    </relatedItem>
    <abstract>Reinforcement learning has shown strong promise for strengthening the reasoning ability of large language models (LLMs), but sparse, delayed rewards over long chains make token-level credit assignment a central challenge. Actor–critic methods like PPO provide token-level credit but require training a value network alongside the policy, which introduces complexity and can encourage overfitting. Critic-free alternatives such as GRPO avoid this burden but rely on sequence-level outcomes, distributing a single reward uniformly across tokens and ignoring structural differences between responses. We propose Prefix-to-Tree (P2T), which organizes the sampled responses of a prompt into a prefix tree and computes nonparametric prefix values by aggregating descendant outcomes. Building on this idea, we develop TEMPO (Tree-Estimated Mean Prefix Value for Policy Optimization), a critic-free algorithm that enriches GRPO with branch-aware temporal-difference (TD) corrections. Across Qwen3-1.7B and Qwen3-4B, TEMPO consistently improves both convergence and final performance over PPO and GRPO on in-distribution benchmarks (MATH, MedQA) and out-of-distribution settings (GSM-HARD, AMC23, MedMCQA, MMLU-Medical), achieving higher validation accuracy within comparable wall-clock time.</abstract>
    <identifier type="citekey">tran-etal-2026-exploiting</identifier>
    <location>
        <url>https://aclanthology.org/2026.findings-acl.524/</url>
    </location>
    <part>
        <date>2026-07</date>
        <extent unit="page">
            <start>10795</start>
            <end>10810</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Exploiting Tree Structure for Credit Assignment in Reinforcement Learning with Large Language Models
%A Tran, Hieu
%A Yao, Zonghai
%A yu, Hong
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Findings of the Association for Computational Linguistics: ACL 2026
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-395-1
%F tran-etal-2026-exploiting
%X Reinforcement learning has shown strong promise for strengthening the reasoning ability of large language models (LLMs), but sparse, delayed rewards over long chains make token-level credit assignment a central challenge. Actor–critic methods like PPO provide token-level credit but require training a value network alongside the policy, which introduces complexity and can encourage overfitting. Critic-free alternatives such as GRPO avoid this burden but rely on sequence-level outcomes, distributing a single reward uniformly across tokens and ignoring structural differences between responses. We propose Prefix-to-Tree (P2T), which organizes the sampled responses of a prompt into a prefix tree and computes nonparametric prefix values by aggregating descendant outcomes. Building on this idea, we develop TEMPO (Tree-Estimated Mean Prefix Value for Policy Optimization), a critic-free algorithm that enriches GRPO with branch-aware temporal-difference (TD) corrections. Across Qwen3-1.7B and Qwen3-4B, TEMPO consistently improves both convergence and final performance over PPO and GRPO on in-distribution benchmarks (MATH, MedQA) and out-of-distribution settings (GSM-HARD, AMC23, MedMCQA, MMLU-Medical), achieving higher validation accuracy within comparable wall-clock time.
%U https://aclanthology.org/2026.findings-acl.524/
%P 10795-10810

Markdown (Informal)

[Exploiting Tree Structure for Credit Assignment in Reinforcement Learning with Large Language Models](https://aclanthology.org/2026.findings-acl.524/) (Tran et al., Findings 2026)

• Exploiting Tree Structure for Credit Assignment in Reinforcement Learning with Large Language Models (Tran et al., Findings 2026)

ACL

• Hieu Tran, Zonghai Yao, and Hong yu. 2026. Exploiting Tree Structure for Credit Assignment in Reinforcement Learning with Large Language Models. In Findings of the Association for Computational Linguistics: ACL 2026, pages 10795–10810, San Diego, California, United States. Association for Computational Linguistics.