@inproceedings{maheshwari-2026-halved,
title = "Halved {CLM} Exposure Mitigates Late-Training Degradation in Small Recurrent Language Models",
author = "Maheshwari, Yash",
editor = "Choshen, Leshem and
Cotterell, Ryan and
Gul, Mustafa Omer and
Jumelet, Jaap and
Linzen, Tal and
Mueller, Aaron and
Salhan, Suchir and
Shah, Raj Sanjay and
Warstadt, Alex and
Wilcox, Ethan Gotlieb",
booktitle = "Proceedings of the Second {B}aby{LM} Workshop",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.babylm-main.1/",
pages = "1--14",
ISBN = "979-8-89176-475-0",
abstract = "Small recurrent language models can degrade late in training under standard causal language modeling (CLM). On the BabyLM Strict-Small corpus, vanilla RWKV-7 peaks near 69{\%} BLiMP early, then loses 2.57 to 3.85 pp by step 18k across all three seeds we ran. Halving how often CLM updates the weights mitigates this decline, via two mechanisms: alternating a backward MLM objective every other step (BiRWKV-7), or skipping every other gradient step (vanilla-skip). Vanilla-skip carries no auxiliary objectives yet beats BiRWKV-7 on peak BLiMP in every seed, so the auxiliaries are not necessary for stability. What halving changes is coupled: it reduces the CLM update count and doubles the learning-rate decay in update space. A constant-rate control still degrades at the two higher rates we tested, so a low terminal rate is not necessary for the decline; but at fixed update count, flooring the rate earlier recovers most of it. The update-space rate trajectory therefore also modulates the decline; it stays coupled to update count, and our design does not separate the two. On the official BabyLM 2026 pipeline our submission scores 68.70{\%} BLiMP, above the GPT-2 baseline (65.23{\%}, 98.4M) at about 28{\%} as many parameters."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="maheshwari-2026-halved">
<titleInfo>
<title>Halved CLM Exposure Mitigates Late-Training Degradation in Small Recurrent Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yash</namePart>
<namePart type="family">Maheshwari</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>Proceedings of the Second BabyLM Workshop</title>
</titleInfo>
<name type="personal">
<namePart type="given">Leshem</namePart>
<namePart type="family">Choshen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ryan</namePart>
<namePart type="family">Cotterell</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mustafa</namePart>
<namePart type="given">Omer</namePart>
<namePart type="family">Gul</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jaap</namePart>
<namePart type="family">Jumelet</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tal</namePart>
<namePart type="family">Linzen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Aaron</namePart>
<namePart type="family">Mueller</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Suchir</namePart>
<namePart type="family">Salhan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Raj</namePart>
<namePart type="given">Sanjay</namePart>
<namePart type="family">Shah</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alex</namePart>
<namePart type="family">Warstadt</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ethan</namePart>
<namePart type="given">Gotlieb</namePart>
<namePart type="family">Wilcox</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">979-8-89176-475-0</identifier>
</relatedItem>
<abstract>Small recurrent language models can degrade late in training under standard causal language modeling (CLM). On the BabyLM Strict-Small corpus, vanilla RWKV-7 peaks near 69% BLiMP early, then loses 2.57 to 3.85 pp by step 18k across all three seeds we ran. Halving how often CLM updates the weights mitigates this decline, via two mechanisms: alternating a backward MLM objective every other step (BiRWKV-7), or skipping every other gradient step (vanilla-skip). Vanilla-skip carries no auxiliary objectives yet beats BiRWKV-7 on peak BLiMP in every seed, so the auxiliaries are not necessary for stability. What halving changes is coupled: it reduces the CLM update count and doubles the learning-rate decay in update space. A constant-rate control still degrades at the two higher rates we tested, so a low terminal rate is not necessary for the decline; but at fixed update count, flooring the rate earlier recovers most of it. The update-space rate trajectory therefore also modulates the decline; it stays coupled to update count, and our design does not separate the two. On the official BabyLM 2026 pipeline our submission scores 68.70% BLiMP, above the GPT-2 baseline (65.23%, 98.4M) at about 28% as many parameters.</abstract>
<identifier type="citekey">maheshwari-2026-halved</identifier>
<location>
<url>https://aclanthology.org/2026.babylm-main.1/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>1</start>
<end>14</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Halved CLM Exposure Mitigates Late-Training Degradation in Small Recurrent Language Models
%A Maheshwari, Yash
%Y Choshen, Leshem
%Y Cotterell, Ryan
%Y Gul, Mustafa Omer
%Y Jumelet, Jaap
%Y Linzen, Tal
%Y Mueller, Aaron
%Y Salhan, Suchir
%Y Shah, Raj Sanjay
%Y Warstadt, Alex
%Y Wilcox, Ethan Gotlieb
%S Proceedings of the Second BabyLM Workshop
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-475-0
%F maheshwari-2026-halved
%X Small recurrent language models can degrade late in training under standard causal language modeling (CLM). On the BabyLM Strict-Small corpus, vanilla RWKV-7 peaks near 69% BLiMP early, then loses 2.57 to 3.85 pp by step 18k across all three seeds we ran. Halving how often CLM updates the weights mitigates this decline, via two mechanisms: alternating a backward MLM objective every other step (BiRWKV-7), or skipping every other gradient step (vanilla-skip). Vanilla-skip carries no auxiliary objectives yet beats BiRWKV-7 on peak BLiMP in every seed, so the auxiliaries are not necessary for stability. What halving changes is coupled: it reduces the CLM update count and doubles the learning-rate decay in update space. A constant-rate control still degrades at the two higher rates we tested, so a low terminal rate is not necessary for the decline; but at fixed update count, flooring the rate earlier recovers most of it. The update-space rate trajectory therefore also modulates the decline; it stays coupled to update count, and our design does not separate the two. On the official BabyLM 2026 pipeline our submission scores 68.70% BLiMP, above the GPT-2 baseline (65.23%, 98.4M) at about 28% as many parameters.
%U https://aclanthology.org/2026.babylm-main.1/
%P 1-14
Markdown (Informal)
[Halved CLM Exposure Mitigates Late-Training Degradation in Small Recurrent Language Models](https://aclanthology.org/2026.babylm-main.1/) (Maheshwari, BabyLM 2026)
ACL