@inproceedings{guo-etal-2026-generating,
title = "When Does Generating More Help? Disentangling Fixed-Source Synthesis from Source Expansion in Synthetic Data Scaling",
author = "Guo, Xu and
Tong, Jian and
Lu, Zhihui and
Guo, Qipeng",
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.839/",
pages = "17767--17783",
ISBN = "979-8-89176-462-0",
abstract = "Synthetic data can be scaled along two routes: Source Expansion (SE), which enlarges the source by adding seed materials or generators, and Fixed-Source Synthesis (FSS), which holds the source fixed and scales the generation budget. Existing scaling studies typically expand the source as the data grows, conflating SE with FSS and leaving FSS underexplored. We isolate FSS within a standard rejection sampling (RS) pipeline and derive a scaling form for it from how repeated sampling covers a fixed source. Empirically, fit on low budgets, the form predicts performance at the held-out highest budget for every evaluated teacher{--}student pair. At matched total-sample budgets, SE and FSS are comparable at small budgets, but adding seed questions outperforms spending the same budget on more responses at large budgets. Within FSS, neither synthesizing additional questions from the existing seeds nor varying the synthesis protocol consistently outperforms plain RS across matched budgets. FSS is thus a bounded scaling axis and a controlled setting for comparing synthesis protocols. We will release our code and data to facilitate further research."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="guo-etal-2026-generating">
<titleInfo>
<title>When Does Generating More Help? Disentangling Fixed-Source Synthesis from Source Expansion in Synthetic Data Scaling</title>
</titleInfo>
<name type="personal">
<namePart type="given">Xu</namePart>
<namePart type="family">Guo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jian</namePart>
<namePart type="family">Tong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhihui</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qipeng</namePart>
<namePart type="family">Guo</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>Synthetic data can be scaled along two routes: Source Expansion (SE), which enlarges the source by adding seed materials or generators, and Fixed-Source Synthesis (FSS), which holds the source fixed and scales the generation budget. Existing scaling studies typically expand the source as the data grows, conflating SE with FSS and leaving FSS underexplored. We isolate FSS within a standard rejection sampling (RS) pipeline and derive a scaling form for it from how repeated sampling covers a fixed source. Empirically, fit on low budgets, the form predicts performance at the held-out highest budget for every evaluated teacher–student pair. At matched total-sample budgets, SE and FSS are comparable at small budgets, but adding seed questions outperforms spending the same budget on more responses at large budgets. Within FSS, neither synthesizing additional questions from the existing seeds nor varying the synthesis protocol consistently outperforms plain RS across matched budgets. FSS is thus a bounded scaling axis and a controlled setting for comparing synthesis protocols. We will release our code and data to facilitate further research.</abstract>
<identifier type="citekey">guo-etal-2026-generating</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.839/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>17767</start>
<end>17783</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T When Does Generating More Help? Disentangling Fixed-Source Synthesis from Source Expansion in Synthetic Data Scaling
%A Guo, Xu
%A Tong, Jian
%A Lu, Zhihui
%A Guo, Qipeng
%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 guo-etal-2026-generating
%X Synthetic data can be scaled along two routes: Source Expansion (SE), which enlarges the source by adding seed materials or generators, and Fixed-Source Synthesis (FSS), which holds the source fixed and scales the generation budget. Existing scaling studies typically expand the source as the data grows, conflating SE with FSS and leaving FSS underexplored. We isolate FSS within a standard rejection sampling (RS) pipeline and derive a scaling form for it from how repeated sampling covers a fixed source. Empirically, fit on low budgets, the form predicts performance at the held-out highest budget for every evaluated teacher–student pair. At matched total-sample budgets, SE and FSS are comparable at small budgets, but adding seed questions outperforms spending the same budget on more responses at large budgets. Within FSS, neither synthesizing additional questions from the existing seeds nor varying the synthesis protocol consistently outperforms plain RS across matched budgets. FSS is thus a bounded scaling axis and a controlled setting for comparing synthesis protocols. We will release our code and data to facilitate further research.
%U https://aclanthology.org/2026.findings-emnlp.839/
%P 17767-17783
Markdown (Informal)
[When Does Generating More Help? Disentangling Fixed-Source Synthesis from Source Expansion in Synthetic Data Scaling](https://aclanthology.org/2026.findings-emnlp.839/) (Guo et al., Findings 2026)
ACL