@inproceedings{rudraraju-2026-function,
title = "Do Function Vectors and Task Vectors Generalize? A Reproducibility Study Across Model Families and Scale",
author = "Rudraraju, Serendeep",
editor = "Arad, Dana and
Belinkov, Yonatan and
Chen, Hanjie and
Feldhus, Nils and
Mueller, Aaron and
Mohebbi, Hosein and
Sarti, Gabriele and
Tutek, Martin",
booktitle = "Proceedings of the 9th {B}lackbox{NLP} Workshop: Analyzing and Interpreting Neural Networks for {NLP}",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.blackboxnlp-1.57/",
pages = "1017--1028",
ISBN = "979-8-89176-476-7",
abstract = "A language model can learn a task from examples in its prompt. Function vectors (FVs) and task vectors (TVs) aim to transfer that behavior to a new prompt by modifying the model{'}s activations. We reproduce both methods on word-pair tasks across GPT-J, Llama-3.1, and Gemma-2. In held-out tests, FVs outperform random controls on GPT-J and Llama-3.1-8B. TVs outperform shuffled-label controls on GPT-J and both base and instruction-tuned Gemma; the Llama-8B comparison is inconclusive. Our FV implementation achieves zero accuracy on both Gemma checkpoints in a two-task follow-up. Sentiment classification shows a further limitation: in-context learning (ICL) achieves 73.5{--}94.0{\%} accuracy with arbitrary labels, while TVs achieve only 49.6{--}52.7{\%}. Successful transfer on word pairs therefore does not extend to this sentiment dataset. Separating vector construction, layer selection, and testing also shows why gains over zero-shot accuracy can overstate the evidence for transfer. Code and data are available online."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="rudraraju-2026-function">
<titleInfo>
<title>Do Function Vectors and Task Vectors Generalize? A Reproducibility Study Across Model Families and Scale</title>
</titleInfo>
<name type="personal">
<namePart type="given">Serendeep</namePart>
<namePart type="family">Rudraraju</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 9th BlackboxNLP Workshop: Analyzing and Interpreting Neural Networks for NLP</title>
</titleInfo>
<name type="personal">
<namePart type="given">Dana</namePart>
<namePart type="family">Arad</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yonatan</namePart>
<namePart type="family">Belinkov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hanjie</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nils</namePart>
<namePart type="family">Feldhus</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">Hosein</namePart>
<namePart type="family">Mohebbi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gabriele</namePart>
<namePart type="family">Sarti</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Martin</namePart>
<namePart type="family">Tutek</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">979-8-89176-476-7</identifier>
</relatedItem>
<abstract>A language model can learn a task from examples in its prompt. Function vectors (FVs) and task vectors (TVs) aim to transfer that behavior to a new prompt by modifying the model’s activations. We reproduce both methods on word-pair tasks across GPT-J, Llama-3.1, and Gemma-2. In held-out tests, FVs outperform random controls on GPT-J and Llama-3.1-8B. TVs outperform shuffled-label controls on GPT-J and both base and instruction-tuned Gemma; the Llama-8B comparison is inconclusive. Our FV implementation achieves zero accuracy on both Gemma checkpoints in a two-task follow-up. Sentiment classification shows a further limitation: in-context learning (ICL) achieves 73.5–94.0% accuracy with arbitrary labels, while TVs achieve only 49.6–52.7%. Successful transfer on word pairs therefore does not extend to this sentiment dataset. Separating vector construction, layer selection, and testing also shows why gains over zero-shot accuracy can overstate the evidence for transfer. Code and data are available online.</abstract>
<identifier type="citekey">rudraraju-2026-function</identifier>
<location>
<url>https://aclanthology.org/2026.blackboxnlp-1.57/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>1017</start>
<end>1028</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Do Function Vectors and Task Vectors Generalize? A Reproducibility Study Across Model Families and Scale
%A Rudraraju, Serendeep
%Y Arad, Dana
%Y Belinkov, Yonatan
%Y Chen, Hanjie
%Y Feldhus, Nils
%Y Mueller, Aaron
%Y Mohebbi, Hosein
%Y Sarti, Gabriele
%Y Tutek, Martin
%S Proceedings of the 9th BlackboxNLP Workshop: Analyzing and Interpreting Neural Networks for NLP
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-476-7
%F rudraraju-2026-function
%X A language model can learn a task from examples in its prompt. Function vectors (FVs) and task vectors (TVs) aim to transfer that behavior to a new prompt by modifying the model’s activations. We reproduce both methods on word-pair tasks across GPT-J, Llama-3.1, and Gemma-2. In held-out tests, FVs outperform random controls on GPT-J and Llama-3.1-8B. TVs outperform shuffled-label controls on GPT-J and both base and instruction-tuned Gemma; the Llama-8B comparison is inconclusive. Our FV implementation achieves zero accuracy on both Gemma checkpoints in a two-task follow-up. Sentiment classification shows a further limitation: in-context learning (ICL) achieves 73.5–94.0% accuracy with arbitrary labels, while TVs achieve only 49.6–52.7%. Successful transfer on word pairs therefore does not extend to this sentiment dataset. Separating vector construction, layer selection, and testing also shows why gains over zero-shot accuracy can overstate the evidence for transfer. Code and data are available online.
%U https://aclanthology.org/2026.blackboxnlp-1.57/
%P 1017-1028
Markdown (Informal)
[Do Function Vectors and Task Vectors Generalize? A Reproducibility Study Across Model Families and Scale](https://aclanthology.org/2026.blackboxnlp-1.57/) (Rudraraju, BlackboxNLP 2026)
ACL