@inproceedings{zhu-han-2026-unlocking,
title = "Unlocking the Unsolvable: Teacher-Guided Curriculum for Data-Efficient {RLVR}",
author = "Zhu, Yukang and
Han, Zhen",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.931/",
doi = "10.18653/2026.findings-emnlp.931",
pages = "19639--19658",
ISBN = "979-8-89176-462-0",
abstract = "Reinforcement Learning with Verifiable Rewards (RLVR) has shown remarkable success in improving the mathematical reasoning of large language models. Yet problems beyond the model{'}s current capability, where rollouts uniformly fail and no learning signal is produced, are structurally wasted despite marking the most informative training frontier. We show that these otherwise-inert problems can be unlocked via teacher-guided curriculum learning: partial reasoning traces from a stronger model create a graded difficulty landscape, and a backward-chaining curriculum progressively withdraws guidance until the model solves problems unaided. Training on only 128 unsolvable problems matches or exceeds GRPO trained on a full 2,000-problem corpus ({\textasciitilde}16{\texttimes} data efficiency) on the nine-benchmark average for both base models, while substantially expanding the reasoning boundary measured by pass@k at large k. Furthermore, we identify a distribution-shift cost that is particularly acute in the unsolvable-only regime and propose Monotone Frontier Curriculum (MFC), a method that monotonically drives training toward unguided solving, consistently outperforming existing curriculum methods."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhu-han-2026-unlocking">
<titleInfo>
<title>Unlocking the Unsolvable: Teacher-Guided Curriculum for Data-Efficient RLVR</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yukang</namePart>
<namePart type="family">Zhu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhen</namePart>
<namePart type="family">Han</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Reinforcement Learning with Verifiable Rewards (RLVR) has shown remarkable success in improving the mathematical reasoning of large language models. Yet problems beyond the model’s current capability, where rollouts uniformly fail and no learning signal is produced, are structurally wasted despite marking the most informative training frontier. We show that these otherwise-inert problems can be unlocked via teacher-guided curriculum learning: partial reasoning traces from a stronger model create a graded difficulty landscape, and a backward-chaining curriculum progressively withdraws guidance until the model solves problems unaided. Training on only 128 unsolvable problems matches or exceeds GRPO trained on a full 2,000-problem corpus (~16× data efficiency) on the nine-benchmark average for both base models, while substantially expanding the reasoning boundary measured by pass@k at large k. Furthermore, we identify a distribution-shift cost that is particularly acute in the unsolvable-only regime and propose Monotone Frontier Curriculum (MFC), a method that monotonically drives training toward unguided solving, consistently outperforming existing curriculum methods.</abstract>
<identifier type="citekey">zhu-han-2026-unlocking</identifier>
<identifier type="doi">10.18653/2026.findings-emnlp.931</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.931/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>19639</start>
<end>19658</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Unlocking the Unsolvable: Teacher-Guided Curriculum for Data-Efficient RLVR
%A Zhu, Yukang
%A Han, Zhen
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F zhu-han-2026-unlocking
%X Reinforcement Learning with Verifiable Rewards (RLVR) has shown remarkable success in improving the mathematical reasoning of large language models. Yet problems beyond the model’s current capability, where rollouts uniformly fail and no learning signal is produced, are structurally wasted despite marking the most informative training frontier. We show that these otherwise-inert problems can be unlocked via teacher-guided curriculum learning: partial reasoning traces from a stronger model create a graded difficulty landscape, and a backward-chaining curriculum progressively withdraws guidance until the model solves problems unaided. Training on only 128 unsolvable problems matches or exceeds GRPO trained on a full 2,000-problem corpus (~16× data efficiency) on the nine-benchmark average for both base models, while substantially expanding the reasoning boundary measured by pass@k at large k. Furthermore, we identify a distribution-shift cost that is particularly acute in the unsolvable-only regime and propose Monotone Frontier Curriculum (MFC), a method that monotonically drives training toward unguided solving, consistently outperforming existing curriculum methods.
%R 10.18653/2026.findings-emnlp.931
%U https://aclanthology.org/2026.findings-emnlp.931/
%U https://doi.org/10.18653/2026.findings-emnlp.931
%P 19639-19658
Markdown (Informal)
[Unlocking the Unsolvable: Teacher-Guided Curriculum for Data-Efficient RLVR](https://aclanthology.org/2026.findings-emnlp.931/) (Zhu & Han, Findings 2026)
ACL